IJAET Volume 7 Issue 3 July 2014

VOLUME-7, ISSUE-3
Date:

International Journal of Advances in
Engineering & Technology (IJAET)

Smooth, Simple and Timely Publishing of
Review and Research Articles!
ISSN : 2231-1963
July2014
International Journal of Advances in Engineering & Technology, Jul. 2014.
IJAET ISSN: 2231-1963

i Vol. 7, Issue 3, pp. i-vi

Table of Content
S. No. Article Title & Authors (Volume 7, Issue 3, July-2014) Page Nos
1.
Data Warehouse Design and Implementation Based on Quality
Requirements
Khalid Ibrahim Mohammed
642-651
2.
A New Mechanism in Hmipv6 to Improve Micro and Macro
Mobility
Sahar Abdul Aziz Al-Talib
652-665
3.
Efficiency Optimization of Vector-Controlled Induction Motor
Drive
Hussein Sarhan
666-674
4.
Flexible Differential Frequency-to-Voltage and Voltage-to-
Frequency Converters using Monolithic Analogue
Reconfigurable Devices
Ivailo Milanov Pandiev
675-683
5.
A Review Search Bitmap Image for Sub Image and the Padding
Problem
Omeed Kamal Khorsheed
684-691
6.
Potential Use of Phase Change Materials with Reference to
Thermal Energy Systems in South Africa
Basakayi J.K., Storm C.P.
692-700
7.
Improving Software Quality in the Service Process Industry
using Agility with Software Reusable Components as Software
Product Line: An Empirical Study of Indian Service Providers
Charles Ikerionwu, Richard Foley, Edwin Gray
701-711
8.
Produce Low-Pass and High-Pass Image Filter in Java
712-722
9.
The Comparative Analysis of Social Network in International
and Local Corporate Business
Mohmed Y. Mohmed AL-SABAAWI
723-732

ii Vol. 7, Issue 3, pp. i-vi

10.
Precise Calculation Unit Based on a Hardware Implementation
of a Formal Neuron in a FPGA Platform
Mohamed ATIBI, Abdelattif BENNIS, Mohamed
BOUSSAA
733-742
11.
Temperature Profiling at Southern Latitudes by Deploying
Microwave Radiometer
A. K. Pradhan, S. Mondal, L. A. T. Machado and P. K.
Karmakar
743-755
12.
Development and Evaluation of Trolley-cum-Batch Dryer for
Paddy
Mohammed Shafiq Alam and V K Sehgal
756-764
13.
Joint Change Detection and Image Registration Method for
Multitemporal SAR Images
Lijinu M Thankachan, Jeny Jose
765-772
14.
Load - Settlement Behaviour of Granular Pile in Black Cotton
Soil
Siddharth Arora, Rakesh Kumar and P. K. Jain
773-781
15.
Harmonic Study of VFDS and Filter Design: A Case Study for
Sugar Industry with Cogeneration
V. P. Gosavi and S. M. Shinde
782-789
16.
Precipitation and Kinetics of Ferrous Carbonate in Simulated
Brine Solution and its Impact on CO2 Corrosion of Steel
G. S. Das
790-797
17.
Performance Comparison of Power System Stabilizer with and
Without Facts Device
Amit Kumar Vidyarthi, Subrahmanyam Tanala, Ashish
Dhar Diwan
798-806
18.
Hydrological Study of Man (Chandrabhaga) River
Shirgire Anil Vasant, Talegaokar S.D.
807-817
19. Crop Detection by Machine Vision for Weed Management
818-826

iii Vol. 7, Issue 3, pp. i-vi

Ashitosh K Shinde and Mrudang Y Shukla
20.
Detection & Control of Downey Mildew Disease in Grape Field
Vikramsinh Kadam, Mrudang Shukla
826-837
21.
Ground Water Status- A Case Study of Allahabad, UP, India
Ayush Mittal, Munesh Kumar
838-844
22.
Clustering and Noise Detection For Geographic Knowledge
Discovery
Sneha N S and Pushpa
845-855
23.
Proxy Driven FP Growth Based Prefetching
Devender Banga and Sunitha Cheepurisetti
856-862
24.
Search Network Future Generation Network for Information
Interchange
G. S. Satisha
863-867
25.
A Brief Survey on Bio Inspired Optimization Algorithms For
Molecular Docking
Mayukh Mukhopadhyay
868-878
26.
Heat Transfer Analysis of Cold Storage
Upamanyu Bangale and Samir Deshmukh
879-886
27.
Localized RGB Color Histogram Feature Descriptor for Image
Retrieval
K. Prasanthi Jasmine, P. Rajesh Kumar
887-895
28.
Witricity for Wireless Sensor Nodes
M. Karthika and C. Venkatesh
896-904
29.
Study of Swelling Behaviour of Black Cotton Soil Improved
with Sand Column
Aparna, P.K. Jain and Rakesh Kumar
905-910
30.
Effective Fault Handling Algorithm for Load Balancing using
Ant Colony Optimization in Cloud Computing
911-916

iv Vol. 7, Issue 3, pp. i-vi

Divya Rastogi and Farhat Ullah Khan
31.
Handling Selfishness over Mobile Ad Hoc Network
Madhuri D. Mane and B. M. Patil
917-922
32.
A New Approach to Design Low Power CMOS Flash A/D
Converter
C Mohan and T Ravisekhar
923-929
33.
Optimization and Comparative Analysis of Non-Renewable and
Renewable System
Swati Negi and Lini Mathew
930-937
34.
A Feed Forward Artificial Neural Network based System to
Minimize DOS Attack in Wireless Network
Tapasya Pandit & Anil Dudy
938-947
35.
Improving Performance of Delay Aware Data Collection Using
Sleep and Wake Up Approach in Wireless Sensor Network
Paralkar S. S. and B. M. Patil
948-956
36.
Improved New Visual Cryptographic Scheme Using One
Shared Image
Gowramma B.H, Shyla M.G, Vivekananda
957-966
37.
Segmentation of Brain Tumour from Mri Images by Improved
Fuzzy System
Sumitharaj.R, Shanthi.K
967-973
38.
Implementation of Classroom Attendance System Based on
Face Recognition in Class
Ajinkya Patil, Mrudang Shukla
974-979
39.
A Tune-In Optimization Process of AISI 4140 in Raw Turning
Operation using CVD Coated Insert
C. Rajesh
980-990
40.
A Modified Single-frame Learning based Super-Resolution and
its Observations
Vaishali R. Bagul and Varsha M. Jain
991-997

v Vol. 7, Issue 3, pp. i-vi

41.
Design Modification and Analysis of Two Wheeler Cooling
Fins-A Review
Mohsin A. Ali and S.M Kherde
998-1002
42.
Virtual Wireless Keyboard System with Co-ordinate Mapping
Souvik Roy, Ajay Kumar Singh, Aman Mittal, Kunal
Thakral
1003-1008
43.
Secure Key Management in Ad-Hoc Network: A Review
Anju Chahal and Anuj Kumar, Auradha
1009-1017
44.
Prediction of Study Track by Aptitude Test using Java
Deepali Joshi and Priyanka Desai
1018-1026
45.
Unsteady MHD Three Dimensional Flow of Maxwell Fluid
Through a Porous Medium in a Parallel Plate Channel under
the Influence of Inclined Magnetic Field
L. Sreekala, M. Veera Krishna, L. Hari Krishna and E.
Kesava Reddy
1027-1037
46.
Video Streaming Adaptivity and Efficiency in Social
Networking Sites
G. Divya and E R Aruna
1038-1043
47.
Intrusion Detection System using Dynamic Agent Selection and
Configuration
Manish Kumar, M. Hanumanthappa
1044-1052
48.
Evaluation of Characteristic Properties of Red Mud For
Possible use as a Geotechnical Material in Civil Construction
Kusum Deelwal, Kishan Dharavath, Mukul Kulshreshtha
1053-1059
49.
Performance Analysis of IEEE 802.11e EDCA with QoS
Enhancements Through Adapting AIFSN Parameter
Vandita Grover and Vidusha Madan
1060-1066
50.
Data Derivation Investigation
S. S. Kadam, P.B. Kumbharkar
1067-1074

vi Vol. 7, Issue 3, pp. i-vi

51.
Design and Implementation of Online Patient Monitoring
System
Harsha G S
1075-1081
52.
Comparison Between Classical and Modern Methods of
Direction of Arrival (DOA) Estimation
Mujahid F. Al-Azzo, Khalaf I. Al-Sabaaw
1082-1090
53.
Modelling Lean, Agile, Leagile Manufacturing Strategies: An
Fuzzy Analytical Hierarchy Process Approach For Ready
Made Ware (Clothing) Industry in Mosul, Iraq
Thaeir Ahmed Saadoon Al Samman
1091-1108

Members of IJAET Fraternity A - N

International Journal of Advances in Engineering & Technology, July, 2014.
IJAET ISSN: 22311963
642 Vol. 7, Issue 3, pp. 642-651

DATA WAREHOUSE DESIGN AND IMPLEMENTATION BASED
ON QUALITY REQUIREMENTS
Khalid Ibrahim Mohammed
Department of Computer Science, College of Computer, University of Anbar, Iraq.

ABSTRACT
The data warehouses are considered modern ancient techniques, since the early days for the relational
databases, the idea of the keeping a historical data for reference when it needed has been originated, and the
idea was primitive to create archives for the historical data to save these data, despite of the usage of a special
techniques for the recovery of these data from the different storage modes. This research applied of structured
databases for a trading company operating across the continents, has a set of branches each one has its own
stores and showrooms, and the company branchs group of sections with specific activities, such as stores
management, showrooms management, accounting management, contracts and other departments. It also
assumes that the company center exported software to manage databases for all branches to ensure the safety
performance, standardization of processors and prevent the possible errors and bottlenecks problems. Also the
research provides this methods the best requirements have been used for the applied of the data warehouse
(DW), the information that managed by such a applied must be with high accuracy. It must be emphasized to
ensure compatibility information and hedge its security, in schemes domain, been applied to a comparison
between the two schemes (Star and Snowflake Schemas) with the concepts of multidimensional database. It
turns out that Star Schema is better than Snowflake Schema in (Query complexity, Query performance,
Foreign Key Joins),And finally it has been concluded that Star Schema center fact and change, while Snowflake
Schema center fact and not change.
KEYWORDS: Data Warehouses, OLAP Operation, ETL, DSS, Data Quality.
I. INTRODUCTION
A data warehouse is a subject-oriented, integrated, nonvolatile, and time-variant collection of data in
support of managements decisions. The data warehouse contains granular corporate data. Data in the
data warehouse is able to be used for many different purposes, including sitting and waiting for future
requirements which are unknown today [1]. Data warehouse provides the primary support for
Decision Support Systems (DSS) and Business Intelligence (BI) systems. Data warehouse, combined
with On-Line Analytical Processing (OLAP) operations, has become and more popular in Decision
Support Systems and Business Intelligence systems. The most popular data model of Data warehouse
is multidimensional model, which consists of a group of dimension tables and one fact table
according to the functional requirements [2]. The purpose of a data warehouse is to ensure the
appropriate data is available to the appropriate end user at the appropriate time [3]. Data warehouses
are based on multidimensional modeling. Using On-Line Analytical Processing tools, decision
makers navigate through and analyze multidimensional data [4].
Data warehouse uses a data model that is based on multidimensional data model. This model is also
known as a data cube which allows data to be modeled and viewed in multiple dimensions [5]. And
the schema of a data warehouse lies on two kinds of elements: facts and dimensions. Facts are used to
memorize measures about situations or events. Dimensions are used to analyze these measures,
particularly through aggregations operations (counting, summation, average, etc.) [6, 7]. Data
Quality (DQ) is the crucial factor in data warehouse creation and data integration. The data
warehouse must fail and cause a great economic loss and decision fault without insight analysis of
643 Vol. 7, Issue 3, pp. 642-651

data problems [8]. The quality of data is often evaluated to determine usability and to establish the
processes necessary for improving data quality. Data quality may be measured objectively or
subjectively. Data quality is a state of completeness, validity, consistency, timeliness and accuracy
that make data appropriate for a specific use [9]. The paper is divided into seven sections. Section 1
introduction, Definition of Data Warehouse and The Quality of Data Warehouse. Section 2 presents
related work, Section 3 presents Data Warehouse Creation and the main idea is that a Data warehouse
database gathers data from an overseas trading company databases. Section 4describes Data
Warehouse Design For this study, we suppose a hypothetical company with many branches around
the world, each branch has so many stores and showrooms scattered within the branch location. Each
branch has a database to manage branch information. Section 5 describes our evaluation Study of
Quality Criteria for DW, which covers aspects related both to quality and performance of our
approach, and the obtained results, and work on compare between star schema and snowflake schema.
Section 6 provides conclusions. Finally, Section 7 describes open issues and our planned future work.
1.1 Definition of Data Warehouse
A data warehouse is a relational database that is designed for query and analysis rather than for
transaction processing. It usually contains historical data derived from transaction data, but it can
include data from other sources. It separates analysis workload from transaction workload and
enables an organization to consolidate data from several sources. In addition to a relational database,
a data warehouse environment can include an extraction, transportation, transformation, and loading
(ETL) solution, an online analytical processing (OLAP) engine, client analysis tools, and other
applications that manage the process of gathering data and delivering it to business users. A common
way of introducing data warehousing is to refer to the characteristics of a data warehouse as set forth
by William Inmon [10]:
1. Subject Oriented.
2. Integrated.
3. Nonvolatile.
4. Time Variant.
1.2 The Quality of Data Warehouse
Data quality has been defined as the fraction of performance over expectancy, or as the loss imparted
to society from the time a product is shipped [11]. The believe was the best definition is the one found
in [12, 13, and 14]: data quality is defined as "fitness for use". The nature of this definition directly
implies that the concept of data quality is relative. For example, data semantics is different for each
distinct user. The main purpose of data quality is about horrific data - data which is missing or
incorrect or invalid in some perspective. A large term is that, data quality is attained when business
uses data that is comprehensive, understandable, and consistent, indulging the main data quality
magnitude is the first step to data quality perfection which is a method and able to understand in an
effective and efficient manner, data has to satisfy a set of quality criteria. Data gratifying the quality
criterion is said to be of high quality [9].
II. RELATED WORK
In this section we will review related work in Data Warehouse Design and Implementation Based on
Quality Requirements. We will start with the former. The paper introduced by Panos Vassiladis,
Mokrane Bouzegeghoub and Christoph Quix, (2000), the proposed approach covers the full lifecycle
of the data warehouse, and allows capturing the interrelationships between different quality factors
and helps the interested user to organize them in order to fulfill specific quality goals. Furthermore,
they prove how the quality management of the data warehouse can guide the process of data
warehouse evolution, by tracking the interrelationships between the components of the data
warehouse. Finally, they presented a case study, as a proof of concept for the proposed methodology
[15]. The paper introduced by Leo Willyanto Santoso and Kartika Gunadi, (2006), this paper
describes a study which explores modeling of the dynamic parts of the data warehouse. This
metamodel enables data warehouse management, design and evolution based on a high level
conceptual perspective, which can be linked to the actual structural and physical aspects of the data
644 Vol. 7, Issue 3, pp. 642-651

warehouse architecture. Moreover, this metamodel is capable of modeling complex activities, their
interrelationships, the relationship of activities with data sources and execution details [16]. The paper
introduced by Amer Nizar Abu Ali and Haifa Yousef Abu-Addose, (2010), The aim of this paper is to
discover the main critical success factors (CSF) that leaded to an efficient implementation of DW in
different organizations, by comparing two organizations namely: First American Corporation (FAC)
and Whirlpool to come up with a more general (CSF) to guide other organizations in implementing
DW efficiently. The result from this study showed that FAC Corporation had greater returns from
data warehousing than Whirlpool. After that and based on them extensive study of these organizations
and other related resource according to CSFs, they categorized these (CSF) into five main categories
to help other organization in implementing DW efficiently and avoiding data warehouse killers, based
on these factors [17]. The paper introduced by Manjunath T.N, Ravindra S Hegadi, (2013), The
proposed model evaluates the data quality of decision databases and evaluates the model at different
dimensions like accuracy derivation integrity, consistency, timeliness, completeness, validity,
precision and interpretability, on various data sets after migration. The proposed data quality
assessment model evaluates the data at different dimensions to give confidence for the end users to
rely on their businesses. Author extended to classify various data sets which are suitable for decision
making. The results reveal the proposed model is performing an average of 12.8 percent of
improvement in evaluation criteria dimensions with respect to the selected case study [18].
III. DATA WAREHOUSE CREATION
The main idea is that a Data warehouse database gathers data from an overseas trading company
databases. For each branch of the supposed company we have a database consisting of the following
schemas:
Contracting schema consists a contract and contractor date.
Stores schema managing storing information.
Showrooms schema to manage showrooms information for any branch of the supposed
company.
At the top of the above schemas, an accounting schema was installed which manages all
accounting operations for any branch or the while company.
All information is stored into fully relational tables according to the known third normal form. The
data integrity is maintained by using a foreign keys relationship between related tables, non-null
constraints, check constraints, and oracle database triggers are used for the same purpose. Many
indexes are created to be used by oracle optimizer to minimize DML and query response time.
Security constraints are maintained using oracle privileges. Oracle OLAP policy is taken in
consideration.
IV. DATA WAREHOUSE DESIGN
As mentioned above a warehouse home is installed on the same machine. The data warehouse is
stored on a separate oracle table spaces and configured to use the above relational online tables as a
source data. So the mentioned schemas are treated as data locations. Oracle warehouse builder is a
java program, which are used warehouse managements. The locations of data sources are:
1. Accounting schema.
2. Stores schema.
3. Contracting schema.
4. Showrooms schema.
For this study, we suppose a hypothetical company with many branches around the world, each
branch has so many stores and showrooms scattered within the branch location. Each branch has a
database to manage branch information. Within each supposed branch database there are the
following schemas which work according to OLAP policies and maintain securities and data
integrity. The schemas are: Accounting schema, Contracting schema, Stores schema and showrooms
schema. All branches databases are connected to each other over WAN.

645 Vol. 7, Issue 3, pp. 642-651

Figure 1. Distributed database for hypothetical company.
overseas company, as a base (or a source) for warehouse database. This paper suppose that each node
belongs to one company branch, so each branch controls its own data. The main office of the
company, controls the while data also with the central node. The warehouse database could be at the
central node, as the company needs. We suppose that all nodes use the same programs, which are
applied the database(s). Within each node, each activity is represented by a database schema, i.e.
stores, showrooms, contracting, and other schemas. The core of all schemas is the accounting schema.
According to jobs load, each schema could be installed on a separate database or on the same
database. All related databases around company branches are cooperated within the same WAN.
V. STUDY OF QUALITY CRITERIA FOR DW
In this study, we will carry out some of the criteria, and these criteria are:
5.1. Data warehouse using snowflake schema
Using oracle warehouse policies, each database has the following snow flaking modules:
1. Sales module.
2. Supplying module.
5.1.1. Sales module
It consist of the following relational tables
Table 1. Explain the relational table
Table name Table type Oracle schema (owner)
Sales_sh Fact table Showrooms
showrooms Dimensional table Showrooms
Items Dimensional table Accounting
Currencies Dimensional table Accounting
Customers Dimensional table Showrooms
Locations Dimensional table Accounting
The following diagram depicts the relations between the above dimensional and fact tables

Figure 2. Sales module.
646 Vol. 7, Issue 3, pp. 642-651

Figure 2 above, represents all entities within the sales module. Any entity is designed using the Third
Normal Form (3NF) rule, so it has a primary key. The most important tools used to implement
integrity and validations are oracle constraints. After supplying data to the above module, and
transferring it to oracle warehouse design center, retrieving data (557,441 rows) from fact table sales
which are shown in the following figure 3, which mentions the detailed information for each single
sales within each showroom, and location. It consists of: Voucher (doc) no. And date, the sold item,
sold quantity and price. This data is available at the corresponding node (branch) and the center. Of
course, the same data would be transferred to warehouse database for historical purpose.

Figure 3. Sales data.
Customers dimensional table consists of some personal date about the customer, like mobile, email,
and location which is useful to contact him. Also it indicates the date of last purchase, and the total
amount purchased for last year. This data is available for the corresponding node and the center; also
it refers to the warehouse database. (See figure 4 customer data) The dimensional table of customers
would is shown below.

Figure 4. Customers data.
5.1.2. Supplying module
Supplying the company by materials according to company usual contracts is managed by this
module according to snowflake design. It consists of the following relational tables.
Table 2. Within supplying module
STR_RECIEVING Fact table Stores
Contracts Dimensional table contracting
Stores Dimensional table Stores
Daily_headers Dimensional table Accounting
647 Vol. 7, Issue 3, pp. 642-651

The following diagram Fig 5 depicts the relations between the above dimensional and fact tables.
They are obey 3NF rule, so they have their primary key constraints, and constrained to each other
using foreign keys constraints. The fact table STR_RECIEVING consists of all charges information
received at company stores (contented by stores table owned by stores schema), according to the
contracts (contented by contracts table owned by contracting schema). Daily headers dimensional
table represent the accounting information for each contract. Using oracle triggers when new record
inserted into the STR_RECIEVING fact table, some other accounting data would be created into
details table row related (through foreign key) to Daily headers dimensional table. Also any charge
value could be converted to the wanted currency using the data maintained by currencies dimensional
table owned by accounting schema.

Figure 5. Supplying module.
For security reasons, direct excess to object fact table is not allowed, un imaginary view is created
(named str_recieving_v) , then all users are allowed to generate a DML (data manipulation language
instructions) on this view. A certain piece of code (oracle trigger) is written to manipulate data,
according to server policies (data integrity and consistency) as user supplies data to the imaginary
view. After supplying data to the above module, and transferring it to oracle warehouse design
center, retrieving data (415,511rows) from fact table str_recieving as shown in the following figure.

Figure 6. Received charges on str_recieving fact table.
During charges insertion, a background process (oracle trigger) should update the stock dimension
data to reflect the latest information about quantities in stock at each node and the center. The stock
data contains the quantity balance, quantity in and out for the current year. Its available at the
648 Vol. 7, Issue 3, pp. 642-651

corresponding node and the center, at the online database, and at warehouse database for previous
years. Stock data could be like figure 7 as viewed by oracle SQL developer.

Figure 7. Stock dimensional table data.
5.2. Data warehouse using star schema
As a study case using warehouse star schema, we have:
1. Stocktaking module.
2. Accounting module.
5.2.1. Stocktaking module
Its manage for the current stock within any store for the company, its data is determined by the
following table.
Table 3. Cooperated within stocktaking module
Stock Fact table Stores
Stores Dimensional table Stores
showrooms Dimensional table Showrooms
contracts Dimensional table Contracting
Str_recieving Fact table Stores
To_show_t Fact table Stores
The stock fact table stands for the actual stock balances within each store belongs to each branch, and
the whole company at the center. The following diagram depicts the relations between the below
dimensions.

Figure 8. Stocktaking module as a warehouse star schema.
649 Vol. 7, Issue 3, pp. 642-651

DML (data manipulation language instructions) is done on stock fact table through oracle triggers
which are the most trusted programs to maintain the highest level of integrity and security, so the
imaginary view ( named stock_v) was created, users are allowed to supply data to that view, then the
server would process the supplied data using oracle trigger. Querying the renormalized stock fact
table within the star schema module, using oracle design center is depicted as below (no. of rows on
stock table within our study case is 15,150). This figure 9 query execution is allowed for all users
(public).

Figure 9. Stocktaking on oracle warehouse design center.
5.2.2. Accounting module
One of the most importance aspects of accounting functions is the calculations of the daily cash
within each showroom belongs to the company. The daily totals for each branch and the grand total
could be calculated. Timely based cash could be accumulated later on demand.
Table 4. The tables needed for this activity
Table name Type Oracle schema (owner)
Daily cash Fact table Accounting
Show_sales Fact table Accounting
Showrooms Dimensional table Showroom
Currencies_tab Dimensional table Accounting
Customers Dimensional table showrooms
The daily cash is a view used to reflect the actual cash with each showroom on daily base.

Figure 10. Daily cash using warehouse star schema.
650 Vol. 7, Issue 3, pp. 642-651

Using inner SQL joins, one could retrieve data about daily cash as follows.

Figure 11. Grand daily cash as depicted by Oracle warehouse design center.
VI. CONCLUSIONS
The following expected conclusions have been drawn:
1. Reduce the query response time and Data Manipulation Language and using many indexes which
are created to be used by oracle optimizer.
2. Star Schema is best of them Snowflake Schema the following points are reached:
Query complexity: Star Schema the query is very simple and easy to understand, while
Snowflake Schema is more complex query due to multiple foreign key which joins between
dimension tables .
Query performance: Star Schema High performance. Database engine can optimize and boost
the query performance based on predictable framework, while Snowflake Schema is more
foreign key joins; therefore, longer execution time of query in compare with star schema.
Foreign Key Joins: Star Schema Fewer Joins, while Snowflake Schema has higher number of
joins.
And finally it has been concluded that Star Schema center fact and change, while Snowflake Schema
center fact and not change.
VII. FUTURE WORKS
1. Using any other criteria in development implementation of the proposed system.
2. Using statistical methods to implement other criteria of Data Warehouse.
3. Applying algorithm Metadata and comparing between bitmap index and b-tree index.
4. Applying this work for a real organization not prototype warehouse.
5. Take advantage of the above standards in improving the performance of the use of the data
warehouse and institutions according to their environment.
ACKNOWLEDGEMENTS
I would like to thank my supervisor Assist. Prof. Dr. Murtadha Mohammed Hamad for his
guidance, encouragement and assistance throughout the preparation of this study. I would also like to
extend thanks and gratitude to all teaching and administrative staff in the College of Computer
University of Anbar, and all those who lent me a helping hand and assistance. Finally, special thanks
are due to members of my family for their patience, sacrifice and encouragement.

651 Vol. 7, Issue 3, pp. 642-651

REFERENCES
[1] H. William Inmon, Building the Data Warehouse. Fourth Edition Published by Wiley Publishing, Inc.,
Indianapolis, Indiana.2005.
[2] R. Kimball, The data warehouse lifecycle toolkit. 1
st
Edition ed.: New York, John Wiley and Sons,
1998.
[3] K.W. Chau, Y. Cao, M. Anson and J. Zhang, Application of Data Warehouse and Decision Support
System in Construction Management. Automation in Construction 12 (2) (2002) 213-224.
[4] Nicolas Prat, Isabelle Comyn-Wattiau and Jacky Akoka , Combining objects with rules to represent
aggregation knowledge in data warehouse and OLAP systems. Data & Knowledge Engineering 70 (2011)
732752.
[5] Singhal, Anoop, Data Warehousing and Data Mining Techniques for Cyber Security. Springer. United
states of America.2007.
[6] Bhanali, Neera,Strategic Data Warehousing: Achieving Alignment with Business. CRC Prees .
United State of America. 2010.
[7] Wang, John. Encyclopedia of Data Warehousing and Mining .Second Edition. Published by
Information Science Reference. United States of America. 2009.
[8] Yu. Huang, Xiao-yi. Zhang, Yuan Zhen , Guo-quan. Jiang, A Universal Data Cleaning Framework
Based on User Model , IEEE,ISECS International Computing, Communication, Control , and Management,
Sanya, China, Aug 2009.
[9] T.N. Manjunath, S. Hegadi Ravindra, G.K. Ravikumar, Analysis of Data Quality Aspects in Data
Warehouse Systems. International Journal of Computer Science and Information Technologies, Vol. 2 (1),
2011, 477-485.
[10] Paul Lane, Oracle Database Data Warehousing Guide. 11g Release 2 (11.2), September 2011.
[11] D.H. Besterfield, C. Besterfield-Michna, G. Besterfield, and M. Besterfield-Sacre, Total Quality
Management. Prentice Hall, 1995.
[12] G.K. Tayi, D.P. Ballou, Examining Data Quality. In Communications of the ACM, 41(2), 1998. pp.
54-57.
[13] K. Orr, Data quality and systems theory. Communications of the ACM, 41(2), 1998. pp. 66-71.
[14] R.Y. Wang, D. Strong, L.M. Guarascio, Beyond Accuracy: What Data Quality Means to Data
Consumers. Technical Report TDQM-94-10, Total Data Quality Management Research Program, MIT Sloan
School of Management, Cambridge, Mass., 1994.
[15] Panos vassiladis, Mokrane Bouzegeghoub and Christoph Quix. Towards quality-oriented data
warehouse usage and evolution. Information Systems Vol. 25, No. 2, pp. 89-l 15, 2000.
[16] Leo Willyanto Santoso and Kartika Gunadi. A proposal of data quality for data warehouses
environment. Journal Information VOL. 7, NO. 2, November 2006: 143 148
[17] Amer Nizar AbuAli and Haifa Yousef Abu-Addose. Data Warehouse Critical Success Factors.
European Journal of Scientific Research ISSN 1450-216X Vol.42 No.2 (2010), pp.326-335.
[18] T.N. Manjunath, S. Hegadi Ravindra, Data Quality Assessment Model for Data Migration Business
Enterprise. International Journal of Engineering and Technology (IJET) ISSN: 0975-4024, Vol 5 No 1 Feb-
Mar 2013.
AUTHOR PROFILE
Khalid Ibrahim Mohammed was born in BAGHDAD, IRAQ, in 1981, received his
Bachelors Degree in computer Science from AL-MAMON UNIVERSITY COLLEGE,
BAGHDAD, IRAQ during the year 2006 and M. Sc. in computer Science from College of
Computer Anbar University, ANBAR , IRAQ during the year 2013. He is having total 8
years of Industry and teaching experience. His areas of interests are Data Warehouse &
Business Intelligence, multimedia and Databases.
652 Vol. 7, Issue 3, pp. 652-665

A NEW MECHANISM IN HMIPV6 TO IMPROVE MICRO AND
MACRO MOBILITY
Sahar Abdul Aziz Al-Talib
Computer and Information Engineering
Electronics Engineering, University of Mosul, Iraq

ABSTRACT
The paper proposes new mechanism in Hierarchical Mobile IPv6 (HMIPv6) implementation for utilization in
IPv6 Enterprise Gateway. This mechanism provides seamless mobility and fast handover of a mobile node in
HMIPv6 networks. Besides it provides continuous communication among a member of HMIPv6 components
while roaming. The mechanism anticipates future movement of mobile node and accordingly provides an
effective updating mechanism. A bitmap has been proposed to help in applying the new mechanism as will be
explained throughout this paper. Instead of scanning the prefix bits which might be up to 64 bits for subnet
discovery, about 10% of this length will be sufficient to determine the hierarchical topology. This will enable
fast Micro and Macro HMIPv6 in Enterprise Gateway.
KEYWORDS: Hierarchical Mobile IPv6, bitmap, Enterprise Gateway, Mobility Anchor Point.
I. INTRODUCTION
IPv6 Enterprise Gateway is an apparatus or device which intercommunicate the public network with
enterprise network. The enterprise network includes many sub-networks with different access routers
or home anchors. For flexibility, HMIPv6 is used to organize these enterprise networks in hierarchal
way. HMIPv6 stands for Hierarchical Mobile Internet Protocol Version 6. HMIPv6 is provisioning
the different access areas by assigning mobile anchor point (MAP) to each area. The MAP stands on
behalf of the enterprise gateway (home agent) in its particular coverage area. Besides, it will reduce
the control messages exchanged and decrease the time when the mobile node roaming to another
network. The mobile node can use the local MAP to keep the communication on without the need to
communicate with enterprise gateway.
It is known that Hierarchical Mobile IPv6 (HMIPv6) provides a flexible mechanism for local mobility
management within visited networks. The main problem in hierarchical mobility is that the
communication between the correspondent node (CN) and the mobile node (MN) suffers from
significant delay in case of the mobile node roaming between different MAPs or between different
access routers (ARs) in the same MAP coverage area. This happens because the MN in the visited
network acquires a local care of address (LCoA), therefore it receives the router advertisement (RA)
and uses its prefix to build its new LCoA. This process causes a communication delay between the
CN and the MN for a while, especially when the roaming happens between different MAPs. This
process involves more binding update messages to be sent not just to the local MAP, but also to the
enterprise gateway as shown in Figure 1, the messages flow to acquire new CoA starts after MN
reached the foreign network.
The patent in [1] anticipates the probable CoAs based on the number of vicinity Access Routers,
which means the number of probable CoA in each single movement, and then the MAP should
multicast the traffic to all of these addresses. In this paper, the Home Anchor creates the anticipated
CoA based on the received bitmap from MN, this Bitmap refers to the strongest signal that the MN
653 Vol. 7, Issue 3, pp. 652-665

was triggered by the foreign router, this means Home Anchor have a single probable address to tunnel
the packet to, besides the current CoA.
RFC4260 [2] proposed the idea of anticipating the new CoA and tunnel the packets between the
previous Access Router and the New Access Router. The differences between the work in this paper
and in [2] is that packet loss will likely occur if the process is performed too late or too early with
respect to the time in which the mobile node detaches from the previous access router and attach to
the new one, or the packet loss is likely occur in case the BU cache in the home anchor is updated but
the layer 2 handover does not complete yet. The work in [3] proposed a scheme that reduced the 82%
of the total cost of the macro mobility handover of the original HMIPv6.
Kumar et al. in [4] proposed an analytical model which shows the performance and applicability of
MIPv6 and HMIPv6 against some key parameters in terms of cost.
An adaptive MAP selection based on active overload prevention (MAP-AOP) is proposed in [5]. The
MAP periodically evaluates the load status by using dynamic weighted load evaluation algorithm, and
then sends the load information to the covered access routers (AR) by using the expanded routing
advertisement message in a dynamic manner.
In this paper, a solution to the explained problem has been proposed in which the mobile node sends
binding update request at the moment when it initiates the movement to the foreign network, and to
add a bitmap field to the binding update request to recognize different components of the HMIPv6.
The paper is organized as follows. Section 2 describes the methodology of the proposed mechanism
supported by figures; section 3 illustrates the new operation of micro and macro-mobility. Binding
update message format is introduced in section 4 where the bitmap idea comes from. Finally, section
5 concludes the work and presents the future work.
II. METHODOLOGY
The objective of this paper is provided by a method for multi-cell cooperative communication
comprising: detecting a beacon from a new access router, translating the beacon received from the
new access router to a bitmap, sending a binding update request message that contains the bitmap to a
current mobile access point, assessing the bitmap, tunneling data packets from a correspondent node
through the enterprise gateway to the current destination and to the new destination simultaneously,
sending an update message once the mobile node reaches the new destination, refreshing binding
cache tables and tunneling the data packets only to the new destination according to the new address
of mobile node.

654 Vol. 7, Issue 3, pp. 652-665

Figure 1 (a): Flow Control of Micro and Macro Mobility Process (part 1)

655 Vol. 7, Issue 3, pp. 652-665

Figure 1 (b): Flow Control of Micro and Macro Mobility Process (part 2)
Some of the most important key points in this paper:
Adding a bitmap which is a number of binary bits proposed to be used instead of the subnet
prefix to specify the subnet.
656 Vol. 7, Issue 3, pp. 652-665

Propose to update the binding update message to include a bitmap field; this field is
occupied by a lowest part which refers to the routers prefix and a highest part which refers
to the home anchors address, the proposed message can be called binding update request
which should be sent as soon as the MN initiates the movement to the anticipated foreign
network.
The mobile node has the ability to measure the beacon strength in order to decide which
anchor point to join as the future home agent.
Binding Update message: it is mentioned in HMIPv6 standard [6, 7], this message is sent
by the mobile node to the home anchor or the home agent (in some cases it is sent to
correspondent node), it is used to register the MNs Care of Address in its home anchor or
its home agent, but this message is sent after the MN acquires a CoA in its foreign network.
Binding cache Table: it is a mapping table that resides in the Enterprise Gateway (home
agent) and home anchor points to translate the value of the bitmap to its equivalent prefix.
Also it is used to track the MNs movement.
The proposed solution contributes to both inter-domain (macro-mobility) and intra-domain
(micro-mobility) in terms of handover delay. This will improves multimedia applications.
1. The New Operation of Micro and Macro-mobility:
a) Operation of Micro Mobility: As shown in Figure 1, the proposed solution suggested sending a
binding update request (BU req.) when the MN initiates its movement to the foreign network. The
MN will sense the strong signal from the foreign network router. It translates this to the bitmap
field by setting up the corresponding bit. The contributed bitmap in this paper requires updating the
present binding update message by using the reserved bits already exist for future use.
Then the MN sends the updated BU req. message to the upstream Home Anchor. When the home
anchor receives the updated BU req., it will check the highest bits in the bitmap, as we can see in
the decision box (Figure 1), if these bits refers to the home anchor itself, the home anchor will learn
that this is a micro-mobility and create a new address (CoA) for the MN according to the lowest
bits of the bitmap.
Now, the Home Anchor cache table has two CoAs for a particular MN one for the previous MNs
location and the other for the new location. It starts to tunnel the packets to both CoAs for a while
till the handover preparation is finalized. Therefore the MN will have a seamless mobility and
continuous communication, even before obtaining the new CoA from the future foreign router.
When the MN reaches the foreign router area and obtains the CoA, it will send a Binding update
message which includes the HoA and the CoA to the upstream Home Anchor, then the Home
Anchor will refresh its binding cache and delete the old CoA.
Finally, the Home Anchor will tunnel the packets only to the new CoA.
b) Operation of Macro Mobility
This section explains the mobility between two routers each of them belongs to different Home
Anchor or different domains. The process flow is shown in Figure 1.
Assume the MN initiates the movement to a router that is connected to different home Anchor. The
MN receives the beacon from the foreign router and recognizes it as different from the old one from
the prefix. It reflects this in the higher bits of the bit map. It updates the BU request message
accordingly with the new bitmap then sends the message to the upstream Home Anchor. When the
Home Anchor receives the BU req., it will check the highest bits in the bitmap, in this case it will
recognize that these bits refers to another Anchor, therefore it forwards the BU req. to the Enterprise
Gateway (home agent). At the same time, the home anchor translates the lowest part of the bitmap to
its prefix and adds it to the MAC address to form the new CoA. The Home Anchor will tunnel the
packets to the current CoA and the new CoA at the same time. On the other hand, the Enterprise
Gateway receives the BU req. and translates the highest bits of the bitmap to the equivalent Home
Anchor address. Besides, it refreshes the cache table with this new value and starts to tunnel the
packets to the current Home Anchor and the new Home Anchor. But the traffic which is destined to
the new Anchor will not be forwarded to the MN till the MN reaches the particular foreign router and
obtained a new CoA. At this time the MN will send a BU message to the Enterprise Gateway. As a
result, the Home Agent refreshes its cache table to delete the previous Anchor and includes the
657 Vol. 7, Issue 3, pp. 652-665

updated one then tunnels the traffic to the new Home Anchor. The new home anchor in turn will
tunnel the packets to the new CoA of the mobile node in the new location.
2. Binding update message format
The main contribution in this work is based on the idea of using the reserved bits in the BU message
shown in Figure 2.
As can be seen in Figure 2 the binding update message has reserved bits starting from bit 5 up to bit
15, so some or all of these bits can be used to implement the proposed mechanism in this paper. For
example, if 6 bits are used, the highest 2- bits can be assigned to the Home Anchors with possibilities
of (00, 01, 10, 11), the lowest 4-bits can be assigned to the access routers (ARs) which can handle 2
4

or 16 routers starting from 0000 to 1111.

0 1 2 3
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| Sequence # |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
|A|H|L|K|M| Reserved | Lifetime |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
| |
| |
+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
Figure 2: Binding update message format [3]
Figure 3 shows the system architecture for the proposed mechanism, here are some abbreviations:
CN: Correspondent Node.
MN: Mobile Node.
BU req.: Binding Update request message.
ACK: Acknowledgement message
HoA: MNs Home Address.
CoA: MNs Care of Address in the foreign network.
Based on the architecture, the MN initiates its movement from the home network to Home Anchor-1.
As shown in Figure 3, MN receives the beacon from R2 in Home Anchor1s coverage area and it
translates this signal to the bitmap by setting the correspondent bit. The MN will send a BU request
message which contains the bitmap to the Home Anchor1. Home Anchor1 builds its bitmap mapping
table by translating the received bitmap to CoA2, then it replies back an acknowledgment message to
MN. The MN receives the acknowledgment and sends another BU request message to the Enterprise
Gateway (home Agent). The Enterprise Gateway receives the BU req. and checks the highest bits in
the bitmap field and translates it to the equivalent home Anchors addresses. In this architecture, the
highest bits refer to the Home Anchor-1, then the Home Agent refreshes its bitmap mapping table to
add the Home Anchors addresses in the table.

658 Vol. 7, Issue 3, pp. 652-665

Enterprise Gateway
/Home Agent
2407:3000:200d:312::2/64
Home Anchor1
2407:4000::2/64
Home Anchor2
2407:5000::2/64
R1
R2 R3 R4 R5 R6
2407:4010::/64 2407:4020::/64 2407:4030::/64
2407:4040::/64 2407:4050::/64 2407:4060::/64
MN
HoA: 2407:3000:200d:312::b6:d2
CoA2: 2407:4020::b6:d2
1. Movement
2. BU req. (HoA: 2407:3000:200d:312::b6:d2, Bit Map:
000010)
Home Address Bitmap
2407:3000:200d:312:b6:d2 000010
Care of Address
2407:4020::b6:d2
3.
Binding cache Table
Lowest bit represents->R
4. Ack
5. BU (HoA: 2407:3000:200d:312::b6:d2,
Bit Map: 000010)
2407:4000::2 (HA1)
Home Address Bitmap
2407:3000:200d:312:b6:d2
Home Anchor
00
Binding cache Table
The Highest bits represent Home
Anchors address
6.
5.
MN

Figure 3: MN moves from Home Network to the R2s Network
Assume there is a correspondent node (CN) in the Home Network vicinity as shown in Figure 4, the
CN will send the packets to the home agent address (HoA), and the Home Agent receives the traffic
and checks the binding cache table. It will find out the HoA refers to the Home Anchor-1, therefore
the Home Agent tunnels the packets to Home Anchor-1, and the Home Anchor-1 in turn tunnels the
packets to CoA2 according to its cache table. Finally the MN receives the packets from R2 as shown
in Figure 4.
Figure 5 shows the Micro-mobility scenario in the proposed mechanism, where MN indicates the
augmentation of the signal received from R3 while it initiates its movement towards R3, thus it
translates the beacon received from R3 to the equivalent bitmap. Then, MN sends a BU req. message
including the bitmap field to the Home Anchor-1 which checks the highest bits of the received
bitmap, the highest bits refers to the same Home Anchor-1. Therefore Home Anchor-1 translates the
received bitmap to the equivalent Care of Address which is CoA3 (new CoA) in this case, and it adds
it to the Home Anchor-1s binding cache table.

659 Vol. 7, Issue 3, pp. 652-665

Enterprise Gateway
/Home Agent
2407:3000:200d:312::2/64
Home Anchor1
2407:4000::2/64
Home Anchor2
2407:5000::2/64
R1
R2 R3 R4 R5 R6
2407:4010::/64 2407:4020::/64 2407:4030::/64
2407:4040::/64 2407:4050::/64 2407:4060::/64
MN
HoA: 2407:3000:200d:312::b6:d2
CoA2: 2407:4020::b6:d2
Home Address Bitmap
2407:3000:200d:312:b6:d2 000010
Care of Address
2407:4020::b6:d2
2407:4000::2 (HA1)
Home Address Bitmap
2407:3000:200d:312:b6:d2
Home Anchor
000010
CN
1. Packet destined to HoA
2. Packet Tunneled to HA1
3. Packet Tunnel to CoA2
4. MN receives and
decapsulate the packet

Figure 4: CN sends Packets to the MN in its Foreign Network (R2)
So far, the Home Anchor-1 tunnels the packets to both CoA2 and CoA3 at the same time, this will
provide a continuous communication between CN and MN. MN still can receive the traffic even
before sending a BU message to Home Anchor-1 through R3.
Figure 6 shows that when MN obtains CoA3 from R3, it will send a BU message to Home Anchor-1
through R3, then Home Anchor-1 checks the Binding cache table again and refreshes it. It will
exclude CoA2 from the table and tunnels the packet only to CoA3 according to the new address of
MN.

660 Vol. 7, Issue 3, pp. 652-665

Enterprise Gateway
/Home Agent
2407:3000:200d:312::2/64
Home Anchor1
2407:4000::2/64
Home Anchor2
2407:5000::2/64
R1 R2 R3 R4 R5 R6
2407:4010::/64 2407:4020::/64 2407:4030::/64
2407:4040::/64
2407:4050::/64 2407:4060::/64
MN
HoA: 2407:3000:200d:312::b6:d2
CoA2: 2407:4020::b6:d2
3. Home Address Bitmap
2407:3000:200d:312:b6:d2
000010
Care of Adress
2407:4020::b6:d2
2407:4000::2 (HA1)
Home Address Bitmap
2407:3000:200d:312:b6:d2
Home Anchor
000010
CN
Packet destined to HoA
Packet Tunneled to HA1
4. Tunneling the packets to CoA2
and Probable CoA3
5. MN receives and
1. Movement
MN
2
2. BU req.
(HoA: 2407:3000:200d:312:b6:d2,
Expected bitmap: 000011)
000011 2407:4030::b6:d2
1. Initiate the movement
toward R3

Figure 5: MN starts to move to R3
Figure 7 shows the macro-mobility between two different home anchors. MN initiates its movement
toward R4, the same procedure previously explained will be followed, and MN will decry the
augmentation of R4s signal and translate it to the equivalent bitmap.
MN sends a BU req. message to Home Anchor-1; here the difference is that the Home Anchor checks
the received bitmap, it finds out that the highest bits of bitmap refers to another home anchor.
Therefore, Home Anchor-1 will forward the BU req. to the upstream Enterprise Gateway (Home
Agent) and at the same time it translates the lowest bits to the equivalent CoA (in this case CoA4).
Now, Home Anchor-1 tunnels the packets to CoA3 and CoA4 as shown in Figure 7, Enterprise
gateway receives the BU req. from Home Anchor-1, and translates the highest bit to the equivalent
home anchors address, as shown in figure 7. The Enterprise gateway will add the address of Home
Anchor-2, and then it will start to tunnel the packets to Home Anchor-2. Home Anchor-2 still does not
have a record in its binding cache table to track the new MNs CoA.

661 Vol. 7, Issue 3, pp. 652-665

Enterprise Gateway
/Home Agent
2407:3000:200d:312::2/64
Home Anchor1
2407:4000::2/64
Home Anchor2
2407:5000::2/64
R1 R2 R3 R4 R5 R6
2407:4010::/64 2407:4020::/64 2407:4030::/64
2407:4060::/64
2407:4070::/64 2407:4080::/64
HoA: 2407:3000:200d:312::b6:d2
CoA3: 2407:4030::b6:d2
2407:3000:200d:312:b6:d2
000010
Care of Adress
2407:4020::b6:d2
2407:4000::2 (HA1)
Home Address Bitmap
2407:3000:200d:312:b6:d2
Home Anchor
000010
CN
Packet Tunneled to HA1
MN receives and
1. BU
(HoA: 2407:3000:200d:312:b6:d2,
CoA3: 2407:3030::b6:d2)
000011 2407:4030::b6:d2
MN
2. Refresh
the table
Figure 6: MN obtained CoA3 from R3
As shown in figure 8, MN acquired CoA4 and sends a BU message to Home Anchor-2 and Home
Anchor-1 simultaneously.
Home Anchor-1 updates its table and sends packets to CoA4 only and at the same time Home
Anchor-2 starts to forward the traffic to CoA4. The MN will send a BU message to the Enterprise
Gateway (Home Agent) telling about CoA4.
Home Agent receives the BU message, checks the MNs CoA then excludes Home Anchor-1 from the
table and sends the traffic to Home Anhcor-2 only which is the next hop to reach CoA4.
662 Vol. 7, Issue 3, pp. 652-665

Enterprise Gateway
/Home Agent
2407:3000:200d:312::2/64
Home Anchor1
2407:4000::2/64
Home Anchor2
2407:5000::2/64
R1 R2 R3 R4 R5 R6
2407:4010::/64 2407:4020::/64 2407:4030::/64
2407:4060::/64
2407:4070::/64 2407:4080::/64
HoA: 2407:3000:200d:312::b6:d2
CoA3: 2407:4030::b6:d2
2407:3000:200d:312:b6:d2
000010
Care of Adress
2407:4020::b6:d2
2407:4000::2 (HA1)
2407:3000:200d:312:b6:d2
Home Anchor
0010
CN
2. BU request
(HoA: 2407:3000:200d:312:b6:d2,
Bitmap: 100110)
000011 2407:4030::b6:d2
MN
1. Initiates movement toward R4
100110 2407:4060::b6:d2
4. checking the Highest bit
5. a)Forward BU request
(HoA: 2407:3000:200d:312:b6:d2,
Bitmap: 100110)
2407:5000::2 (HA2) 100110
7. Tunnel the packet to HA1 & HA2
5. b)Tunnel packets to CoA3 & CoA4
MN

Figure 7: MN initiates movement to Home Anchor-2 (R4)

663 Vol. 7, Issue 3, pp. 652-665

100110
Enterprise Gateway
/Home Agent
2407:3000:200d:312::2/64
Home Anchor1
2407:4000::2/64
Home Anchor2
2407:5000::2/64
R1
R2 R3 R4 R5 R6
2407:4010::/64 2407:4020::/64 2407:4030::/64
2407:4060::/64
2407:4070::/64 2407:4080::/64
HoA: 2407:3000:200d:312::b6:d2
CoA4: 2407:4060::b6:d2
2407:3000:200d:312:b6:d2
000010
Care of Adress
2407:4020::b6:d2
2407:4000::2 (HA1)
Home Address Bitmap
2407:3000:200d:312:b6:d2
Home Anchor
000010
CN
1. BU
(HoA: 2407:3000:200d:312:b6:d2,
CoA4: 2407:4060::b6:d2)
000011 2407:4030::b6:d2
MN
100110 2407:4060::b6:d2
2. a) Refresh the table
2407:5000::2 (HA2)
Tunnel the packet to HA1 & HA2
Home Address Bitmap
2407:3000:200d:312:b6:d2 0010
Home Address
2407:4060::b6:d2
3. Forwards the packet from Home Agent
to CoA4

Figure 8: Home Anchor-2 forwards the traffic to CoA4

664 Vol. 7, Issue 3, pp. 652-665

100110
Enterprise Gateway
/Home Agent
2407:3000:200d:312::2/64
Home Anchor1
2407:4000::2/64
Home Anchor2
2407:5000::2/64
R1 R2 R3 R4 R5 R6
2407:4010::/64 2407:4020::/64 2407:4030::/64
2407:4060::/64
2407:4070::/64 2407:4080::/64
HoA: 2407:3000:200d:312::b6:d2
CoA4: 2407:4060::b6:d2
2407:3000:200d:312:b6:d2
000010
Care of Adress
2407:4020::b6:d2
2407:4000::2 (HA1)
2407:3000:200d:312:b6:d2
Home Anchor
000010
CN
1. BU
(HoA: 2407:3000:200d:312:b6:d2,
CoA4: 2407:4060::b6:d2)
000011 2407:4030::b6:d2
MN
100110 2407:4060::b6:d2
2407:5000::2 (HA2)
Home Address Bitmap
2407:3000:200d:312:b6:d2 0010
CoA
2407:4060::b6:d2
Forwards the packet from Home Agent to
CoA4
2. refresh the Home Agent table
Packet tunnel to HA2

Figure 9: Updating the Home Agent Table
III. CONCLUSIONS AND FUTURE WORK
From the proposed and explained scenarios, the MN will have a seamless mobility by anticipating the
new location and new CoA and take some actions as soon as it initiates the movement. On the other
hand, this mechanism will save the network resources by updating the cache table of the server.
Which means the traffic will not keep multicasting to multi probable addresses but only to one
probable address in each single movement. As a result, the real time (multimedia) applications will
experience better service flow.
A future work will study how the multimedia applications are affected when applying the proposed
mechanism.
ACKNOWLEDGEMENT
The author would like to thank Mr. Aus S. Matooq for his help reviewing, commenting and his
feedback on this paper.
665 Vol. 7, Issue 3, pp. 652-665

REFERENCES
[1]. Omae K., Okajima I., Umeda N., Mobility Management System, and Mobile Node Used in the
System, Mobility Management Method, Mobility Management Program, And Mobility Management
Node, International Patent US 7,349,364 B2, published date: 25 March 2008.
[2]. McCann P., Mobile IPv6 Fast Handovers for 802.11 Networks, Lucent Technologies, RFC4260,
2005.
[3]. Lee K. et. al., A Macro Mobility Handover Performance Improvement Scheme for HMIPv6, Lecture
Notes in Computer Science, Volume 3981, 2006, pp 410-419.
[4]. Kumar V., Lall G C and Dahiya P., Performance Evaluation of MIPv6 and HMIPv6 in terms of Key
Parameters, International Journal of Computer Applications 54(4):1-3, September 2012.
[5]. Tao M., Yuan H. and Wei W., Active overload prevention based adaptive MAP selection in HMIPv6
networks, Journal of Wireless Networks, February 2014, Volume 20, Issue 2, pp 197-208.
[6]. Soliman H., Castelluccia C.,El Malki K., Bellier L., Hierarchical Mobile IPv6 Mobility Management
(HMIPv6), RFC 4140, 2005.
[7]. Soliman H., Castelluccia C.,El Malki K., Bellier L., Hierarchical Mobile IPv6 (HMIPv6) Mobility
Management, RFC 5380, 2008.
AUTHORS BIOGRAPHY
Sahar A. Al-Talib received her BSc. in Electrical and Communication Engineering from
University of Mosul in 1976, her MSc, and PhD in Computer and Communications
Engineering from University Putra Malaysia (UPM), in 2003 and 2006 respectively. She
served at MIMOS Berhad, Technology Park Malaysia, for about 4 years as Staff Researcher
where she enrolled in infrastructure mesh network project (P000091) adopting IEEE
802.16j/e/d wireless technology. She became a lecturer at the Department of Computer and
Information Engineering, Electronics Engineering, University of Mosul, Iraq in 2011. She
was the main inventor of 10 patents and published more than 20 papers in local, regional, and international
journals and conferences. She certified CISCO-CCNA, WiMAX RF Network Engineer, WiMAX CORE
Network Engineer, IBM Basic Hardware Training Course- UK, Cii-Honeywell Bull Software Training
Course on Time Sharing System and Communication Concept- France, Six-Sigma white belt, Harvard Blended
Leadership Program by Harvard Business School Publishing USA, Protocol Design and Development for
Mesh Networks by ComNets Germany.

666 Vol. 7, Issue 3, pp. 666-674

EFFICIENCY OPTIMIZATION OF VECTOR-CONTROLLED
INDUCTION MOTOR DRIVE
Hussein Sarhan
Department of Mechatronics Engineering,
Faculty of Engineering Technology, Amman, Jordan

ABSTRACT
This paper presents a new approach that minimizes total losses and optimizes the efficiency of a variable speed
three-phase, squirrel-cage, vector-controlled induction motor drive through optimal control, based on combined
loss-model control and search control. The motor power factor is used as the main control variable. An
improvement of efficiency is obtained by online adjusting the stator current according to the value of power
factor corresponding to the minimum total losses at a given operating point. The drive system is simulated using
Matlab SIMULINK models. A PIC microcontroller is used as minimum loss power factor controller. Simulation
results show that the proposed approach significantly improves the efficiency and dynamic performance of the
drive system at all different operating conditions.
KEYWORDS: Efficiency Optimization, Induction Motor Drive, Optimal Power Factor, Simulation, Vector
Control.
I. INTRODUCTION
Squirrel-cage three-phase induction motors IMs are the workhorse of industries for variable speed
applications in a wide power range. However, the torque and speed control of these motors is difficult
because of their nonlinear and complex structure. In general there are two strategies to control these
drives: scalar control and vector control. Scalar control is due to magnitude variation of the control
variable only. The stator voltage can be used to control the flux, and frequency or slip can be adjusted
to control the torque. Different schemes for scalar control are used, such as: constant V/f ratio,
constant slip, and constant air-gap flux control. Scalar controlled drives have been widely used in
industry, but the inherent coupling effect (both torque and flux are function of stator voltage or current
and frequency) give sluggish response and system is easily prone to instability [1-3]. To improve the
performance of scalar-controlled drives, a feedback by angular rotational speed is used. However, it is
expensive and destroys the mechanical robustness of the drive system. Performance analysis of scalar-
controlled drives shows that scalar control can produce adequate performance in variable speed
drives, where the precision control is not required. These limitations of scalar control can be
overcome by implementing vector (field oriented) control.
Vector control was introduced in 1972 to realize the characteristics of separately-excited DC motor in
induction motor drives by decoupling the control of torque and flux in the motor. This type of control
is applicable to both induction and synchronous motors. Vector control is widely used in drive
systems requiring high dynamic and static performance. The principle of vector control is to control
independently the two Park components of the motor current, responsible for producing the torque
and flux respectively. In that way, the IM drive operates like a separately-excited DC motor drive
(where the torque and the flux are controlled by two independent orthogonal variables: the armature
and field currents, respectively) [4-8].
Vector control schemes are classified according to how the field angle is acquired. If the field angle is
calculated by using stator voltage and currents or hall sensors or flux sensing winding, then it is
known as direct vector control DVC. The field angle can also be obtained by using rotor position
measurement and partial estimation with only machine parameters, but not any other variables, such
667 Vol. 7, Issue 3, pp. 666-674

as voltages or currents. Using this field angle leads to a class of control schemes, known as indirect
vector control IVC [1, 9].
From energetic point of view, it is well known that three-phase induction motors, especially the
squirrel-cage type are responsible for most of energy consumed by electric motors in industry.
Therefore, motor energy saving solutions by increasing its efficiency has received considerable
attention during the last few decades due to the increase in energy cost. Energy saving can be
achieved by proper motor selection and design, improvement of power supply parameters and
utilizing a suitable optimal control technique [3-5, 10,11]
Induction motor operation under rated conditions is highly efficient. However, in many applications,
when the motor works at variable speed, it has more losses and less efficiency, so it operates far from
the rated point. Under these circumstances, it is not possible to improve the motor efficiency by motor
design or by supply waveform shaping technique. Therefore, a suitable control algorithm that
minimizes the motor losses will rather take place.
Minimum-loss control schemes have can be classified into three categories: search method, loss
model, power factor control. The power factor control scheme has the advantage that the controller
can be stabilized easily and the motor parameter information is not required. However, analytical
generation of the optimal power factor commands remains tedious and restrictive because empirical,
trial and error methods are generally used [4,5]. For this reason, search control using digital
controllers is preferable.
In this paper, a combined minimum-loss control and search control approach is used to find the power
factor, corresponding to the minimum losses in the drive system at a specified operating point. A PIC
microcontroller is used as an optimal power factor controller to online adjust the stator current
(voltage) to achieve the maximum efficiency of the drive system.
II. EFFICIENCY OPTIMIZATION OF VECTOR-CONTROLLED DRIVE SYSTEM
The generalized block diagram of vector-controlled induction motor drive is shown in Figure 1.

Figure 1. Generalized block diagram of vector-controlled induction motor drive.
To perform vector control, the following steps are required [1]:
1. Measurements of motor phase voltages and currents.
2. Transformation motor phase voltages and currents to 2-phase system ( | o, ) using Clarke
transformation, according to the following equation:
668 Vol. 7, Issue 3, pp. 666-674

) (
2
3
],
2
1
2
1
[
sc sb s
sc sb sa s
i i k i
i i i k i
=
=
|
o
(1)
where:
=
sa
i actual stator current of the motor phase A
=
sb
i actual stator current of the motor phase B
=
sc
i actual current of the motor phase C
const k =
3. Calculation of rotor flux space vector magnitude and position angle through the components of
rotor flux.
4. Transformation of stator currents to q d coordinate system using Park transformation, according
to:
field s field s sq
field s
ield
f s sd
i i i
i i i
u u
u u
| o
| o
cos sin
, sin cos
+ =
+ =
(2)
where
field
u is the rotor flux position.
The component
sd
i is called the direct axis component (the flux-producing component) and
sq
i is
called the quadrature axis component (the torque-producing component). They are time invariant; flux
and torque control with them is easy.
The values of
field
u sin and
field
u cos can be calculated by:
rd
r
field
rd
r
field
+
+
=
+
+
=
o
|
u
u
cos
, sin
(3)
where:
| o r r
rd
2 2
+ + + = + (4)
The rotor flux linkages can be expressed as:
o o o
o | |
s m r r r
s m r r r
i L i L
i L i L
+ = +
+ = + ,
(5)
5. The stator current torque-
sq
i and flux-
sd
i producing components are separately controlled.
6. Calculation of the output stator voltage space vector using decoupling block.
7. Transformation of stator space vector back from q d coordinate system to 2-phase system fixed
with the stator using inverse Park transformation by:
field sq field sd s
field sd s
i i i
field isq i i
u u
u u
|
o
cos sin
, sin cos
+ =
=
(6)
8. Generation of the output 3-phase voltage using space modulation.
The developed electromagnetic torque of the motor T can be defined as:
qs dr
r
m
i
L
L
P T |
4
3
= (7)
where P is the number of poles of the motor.
From Equation (7) it is clear that the torque is proportional to the product of the rotor flux linkages
and q-component of the stator current. This resembles the developed torque expression of the DC
motor, which is proportional to the product of the field flux linkages and the armature current. If the
669 Vol. 7, Issue 3, pp. 666-674

rotor flux linkage is maintained constant, then the torque is simply proportional to the torque
producing component of the stator current, as in the case of the separately excited DC machine with
armature current control, where the torque is proportional to the armature current when the field is
constant.
The power factor p.f is also a function of developed torque, motor speed and rotor flux, and can be
calculated as the ratio of input power to apparent power [3-4]:
ds qs ds qs
ds ds qs qs
i i v v
i v i v
f p
2 2 2 2
.
+ +
+
= (8)
The power factor can be used as a criterion for efficiency optimization. The optimal power factor
corresponding to minimum power losses in the drive system can be found analytically or by using
search method of control. To avoid tedious analytical calculations of power factor, a search control is
implemented in this paper. The stator voltage is incrementally adjusted until the controller detects the
minimum total losses at a given operating point. The total power losses P A can be calculated as
the difference between the input power
in
P and the output mechanical power
out
P :
e T i v i v P
ds ds qs qs
+ = A ) (
2
3
(9)
The block diagram of proposed efficiency optimization system for vector-controlled induction motor
is shown in Fig. 2. The system consists of 400V,4kW, 1430 rpm, 50Hz three-phase, squirrel-cage
induction motor. The motor is fed from three phase AC-to-AC energy converter, based on voltage
controlled pulse-width modulated inverter. A PIC 16f877A microcontroller is used as a controller in
the system.

Figure 2. Block diagram of vector-controlled optimized system.
To investigate the proposed optimized system, two Matlab SIMULINK models were constructed. The
first model, which is shown in Figure 3, represents the original vector-controlled drive system without
efficiency optimization. The second model, Figure 4, illustrates the efficiency-optimized, vector-
controlled system.
670 Vol. 7, Issue 3, pp. 666-674

Figure 3. Matlab SIMULINK model of vectorcontrolled drive system.

Figure 4. Matlab SIMULINK model of efficiency-optimized, vector-controlled drive system.
The PIC microcontroller was linked to the Matlab model by interface circuit shown in Figure 5.

Figure 5. Block diagram of interface circuit between Matlab model and PIC microcontroller model.
671 Vol. 7, Issue 3, pp. 666-674

III. SIMULATION RESULTS
To show the effect of proposed efficiency optimization technique based on power factor optimization,
a comparative analysis between original drive system and optimized one has been done. Investigation
was provided under the same operating conditions: frequency (speed) and load torque. Figures (6-8)
show the relationship between efficiency and load torque under constant frequency.

Figure 6. Relationship between efficiency and load torque at constant frequency Hz f 50 = .
It is clear that the implemented efficiency optimization technique improves the efficiency of the drive
system for the whole frequency range. Improvement is significant at light loads.


672 Vol. 7, Issue 3, pp. 666-674

The relationship between efficiency and frequency at constant load torque is presented in Figures (9-
10). At light load torques the efficiency improvement is very significant for the whole frequency
range.

Figure 9. Relationship between efficiency and frequency at constant load torque m N T . 12 =

Figure 10. Relationship between power factor and frequency at constant load torque m N T . 2 =
Figure (11) is an example of the relationship between power factor and frequency at constant load. It
is clear that the optimized system has better power factor for all frequencies.

Figure 11. Relationship between power factor and frequency at constant load torque m N T . 10 =
It was noticed that the dynamic response of the optimized system has been improved. The oscillations
in electromagnetic torque and angular speed disappeared and the response becomes faster. Examples
673 Vol. 7, Issue 3, pp. 666-674

of dynamic response at load torque N.m 5 = T and frequency Hz 30 = f are shown in Figures (12,
13).

Figure 12. Dynamic response of electromagnetic torque at load torque N.m 5 = T and frequency
Hz 30 = f .

Figure 13. Dynamic response of angular speed at load torque N.m 5 = T and frequency Hz 30 = f .
674 Vol. 7, Issue 3, pp. 666-674

IV. CONCLUSIONS
An online efficiency optimization control technique based on detecting optimal power factor, which
minimizes the total operational power losses in vector-controlled induction motor drive is proposed in
this paper. The power factor is used as the main control variable and manipulates the stator current in
order for the motor to operate at its minimum-loss point. Simulation results show that the
implemented method significantly improves the efficiency and dynamic performance of the drive
system, especially when the system operates at light loads and low frequencies.
V. FUTURE WORK
The obtained simulation (theoretical) results should be experimentally tested and validated for
industrial drive systems, operating at variable speeds with different load types. Also, the results
should be compared with that, for scalar-controlled drive system to insure that vector control approach
gives better results and ease to implement.
REFERENCES
[1] Feng-Chieh Lin and Sheng-Ming Yang (2003), On-line tuning of an efficiency-optimized vector
controlled induction motor drive, Tamkang Journal of Science and Engineering, Vol. 6, No. 2, pp. 103-110.
[2] C. Thanga Raj, S. P. Srivastava and Pramod Agarwal (2009), Energy efficient control of three-phase
induction motor- a review, International Journal of Computer and Electrical Engineering, Vol. 1, No. 1, pp.
1793-8198.
[3] Hussein Sarhan (2011), "Energy efficient control of three-phase induction motor drive", Energy and
Power Engineering, Vol. 3, pp. 107-112.
[4] Hussein Sarhan (2011), "Online energy efficient control of three-phase induction motor drive using
PIC microcontroller", International Review on Modeling and Simulation (I.RE.MO.S), Vol. 4, No. 5, pp. 2278-
2284.
[5] Hussein Sarhan, (2014) "Effect of high-order harmonics on efficiency-optimized three-phase induction
motor drive system performance", International Journal of Enhanced Research in Science Technology and
Engineering, Vol. 3, No. 4, pp. 15-20.
[6] Seena Thomas and Rinu Alice Koshy (2013), Efficiency optimization with improved transient
performance of indirect vector controlled induction motor drive, International Journal of Advanced Research
in Electrical, Electronics and Instrumentation Engineering, Vol. 2, Special Issue 1, pp. 374-385.
[7] K. Ranjith Kumar, D. Sakthibala and Dr. S. Palaniswami (2010), Efficiency optimization of induction
motor drive using soft computing techniques, International Journal of Computer Applications, Vol. 3, No. 1,
pp. 8875-8887.
[8] Branko D. Blanusa, Branko L. Dokic and Slobodan N. Vukosavic (2009), Efficiency optimized
control of high performance induction motor drive Electronics, Vol. 13, No. 2, pp. 8-13.
[9] G. Kohlrusz and D. Fodor (2011), Comparison of scalar and vector control strategies of induction
motors, Hungarian Journal of Industrial Chemistry, Vol. 39, No. 2, pp. 265-270.
[10] Rateb H. Issa (2013), Optimal efficiency controller of AC drive system, International Journal of
Computer Applications, Vol. 62, No. 12, pp. 40-46.
[11] A Taheri and H. Al-Jallad (2012), Induction motor efficient optimization control based on neural
networks, International Journal on Technical and Physical Problems of Engineering, Vol. 4, No. 2, pp. 140-
144.
AUTHORS
Hussein S. Sarhan was born in Amman, Jordan, in 1952. He received the Master and Ph.D
degrees in Electric Drive and Automation from Moscow Power Engineering Institute,
USSR, in 1978 and 1981, respectively.His research areas are induction motor optimization
techniques and energy efficient control of electric drives.Dr. Sarhan is a faculty member/
associate professor in Mechatronics Engineering Department/ Faculty of Engineering
Technology.
Sarhan is a member of Jordanian Engineering Association.

675 Vol. 7, Issue 3, pp. 675-683

FLEXIBLE DIFFERENTIAL FREQUENCY-TO-VOLTAGE AND
VOLTAGE-TO-FREQUENCY CONVERTERS USING
MONOLITHIC ANALOGUE RECONFIGURABLE DEVICES
Ivailo Milanov Pandiev
Department of Electronics, Technical University of Sofia, Sofia, Bulgaria

ABSTRACT
This paper presents differential frequency-to-voltage (F/V) and voltage-to-frequency (V/F) converters
employing single CMOS Field Programmable Analog Array (FPAA) device. The proposed electronic circuits
are based on charge-balance conversion method. The F/V converter basically consists of analogue comparator,
differentiator that detects the rising and falling edges of the shaped waveform, voltage-controlled switch and
output low-pass filter, working as an averaging circuit. The V/F conversion circuit includes the same modules.
The input signal for the V/F circuit is applied to an integrator (realized with low-pass filter) through two-
position voltage-controlled switch and the positive feedback is closed by external electrical connections. The
proposed converters can be redesigned during operation, in dependence on the amplitude, frequency range and
noise level of the input signal, without using of external components or applying of external current or voltage.
The functional elements of the converters are realized by employment of the available configurable analogue
modules (CAMs) of the FPAA AN231E04 from Anadigm. The converters have wide-band frequency response
and can operate with single supply voltage of 3.3V. The experimental results show that the linearity error is less
than 0.5% at the frequency range of 0 to 2025kHz and differential input range operation of 0 - 3V.
KEYWORDS: Mixed-signal circuits, Charge-balance voltage-to-frequency conversion method, Frequency-to-
voltage converter (FVC), Voltage-to-frequency converter (VFC), FPAA, System prototyping.
I. INTRODUCTION
The voltage-to-frequency converters (VFC) are voltage-controlled oscillators whose frequency is
linearly proportional to a control input voltage. The VFCs are useful devices for variety of mixed-
signal (analogue and digital) processing systems, such as A/D conversion systems with a
microcontroller, temperature meters, long-term integrators with infinite hold and telemetry systems
[1-4]. The dual process of the voltage-to-frequency (V/F) conversion is the frequency-to-voltage
(F/V) conversion. Besides, its independent applications to a speed indicator and motor speed control
system (tachometer), the FVC paired with the VFC is used for high noise immunity data transmission
systems and FSK generation and decoding.
The basic electrical parameters that define the quality of the F/V and the V/F converters are maximum
working frequency, sensitivity and linearity error. For the commercial monolithic F/V and the V/F
converters, such as AD650 (from Analog Devices), LM2907 (from Texas Instruments) and LM331
(from Texas Instruments) the transfer function and the related electrical parameters are formed by the
internal active building blocks and a group of external passive components. Moreover, the accuracy of
the electrical parameters is largely determined by the manufacturing tolerances and the temperature
drift of the values of the external passive components. A variety of F/V and the V/F converter
prototype circuits for mixed-signal signal processing, are available in the literature [5-15].
Furthermore, some of them realize only V/F conversion or F/V conversion, while others implement
both functions of conversion [5]. The majority of the published circuits of F/V and the V/F converters
have fixed value of the sensitivity and possibility for modification of this value is not demonstrated. In
many cases the amplification of small signals (such as bio-signals) requires modification of the
676 Vol. 7, Issue 3, pp. 675-683

sensitivity according to the variation of the amplitude, the bandwidth or the noise level. To solve this
problem, it can be used Field Programmable Analog Array (FPAA) devices. The FPAA are
programmable CMOS analogue integrated circuits based on SC technology, which can be configured
not only in the design process of a device, but also during the operation. As well as FPGA the
programmable analogue arrays provide cost-efficient and relatively fast design of complex electronic
circuits.
The use of these reconfigurable analogue devices provides the F/V and the V/F converters with ability
to modify the electrical parameters according to the parameters of the input signals. Furthermore, by
using of FPAAs the noise level can be obtained with significantly lower values due to the differential
inputs and outputs. Among different devices and technologies in the market [16-
offers dynamically programmed FPAAs (or dynamically programmed analogue signal processors -
dpASPs), in particular the AN231E04 that is a reconfigurable programmable analogue device based
on switched-
configurable analogue blocks (CABs), surrounded by programmable interconnect resources and
analogue input/output cells.
The VFC in [10] is realized by using single FPAA AN221E04 and convert the input voltage from 0 to
3V into square waves within frequency range from 0 to 7kHz. Thus, the sensitivity of [10] is 2kHz/V
and the achieved relative linearity error is approximately equal to 2%. As a flexible differential
building block, however, it is desirable to realize the V/F converter in a form of compatible with the
F/V converter. With this in mind, novel FPAA-based circuits are developed for F/V and V/F
conversion.
The organization of the paper is as follows: the structures and the principles of operation of the F/V
and the V/F converters are described in section II; for the proposed electronic circuits in section II the
transfer functions and methods for modification of the transmission coefficients are given; in section
III are illustrated the experimental test of the proposed converters based on AN231K04-DVLP3
development board [20], which is built around the AN231E04 device and finally in section IV the
concluding remarks and directions for future work are discussed.
Nomenclature
os
t Pulse with duration, determined by the master clock frequency for all CAMs
in
f Frequency of input signal
c
f Corner frequency of low-pass filter
2
K Transmission coefficient of the SumFilter1 CAM
ref
U
Reference voltage
V
K Sensitivity of F/V converter
out
f Frequency of output signal
in
U Input voltage
int
t
os
out
t
f
1
- amount of time required to reach comparator threshold
f
K
Sensitivity of V/F converter
dc Direct current
II. CIRCUITS DESCRIPTION
2.1. F/V converter
The circuit diagram of the F/V converter is shown on Fig. 1. It is built by selection among the
available configurable analogue modules (CAMs) of the AN231E04, shown in Table 1. The
differential input signal is applied between terminals 11 and 12 and average output voltage is obtained
between terminals 15 and 16. After shaping of the input signal by a Comparator1, the Differentiator1
detects the rising and falling edges of the shaped waveform in synchronism with the non-overlapping

the master clock quartz stabilized signal, used for the proper work of the whole FPAA. The
677 Vol. 7, Issue 3, pp. 675-683

differentiator produces short pulses with duration
os
t , determined by the master clock frequency for
all CAMs, and the polarity depending on the direction of change of the comparators output voltage.
When the output of the differentiator is equal to zero, the voltage steering switch GainSwitch1 diverts
differential voltage equal to 0V to the output of the gain stage; this is called the integration period.
When the differentiator produces short pulse with negative value, the switch GainSwitch1 diverts the
voltage K2Uref (Uref produces by the voltage source Voltage1 CAM with fixed value equal to +2V)
from the SumFilter1 CAM to the output of the gain stage; this is called the reset period. The output
low-pass filter implemented by FilterBilinear1 CAM works as an averaging circuit (
c in
f f >> , where
c
f is the corner frequency). As the frequency
in
f of the input signal increases, the amount of charge
injected into the averaging circuit increases proportionally.
The average output voltage
av o
U
,
, obtained between terminals 17 and 18, that is proportional to the
input frequency is given as
(1)
in ref os
t
ref
in
av o
f U t K dt U K
T
U
os
2
0
2 ,
.
1
= =
}
,
where
in in
f T / 1 = is the period of the input signal.
Therefore, based on the formula (1), the sensitivity of the proposed F/V converter is obtained as
(2)
ref os V
U t K K
2
= .
Output
voltage
av o
U
,
VMR
Input
signal
VMR
in
f
SumFilter1
Voltage1
GainSwitch1
Comparator1
Differentiator1
Comparator2
FilterBilinear1

Figure 1. FPAA configuration of F/V converter.
The analysis of equation (2) shows that the change of the F/V converters sensitivity can be
accomplished in two ways. First, for a fixed value of
os
t , for example equal to s 20 (at master clock
frequency equal to kHz 50 ), for modification of the transmission coefficient
2
K from 0.1 to 1 the
sensitivity will vary from kHz mV/ 4 to kHz mV/ 40 . Second, for a fixed value of the
2
K , for
example equal to 0.5 when change the clock frequency from kHz 10 ( s t
os
=100 ) to kHz 100
( s t
os
=10 ) the sensitivity varies from kHz mV/ 100 to kHz mV/ 10 . Moreover, the change of the
coefficients of the CAMs can be implemented in the operating mode of the F/V converter.

678 Vol. 7, Issue 3, pp. 675-683

Table 1. FPAA CAMs for F/V and V/F conversion.
Name Options Parameters Clocks
SumFilter1

Output changes On: Phase 1
Input 1: non-inverting
Input 2: inverting
Corner frequency [kHz]: 1
Gain1 (Upper input): K1=1
Gain 2: (Lower input): K2
Clock A: 50kHz
(Clock 3)
Voltage1

Polarity: Positive (+2V) - -
GainSwitch1

Compare control to: Signal
ground
Select input 1 when: Control low
Comparator sampling phase:
Phase 2
Gain stage: half cycle
Gain1 (Upper input): 1.00
Gain 2: (Lower input):
1:00
Clock A: 50kHz
(Clock 3)
Comparator1

Compare to: Dual input
Input sampling: Phase 1
Output polarity: Non-inverted
Hysteresis: 0mV
- Clock A: 50kHz
(Clock 3)
Differentiator1

- Differentiation constant
[us]: 1
Clock A: 50kHz
(Clock 3)
FilterBilinear1

Filter type: Low pass
Input sampling phase: Phase 2
Polarity: Non-inverted
Corner frequency [kHz]:
0.055
Gain: 1.00
Cint1=Cint2=Cint=7.9pF
Clock A: 50kHz
(Clock 3)
Comparator2

Compare to: Signal ground
Input sampling: Phase 2
Output polarity: Non-inverted
Hysteresis: 0mV
- Clock A: 50kHz
(Clock 3)
The stability of the
V
K in equation (2) is determined by the stability of the master clock frequency,
which is derived from the crystal oscillator and the accuracy of the parameters of the CAMs, whose
values are derived from the same clock frequency. Therefore, the variation of the sensitivity
V
K of
the proposed F/V converter will not cause changing of the nonlinearity error.
For normal operation of the F/V converter the inverting input (terminals 23 and 24) of the comparator
and the input (terminals 01 and 02) of the SumFilter1 are connected to voltage mid-rail (VMR) equal
to V 5 . 1 + .
2.2. V/F converter
The circuit diagram of the V/F converter is shown on Fig. 2. As can be seen, for the implementation
of the V/F converter is used the F/V converter shown on Fig. 1. For this purpose, the input voltage
in
U is applied between terminals 01 and 02, terminals 11 and 12 are connected to a VMR and the
voltage between terminals 17 and 18 is applied directly to the terminals 24 and 23. Thus it closes the
positive feedback between the output of the averaging circuit and the inverting input of the
Comparator1. The output differential signal with frequency
out
f is formed by the Comparator2 and
can be obtained between terminals 19 and 20.
When the circuit operated as a V/F converter, the transformation from voltage to frequency is based
on a comparison of the input voltages value
in
U to the internal voltage equal to V 2 + , implemented
by Voltage1 CAM.
The principle of operation of the circuit shown on Fig. 2 can be described as follows. At the beginning
of a cycle, the input voltage is applied to the input of the FilterBilinear1 and its output voltage
decreases. This is the integration period. When the FilterBilinear1 output voltage (terminals 17 and
18) crosses signal ground, the Comparator1 triggers a one-shot whose time period
os
t is determined
by the clock frequency. During this period, a voltage of (
ref in
U K U
2
) is applied to the input of the
FilterBilinear1 and the output voltage starts to increase. This is the reset period of the circuit. After
the reset period has ended, the circuit begins another integration period, and starts ramping downward
again. The amount of time required to reach the comparator threshold is given as
679 Vol. 7, Issue 3, pp. 675-683

(3)
int
2
int
18 17
18 17
int
) (
C
U
U K U
C
t
dt
dU
U
t
in
ref in
os
=
A
=
,
where the pF C 9 . 7
int
= is the internal capacitor of the FilterBilinear1.
The output frequency can be found as
(4)
ref os
in
os
out
U K t
U
t t
f
2 int
1
=
+
= .
The integration capacitor
int
C has no effect on the transfer function, but merely determines the
amplitude of the sawtooth signal out of the averaging circuit.
Output
signal -
out
f
Input
voltage
VMR
in
U
SumFilter1
Voltage1
GainSwitch1
Comparator1
Differentiator1
Comparator2
FilterBilinear1

Figure 2. FPAA configuration of V/F converter.
From (4) for the instability of output frequency is found
(5)
ref
ref
os
os
in
in
out
out
U
U
K
K
t
t
U
U
f
f
A
A
=
A
2
2
.
The first term in formula (5) is the regulating effect. The second, third and fourth term reflect
confusing influence in the circuit. The output frequency is linearly dependent on the input voltage.
The transmission coefficient
2
K is interrelated with the corner frequency value of the CAM and the
ratios of the two switched capacitors with values pF C C
out in
9 . 7 = = . The voltage
ref
U is produced by
a dc voltage source, which making a connection to the on-chip voltage references. There are no
capacitors or dynamic switches. It produces a dc output so that the output is continuous and valid
during the both phases of the clock signals [20].
According to the formula (4), it is concluded that the sensitivity of the proposed V/F converter is
governed by the following equation
(6)
ref os f
U t K K
2
/ 1 = .
Similar to equation (2), the analysis of equation (6) shows that the variation of the sensitivity can be
achieved by changing the value of the
2
K or by changing the master clock frequency of the circuit.
III. EXPERIMENTAL RESULTS AND DISCUSSION
The experimental test that has been used for validation of the proposed converters is based on
AN231K04-DVLP3 Development board [20], which is built around the AN231E04 device. The
power supply voltage of the AN231E04 is equal to +3.3V.
680 Vol. 7, Issue 3, pp. 675-683

The measured average output voltages of the prototype V/F converter are plotted on Fig. 3 against the
frequency of the output signal at two values of the conversion sensitivity. The conversion sensitivities
1 V
K and
2 V
K are obtained for gain
2
K equal to 0.5 and 1, respectively. The clock frequency of the
circuit is chosen to be 50kHz. However, for the frequency change from 0 to 25kHz the average output
voltage varies from 0 to 0.5 V (or 1V). In this operating range an error of less than 0.5% is achieved.
In comparison with the F/V converters, presented in [5] and [7], the proposed converter has higher
operating frequency and greater sensitivity. The F/V converters reported in [6] and [8] has a wider
bandwidth (> 1MHz) and good linearity, but the sensitivity is a fixed value and the modification of the
electrical parameters required redesigning of the electronic circuits. Also the F/V converter presented
in [6] can use only digital input signals, while the created F/V converter can operate with both
analogue and digital signals.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 5 10 15 20 25
measured
calculated
V U
av o
,
,
kHz
mV
K
V
20
1
=
kHz
mV
K
V
40
2
=
kHz f
in
,

Figure 3. Average output voltage versus the frequency of the input signal at two values of the conversion
sensitivity obtained by the F/V converter.
The oscillation frequency
out
f of the prototype V/F converter against the input voltage is plotted on
Fig. 4. The gain
2
K is set to 0.5 and 1, respectively. The clock frequency of the circuit is set equal to
50kHz. When the input voltage changes from 0 to 0.4 V (or 0.8 V) the oscillation frequency varies
from 0 to 20kHz, the linearity error is again not higher than 0.5%. Thus, a figure of merit (FOM =
sensitivity (MHz) / error (%)) is equal to 0.1. The result is in a good agreement with the calculated by
(4). Fig. 5 and 6 present the measured output waveform at the input voltage equal to mV 20 and
mV 400
, respectively. The clock frequency is set equal to kHz 50 and the sensitivity is
V kHz K
f
/ 25 = . The output frequencies at these input voltages are 500.572Hz and 10.0013kHz,
respectively. For the both waveforms to channel 2 (CH2) the output signal from the inverting output is
applied (terminal 19 of the FPAA) and to the channel 1 (CH1) the signal from the non-inverting
output is applied (terminal 20 of the FPAA). The differential output signal (M) is the difference of the
voltages applied to the both channels. Notably, that with variation of the input voltage the output
frequency varies proportionally, as the pulse duration is unaffected ( s t
os
= 20 ) and the error is less
than 0.2%.
681 Vol. 7, Issue 3, pp. 675-683

0
5
10
15
20
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
measured
calculated
measured
calculated
V U
in
,
kHz f
out
,
V
kHz
K
f
25
1
=
V
kHz
K
f
50
2
=
V
kHz
K
f
50
2
=

Figure 4. The oscillation frequency versus the input voltage at two values of the conversion sensitivity obtained
by the V/F converter.

Figure 5. The output waveform of the V/F converter at V kHz K
f
/ 25 = and mV U
in
20 = .

Figure 6. The output waveform of the V/F converter at V kHz K
f
/ 25 = and mV U
in
400 = .

682 Vol. 7, Issue 3, pp. 675-683

Table 2. Comparison with prior designs of V/F converters.
This work [5] [9] [10] [11] [13] [14] [15]
Technology CMOS CMOS CMOS CMOS N/A CMOS CMOS CMOS
Max. fout 20kHz 20kHz 8kHz 7kHz 3.5kHz 52.95MHz 260kHz 100MHz
Uin 0-3V 0-1V 0.1-10V 0-3V 0-5V 0-0.9V 0-2V
0-10V in
100mV
steps
Sensitivity 25250kHz/V 1kHz/V 0.8kHz/V 2kHz/V 0.7kHz/V 58MHz/V 130kHz/V -
Error s 0.5% N/A 0.02% 1.95% < 1% < 8.5% < 7%
within
5ppm
FOM 0.050.5 N/A 0.04 0,00102 0.0007 6.28 0.019 -
Power at
max fout
65mW N/A N/A 95mW N/A 0.218mW N/A
N/A
Year 2014 1986 1997 2004 2005 2007 2010 2013

The measured results compared with several prior works are summarized in Table 2. In comparison
with the V/F converter, presented in [9], the proposed V/F converter has higher error (s 0.5%), but the
maximum frequency is higher and the sensitivity with significantly greater value can be obtained. The
V/F converter presented in [13] is characterized by higher operating frequency and sensitivity, but the
error accept relatively large value (equal to 8.5%), which may affects to the conversion of weak
signals from various sensors. Obviously, the error for the converters reported in [5], [10], [11] and
[14] are higher than the proposed V/F converter, while the input voltage range for the most circuits is
from 0 to several volts. For V/F converter reported in [15] the operating frequency bandwidth is
higher and the error is relatively small (within 5ppm), but the sensitivity is with fixed value and
under a step equal to 100mV are obtained 100 discrete values for the frequency of the output signal in
the range up to 100MHz.
IV. CONCLUSION AND FUTURE WORK
In this paper a differential F/V and V/F converters by using of a charge-balance method was
proposed. The selected FPAA is an Anadigm AN231E04, where the mixed-signal processing is
implemented. Moreover, for the created V/F converter the linearity error not exceeding 0.5% was
achieved at sensitivity changes from 25kHz/V to 250kHz/V. In comparison with prior works the
variation of the sensitivity depending on the parameters of the input signal could be done by changing
the parameters of configurable analogue blocks without changing the structure of the circuit or
without adding of external passive components. The stability of the sensitivity was determined by the
stability of the master clock frequency, which is derived from the external quartz crystal oscillator and
the accuracy of the parameters of the configurable blocks, whose values are derived from the same
clock frequency. Therefore, these converters can be used as flexible mixed-signal building blocks in
extracting weak differential signals from transducers and other signal sources.
Future work will be focused on implementation of the proposed F/V and V/F converters in
programmable systems-on-chips (PSoCs), which combine analogue and digital functions.
Furthermore, the development of the proposed electronic circuits will be oriented to modelling and
design of PLL-based frequency synthesizers and synchronizers.
REFERENCES
[1]. Kester, W., & Bryant J. (2004). Analog-digital conversion, USA Analog Devices.
[2]. Paiva, M. (2002) Voltage-to-frequency/frequency-to-voltage converter (AN795). Retrieved from
Microchip website: http://ww1.microchip.com/downloads/ en/appnotes/00795a.pdf.
[3]. Bryant J. (1989) Voltage-to-frequency converters (AN-361). Retrieved from Analog Devices
website: http://www.analog.com/static/imported-files/application_notes/84860375AN361.pdf
683 Vol. 7, Issue 3, pp. 675-683

[4]. Klonowski, P. (1990) Analog-to-digital conversion using voltage-to-frequency converters (AN-276).
Retrieved from Analog Devices website: http://www.analog.com/static/imported-
files/application_notes/185321581AN276.pdf.
[5]. Matsumoto, H., & Fujiwara, K. (1986) Switched- capacitor frequency-to-voltage and voltage-to-
frequency converters, IEEE Transactions on Circuits and Systems, Vol. 33 (8), pp 836838.
[6]. Djemouai, A., Sawan, M., & Slamani, M. (2001) New frequency-locked loop based on CMOS
frequency-to-voltage converter: design and implementation, IEEE Transactions on Circuits and
Systems II: Analog and Digital Signal Processing, Vol. 48 (5), pp 441449.
[7]. Gupta, K., Bhardwaj, M., & Singh, B. (2012) Design and analysis of CMOS frequency to voltage
converter using 0.35m technology, International Journal of Scientific & Engineering Research, Vol.
3 (5), pp 19. Retrieved from www.ijser.org/researchpaper%5CDesign-and-Analysis-of-CMOS-
Frequency- to-Voltage-Converter.pdf.
[8]. Singh, M., & Sahu P.P. (2013) A wideband linear sinusoidal frequency to voltage converter with fast
response time, Procedia Engineering, Vol. 64, pp 26-35.
[9]. Trofimenkoff, F.N., Sabouri, F., Jichang, Qin & Haslett, J.W. (1997) A square-rooting voltage-to-
frequency converter, IEEE Transactions on Instrumentation and Measurement, Vol. 46 (5), pp 1208
1211.
[10]. Yakimov, P., Manolov, E., & Hristov, M. (2004) Design and implementation of a V-f converter using
FPAA, 27th International Spring Seminar on Electronics Technology: Meeting the Challenges of
Electronics Technology Progress, Vol. 1, pp 126-129.
[11]. Stork, M. (2005) Sigma-delta voltage to frequency converter with phase modulapossibility, Turkish
Journal of Electrical Engineering & Computer Sciences, Vol. 13 (1), pp 6167.
[12]. Wang, C.-C, Lee, T.-Je, Li C.-C. & Hu, R. (2006) An all-MOS high-linearity voltage-to-frequency
converter chip with 520-kHz/V sensitivity, IEEE Transactions on Circuits and Systems II: Express
Briefs, Vol. 53 (8), pp 744747.
[13]. Wang, C.-C, Lee, T.-Je, Li C.-C. & Hu, R. (2007) Voltage-to-frequency converter with high
sensitivity using all-MOS voltage window comparator, Microelectronics Journal, Vol. 38 (2), pp
197202.
[14]. Valero, M.R., Aldea, C., Celma, S. Calvo, B., & Medrano, N. (2010) A CMOS differential voltage-to-
frequency converter with temperature drift compensation, 2010 European Workshop on Smart
Objects: Systems, Technologies and Applications (RFID Sys Tech), Ciudad, Spain.
[15]. Hino, R., Clement, J. M., & Fajardo, P. (2013) A 100MHz voltage to frequency converter, Journal of
Physics: Conference Series, Vol. 425 (21), pp 1-4.
[16]. Lattice ispPAC-POWR1014/A (2012). Retrieved from http://www.latticesemi.com/
documents/DS1018.pdf.
[17]. Cypress PSoC Mixed-Signal Array (2012). Retrieved from http://www.cypress.com
[18]. Zetex TRAC-S2 (2012), Retrieved from http://www.datasheetcatalog.org/datasheets/
208/306841_DS.pdf.
[19]. Anadigm dpASP (2013). Retrieved from http://www.anadigm.com/dpasp.asp
[20]. AN231K04-DVLP3 AnadigmApex Development Board (2013). Retrieved from
http://www.anadigm.com/_doc/UM231000-K001.pdf.
AUTHOR
Ivailo Pandiev received his M.Sc and Ph.D. degrees in Electronic Engineering from
Technical University of Sofia (TU-Sofia), Bulgaria, in 1996 and 2000, respectively. He is
currently an associate professor in the Department of Electronics at the TU-Sofia. His
research interests include analysis and design of analogue and mixed-signal circuits and
systems.
684 Vol. 7, Issue 3, pp. 684-691

A Review SEARCH BITMAP IMAGE FOR SUB IMAGE AND
THE PADDING PROBLEM
School of Engineering, Dept. of S\w Engineering
Koya University, Iraq - Erbil

ABSTRACT
In this paper we will discuss the algorithm of search Bitmap image for sub image or duplicate images ,It is
difficult for people to find or recognize if sub images are Found in large Image .This algorithm is a part of
Image Processing Technique to help people solve Search problem .This algorithm has possibility to Convert
the image to Binary Matrix and to individualize The Reading Buffer Size according to the padding edge. This
algorithm is the first step to create Image Search Engine which can find matching image without looking to
image name text or alternative text tag.
KEYWORDS: Sub Image, Image Search Engine, Image Processing, Bitmap, Padding, algorithm, Duplicate,
Matrix, Bitmap file header, Java, Looping, Coordinate.
I. INTRODUCTION
After The Huge a evolution in multimedia world ,Images search and images manipulation process
must grow up to be powerful and effective especially with Image sharing web site , Millions of
images are published on the web every day , How we can make search for an image or a part of it
easy like we search text in Google?
Images are not usually named with conveniently text, and the alternative Tag all the time is empty. So
the images need to be searched by their pixel.
Now it is important to compare images with images or sub image and to locate images that contain the
pattern which we are looking for it. The pattern can be a picture of anything Number, Character,
person, building or just duplicate to the same image, and the search operation is to retrieve matching
images. This retrieval is called the sub image searching.
In this paper, we propose an algorithm for searching sub image into large image and to locate the
indexes to all matching Pattern. This algorithm can solve any Image search for 24-bit Bitmap image
format , And we will discuss The padding problem in bitmap file .The search process need big loops
in four levels deep and the Inner condition that find any different in pixel must rest the looping
and exit out the Inner loop to the main loop which start with New Line[7] .
II. SURVEY OF BITMAP IMAGE
The bitmap is very strictly bound to hardware platform architecture, Bitmap was created by Microsoft
and IBM, Bitmap file format compatible with Intel File format which called the little endian format
[1, 2, 3, 5, and 8] because of the byte order that an Intel processor uses internally to store values.
Bitmap file has four types of headers; those types are differentiated by the size member, which is the
first dword in each of the structures.
1. BITMAP CORE HEADER
2. BITMAP INFO HEADER
3. BITMAP V4 HEADER
4. BITMAP V5 HEADER

685 Vol. 7, Issue 3, pp. 684-691

In BITMAP INFO HEADER there is a color table describes how pixel values correspond to RGB
color values. RGB is a model for describing colors that are produced by emitting light [4].Bitmap
pixels stored in order binary integers format this order can by (red, blue, green) but in 24-bit there is
reverse ordering (blue, green, and red) [5] so each image format has deferent bits order which known
as color Mask . Bitmap pixels stored in upside way (down to up) and that mean The first line in the
file is the bottom line of the image and the last line in the file is the first line of the image.[6]
2.2. Bitmap Image Most popular Format:
8-bit image is 8 bits file and it use 1 bytes for color information.
2.3. 24-bit Bitmap Image Padding Problem:
Bitmap file BITMAPINFOHEADER Header has two parts (14-byte file header and a 40-byte
information header) , These two parts Contain information about Data stored in the file and the
order of that data .IF we look at The 24-bit Bitmap Header We will find Describe For Image Size,
Row Length (width) , Column Length (height) and the RGB (Red/Green/Blue) color information
values (pixel) which are used in the file to present the image Details , RGB need 3 bytes in the 24-
bit Bitmap image. [1, 2, 3, 5, 6, 8]
Each pixel is represented by Triple of RGB. Depending on the width of the image, pad maybe used at
the end of each Row. If width is a multiple of 4, no pad is required, otherwise at the end of each line
we need to add a number which will make the width divisible by 4 this number is Gap with Zero
Value.
When OS Load an Image to the memory there is a scan Operation, each scan line is padded out to an
even 4-byte boundary (even 4-byte has odd 3 bytes of RGB). That mean every line in the 24-bit image
file has point to determine the end of line this point called Pad and it must be even number of 4-bytes
Triple, in the same time and at the same scan line it must has RGB of 3-bytes number, and we can
calculate the scan line length by counting the RGB 3- bytes according to this equation:
Scan Row length=Scan Row width * 3
Dependence on the above equation the result could be odd number, but the boundary must be even
number which means maybe the Row has gap between Row length and Row width, this gap known as
Padding Problem, we can get the Image size, Width and Height from the Bitmap Header.
2.4. Example for padding:
If we have a Bitmap image with depth 24-bit and the area (X * Y) of this image is 15 * 15.Now we
can calculate the color pixel information like that (15 * 3=45 bytes) in each row.
45 bytes is the length of every row so now we have to compute if row has pad or not, since 24-bit
image is an odd numbers 3 bytes color depth we have to find the modulo of row by 4 like this (45
MOD 4=1), by looking to the modulo result if its Zero that meaning image format depends on
Padding and no gap found.
In our example the modulo give us the result 1, now we have to find out how many bytes of padding
in this image.
Image_row_length=image_width * 3;
Image_row_length=15 * 3;
Padding= 4 ( image_row_length %4);
Padding=4 ( 1);
Padding=(3);
Image_real _Row_Length= Image_row_length +Padding
Image_real _ Row _Length= 45+3;

Figure 1:24-bit Image Row With Padding Gap
686 Vol. 7, Issue 3, pp. 684-691

Padding result tell us we need to use 3 bytes to fill the gap between image row color info and the end
of the each row, that gap has the value of zero[9].
So we can find The Image Length By this equation:
Image_Length=(( Image_row_length + Padding) * Image_ column _length *3)
2.5. Another Example:
In this Example we will see an 24-bit bitmap image with area 12 * 12.
When we calculate the color pixel info it will be (12 * 3=36 bytes) in each row 300 bytes is the
length of each row so now we have to compute if row has padding or not like this (36 MOD 4=0), the
modulo result is Zero that meaning Padding will be Zero and No gap between image row color
information and the end of row [9].
Image_real _Row_Length= Image_row_length +0.
Image_real _Row_Length= 36 +0 =36.

Figure 2:24-bit Image Row with No Padding Gap
The Image Length:
Image_Length=(( Image_row_length) * Image_ column _length *3)
2.6. Algorithm to solve 24-bit Padding Problem:
//Find 24-bit Image Real Length Padding
Image_row_length=image_width * 3;
if (image_row_length % 4 !=0){
Padding= 4 ( image_row_length %4);
}
else{
Padding=0;
}
Image_real _Length=(( Image_row_length +Padding) *Image_ column _length *3)
//End
After we have explained the problem of Padding in 24-bit image we can suggest a way to calculate
the padding for many deferent format of Bitmap Image like 8-bit ,16-bit,24-bit and 32-bit depends on
Bit Count :
2.7. Algorithm to solve 8-bit, 16-bit, 24-bit and 32-bit Padding:
Image_row_length=image_width *( image_Bit_Count / 8);
if (image_row_length % 4 !=0){
Padding= 4 ( image_row_length % 4);
}
else{
Padding=0;
}
Image_real _Length=(( Image_row_length +Padding) *Image_ column _length *3);
III. IMPLEMENTATION
In this section we will describe the implementation details of our algorithm . This algorithm has two
class .Sub Image Search class and Convert Image To Matrix class, Sub Image Search class will
implement the search operation to find the matching indexes, Convert Image To Matrix class has the
responsibility to Read The Bitmap image file header and individualize the padding and File Buffer
Size , After that it will convert the image RGB color data to a matrix of integer .
3.1.Test Media:
We have an image for Numbers from 1 to 20 ,this image is 24-bit bitmap image
687 Vol. 7, Issue 3, pp. 684-691

Figure 3: LargeImageNumbers.bmp
And the sub image could by any Number image, Lets take number five as example

Figure 4 :SubImageN5.bmp
3.2 Java Class:
3.2.1 ConvertImageToMatrix Class:
Our first step is converting the images to matrix of integers , In order to create the matrix we must
read the image header to get the file size ,Image width , Image height and the Bit count , After that
we have to Check if its 24-bit Image or not [9].
There is a java code designed to read image header and get the Bitmap details [9], You can find this
code on:
http://www.javaworld.com/javaworld/javatips/jw-javatip43.html
3.2.1.2 Read Bitmap Header:
In order to get the bitmap details we have to read the two parts of bitmap header
BITMAPFILEHEADER(14-byte) and BITMAPINFOHEADER(40-byte), All the date in header
stored in binary and that mean we have to use the binary operators {|,&,<} [6,7,9] .
Get File Size:
try {
FileInputStream fs = new FileInputStream(imageFile);
int bflen = 14; // 14 byte BITMAPFILEHEADER
byte bf[] = new byte[bflen];
fs.read(bf,0, bflen); // Read the file header
int nsize = (((int)bf[5] & 0xff) << 24)
| (((int)bf[4] & 0xff) << 16)
| (((int)bf[3] & 0xff) << 8)
| (int)bf[2] & 0xff;
}
catch (Exception e) {
688 Vol. 7, Issue 3, pp. 684-691

System.out.println(e.getMessage());
}
Get image width , Height ,Bit count:
try {
FileInputStream fs = new FileInputStream(imageFile);
int bilen=40; // 40-byte BITMAPINFOHEADER
byte bi[] = new byte[bilen];
fs.read(bi, 0, bilen); // Read the bitmap header
int nbisize = (((int)bi[3] & 0xff) << 24)
| (((int)bi[2] & 0xff) << 16)
| (((int)bi[1] & 0xff) << 8)
| (int)bi[0] & 0xff;
int nwidth = (((int)bi[7] & 0xff) << 24)
| (((int)bi[6] & 0xff) << 16)
| (((int)bi[5] & 0xff) << 8)
int nheight = (((int)bi[11] & 0xff) << 24)
| (((int)bi[10] & 0xff) << 16)
| (((int)bi[9] & 0xff) << 8)
int nbitcount = (((int)bi[15] & 0xff) << 8) | (int)bi[14] & 0xff;
int nsizeimage = (((int)bi[23] & 0xff) << 24)
| (((int)bi[22] & 0xff) << 16)
| (((int)bi[21] & 0xff) << 8)
| (int)bi[20] & 0xff;
if (nbitcount != 24){
System.err.println("Use only 24bit color .bmp files");
System.exit(1);
}
}
catch (Exception e) {
System.out.println(e.getMessage());
}
In this step we checked if the image is 24-bit , If not the code will print error Message and exit.
3.2.1.3 Handling Padding problem and Buffer Size:
After we get the image details, We must handling Padding problem in order to set the Buffer Reader
Size:
int padding =(nwidth * 3)%4;
int npad=0;
if(padding!=0){
npad = 4-( (nwidth * 3)%4); // the Gap Size
}
else{
npad=0;// No Gap
}
//-----Set Buffer Size
int imageBufferSize=( nwidth + npad) * 3 * nheight;
3.2.1.3 Reading Image Data And Store it in matrix :
Now we can store image data into imageFileRGBData matrix depend on the buffer reader size
imageBuffersize.
int[ ][ ] imageFileRGBData = null; is the Matrix to representing the image pixels
try {
int ndata[] = new int [nheight * nwidth];
this.imageFileRGBData = new int[nheight][nwidth]; // [line][column]
byte brgb[] = new byte [imageBufferSize];
689 Vol. 7, Issue 3, pp. 684-691

fs.read (brgb, 0, imageBufferSize); // Read the bitmap
int nindex = 0; // current postion in brgb
for (int j = 0; j < nheight; j++){ // by lines, from bottom to top
for (int i = 0; i < nwidth; i++){ // by columns
int rgbValue = (255 & 0xff) << 24
| (((int)brgb[nindex + 2] & 0xff) << 16)
| (((int)brgb[nindex + 1] & 0xff) << 8)
| (int)brgb[nindex] & 0xff;
ndata[nwidth * (nheight - j - 1) + i] = rgbValue;
this.imageFileRGBData[nheight - j - 1][i] = rgbValue;

nindex += 3;
}
nindex += npad;
}
}
catch (Exception ee) {
System.out.println(ee.getMessage());
}

Conversion step to both Image (the large image and sub image) will done as object of
ConvertImageToMatrix class.
3.2.2 SubImageSearch Class:
In this class the first step is to create the two object of image matrix like this:
ConvertImageToMatrix largeImage = new ConvertImageToMatrix("LargeImageNumbers.bmp");
ConvertImageToMatrix subImage = new ConvertImageToMatrix("SubImageN5.bmp");
int[][] large=largeImage.imageFileRGBData;//The Matrix of Large Image
int[][] sub=subImage.imageFileRGBData;//The Matrix of Sub Image
Now we have two matrix of integers, large matrix and sub matrix so we can now search for sub
matrix in large matrix.To start search for matching we must get the first pixel value from the sub
matrix as start point for search.
int pixelIndex = sub[0][0];//Search Start Point is the First Pixel(x,y) From Sub Image
3.2.2.1 SubImageSearch Four Level Looping:
We have to start our search loop by row lines scan, In each row we will scan all column with nested
loop to find matching point (x,y), And from the matching point we will search for any different
between sub and large depend on sub image size ,that mean we need new two nested loop looking for
different .
If we found the first different we have to leave the row looping index and start over with next index.
In this case we must to use Branching Statement ( Continue Label) [7] between row loop and column
so we can exit the internal fourth loop to the main loop.
If different not found we will print out the point of matching (x,y) and continue searching for another
matching point in case the sub image exists more than one time.
We used foundCount as matching Counter. And rowIndex to save matching Row index, and
columnIndex to save matching Column index , Both have the initial value -1 to be sure index is out of
large image.
public static int foundCount = 0;//counter for Matching if found
public static int rowIndex = -1;//-1 is out of image
public static int columnIndex = -1;//-1 is out of image
//-----------------Row lines scan
for ( int row = 0 ; row <= large.length - sub.length ; row++ ){
// Column Scan
continuescan://Column Scan Label
for ( int column = 0 ; column < large[0].length - sub[0].length; column++ ){
if (large[row][column] != pixelIndex) continue; // No Match With Start Point Go to Next
row
690 Vol. 7, Issue 3, pp. 684-691

// If there is matching With Start Point we have to Check All Pixel in sub with Pixel in large From
this Point
for ( int r2 = 0 ; r2 < sub.length ; r2++ )
for ( int c2 = 0 ; c2 < sub[0].length ; c2++ )
// Now we are looking for Any different Pixel
if ( large[row + r2][column + c2] != sub[r2][c2]){
continue continuescan;//there is no match Go to New Line (Next Y)
}//end if Any different
//-------Save The Matching Point Index
rowIndex = column;
columnIndex = row;
//--------Check if Index is positive and the Point is In Image
if ( rowIndex != -1 && columnIndex !=-1){

System.out.println(++foundCount+" : The Sub Image Match Found at (" + rowIndex + " , " +
columnIndex+")");
}//end if positive
}//end for column
}//end for row
//--------Check if Index is negative and the Point is Out Image
if( rowIndex == -1 || columnIndex ==-1){
System.out.println("No Match Found");

}//end if negative

3.2.2.2 The Test Result:

When we run the algorithm with the tow image the result will be :

Figure 5 :result of searching to the sub image
Now we can see the coordinate of matching point like this:

Figure 6 :Matching found in two Places
691 Vol. 7, Issue 3, pp. 684-691

IV. LIMITATIONS
Our algorithm can work only with 24-bit bitmap image format. But as we know there are many ways
to convert none 24-bit images to this format.
V. CONCLUSION
In this algorithm, we handled the problem of Searching for Sub Image and how to find out the
Padding and real Buffer size when we are reading bitmap image .
Our proposal algorithm Can detect the matching Indexes in the large image for sub image however it
occurs in the target large image .For the implementation of this algorithm we used the java language.

VI. Future Work
Our future work will be to create and development an image search engine which can search in image
contains instead of Text or HTML meta data, This search engine can be helpful to find the target
image from internet or any image storage, after that we will try to handle video frame search to
lookup for still image. The next future research will be about Low-Pass and High-Pass Image filters.
REFERENCES
[1] Dr. William T. Verts, "An Essay on Endian Order" , April 19, 1996
[2] Dr. David Kirkby (G8WRB) , atlc "bmp format", http://atlc.sourceforge.net/bmp.html
[3] Microsoft MSDN , "Bitmap Storage,
[4] Microsoft MSDN ,"Bitmap Header Types" ,
[5] John Miano, "Compressed Image File Formats: JPEG, PNG, GIF, XBM, BMP"
[6] (DUE 9816443) and the Hypermedia and Visualization Laboratory, "24-bit BMP FILES", Georgia State University, G.
Scott Owen
[7] The Java Tutorials, Branching Statements",
[8] Charlap, David, "The BMP File Format: Part I," Dr. Dobb's Journal, March 01, 1995
[ 9] John D. Mitchell , "How to read 8- and 24-bit Microsoft Windows bitmaps in Java ", JavaWorld ,Dec 1, 1997

AUTHORS BIOGRAPHY
O. K. Khorsheed born in Baghdad in 1974. BSC computer science from al-mustansiriya
University. Higher diploma degree from Informatics Institute of Higher Studies. Master
degree from Jordan (Arab academic for banking and financial science) school of (IT)
computer information system. Lecturer in Koya University since 2005 in school of
engineering S\W engineering department.

692 Vol. 7, Issue 3, pp. 692-700

POTENTIAL USE OF PHASE CHANGE MATERIALS WITH
REFERENCE TO THERMAL ENERGY SYSTEMS IN SOUTH
AFRICA
Basakayi J.K.
1
; Storm C.P.
2

1
Department of Mechanical Engineering, Vaal University of Technology,

2
School of Mechanical and Nuclear Engineering, North West University, Potchefstroom

ABSTRACT
The supply of fossil energy source such as coal is becoming less reliable. The CO2 has to be reduced. At the same
time energy consumption is increasing in South Africa.
One of the solutions is to make use of phase change materials in the thermal systems. Phase change materials
absorb and release thermal energy at a constant temperature. In addition, they have high density energy storage.
Integrating phase change materials in thermal systems can improve the efficiency and reliability of those systems.
Different potential applications of Phase change materials are provided in this paper that can be valuable in the
context of South Africa for storing energy and control temperature purpose.
KEYWORDS: Phase change material, thermal energy &latent heat storage.
I. INTRODUCTION
Lack of reliable and cost effective energy is one of the major problems facing social and economic
development of each country. South Africa has an increasing population. The excessive demand of
energy consumption keeps on increasing with the growing of the population. Industries and production
companies also require abundant and reliable energy to run their activities.
In South Africa, electricity is essentially generated from coal [1]. The disadvantages associated with
this traditional type of energy are numerous: the pollution of the atmosphere, depletion of the Ozone
and limited amount of coal.
To deal with the high demand of energy and the problems linked to the use of energy produced by fossil
fuel, the two possible ways are: investigate other forms of energy such as solar energy and well manage
the existing energy.
Since the crisis of energy during the period of 1970s, other types of energy have been considered.
Renewable energy is one type of energy that is constantly and regularly replenished and will never
come to an end anytime soon. Among the types of renewable energy investigated there are: Bio-mass,
wind, solar energy, geothermal.
Solar energy is one of the most promising types of renewable energy that offers a better alternative to
the energy generated by the fossil fuel. It is available, clean and abundant [2].
The excessive solar energy can be also stored when it is available for later use. It is about well managing
the existing energy by making use of some form of storage when sun is shining and more solar radiation
is accessible.
Different techniques used for storing thermal energy are: Sensible energy, latent heat energy and
thermo-chemical reaction. Among those three methods considered for storing thermal energy, the latent
heat energy emerges as the best option thanks to the advantages of releasing and absorbing energy at
constant temperature for the storage material used and also to the possibility of having high energy
693 Vol. 7, Issue 3, pp. 692-700

density storage compared to the sensible heat storage. Thermo-chemical storage system can provide
high density energy but it is still under development [3].
Latent heat storage systems make use of the material called Phase Change Material (PCM) due to the
change of phase that occurs during the discharge and the charge processes. In addition of storing and
releasing the latent thermal energy during the charging and discharging processes, PCM can be also
used for thermal control for different applications purposes.
Since solar energy is not available during night or cloudy day, the integration of a storage system in a
solar thermal energy as a latent heat can increase the reliability and the performance of thermal systems.
In applications for temperature control, the focus is on the temperature regulation and for storage of
heat or cold with high storage density; the emphasis is on the amount of heat supplied.
The applications found in literature for PCMs include solar hot water heating, solar cooking, solar
absorption, solar green house, thermal control of instruments, heating of building. Some applications
for thermal control are: spacecraft thermal control environment, integration of PCMs in delicate and
precise instrument thermal control.
Many investigations have been carried out elsewhere and successful applications are increasing in
different developed countries. No practical applications are found in the market here in South Africa
expect for few studies that are done in academic institutions for research purposes.
The immediate objective of this paper is to provide a basic understanding of the behaviour of PCM,
indicate how this material can be selected for a given application and finally provide different potential
applications of PCM with reference to thermal energy systems in South Africa, so that the integration
of the PCM can contribute in solving the problem of energy in this country.
This paper focuses on the potential applications of PCM in South Africa. The second section gives some
constructive ideas for using PCM. The basic concept of Latent heat storage systems is explain in the
third section. In the fourth section, the selection of PCM for different applications is presented. The
following section covers the potential applications of PCM here in South Africa. The last section
includes the future work which will be on the modelling of a latent heat storage.

II. WHY IS THE APPLICATION OF PCM RELEVANT IN SOUTH AFRICA?
More energy from sunlight strikes the Earth in one hour (4.3 10
20
J) than all the energy consumed on
the planet in a year (4.1 10
20
J) [1]. Most areas in South Africa average more than 2 500 hours of
sunshine per year, and average solar radiation levels range between 4.5 and 6.5 kWh/m
2
in one day [4].
South Africa has therefore one of the highest solar radiations in the world.
Today, many solar energy systems able to convert solar radiation directly into thermal energy have been
developed for low, medium and high temperature heating applications in different developed countries.
Solar thermal power generation and solar space heating are examples of industrial applications;
domestic applications include solar water heating and solar absorption refrigeration. The previously-
mentioned examples are only a sample of applications of solar energy to name but a few.
Although solar energy has its advantages, it remains, however, intermittent and unpredictable. Its total
available value is seasonal and often dependent on meteorological conditions of a location. Therefore,
solar energy cannot, for example, be trusted to produce cooling during periods of low solar energy
irradiation for a solar absorption machine. Some form of thermal energy storage is necessary for
effective utilisation of such an energy source, to meet demand on cloudy days and at night. Thermal
energy storage, therefore, plays an important role in conserving thermal energy, leading to an
improvement in the performance and reliability of a range of energy systems.
Thermal energy storage refers to a number of technologies which store energy in a thermal accumulator
for later re-use. Integrating thermal energy storage in a thermal system has a potential to increase
effective use of thermal energy equipment and it can facilitate large scale switching [5]. This energy
storage is used for correcting the mismatch between the supply and the demand of energy.
The use of PCM as an energy storage medium is now worldwide considered as an option with a number
of advantages. The combination of solar energy and the use of PCMs in any thermal energy system
may result on alleviating the problem of pollution due to the use of fossil sources energy, assisting the
management of energy with the high demand of energy increasing by storing when it is available and
694 Vol. 7, Issue 3, pp. 692-700

use later or during cloudy and low solar radiation. It is also important to make use of PCM wherever
there is a need to control the temperature of space, equipment or human body.
III. LATENT HEAT STORAGE SYSTEMS
Latent Heat Thermal Energy Storage (LHS) is based on the heat absorption or release when a storage
material undergoes a phase change from solid to liquid, solid to solid or liquid to gas or vice-versa.
Figure 1 summarises the classification of LHS materials.
Fig.1 Classification of the LHS materials (modified after Shama et al., 2004)
Latent heat storage can be accomplished through solid-liquid, liquid-gas, solid-gas, and solid-solid
phase transformations. Only two are of practical interest, solid-liquid and solid-solid [6]. Solid-gas and
liquid-gas transition have a higher latent heat of fusion, but their large volume changes on phase
transition are associated with containment problems and rule out their potential utility in thermal storage
systems. Solid-liquid PCMs present the advantages of smaller volume change during the phase change
process and longer lifespan [6]:
The storage capacity of a LHS system with a PCM is given by:
)] T (T C ) T (T C H m[a Q
m f lp i m sp m
- -
Where Q = quantity of heat stored;
m = mass of the PCM;
am = fraction melted;
H = latent heat of fusion;
Csp = average specific heat between temperatures Ti and Tm;
Clp = average specific heat between temperatures Tm and Tf;
Ti = Initial temperature;
Tf = Final temperature;
Tm = Melting point

Latent heat storage materials
Organic
Paraffins
Non-
Paraffins
Inorganic
Salt
Hydrates
Metallic
Eutectics
Inorganic -
Inorganic
Inorganic-
Inorganic
Organic-
Organic
Miscellaneous
695 Vol. 7, Issue 3, pp. 692-700

T

SOLID MELTING LIQUID

Fig.2 Thermal energy stored in a PCM as a function of temperature (modified after Rubitherm, 2008)
Figure 2 illustrates the change in stored energy (internal energy) as a function of temperature. At the
beginning of the heating process the material is in a solid state. Before it reaches the melting point Tm,
the heat absorbed is sensible heat. Starting at melting point, the material undergoes a change of state
from a solid to a liquid. During this process, the material absorbs heat known as enthalpy melting. The
temperature remains constant. If the material is heated up further after this process, there will be
sensible heat added on completion of melting.
IV. SELECTION CRITERIA OF A PCM
A PCM is selected depending on the application. The following criteria are considered for the selection
of a PCM [7]:
1. The thermo-physical properties:- Melting temperature in the desired operating temperature range, -
High latent heat of fusion per unit volume, - High specific heat and high thermal conductivity (the low
thermal conductivity is one of major problems in use of PCM), -Small volume changes on phase
transformation and congruent melting for a constant storage capacity.
2. Kinetic properties: Among kinetic properties to be considered are: High nucleation rate and high rate
of crystal growth.
3. Chemical properties: Chemical stability, complete reversible freeze/melt cycle, no-degradation after
a larger number of freeze/melt cycles, non-corrosiveness to the construction materials, non-toxic, non-
flammable and non-explosive for safety.
4. Economic: PCMs should be cost effective, abundant and be available on a large scale.
There are different ranges of PCMs. Table 1 presents some commercial PCMs available on the market.
Table 1: Some commercial PCMs available on the market (Zalba et al., 2003)
Commercial Name Type of product Melting point (
o
C) Heat of fusion (kJ/kg) Source (Producer)
RT 20 Paraffin 22 125 Rubitherm
SP 22 A4 Eutectic 22 165 Rubitherm
SP 25 A8 Eutectic 25 180 Rubitherm
ClimSEL C22 Salt hydrate 22 144 Climator
ClimSEL C28 Salt hydrate 28 126 Climator
In Table 2, the advantages and disadvantages of organic and inorganic materials are presented. From
the behaviour of the PCM shown in Table 2, an appropriate PCM can be selected for a given application.
Table 2: Comparisons of organic and inorganic PCMs (Zalba et al., 2003)
ORGANIC INORGANIC
ADVANTAGES
non-corrosive great change enthalpy

Q
Tm
696 Vol. 7, Issue 3, pp. 692-700

non-toxic good thermal conductivity
little or no supercooling cheap and non-flammable
chemically and thermally stable
DISADVANTAGES
lower phase change enthalpy undercooling
low thermal conductivity corrosion to most metals
phase separation
phase segregation
lack of thermal stability
V. APPLICATIONS OF PHASE CHANGE MATERIALS
In this section, some potential applications of PCM that can be useful in South Africa to address the
issue of energy are presented.
5.1. Solar water heating and Space heating
Water heating accounts for up to 40% of household energy consumption in South Africa [8]. Up to now,
solar water heating system use in South Africa makes use mostly of sensible heat technology in their
systems. It has been proved that the integration of PCM in a solar water heating system can improve
the efficiency of the system. By decreasing the volume of geyser and by providing thermal energy stored
at all most constant temperature for a long period compared to sensible heating system. 5 - 14 times
more energy can be stored by using latent heat than the sensible heat [9].
Using the latent heat storage with PCM like in hospital, restaurants and residential area will
tremendously improve the life of population with an efficient energy system [10].
In a latent heat storage solar water heating systems, water is heated up during the sunshine period. The
heat is transferred to the PCM. During cloudy or period of low solar radiation, the hot water is taken
out and is replaced by cold water, which absorbs energy from the PCM. The energy is released by the
PCM during the solidification process (changing from liquid to solid).
A study done by Mehling et al. [11] showed that a PCM module at the top of a stratified water tank
increased the energy storage and improved the performance of the tank. Paraffin wax, stearic acid, Na2
SO4. 10 H2O have been tested and good results were obtained for solar heating systems [12]. Hasan et
al. [13] recommended that myristic acid, palmitic acid and stearic, with melting temperature between
50
o
C 70
o
C are the promising PCMs for water heating.
In South Africa, space heating is one of the largest energy loads on a typical house. Using a LTE system
can reduce the amount of energy consumed for these loads by storing excess thermal energy that is
either available during the day (diurnal storage) or available in the summer (seasonal storage).
Solar air heater integrated with PCM can be used for heating space. Strich et al. [14] used transparent
insulation material and translucent PCM in the wall for heating the air for the ventilation of the house.
Paraffin wax was used as PCM. The efficiency of solar energy absorbed into the PCM and transferred
to the ventilation air was 45% on average.
5.2 Solar cookers
The use of solar stoves can have great significance in South Africa if they are used in combination with
PCM. The utilization of PCM in the solar cooker would assist to conserve the environment by not using
fuel wood for cooking as it is currently used. According to CSIR, rural households in South Africa use
between 4.5 and 6.7 million tonnes of fuelwood per year [15].
Some box-type solar cooker with PCM has been tested and proved to be efficient to cook food even
during the evening (Buddhi and Sahoo [16] and Sharma et al. [17]). Commercial grade stearic acid,
acetamide, erythritol are some of PCMs that can be used for this application.
5.3 Solar cooling
Often during summertime, there is a need to cool the buildings and houses. Cooling TES systems can
be used to reduce the amount of energy consumed for space cooling. PCMs are used as a storage media
for space cooling. As examples of PCMs used for space cooling, there are: inorganic salt hydrates,
organic paraffin waxes and mixture of these, Rubitherm RT 5.
697 Vol. 7, Issue 3, pp. 692-700

Typically building services that include lighting, electronic devices and appliances account for 57% of
the electricity consumption and thermal comfort systems (Heating, Ventilation and Air Conditioning -
HVAC) make up the remaining 43% [18].
Agyenim et al. [19] investigated the possibility of integrating latent thermal energy storage to the hot
side of a LiBr absorption cooling system to cover 100% of the peak cooling load for a three bedroom
house on the hottest summer day in Cardiff, Wales. A 100 l of Erythritol was required to provide about
4.4 hours of cooling at peak loads based on the optimum Coefficient Of Performance (COP) of 0.7
PCMs can be used to store coolness. In this application, the cold stored from ambient air during night
is relieved to the indoor ambient during the hottest hours of the day [7].
5.4 Intermittent Power Plants
Integrating PCM in a latent heat storage in a power plant for example to smooth out the supply of
electricity from intermittent power plants for the utility grid can be beneficial. This can result on
generating electricity more effectively during the distribution and to take control and manage the energy
more adequately. Electricity can be produced by solar energy that was stored several days earlier by
storing excess energy in large thermal reservoirs. Nitrate salt represents possible PCM for applications
beyond 100 C [19].
In his investigation, Hunold et al. [20] proved that phase change storage is technical feasible and
proposed a storage design for a power plant. South Africa generates electricity from different power
plants. By incorporating PCM in the power plant, it is possible to improve the distribution of electricity
in South Africa.
5.5 Waste heat recovery systems
Air conditioning system ejects heat. The exhausting temperature of the compressor is relatively high
when using Freon as refrigerant. Therefore, it can be recovered using an accumulator and gets heat of
higher temperature.
Zu et al. [21] developed a heat recovery system using PCM to recover the ejected heat of air
conditioning system and producing low temperature hot water for washing and bathing. It was observed
that the efficiency ratio of the system is improved effectively when all the rejected heat from air
conditioning systems is recovered. Erythritol, stearic acid have been used and tested as PCMs for this
application.
5.6 Greenhouse heating
Greenhouse requires control of temperature, humidity, solar irradiance and internal gas composition
with rational consumption of energy.
Some PCMs were used for this application: CaCl2.6H2O, Na2SO4.10H2O, paraffins. Nishina and
Takakura [22] compared conventional greenhouse with PCM storage type storage. Efficiency of the
greenhouse with PCM storage integrated with solar collector was 59% and able to maintain 8
o
C inside
the greenhouse at night.
5.7 Electrical Power Devices
PCMs with high melt temperatures can be used in conjunction with electronic power-producing
systems. Radiators used to collect solar energy can be packed with PCM to store the energy via phase
change at the melt temperature.
This stored energy can then be converted into electrical power by using the large temperature difference
between the radiator and deep space in either thermionic or thermoelectric devices.
Preliminary analytical and experimental studies reported in [23] indicate the feasibility of PCM
application, and materials have been found with suitable properties for such PCM systems.

To maintain high photovoltaic (PV) efficiency and increasing operating PV life by maintaining them at
a lower temperature, PCMs are integrated into PV panels to absorb excess heat by latent heat absorption
mechanism and regulate PV temperature. The results show that such systems are financially viable in
higher temperature and high solar radiation environment [24].

698 Vol. 7, Issue 3, pp. 692-700

5.8 Transport
Containers with PCMs can be used to transport film, food, waste products, biological samples, etc.
Such containers represent isothermal protection systems for perishable cargo.
Applications in transport and storage containers are in most cases basically an insulated environment
with heat capacity increased by PCM.
When food or medications are transported, they must be kept in a certain range of temperature. PCM
are very suitable in transportation of these products (food and medications).
Companies such as Rubitherm GmbH, Sofrigam, PCM thermal Solutions design transport boxes for
sensitive materials [7].
5.9 Delicate instrument thermal control
For delicate, highly temperature-sensitive instruments, PCM can be used to maintain these instruments
within extremely small temperature ranges.
Precise Thermal Control of Instruments Temperature - sensitive instruments required to deliver highly
accurate responses have been protected by PCM thermal control systems. Russian investigators have
studied the feasibility of using PCM to precisely control the temperature of gravity meters which require
a relative accuracy of 10-8[24].
5. 10.Applications for the human body
The common approach to use PCM to control and stabilize the temperature of the human body is to
integrate the PCM into clothes. Approaches are: macro encapsulation, microencapsulation, composite
materials.
Pocket heater (used to release heat when a person is freezing), Vests for different applications,
developed by Climator [25]. Those vests are developed to cool the body of people who work in hot
environments (such as mine), or with extreme physical exercise. Clothes and underwear are used to
reduce sweating. Other applications include: Kidney belt, plumeaus and sleeping bags and shoe inlets.
Another important application can be found in medical field: the use of ice for cooling as a treatment
of different sports injuries.
In Textile, PCMS make it possible to engineer fabrics that help regulate human body temperature.
Microencapsulation makes phase change materials light and portable enough to incorporate into
textiles, either as a top-coat or as an addition to the fabric fibers [26].
VI. CONCLUSION AND FUTURE WORK
Make use of the PCM can be useful and relevant in South Africa if this material is used in conjunction
with solar energy and waste thermal energy system. It can significantly contribute to solve some
problems related to energy. Several problems may be resolved such as pollution of environment, load
shading, management of electrical power, deforestation and energy consumption.
There is a necessity to turn to the PCM with some of applications point out in this paper; life of millions
of South Africans can be alleviated. To attain this result, there is a need to do more research in the field
of energy, to inform people about other possible solutions to the problem of energy and apply those
solution such as the integration of PCM in different thermal systems.
In the next paper, the focus will be on modelling a latent heat storage system for solar refrigeration
system. A mathematical model will be proposed for a latent shell and tube heat exchanger in order to
understand the mechanics of a latent heat storage.
REFERENCE
[1] Energy department: Energy n.d., Republic of South Africa Electricity basic, viewed 14 May 2014, from
http://www.energy.gov.za/files/electricity_frame.html.
[2] Basic research needs for solar energy utilization n.d., Report on the Basic Energy Sciences Workshop on
Solar Energy Utilization, viewed 15 May 2014 from http:// www.sc
.doe.gov/bes/reports/files/SEU_rpt.pdf
[3] Stine, W.B. & Harrigan, R.W., (1986) Solar energy systems. John Wiley and sons Inc.
699 Vol. 7, Issue 3, pp. 692-700

[4] Naidoo, N., 2013. Light the way for solar, viewed 14 January 2014 http://www.cncafrica.com/special-
report/2013/09/12/africa-well-positioned-to-light-the way-for solar/
[5] Dincer, I. & Rosen, M.A., (2002) Thermal energy storage. Systems and applications. John Willey &
Sons,Ltd, Baffins Lane, Chichester, West Sussex PO 19 1 UD, England.
[6] Sharma, S., & Kazunobu, S., Latent heat storage materials and Systems: a review, International Journal
of Green Energy 2, 1-56, 2005
[7] Zalba, B. Marin, J.M., Cabeza, L.F & Mehling, H., Review on thermal energy storage with phase change:
materials, heat transfer analysis and applications, Applied Thermal Engineering 23, 251-283, 2002
[8] Its going to be a bright, sunshiny day. Solar solutions n.d, viewed 18 May 2014, from
http:www.nedbank.co.za/website/content/nedbank_solar/Solar-Solutions.aspx
[9] Sharma A., Tyagi, V.V & Buddhi D., Review on thermal energy storage with phase change materials
and applications, Renewable and sustainable Energy Reviews 13, 318-345,2009.
[10] Basakayi, J.K, (2012) Modelling and design of a latent heat storage system with reference to solar
refrigeration. Master thesis. University of
Johannesburg.https://ujdigispace.uj.za/handle/10210/7883/Kantole.pdf
[11] Mehling, H., Cabeza, L.F, Hippeli , S. & Hiebler, J.,, Improvement of stratified hot water heat stores
using a PCM-module, Proceedings of EuroSun 2002, Bologna (Italy), 2002.
[12] Gowtham, M., Chander, M.S, Mallikarujanan, S.S, & Karthikeyan, N.,Concentrated Parabolic Solar
Distiller with latent heat storage capacity, International Journal of Chemical Engineering and
Applications, 2(3),2011.
[13] Hasan A., & Sayig A.A., Some fatty acids as Phase Change Thermal Energy Storage Materials,
Renewable Energy 4 69-76. Elsevier Science Ltd, 1994.
[14] Strith, U., Novak, P.,Thermal storage of solar energy in the wall for building ventilation. IEA, ECES IA
Annex 17. Advanced Thermal Energy storage Techniques Feasibility Studies and Demonstration
Projects 2
nd
Workshop. Ljubljana, Slovenia, 2002.
[15] Sustaining the wood for the trees, viewed 15 May 2014 from
http://www.csir.co.za/enews/2013_mar/07.html
[16] Buddhi, D., Sahoo, L.K.,Solar cooker with latent heat storage design and experimental testing, Energy
Convers. Manage, 38(5):493-501, 1997.
[17] Sharma, S.D., Iwata, T., Kitano, H. & Sagara, K.,Thermal performance of a solar cooker based on an
evacuated tube solar collector with a PCM storage unit. Solar energy 78, 416-426, 2005.
[18] Agyenim , F.B., n.d, Solar air conditioning and the need for energy storage for meeting domestic
cooling, viewed 15 April 2014, from www.freebs.com/agyenim-aboffourwebsite.
[19] Doerte, L., (2013) Thermal Energy Storage for Concentrated Solar Thermal Power Plants. Seminars in
engineering science. Lehigh University, viewed 16 May 2014, from
http://elib.dlr.de/86888/1/Laing_Lehigh-University _13-09-2013.pdf
[20] Hunold, D., (2014) Parabolic trough solar power plants- The largest thermal oil plants in the world,
viewed 16 May 2014,from http:// www.heat11.com/fileadmin/uploads/Whitepaper/heat11-CSP_HTF-
Plants_whitepaper.pdf?PHPSESSID=cd921472a801413eOaO253a14b513545
[21] Zhu,N., & Wang,S., Dynamic characteristics and energy performance of buildings using phase change
materials: A review , Energy Conversion and Management, 50 (12) 3169-3181,(2009)
[22] Nishina, H. & Takakura. T. Greenhouse heating by means of latent heat storage
units. Acta Hortic.148:751-754,1984
[23] Kiziroglou, M.E.,n.d, Design and fabrication of heat storage thermoelectric harvesting Devices, viewed,
18 May 2014, from
https://spiral.imperial.ac.uk/bitstream/10044/1/11116/6/13_IEEE_TIE_FinalSubmission_v4%20
(EPSfigures).pdf
[24] Hasan, A., McCormack, S., J, Huang, M.J., Norton, B., Energy and cost saving of a photovoltaic Phase
change materials (PV-PCM) system through temperature regulation and performance enhancement of
Photovoltaics, Energies 7, 1318-1331,2014
[25] Hale, D.V, Hoover, M.J & ONeill, M.J., (1971) Phase change materials handbook, Alabama: NASA-
CR-61313.Lockeed missiles and Space Company.
[26] Mehling, H. & Cabeza, L.F., (2008) Heat and Cold storage with PCM. Berlin- Heldbergy: Springer.
[27] Phase Change Materials in Textiles. StudyMode.com. Retrieved 04, 2012, from
http://www.studymode.com/essays/Phase-Change-Materials-In-Textiles-958775.htm.

700 Vol. 7, Issue 3, pp. 692-700

AUTHORS
Basakayi, J.K graduated with a BEng degree from the Universite de Mbujimayi (U.M), The
Democratic Republic of Congo in 2001. He then completed a Master of Engineering degree
in 2012 from the University of Johannesburg in South Africa.
In 2007, Basakayi has worked at the University of Johannesburg as a Part Time Lecturer in
Mechanical Engineering before joining the Cape Peninsula University of Technology from
2008 until 2013. He is a currently lecturer at Vaal University of Technology (VUT) in
mechanical engineering Department. He is a member of South African Institute of
Mechanical Engineering (SaiMech).

Storm C.P. received the B.Eng degree from the University of Pretoria and M. Eng degree
from the North West University, South Africa, in 1982 and 1994 respectively, and the Phd.
Degree from the North-West University in 1998. He was Director of the School of
Mechanical and nuclear engineering from 2007 until 2013 at the
North West University. He is currently Professor of Engineering. Since 1997, he has been
actively involved as a researcher and lecturer in the area of thermodynamics and
Optimization of thermal systems. He did publish a number of papers in the field of
thermodynamics and power plant.
Prof Storm worked at Eskom from 1982 until 1999 as assistant Engineer, manager and Senior Engineer
respectively.

701 Vol. 7, Issue 3, pp. 701-711

IMPROVING SOFTWARE QUALITY IN THE SERVICE PROCESS
INDUSTRY USING AGILITY WITH SOFTWARE REUSABLE
COMPONENTS AS SOFTWARE PRODUCT LINE: AN
EMPIRICAL STUDY OF INDIAN SERVICE PROVIDERS
Charles Ikerionwu, Richard Foley, Edwin Gray
School of Engineering and Built Environment, Glasgow Caledonian University,
Cowcaddens Road, Glasgow, G4 0BA, Scotland, UK

ABSTRACT
In a software based business process outsourcing (BPO) environment, the quality of services delivered to clients
hinges on the quality of software used in processing the service. Software quality attributes have been defined by
ISO/IEC standards but different organisations give priorities to specific attributes based on clients requirements
and the prevailing environment. The aim of this study is to identify and demonstrate the particular software
development process that guarantees an acceptable level of software quality within a specific domain that would
translate to desired quality of services delivered to clients. Therefore, this study through a mixed method approach
investigated BPO service providers in India to ascertain what software quality means to their respective
organisations, what software quality attributes are given priority and how it could be improved. The findings
suggest that software quality is highly dependent on the software development process. The vast majority of
successful organisations operated in-house software development through the establishment of software product
line as a platform to embed software reusable components within an agile framework. Through this process, there
is significant reduction in defect density, thereby improving the software quality. This software quality is also
translated to the quality of services delivered to clients
KEYWORDS: software quality, reusable components, agile framework, software product line engineering,
service provider.
I. INTRODUCTION
Over the years, we have witnessed organisations outsourcing business processes, knowledge processes,
and most recently business process management outsourcing. Given the increasing demand for these
services, and the increased reliance on IT platforms for its processing and delivery; service providers
need to respond by providing quality services at a reduced cost [1]. In this way, clients competitiveness
is increased through good quality of services provided by service providers. However, [2] demonstrated
through their research findings that quality services that guarantee clients satisfaction must evolve from
the quality of the application software used in processing clients services.
In recent times there have been reported incidents of clients bearing the brunt of poor services from
service providers. Such poor services have been traced to unreliable application software [3]. For
example, Royal Bank of Scotland outsources its IT and business processes to Indian service providers
and recently experienced a setback when their customers across the UK could not avail many services
due to poor services emanating from unreliable application software used by their service provider in
processing services [4]. According to [2], significant number of service providers in India integrated IT
into their business process organisation in order to be competitive by providing better services through
702 Vol. 7, Issue 3, pp. 701-711

in-house development of application software with an acceptable level of software quality. This entails
that the service provider develops the application software it requires to process clients services.
However, there have been no indicator within a specific domain like the business process service
providers what software quality means or to ascertain the level of reliability of application software
required to process clients services. An earlier study by [5], demonstrated that there is little evidence
that conformance to process standards guarantee good software product quality. [6] extended the study
by empirically studying the life cycle productivity and conformance quality in software products but
concluded by including all the software quality attributes as defined by the ISO standards to be
applicable to every domain. The key problem with this finding is that generalising software quality to
every domain could not be feasible because each organisation indicates what their expectations from
the software product are, based upon specific requirements and prevailing environment. Therefore, this
study would specifically limit its scope to BPO services providers that provide services to clients within
the financial domain. The study comprises of the following specific research questions:
Research question (RQ) 1: Which development paradigm guarantees an acceptable level of software
quality? Primarily this question identifies and demonstrates the development paradigm applied by the
dedicated software development team within the BPO service provider that ensures quality of
application software.
Research question (RQ) 2: What does software quality mean to a service provider? The motivation for
this question is to identify how organisations prioritise specific software quality attribute(s) that suits
their organisations requirements.
Research question (RQ) 3: Does the organisational model play any role in determining the quality of
the application software? In answering this question, the researchers could ascertain the effect of
integrating dedicated software development team into a service provider on the quality of services
provided to clients.
II. SOFTWARE QUALITY
To address the issues of software product quality, the International Standard Organisation and
International Electrotechnical Commission recently published a set of software product quality
standards known as ISO/IEC 25010. According to this standard, software product quality is the degree
to which a set of static attributes of a software product satisfies stated and implied needs when the
software product is used under specified conditions [7]. The definition went further to include eight
characteristics of product quality model and specific five key characteristics of the model of software
quality. According to this ISO/IEC release, the five key characteristics of a software quality includes
effectiveness, efficiency, satisfaction, safety, and usability. The definition went further to say; usability
is the extent to which a product can be used by specific users to achieve specified goals with
effectiveness, efficiency and satisfaction in a specified context of use. However, mere building of
standards alone is not sufficient to guarantee software quality because what is considered quality
attribute in one domain might differ from another domain. This suggests that, for a specific domain, the
software development team members would apply a development process that would ensure the
realisation of the organisations software quality perception.
[8] is of the view that, software quality is subjective; i.e., the quality of every software is based on its
non-functional characteristics This implies that software quality is not only measured by the
implementation of the functional attributes but also the non-functional system attributes. Among the 15
software quality attributes presented by [9], [8] contends that it would be very difficult for every
software to attain optimal level in all the attributes. For this reason, [8] suggested that the management
team of the software development team should at the earliest opportunity identify which attributes best
answers to clients requirements. For example, the service providers that develop application software
required for service processing would recommend attributes like dependability, reliability, usability,
efficiency and maintainability to the development team to be in-built into the software. The
identification and inclusion of these attributes ensure that the service providers perception of software
quality are built into the software and the desired level of service quality is delivered to clients.
Every software system must have evolved from a development process. According to [2], the
development process of an application software has a significant role on the quality of the ensuing
703 Vol. 7, Issue 3, pp. 701-711

software. [10] introduced the agile software development process with the release of its manifesto that
paid more attention to increasing software quality. Since then, many research findings [11], [12], [13]
have suggested the use of agile software development method for an optimal software quality. The most
significant attribute of agile method that ensures an acceptable level of software quality is its test driven
approach [14]. It is also noteworthy to mention that service providers with several number of clients
would have variety of client requirements which are subject to change even when the requirement is
been processed. Agile method provides agility which ensures that the issues arising from variability of
clients requirements are significantly contained. Because of these reasons, many organisations have
adopted agile development approach.
III. RESEARCH METHOD
In order to understand the underlying attributes behind the development of application software with an
acceptable level of quality that would guarantee fulfillment of clients requirements within the financial
domain, we adopted a study of Indian business process service providers through an exploratory
sequential mixed method approach. Through this approach, qualitative data collection was first done
through a case study of seven medium large scale appropriate service providers representing service
providers in India. This was followed up by a quantitative data collection (which was informed by the
results of the case studies) through a larger scale survey of a significant number of Indian providers.
Mixed method research has become the most popular term for mixing qualitative and quantitative data
in a single study [15]. It has been successfully applied in information system domain [16], and software
engineering [17], [18]. In the words of [19], mixed method research is the type of research in which
qualitative and quantitative research approaches are used (e.g. use of qualitative and quantitative
viewpoints, data collection, analysis, inference techniques) for the broad purpose of breath and depth
of understanding and corroboration. [20b] went further to suggest that the aim of mixing methods is
not necessarily for corroboration but rather to expand ones understanding of the phenomenon.
According to [21], the use of results from a qualitative study to inform a survey will enhance the
sensitivity and accuracy of survey questions. Then, as suggested by [22], the findings of these two
phases are to be integrated to provide the final analysis and conclusion. For these very reasons, in this
study, the selection of mixed method has legitimised the use of multiple approaches in achieving the
research objectives rather than restricting or constraining researchers choices.
In this study, at the qualitative level, interviews were tape recorded, manually transcribed by the
researchers, and imported to qualitative data analysis software Nvivo 10.0 for coding and subsequent
analysis. [23] suggests that in exploratory analysis, the researchers primary concern is on what emerges
from the interaction between the researcher and the respondent. Subsequently, the content of this
interaction drives the development of codes and identification of themes. Patterns which were identified
from the emanating themes and sub themes from the coding were later subjected to thematic analysis
in order to locate meaning in the data. Thematic analysis is a method for identifying, analysing and
reporting themes and patterns within data [24]. In the process, the researcher is examining commonality,
examining differences and relationships [25]. One of the main reasons for applying thematic analysis
is that it does not attach to any predetermined theoretical framework. It is flexible, organises data to the
possible minimal form and reveals the richness of data.
3.1 Case study
The researchers identified 7 business process service providers in India through purposeful sampling.
The sampling was based on service provider organisations with onshore and offshore clients with a
minimum of 5 years in operation. The data collection was carried out through face-to-face semi-
structured interview across senior software development team members at their respective offices in
India. Most of these service provider organisations were identified to be third party organisation (owned
by Indian companies) whilst only one is a captive service provider organisations (owned by a large
multinational IT company based in the Silicon Valley, USA). As part of the research study, participants
were asked questions relating to areas such as: the nature of the application software required to process
clients services; software development paradigm used by the development team; what software quality
means to their organisation; how software quality is maintained at the development level; and its effect
on the services delivered to clients. After the interview transcription, it was imported to Nvivo software
704 Vol. 7, Issue 3, pp. 701-711

to assist in reducing the text to themes and subthemes. To identify the fine-grained answers to questions
raised during the interview, the researchers applied thematic analysis. Most significant number of
service providers provide such services like analytics, financial services, backend, insurance, animation
and e-commerce to their respective clients.
3.2 Questionnaire
The outcomes of the thematic analysis from the case study formed the basis for the development of the
subsequent online questionnaire. In all, the questionnaire study had 27 Likert scale questions which
specifically dwelt on the software development process and software quality attributes. This aimed to
reach a larger population of service providers and validate the main thematic issues identified in the
case study. Through NASSCOM and BPO Association of India, the questionnaire were distributed to
358 potential participants. As part of the study, significant number of the questions centered on the
quality of the application software required for processing clients services. The questionnaire was
aimed at high level individuals within the IT management, application software development team
members and service process team leaders. Specific questions relating to issues about software quality
identified in the thematic analysis were asked. Such question includes service providers perception of
software quality attributes with regard to their organisation and software development paradigm that
guarantees software quality. A total of 156 valid responses were received. The data collected through
the questionnaire was subjected to statistical and descriptive analysis and the outcome was combined
with the results from the thematic analysis to arrive at the research findings.
IV. RESEARCH ANALYSIS
4.1 Case study
Dedicated software development team members: Although the research participants are service
providers, all but one of them indicated that their organisations develop the application software
required for service processing. Thus, the lone service provider organisation either obtain application
software from their client, outsource to an IT company or from a vendor. The reason for developing the
software in-house could be summed up when an interviewee said as follows: Our primary objective is
to provide quality service to our numerous clients and we can only achieve this through the quality of
the application software we use for service processing. And the best way to achieve an optimum level
of software quality is to develop it in-house. We can only trust our product. For every client, our
reputation is at stake and we cannot compromise. The study indicates that both medium and large
scale service providers have dedicated software development teams who are responsible for the
development of specific application software required for processing of clients services. These
individuals have, over the years acquired significant experience within the domain and could also add
to the clients participation in the software development process.
Software development paradigm: In table 1 column 2 below, respondents indicated the development
paradigm they have applied in the development of the application software. With the exception of BB4
organisation that does not develop software, others indicated that their development team have used a
combination of paradigms. It is evidenced that software product line engineering (SPLE) is a
composition of reusable components and software development life cycle [26] and in this study, across
the cases, respondents indicated the use of reusable software components either contained in SPLE or
explicitly applied. Four of the research participants (BB1, BB2, BB3, and BB5) indicated the
combination of SPLE and Agile with reuse while service provider BB7 combined Agile with reuse
components as development paradigm.
Through the thematic analysis presented in table 1 column 3 below, respondents were of the view that
the combined paradigms provided the development team with appropriate measures that guaranteed
software quality. In order to ensure an acceptable level of software quality, the development teams in
their respective service provider organisations adopted software quality measures that include; the test
driven approach provided by agile framework; reusable components that ensure the use of tested and
trusted components; and the overall rigorous testing within the software product line.

705 Vol. 7, Issue 3, pp. 701-711

Table 1: Software development paradigm and adopted quality measures

Software quality perception: Table 2 below summarises the software quality key attributes identified
by their respective organisations. Respondents indicated what their respective perception of software
quality means with specific reference to their organisations. As evidenced in table 2 below, respondents
indicated their respective software quality perceptions to include; the software must be absolutely
reliable by delivering the clients requirements; must satisfy the core attribute of usability; must deliver
services within a timescale; must be secured and safe; and must guarantee an expected level of service
quality provided to clients. Although the service provider BB4 does not development software but it
sources software that would provide optimal quality required to providing quality of service to clients.
This means that they would rely on outsiders for successful implementation of clients requirements.
Table 2: Software quality perception
Service
provider
Software quality perception
BB1 Absolutely reliable software that would deliver client's
requirements
BB2 Usability is a core factor within our service process team.
It must deliver within the shortest possible time.
BB3 Must guarantee quality services to our numerous clients.
Provide answers to clients requirements
BB4 Reliable and able to reduce service processing and
delivery turnaround time
BB5 Must provide security to clients data and safety of users.
The software must be reliable
BB6 Clients satisfaction and absolute reliable
BB7 The software must be reliableand could be easily used by
our processing team members
4.2 Questionnaire
The aim of this questionnaire is twofold: first, to validate the key thematic findings in the case study,
and secondly, to reach a larger number of service providers for a broader study. The questionnaire
returned 156 valid responses, and which represents 44% response rate. This value is consistent with
[27] assertion that the average response rate required for studies at organisational level is 37.2%. In
relative with the key attributes identified in the case study, the following analyses are presented below.
Dedicated software development team members: There are 156 medium and large-scale service
providers that participated in the questionnaire study with a combined 9856 members within their
respective software development teams. These members are specifically integrated into a service
provider organization for the very purpose of developing required application software to process client
services. Their organisations have a total of 12890 clients. This is also in line with the case study where
medium and large-scale service organisations have large number of development team members and
large number of clients.
Service
provider
Software development paradigm Adopted Software quality
measures
BB1 SPLE,Agile Test driven process and SPLE
quality measures
BB2 Our SPLE provides the required platform
for application development through
reusable components in Agile framework
Rigorous testing, reused tested
components
BB3 SPLE, Agile and reusable software
components
Reused components, SPLE
quality features
BB4 NA NA
BB5 Agile, SPLE, reusable software
components
SPLE quality features, test driven
of Agile, tested components
BB6 SPLE SPLE quality features
BB7 Agile + Reusable software components test driven, reused components
706 Vol. 7, Issue 3, pp. 701-711

Software development paradigm: Figure 1 below presents the software development paradigms service
providers development team use in the application software development. In the question, respondents
were specifically asked to select a paradigm(s) that guarantees software quality during the development
process. A total of 76 organisations indicated that their organisations have established a software
product line where they always embed software reusable components into an agile framework. Through
this process, at different times, the development team develops reliable software that provides quality
services to their numerous customers. To the same question, 57 research participants indicated that they
only embed reusable software components into an agile framework to develop quality software.

The number of service providers that indicated the use of SPLE without a fixed development paradigm
is 21 whilst 2 said that they use reusable software components only. Significant majority of the service
providers indicate an initial establishment of a software product line that embedded reusable software
components into an agile framework. This is a key characteristic of software development by service
providers that develop the application software in-house and it is a validation of the same attribute found
in the case study.
Software quality attributes: In order to establish each organisations perception of software quality,
respondents were asked to indicate by selecting the appropriate software quality attribute that defines
the quality of the application software their respective organisations use in service processing. In figure
2 below, respondents indicate how they view each software quality attribute with regards to service
processing within their organisation. Attributes like usability, reliability and maintainability are given
the highest priority (132,136,126) at the development level. This implies that the presence of these
attributes in the software developed in-house provides a guarantee that services processed would also
be of good quality.

Figure 1: Software development paradigm
Figure 2: Service providers perception software quality attributes
707 Vol. 7, Issue 3, pp. 701-711

According to ISO/IEC 25010, safety is an attribute of software quality. However, in figure 2 above,
vast majority of respondents were of the view that safety is not or is the least considered software quality
attribute in their respective organisations. This implies that different domains do have different software
quality attribute depending on the prevailing environment and specifically, clients requirements. This
is an indication of validating the varied software quality attributes, which were indicated by
interviewees in the case study.
V. RESEARCH FINDINGS AND DISCUSSION
5.1 Software development paradigm that guarantees software quality
This finding provides the answer to RQ1. Figure 3 below shows the software product line platform for
the application software development identified in this research. Service providers within a specific
domain with numerous clients use a common platform for quick and reliable software development.
For example, within the financial sector, a service provider would define domain requirements, domain
architecture and the software development life cycle (SDLC) that would be generally applied in
developing several application software suitable for several clients. But different clients would have
different requirements. Thus, the problem of variability caused by differing requirements and changes
in specific clients requirements could be detected during the specific clients requirement engineering
in SDLC. Although the software product line has a defined domain requirements and architectural
design, carrying out specific clients requirement engineering and architectural design within the agile
framework is a significant approach to reducing problems caused by variability. In figure 3 below, the
arrow labeled 1 start the agile framework from where the variability is detected and addressed during
the clients specific requirement engineering. In this way, the negative effect caused by variable client
requirements would be reduced which would also reduce the defect density of the ensuing software
product.

The software product line in figure 3 comprises of 3 specific components: the domain requirement
engineering, domain architecture design and the software development life cycle. The combined
activities and processes within the SDLC are demonstrated in detail in figure 4 below. In figure 3 above,
the SPLE platform indicates that from the third element (labeled 1, is the beginning of SDLC) of the
software product line, there is continuous testing of the components till the final software product is
produced. There is also a demonstration within the software product line that reusable software
components are embedded into the software development life cycle.
Figure 4 presents a greater detail of the SDLC within the software product line. The component labeled
1 in the figure 4 below corresponds to the client requirement engineering within the software product
line. The development team embedded reusable software components into an agile framework
purposely for increasing the software quality and reducing the turnaround time for development and
processing of client services. Changes in clients requirement are detected during the normal clients
requirements engineering and architectural design in agile framework. In the normal sprint cycle of
Domain Requirement
Engineering
Domain Architectural
Design
Domain reuse software
component
Agile framework Assemble & Integrate
Testing Testing
Application
Software
Testing
Software
development life
cycle
1
2
Figure 3: Software product line engineering
708 Vol. 7, Issue 3, pp. 701-711

Scrum or release cycle of XP, instead of the normal coding witnessed in agile after the initial meetings,
specific software components are identified and selected from an established repository. These
components are retested, assembled, integrated and again tested as the software product is produced.
There is a noticeable flexibility in this process model, if the required component is not available, the
development team goes back and develops it (or produces it via the enhancement of an existing
component) before proceeding to the next level of assembling. The element labeled 2 (application
software) is the final product, which the service processing team of the service provider would use to
process client services.

Although the application software could only have been developed through agile method or simply the
reusable component, but for the very essence of software and service qualities, service providers tend
to combine different development paradigms. Researchers [28,], [29], [30] have also identified
variability as a bottleneck in the application of software product line engineering. However, it has been
successful in domains producing physical products like mobile phones. For example, Nokia Inc.
Company popularised software product line engineering when it successfully implemented SPLE and
churned out thousands of products within a very short time [31]. In contrast, the service industry is
bedeviled with varying client requirements and frequent changes even while the processing of the
service is ongoing. This particular variation in clients requirement brought variability to the fore while
applying software product line in the service industry. In order to solve this problem emanating from
variability, agile with reuse is introduced in the software product line. First, the second level of client
requirement engineering undertaken at the beginning of the agile framework detects any variation from
the established domain requirements. Once detected, it is immediately addressed and incorporated in
the design phase. During the sprint cycle of the agile framework, any variation that might have
mistakenly filtered in, is also corrected before the identification and subsequent integration of relevant
reusable components. Through this process varying clients requirements are promptly addressed and
software with an acceptable level of quality is produced.
5.2 Domain specific software quality attributes
RQ2: What does software quality mean to a service provider?
As indicated in table 2 and figure 2 above, respondents specifically identified four major software
quality attributes that service providers use for the measure of software quality. These include usability,
reliability, maintainability and reusability. Most importantly, the software must provide an acceptable
level of service quality delivered to their respective clients who are within the financial domain. Thus,
Planning and
architectural
design
Clients
requirements
Application
Software
R
e
u
s
a
b
l
e

S
o
f
t
w
a
r
e

C
o
m
p
o
n
e
n
t

R
e
p
o
s
i
t
o
r
y
Assemble
components
Develop
reusable
components
Integrate
components
Component
NO YES
Service
Assessment
Build/Test Review
Select
Sprint Cycle
Process
team
Processes
1
2
Figure 4: Expanded SDLC for embedded software reusable components in agile framework
709 Vol. 7, Issue 3, pp. 701-711

during the development process, these specific quality attributes are identified and incorporated into the
software.
This entails that different users or organisations have different software quality measurement attributes.
[32] were of the view that there is no universal set of quality concepts and measurements, because
different organisations or systems will require different attributes and measurements. And in this case
what is considered very important attributes due to organisations specific needs- the services they
deliver to their clients, are usability, reliability and maintainability. Although safety and reusability are
mentioned by some respondents, the combined five attributes are in line with ISO/IEC 25010 standard
and its characteristics but not all are relevant within this financial domain. The quality of service
delivered to clients has its dependence on the software quality which also depends on the software
development paradigm. This implies that the SDLC is directly proportional to both the quality of
software and the quality of processed services delivered to clients.
5.3 Organisational model
RQ3: Does the organisational model play any role in determining the quality of the application
software?
In answering this question, the research finding indicates the significant role of organisational modelling
in ensuring that quality services are delivered to clients. This is achieved through the integration of IT
into BPO service provider organisation to form BPO-IT organisational model. Through this model, the
organisation initially establishes dedicated software development team purposely to develop application
software for the service provider. In table 1, with the exception of BB4, all the service providers
implanted BPO-IT in order to determine the quality of services delivered to clients. The organisation
BB4 would, therefore, depend on vendors or IT companies to provide the application software required
to process client services. This integration ensures that an acceptable level of service quality is delivered
to clients through the provision of quality application software used in processing client services.
VI. CONCLUSIONS AND FUTURE WORK
In the course of this study, it was found that successful business process service providers specifically
indicated that software quality means delivering of quality services to their numerous clients. And the
only way this level of service quality could be achieved is to develop the software in-house through the
establishment of software product line as a development platform. Through this process, the
development team embedded reusable software components into an agile framework as a chosen
development paradigm. The software quality is enhanced through the reuse components, agile
framework and the software product line. [2] asserts that an agile framework with numerous advantages
provide the platform for the use of reusable components in order to quickly develop reliable software
for service processing.
Firstly, the presence of tested and retested components available in the repository during the
development cycle ensures that the end product produces the level of quality that would provide quality
service to clients. Through these components, there is significant reduction in defect density because
they are trusted components as they have been used severally. As shown in Figure 4 above, at the time
of build/test within a sprint cycle of agile, the developers would usually search for the required
components in the repository.
Secondly, agile framework with its test driven approach ensures that the level of software quality is
significantly improved by producing a software with less defects. The iterative and incremental
principle found in agile makes it easier to get feedback from customers and users who are also within
the service provider organisation. Thirdly, the introduction of software product line ensures that the
software is developed quickly to process different client requirements. The element of agility ensures
that the problem of variability arising from varying client requirements are detected early and
controlled.
This whole process of in-house software development guarantees full involvement of all the
stakeholders especially the users. Clients requirements are recognised and built-in during the
development process and users are also involved in every stage of the development level. Issues arising
from maintainability are also treated in-house. The organisational model where application software are
710 Vol. 7, Issue 3, pp. 701-711

developed in-house is a key component of developing an acceptable level of software quality. It ensures
that the required level of software quality is produced which translates to delivery of expected quality
of service to clients. Clients requirements are delivered in time and service providers competitiveness
is significantly enhanced.
Future work could be in the area of developing a central repository of reusable software components
for a specific domain. Through this repository, service providers within the same domain like the
financial sector could quickly avail required components and develop application software for
processing client services.
ACKNOWLEDGEMENT
The authors gratefully acknowledge the full funding of this study by the National Information
Technology Development Agency (NITDA) of Nigeria.
REFERENCES
[1] Javalgi, R. R. G., Joseph, W. B., Granot, E., and Gross, A. C. (2013). Strategies for sustaining the edge in
oshore outsourcing of services: the case of India. Journal of Business & Industrial Marketing, 28(6):475
486.
[2] Ikerionwu, C., Edwin, G., and Foley, R. (2013). Embedded software reusable components in agile framework:
The puzzle link between an outsourcing client and a service provider. In
Quality Comes of Age, SQM XXI, pages 6378. BCS, London
[3] Willcocks, L. (2010). The next step for the CEO: Moving it-enabled services outsourcing to the strategic
agenda. Strategic Outsourcing: An International Journal, 3(1):6266.
[4] Allen, V., Barrow, B and Salmon, J. (2012). Did one high-tech worker bring RBS to its knees? Dailymail-
28
th
June, http://www.dailymail.co.uk/news/article-2165202/Did one-high-tech-worker-bring-RBS-knees-
Junior-technician-blamed-meltdown-froze-millions-accounts.html
[5] Kitchenham, B. A., Pfleeger, S. L., Pickard, L. M., Jones, P. W., Hoaglin, D. C., El Emam, K., and Rosenberg,
J. (2002). Preliminary guidelines for empirical research in software engineering. Software Engineering,
IEEE Transactions on, 28(8):721734.
[6] Krishnan, M. S., Kriebel, C. H., Kekre, S., and Mukhopadhyay, T. (2000). An empirical anal-ysis of
productivity and quality in software products. Management science, 46(6):745759.
[7] ISO/IEC25010. (2010). Systems and software engineering systems and software quality requirements and
evaluation (square) system and software quality models.
http://www.iso.org/iso/home/store/catalogue_tc/catalogue_detail.htm?csnumber=35733
[8] Sommerville, I. (2011). Software Engineering . Pearson Education, UK, 9th edition.
[9] Boehm, B. W., Brown, J. R., Kaspar, H., Lipow, M., MacLeod, G. J., and Merrit, M. J. (1978). Characteristics
of software quality, volume 1. North-Holland Publishing Company.
[10] Beck, K., Beedle, M., Van Bennekum, A., Cockburn, A., Cunningham, W., Fowler, M., Grenning, J.,
Highsmith, J., Hunt, A., Jeries, R., et al. (2001). Manifesto for agile software development.
[11] Dyb, T. and Dingsyr, T. (2008). Empirical studies of agile software development: A systematic review.
Information and software technology, 50(9):833859.
[12] Erickson, J., Lyytinen, K., and Siau, K. (2005). Agile modeling, agile software development, and extreme
programming: the state of research. Journal of Database Management (JDM), 16(4):88100.
[13] Kumar, K., Gupta, P., and Upadhyay, D. (2011). Change-oriented adaptive software engineering by using
agile methodology: In Electronics Computer Technology (ICECT), 2011 3rd International Conference on,
volume 5, pages 1114. IEEE.
[14] Chow, T. and Cao, D.-B. (2008). A survey study of critical success factors in agile software projects. Journal
of Systems and Software, 81(6):961971.
[15] Onwuegbuzie, A.J., Johnson, R.B. (2006). The validity issue in mixed research. Research in the Schools
13 (1), 4863.
[16] Themistocleous, M., Irani, Z., and Love, P. E. (2004). Evaluating the integration of supply chain information
systems: A case study. European Journal of Operational Research, 159(2):393405.
[17] Sjoberg, D. I., Dyba, T., and Jorgensen, M. (2007). The future of empirical methods in software engineering
research. In Future of Software Engineering, 2007. FOSE07, pages 358378. IEEE.
[18] Easterbrook, S., Singer, J., Storey, M.-A., and Damian, D. (2008). Selecting empirical methods for software
engineering research. In Guide to advanced empirical software engineering, pages 285311. Springer.
[19] Johnson, R. B., Onwuegbuzie, A. J., and Turner, L. A. (2007). Toward a definition of mixed methods
research. Journal of mixed methods research, 1(2):112133.
711 Vol. 7, Issue 3, pp. 701-711

[20b] Onwuegbuzie, A. J. and Leech, N. L. (2004). Enhancing the interpretation of significant findings: The
role of mixed methods research. The Qualitative Report, 9(4):770792.
[21] Malterud, K. (2001). Qualitative research: standards, challenges, and guidelines. The lancet, 358(9280):483
488.
[22] Creswell, J. W. (2013). Research design: Qualitative, quantitative, and mixed methods approaches. Sage.
[23] Guest, G., MacQueen, K. M., and Namey, E. E. (2011). Applied thematic analysis. Sage.
[24] Braun, V. and Clarke, V. (2006). Using thematic analysis in psychology. Qualitative research in psychology,
3(2):77101.
[25] Gibson, W. and Brown, A. (2009). Working with qualitative data. Sage
[26] Northrop, L., Clements, P., Bachmann, F., Bergey, J., Chastek, G., Cohen, S., Donohoe, P., Jones, L., Krut,
R., Little, R., et al. (2007). A framework for software product line practice, version 5.0. SEI.2007
http://www. sei. cmu. edu/productlines/index. html.
[27] Baruch, Y. and Holtom, B. C. (2008). Survey response rate levels and trends in organizational research.
Human Relations, 61(8):11391160.
[28] Mannion, M. (2002). Using first-order logic for product line model validation. In Software Product Lines,
pages 176187. Springer
[39] Metzger, A. and Pohl, K. (2007). Variability management in software product line engineering. In
Companion to the proceedings of the 29th International Conference on Software Engineering, pages 186
187. IEEE Computer Society
[30] Ziadi, T. and Jzquel, J.-M. (2006). Software product line engineering with the UML: Deriving products.
In Software Product Lines, pages 557588. Springer
[31] Clements, P. and McGregor, J. (2012). Better, faster, cheaper: Pick any three. Business Horizons, 55(2):201
208.
[32] Gilb, T. and Finzi, S. (1988). Principles of software engineering management, volume 4. Addison-Wesley
Reading, MA.
AUTHORS
Charles Ikerionwu is a PhD candidate in the Department of Computer, Communications and
Interactive Systems at Glasgow Caledonian University. He received his MSc (IT) in 2002
from Punjab Technical University and MCA from Indira Gandhi National Open University
in 2004, both in India. His research interest includes but not limited to Software Product Line
Engineering, Software Quality, Agile development and big data analytics. He is a member of
British Computer Society (MBCS) and International Council on System Engineering
(INCOSE).

Richard Foley is a Senior Lecturer in the Dept. of Computer, Communications and
Interactive Systems at Glasgow Caledonian University. His current research interests are in
Software Quality and Process Improvement. He has been a member of academic staff at the
University for 30 years and received his PhD from the same institution in 1994. He can be
contacted on r.foley@gcu.ac.uk.

Edwin M Gray is Senior Lecturer in the School of Engineering and Built Environment at
Glasgow Caledonian University, United Kingdom. He is a national and international
consultant and lecturer in software engineering. Since 1981, Gray has given seminars on
software engineering to a worldwide audience of engineers and managers in Europe, Asia,
Africa and the Americas. His research interests are in software engineering, particularly
software quality and configuration management systems. Gray has 33 years experience in
applied research in the area of Software Engineering and is recognised nationally for his work
in this area and has a long international standing in software engineering, especially software quality.
His specific research expertise lies in the area of software quality and software process improvement and most
recently in the use of agile methods to support these processes. He has been directly involved with a number of
organisations responsible for developments and advancements in this area over the past 33years, e.g. Software
Engineering Institute (SEI), British Computer Society (BCS). He is also the author of several books and over 35
papers on systems analysis and software engineering.

712 Vol. 7, Issue 3, pp. 712-722

PRODUCE LOW-PASS AND HIGH-PASS IMAGE FILTER IN
JAVA
School of Engineering, Dep of s\w Engineering
Koya University, Iraq - Erbil

ABSTRACT
Image processing is a computers procedures to convert an image to digital data and implement a lot of
operations on the converted data, Processing to enhanced the image details or to extract some information
from it, Low-Pass and High-Pass are the two famous filter in image enhanced processing, In this paper we will
try to review and compare between two types of image filtering algorithm Low-Pass Filter and High-Pass Filter
and how we can implement them using java , Low-Pass and High-Pass are contradictory Image filtering with
the same executive step but the result is opposite, Low-Pass will Produce a Gaussian smoothing blur image, in
the other hand High-Pass filter will increases the contrast between bright and dark pixel to produce a sharpen
image.
KEYWORDS: High-Pass, Low-Pass, Image Processing, Pixel, Gaussian, Convolution Kernel, Smoothing,
Blur, Sharpening, Java.
I. INTRODUCTION
In Image processing and filtering the most important technique is convolution, convolution is an
expression which means to convolve digital Image original information with convolution kernel as
dimensional convolution. [1]
When we take a digital photo with camera or scanner, often images will be noisy. Noise always
changes rapidly from pixel to pixel because each pixel generates its own independent noise. So we
need to improve the image detail, this improve require many image filter effects [2], Filtering effects
can be used to:
Sharpen image.
Blur or smoothing image.
Image noise reduction.
Detect and enhance edges.
Alter the contrast of the image.
Alter the brightness of the image.
In this paper we will discuss two type of image filter Low-Pass Filter and High-Pass filter, the
two Filters have the same Convolution technique with deferent convolution kernel.
II. SURVEY OF LOW-PASS FILTER
Low-Pass filter is known as (smoothing or blurring) filter, Low-pass used for image smoothing and
noise reduction. Low-pass effect is to calculate the average of a pixel and all of its eight neighbors
[3].Low- Pass filtering is convolution that attenuates high frequency of an image while allowing low
frequency passing (leaving) [4].High frequency introduced noise in image that decrease image quality,
The Low-Pass will smooth the image and reduction noise [2].
2.1 Low-Pass Types
1. Ideal low-pass filters (ILPF)
713 Vol. 7, Issue 3, pp. 712-722

ILPF is the most simple low-pass filter, it remove all frequencies higher than the cut-off (a specified
distance) frequency and leaves smaller frequencies unchanged.
ILPF transfer response equation is:

Where:
M is number of rows in image
N is number of columns in image
(M/2, N/2) is center of frequency rectangle

Is the transform point (u, v) from the center
ILPF frequency Curve

Figure 1: Ideal low-pass filters frequency
ILPF transfer response placed the (one) inside and the zero is placed outside in this case we will get
a blurred smoothing image in the same time the image edge content will be reduced .
2. Butterworth low-pass filters (BLPF)
BLPF keeps frequencies inside radius cut-off and discards values outside, the transfer equation of
order ( n ) with cut-off frequency defined as:

Where:
(M/2 , N/2) is center of frequency rectangle

Is the transform point (u ,v) from the center
N is the transfer order
BLPF frequency Curve:
(1)
(2)
(3)
714 Vol. 7, Issue 3, pp. 712-722

Figure 2: Butterworth low-pass filter frequency
3. Gaussian low-pass filters (GLPF)
GLPF is highly effective in blurring and removing Gaussian noise from the image. The transfer
equation of GLPF is defined as :

Where:
D(u ,v) Is the transform point (u ,v) from the center
GLPF frequency Curve:

Figure 3: Gaussian low-pass filter frequency
III. SURVEY OF HIGH-PASS FILTER
High-Pass filter is known as (Sharpening) filter, High-pass used for image Sharpening and emphasize
Image details. High-Pass same as Low-pass calculate the average of a pixel and all of its eight
neighbors [3] but High-Pass convolution attenuates low frequency of an image by allowing high
frequency passing and changing the Low frequency with convolution average [4].
3.1 High-Pass Types:
1. Ideal High-pass filters (IHPF)
(4)
(5)
715 Vol. 7, Issue 3, pp. 712-722

IHPF is The simplest High-pass filter, It remove all frequencies below the cut-off frequency and
leaves higher frequencies unchanged.
HLPF transfer response equation is:

Where:
IHPF frequency Curve:

Figure4: Ideal High-pass filters frequency

2. Butterworth High-pass filters (BHPF)
BHPF keeps frequencies outside radius cut-off and discards values inside, The transfer equation of
order ( n ) with cut-off frequency defined as:

Where:

(6)
(7)
(8)
(9)
(10)
716 Vol. 7, Issue 3, pp. 712-722

N is the transfer order
D0 is Cutoff
BHPF frequency Curve:

Figure 5: Butterworth High-pass filters frequency
3. Gaussian High-pass filters
GHPF has the same concept of ideal high pass filter, but the transition is smoother ,GHPF is defined
as :
GHPF transfer response equation is:

Where:
D(u ,v) Is the transform point (u ,v) from the center
GHPF frequency Curve:

Figure 6: Gaussian low-pass filter frequency
4. Convolution Kernel Method
(11)
(12)
717 Vol. 7, Issue 3, pp. 712-722

Convolution is mathematical operation used in image processing operators. Convolution is a way of
multiplying together two arrays of numbers with the same size or in different sizes producing a third
array of numbers of the same dimensionality. This can be used in image processing to implement
operators whose output pixel values are simple linear combinations of certain input pixel values
[5,4,8].
Kernel is a synonym for a matrix of numbers that is used in image convolutions, Kernels can have
different size matrix to present different convolution numbers pattern. The kernel multiplying
neighborhood pixels to get the average, and after we get the average each pixel is replaced by the
average .
To understand the convolution kernel let us look to this example:
We have the 3*3 (h) kernel and f(x,y) as original Image element:

Figure 7: Convolution Kernel
Now the convolution will be equation:
H[f(x,y)]={ k1.f(x-1,y-1)+ k2.f(x-1,y)+ k3.f(x-1,y)+
k4.f(x-1, y+1) + k5.f(x, y) + k6.f(x, y+1) +
k7.f(x+1, y-1) + k8.f(x+1, y1) +k9.f(x+1, y+1)}
H [f(x, y)] its the convolution of original Image element (pixel).
Now we can get the pixel average by divide H [f(x, y)] by 9
The most general Convolution average equation is:

(m is the kernel ,p is the image element).
(i is row and j is column).
4.1 convolution Algorithm Step:
1. Convert the image to array.
2. Reads the (x , y) value of each pixel in the array
3. Multiplies (x , y) value by appropriate weight in kernel
4. Sums the products of ((x , y) value x weight) for the nine pixels, and divides sum by 9 (3*3
Kernel)
5. Derived value applied to center cell of array
6. Filter moves one pixel to right, and operation is repeated, pixel by pixel, line by line
4.2 Convolution Kernel Values:
The Kernel Values are not specific and not isotropic ,we can change this values dependence on the
purpose of convolution ,in fact we can use the same kernel to smooth image or to sharpen image [8] .
For example this Kernel is to detect isolated points from background.
(13)
(14)
718 Vol. 7, Issue 3, pp. 712-722

Table 1: detect isolated Kernel
-1 -1 -1
-1 8 -1
-1 -1 -1

And this kernel is for smoothing image:
Table 2: smoothing Kernel
1 1 1
1 1 1
1 1 1

And this kernel is for harshly smooth image:
Table 3: harshly smooth Kernel
1/9 1/9 1/9
1/9 1/9 1/9
1/9 1/9 1/9

If we went subtle smooth effect we can use this kernel
Table 4: subtle smoothing Kernel
0 1/8 0
1/8 1/2 1/8
0 1/8 0
Briefly By choosing different kernel, we can create filter has enough noise smoothing, without
blurring the image too much. If any one of the pixels in the neighborhood has a noisy value, this noise
will be besmeared over nine pixels and we will get blur image. Wherefore we must use a median
convolution kernel.
5. Convolution Implementation in Java:
5.1.Java Kernel Class
The Java Kernel class defines a matrix that describes how a spatial domain pixel and its neighborhood
pixels affect the average value computed for the pixel's position in the output image of a filtering
operation. The X origin and Y origin indicate the kernel matrix element that corresponds to the pixel
position for which an output value is being computed [6, 4].
5.1.1Kernel Constructors:
Kernel(int width, int height, float[] data)
Where :
Width Convolution Kernel width
Height is Convolution Kernel height
Data is the Convolution Kernel matrix of float numbers.
*To use image kernel we must import java.awt.image.Kernel
5.2 java ConvolveOp Class
Java ConvolveOp class coded to execute the image convolution from original image to the filtering
output image. Convolution applies a convolution kernel in a spatial domain operation that computes
the average value for the output pixel ,the multiplying operation take in repetitions step an input pixel
and multiply by the kernel matrix . Convolution mathematically affected the output pixel with its
immediate neighborhood according to the kernel values.
ConvolveOp class invoke Buffered Image object to present the image pixels data. Pixels data is the
RGB color components multiplied with alpha component. If the Source Image has an alpha
component, and the color components are not pre-multiplied with the alpha component, then the data
are pre-multiplied before being convolved. If the Destination has color components which are not pre-
multiplied, then alpha is divided out before storing into the Destination (if alpha is 0, the color
719 Vol. 7, Issue 3, pp. 712-722

components are set to 0). If the Destination has no alpha component, then the resulting alpha is
discarded after first dividing it out of the color components [7,4].
5.2.1 filter method
Filter Method Performs a convolution on BufferedImages. Each component of the source image will
be convolved (including the alpha component, if present). If the color model in the source image is
not the same as that in the destination image, the pixels will be converted in the destination. If the
destination image is null, a BufferedImage will be created with the source ColorModel. The
IllegalArgumentException may be thrown if the source is the same as the destination [7,4].
5.3 Convolution Coding Example
In this example we will work on a noisy low quality image like the following baby image

Figure 8: noisy low quality image
Now we will Create the Convolution kernel to smoothing the baby imge
Table 5: smoothing Kernel
1/9 1/9 1/9
1/9 1/9 1/9
1/9 1/9 1/9

float val=1f/9f;
float[ ]data={ val, val, val,val, val, val, val, val, val };
Kernel kernel = new Kernel(3, 3,data);
Now we can load the original image using bufferedImage like this:
BufferedImage buff_original;
buff_original = ImageIO.read(new File("Baby.jpg"));
Its the time to use the BufferedImageOp class to create the ConvolveOp object which hold and
buffer the a convolution kernel.
BufferedImageOp ConOp = new ConvolveOp(kernel);
The last step is to call the filter method:
buff_original = ConOp.filter(buff_original, null);
the Convolution result will be:
720 Vol. 7, Issue 3, pp. 712-722

Figure 9: Convolution smooth image
If we change the convolution Kernel to detect isolated points from background like this kernel:
Table 6: detect isolated Kernel
-1 -1 -1
-1 8 -1
-1 -1 -1
We will get this result

Figure 10: Convolution Detect isolated point image
5.4 Convolution Test Class:
import java.awt.image.BufferedImage;
import java.awt.image.BufferedImageOp;
import java.awt.image.ConvolveOp;
721 Vol. 7, Issue 3, pp. 712-722

import java.awt.image.Kernel;
import java.awt.Image;
import javax.swing.ImageIcon;
import java.awt.*;
import javax.swing.*;
import javax.swing.JFrame;
import javax.swing.JPanel;
import java.awt.event.*;
import java.io.*;
import javax.imageio.*;
public class Convolution {
public static void main(String[] argv) throws Exception {
try {
BufferedImage buff_original;
buff_original = ImageIO.read(new File("Baby.jpg"));
float val=1f/9f;
float[]data={ val, val, val,val, val, val, val, val, val };
Kernel kernel = new Kernel(3, 3,data);
BufferedImageOp ConOp = new ConvolveOp(kernel);
buff_original = ConOp.filter(buff_original, null);
JPanel content = new JPanel();
content.setLayout(new FlowLayout());
// label to load image
content.add(new JLabel(new ImageIcon(buff_original)));
JFrame f = new JFrame("Convolution Image ");
f.addWindowListener(new WindowAdapter() {
public void windowClosing(WindowEvent e) {
System.exit(0);
}
});
f.setContentPane(content);
f.pack();
f.setVisible(true);
} catch (IOException e) {
}
}}//end class
IV. CONCLUSION
In this paper, we discussed the Low-Pass and High-Pass Image filtering, after that we explained the
Convolution Kernel and how to use java utilities to implement the convolution technique in easy
way. In convolution operation we can produce both filters (Low and High) with the same step but
there is a substantial deferent in the kernel, each type of filter and its sub type need deferent kernel
data, in fact even with the same type of filter we need various data kernel that suitability the original
image frequency domains data, by other word the Kernel must be adjustable in easy compatible way
to adjust the filtering purpose that fit the original image. Smoothing image without low-pass filter or
sharpening image without high-pass filter is not that easy. But fortunately Java provide us image
filtering utilities which can process and smoothing or sharpening image in very simple easy way ,
All required from us to create the kernel matrix and let java ConvolveOp to complete the complex
convolution lopping steps to image frequency domains and replace the pixels with average .
V. Future Work
For future work we will implement the convolution java technique in tangible application with user
friendly interface which help the user to suggests and modify the kernel array data in easy way to fit
722 Vol. 7, Issue 3, pp. 712-722

the filtering purpose, with this application user can select the original image from his pc and to save
the filtered image. The next future research will be about image encryption and decryption algorithms.
REFERENCES
[1] The Scientist and Engineer's Guide to Digital Signal Processing
[2] Randall B. Smith, Ph.D 5 January 2012MicroImages, Inc, 1997-2012 Filtering Images .
[3] Image Processing - Laboratory 9: Image filtering in the spatial and frequency domains , Technical
University of Cluj-Napoca
[4] Kenny Hunt, Ph.D., The Art of Digital Image Processing .
[5] Robert Fisher, Simon Perkins, Ashley Walker, Erik Wolfart ,2004, The Hypermedia Image Processing
Reference . http://homepages.inf.ed.ac.uk/rbf/HIPR2/hipr_top.htm
[6] (Java SE Documentation), java.awt.image.Kernel
http://docs.oracle.com/javase/7/docs/api/java/awt/image/Kernel.html
[7] Java SE Documentation , Class ConvolveOp ,
http://docs.oracle.com/javase/7/docs/api/java/awt/image/ConvolveOp.html
[8] John C. Russ , The Image Processing Handbook, Third Edition ,07/01/98

AUTHORS BIOGRAPHY

O. K. Khorsheed born in Baghdad in 1974. BSC computer science from al-mustansiriya
University. Higher deploma degree from Informatics Institute of Higher Studies. Master
degree from Jordan (Arab academic for banking and financial science) school of (IT)
computer information system. Lecturer in koya University since 2005 in school of
engineering S\W engineering department.

723 Vol. 7, Issue 3, pp. 723-732

THE COMPARATIVE ANALYSIS OF SOCIAL NETWORK IN
INTERNATIONAL AND LOCAL CORPORATE BUSINESS
Mohmed Y. Mohmed AL-SABAAWI
College of Administration & Economic
University of Mosul, Iraq

ABSTRACT
Social networks are used by various corporate companies all over the world as tools for enhancing their
marketing strategy through online communication. The range of application of social networks by different
corpotare organizations is wide but common ones are for management to effectively comminucate with
employees and customers, and to keep an eye on competition. The main objective of this research is to compare
the utilisation of social network tools between international and local corporate businesses with a view to
identifying how to improve in the latter. In this paper we conducted a comparative study on how selected
international and local corporate companies use social networks. Specific items compared are the social
network tools used and the market strategy adopted by the different companies. The study identified potential
benefits of using social networks for corporate businesses. The study established more extensive use of diverse
social networks compared to the local companies. This study is among the few of its kind that tries to make
comparison between well established international and emerging local corporate businesses. The implication of
this study is that the local companies can benefit more by adopting some of the strategies used by the
international corporate companies.
KEYWORDS: social network1; corporate business2, Comparative Analysis3.
I. INTRODUCTION
People share similar social network groups. Career interests, social interests, religious subdivisions,
common friends and shared beliefs are among the typical bonds which members of a community
share and live with. Social network facilitate links among different people with similar interest to
become friends among the members. Social networks have been studied by many scholars in various
fields and across a wide variety of topics such as privacy and identity, and the social capital of
communities.
Social networks are use mostly by young people. Adolescents use of the social networks is mainly to
link up with friends. Apart from making contact with friends, social networks provide link between
network-makers and business owners and employees. The most popular online social network sites
bring together more than 20 million users and more than 150 different crafts. Social networks user can
write his autobiography in the field of education and work and can invite friends to recommend it to
others or to collaborate start new areas of work. So these networks are one of the areas that are away
from the future of large social network, the major conflict.
Initially, social networks serve as a media that links business websites, which is considered as
ultimate search engine optimization techniques. It is now recommended that most of the social
network sites nowadays have made some adjustments in order to make links for enhancement in
ranking. However, increase in web traffic can occur.
Current social networks are predominantly constituted by users who might meet face-to-face on
mostly online network such as hi-5, Netlog, MySpace, linkeln, Facebook and others. With the
increased use of enhanced mobile phone technology and the popular mobile phones, Twitter can also
be considered as a social network. This has the advantage of allowing users to find and know what
their friends and relative are doing at certain time of the day. The social networks are virtually free for
724 Vol. 7, Issue 3, pp. 723-732

everyone to join. This has made them to become popular among other networks. Nowadays, senior
executives in many part of the world are considering and implementing strategies to create business
value for their organisations through the use of social networking tools. It is evident that very large
organizations are at various stages of considering and grappling with issues related to the use of
online social networking. While some companies are deliberating on whether to restrict external
social networks, others are proactively exploring the tools. They are adopting these tools for
applications such as project collaboration, recruitment, learning and development, and other business
applications. In this paper we tried to compare the use of social networks between selected
international and local corporate businesses. Specific items compared are the social networ tools used
and the market strategy adopted by the different companies. The study identified potential benefits of
using social networks for corporate businesses. We limited our research to four international
companies Coca-Cola, Ford, IBM and Sears,some selected manufacturing and services companies
and five local companies in Malaysia - Proton, Celcom, TM, J- Biotech and Inter-Pacific Securities.
The research was conducted based on six online social networks; Face-book, LinkedIn, MySpace,
Twitter, Flickr and YouTube.
II. LITERATURE REVIEW
Definition of Social Network
Social networking is defined as the bringing individuals together into to specific groups, often like a
small community or a neighborhood. Although social networking is possible in person, especially in
schools or in the workplace, it is most popular on the internet. This is because unlike most high
schools, colleges, or workplaces, the internet is filled with billions of individuals who are looking to
meet other internet users and develop friendships.( Directory Submissions 2009).
Social network has also been defined as "individuals or groups linked by some common bond, shared
social status, similar or shared functions, or geographic or cultural connection. Social networks form
and discontinue on an ad hoc basis depending on specific need and interest." (Barker, 1999).
Web-based social networking occurs through a variety of websites that allow users to share content,
interact and develop communities around similar interests. Examples include websites such as
Facebook and LinkedIn.( Susan Gunelius 2011).
Social Networking Sites Growth in Marketing
As the use of social networks continue to expand, marketers view this medium as a potential tool
for marketing communications. The number of people visiting networking Web sites keeps increasing.
In a survey shows that social networking Web sites have witnessed a substantial increase in the
number of visitors(Marshall Kirkpatrick 2006).
As more time is being spent on social networking Web sites, many companies are now giving
much attention to marketing on these social media. In a survey, Suzanne Vranica, (2007), predicted
that between 2007 and 2011, spending on social networks in the U.S. will grow by 180% from $900
million to $2.5 billion. Danah (2007) stated that there are over a hundred social network sites, with
various technological affordances, supporting a wide range of interests and practices. While their
principal technological features are fairly consistent, there are different cultures that emerge
around social network sites. Most sites support the maintenance of pre-existing social networks,
but others help new users to connect based on shared interests, political views, or activities.
Krista (2009) itemized some of the key social networking sites. However, because this is such a
dynamic field, there are bound to be new social networks evolving, merging or vanishing all the
time. Trust is one of the key attributes expected of every social networking.
Trust and Privacy in Social Network Sites
Mayer, Davis, and Schoolman (1995) defined trust as the willingness of a party to be vulnerable
to the actions of another party based on the expectation that the other will perform a particular
action important to the tractor, irrespective of the ability to monitor or control that other party.
Electronic commerce research has established that there is a strong relationship between trust and
information disclosure (Metzger, 2004).
725 Vol. 7, Issue 3, pp. 723-732

Social Network Sites at Work
The first category is general social network sites for which registration is open to members of public.
An example is Facebook. The second category is enterprise social network site that is internal to the
particular corporate organisation and therefore only accessible to its employees, e.g., IBM Beehive
DiMicco et al (2008).
Social Network Services
Boyd & Ellison (2007) defined social network service as Social networking services can be
widely defined as internet- or mobile-based social spaces designed to facilitate communication,
collaboration, and content sharing across networks of contacts."
Online Social Networking
Christopher (2008) asserted that "Online social networking has been around in various forms for
over a decade, and has continued to be widely notice in the recent years". Online social networks
take various forms, and are created for several purposes.
Online groups or online study clubs provide a way to share knowledge, meet other dentists with
similar interests, access a calendar of events, share files, and upload pictures and videos. For
dentists, this means we can now practice solo but have the benefit of knowledge sharing and
connectivity to a specific group that is of interest to us Dan Marut (2009).
Tools for Online Corporate Business
The sharing of information and professional work are important for communication and productivity
of an organisation. Using some tools on the internet, employees can create, share and edit files of
work (e.g. calendar, word files, excel) in real time. These web applications are not difficult to
use. They enable people to improve internal communication within an organisation (Kioskea.net,
2009).
According to Faith (2009), the best features of using social networking to promote a business are
free and easy business building tools. Most social networking sites are free to register. This may be
important for a business that is working with a limited advertising budget. It is also suitable for those
without an internet specialist or marketing expert to use them for building your customer and/or
supplier list of contacts. It is worthwhile trying for any business that is not yet using social
networking to explore and promote their enterprise".
Social Network Tools for Online Corporate Business
There are several online social network tools which are beyond the scope of this paper to cover.
The most common social networking tools adopted for business applications include the
following: LinkedIn, Facebook, MySpace, Twitter, Flickr and Youtube.
i. Facebook
Nicole S (2008) stated that Facebook is structured around linked personal profiles based on
geographic, educational, or corporate networks. Profile of members can show a range of personal
information, including favorite books, films, and music; e-mail and street addresses; phone numbers;
education and employment histories; relationship status (including a link to the profile of the person
with whom one is involved); political views; and religion. According to Facebook, an average of 20
minutes per day is spent by members once they log in. tasks performed by members include linking to
friends profiles, uploading and tagging (or labeling) friends in photos, creating and joining groups,
posting events, website links, and videos, sending messages, and writing public notes for each other.
Ads can be posted in either available or customized format. The market place is available for all
Facebook users and is currently free (Facebook Adds 2007).
Facebook has variety of resources to use. Taking approach of learning it step by step to create
profile that is interesting and start engaging with your friends, make frequent updates and work toward
building your profile. It permits creating ones own sub groups within the community where one
can promote events to a specified inner circle of entrepreneurs and professionals Pagiamtzis,
(2009).

726 Vol. 7, Issue 3, pp. 723-732

ii. LinkedIn
This is a business-oriented social networking site created in December 2002 and became
operational in May 2003 mainly used for professional networking. The network consists of more
than 43 million registered users, across 170 industries. The goal of the site is to allow registered
users to maintain a list of contact details of persons they know and trust in business. The persons on
the list are referred to as Connections. The network permits users to invite anyone (whether a site
user or not) to become a connection. "LinkedIn takes your personal business network online and
gives you access to people, jobs and opportunities like never before" (Sonja, 2008). It is mainly
designed and targeted for professionals. Because of this, most of the users are above the age of 24.
Many users have been in employment for a relatively long period and are mostly well-educated.
i i i . Twi t t er
This is a Web 2.0 micro-blogging tool that requests users from the general public to answer a
simple question, What are you doing? The web-based software has been described as a
combination of blogging and Instant Messaging. It encourages brevity by limiting the postings to
140 characters, (Steven (2008). Twitter is a deceptively simple system in which users create an
account and post messages, or tweets, comprised of up not more than 140 characters, which can be
viewed by a network of associates.
Twitter was created in March 2006. While Twitter does not publish statistics on usage, Forrester
Research estimated that there are about 5 million Twitter users in March 2009. It is one of the recent
communication tools parallel growing social and cultural changes (in most parts of the world)
toward greater transparency and democratization. These changes are also emerging and becoming
visible within individual businesses and non-profit organizations.
Tweeting is not a difficult task. One can follow or be followed by others on the social networking
journey. People from across various backgrounds can be found twitting on daily basis about their
happenings Pagiamtzis, (2009). However, Pagiamtzis stated that "Recently many corporations have
used the Twitter to keep their customer informed on company news and product and service
launches". One can also post links to other networking sites that have posted information for
followers to learn more about him in the social networking arena.

iv. Flickr
Flickr is a web site owned by Yahoo! that offers photo sharing and a host of related services for users
of any level, from beginners to professionals. It is tremendously easy to use, and when used with
friends or associates it can be great fun.
According to crunchbase.com, (2010) Flickr.com allows users to utilize a number of useful features.
Unlike many photo sharing sites, no fee is required to start loading and sharing pictures. Users have
the option to make their pictures private or public. However, around 80% of the pictures on Flickr are
public. Other features of Flickr include maps to track where and when photos were taken, tagging and
commenting on pictures, and the ability to create postcards and other products.
Flickr is absolutely the most exciting place for pictures of cute puppies, slightly out-of-focus sunsets
and wedding cakes. It is not very famous as business hub use Flickr through Perl that are more
business oriented. A common business use for Flickr has been to show a product. This works well
enough for companies with a physical product to sell, especially one that benefits from visual
exposure (Teodor Zlatanov, 2009). This is achieved by creating a group to invite fans to join and
share photos.

v. Youtube
YouTube is a video sharing website on which users can upload and share videos. Everyone can watch
videos on YouTube. People can see first-hand accounts of current events, find videos about their
hobbies and interests, and discover the quirky and unusual. As more people capture special moments
on video, YouTube is empowering them to become the broadcasters of tomorrow
(www.crunchbase.com, 2010). Thus, YouTube can make a fortune in the business setting, as it could
help facilitate a new mode of business communication.
Executives at many companies provide regular presentations to department groups or even the entire
workforce, to discuss financial performance, share information, recognize outstanding workers and so
on. While the bulk of the audience may be at headquarters, many others may be in remote offices. A
727 Vol. 7, Issue 3, pp. 723-732

lot of workers may be absent the day of the presentation or unable to attend for other reasons. By
videotaping the highlights of the presentation and putting them on YouTube, they can be viewed by
everyone who is authorized to do so (Ramel, 2007).
Moreover, Ramel, (2007) mentioned that, a picture is worth a thousand words, and thousands of
pictures streamed together at 29.97 frames per second are worth a lot more. YouTube can be used to
show off new shiny hardware to prospective customers or demonstrate new software and provide
product data, statistics or any other information.
III. SOCIAL NETWORK STRATEGY
Having a good Social Networking Strategy for corporate organisation is helpful if it wants to become
successful in it business and consequently be able to develop truly complex business strategies using
online social networking. It is very important to develop a real verifiable presence on all the key.
Organization networking sites and all the major business networking sites. A lot of people spend so
much time ego networking trying to become known by everyone that they actually rarely leverage
their position and do not get to know anyone well enough to actually transact any business.
According to Pagiamtzis (2009), there has been many webinars and websites discussing the value
added reasons in combining communications as a key social strategy for your online presence. They
further mentioned some practical advice on how to use the social networking as a key tool to build
organizational resources and promote its expertise or marketing ideas in easy and effortless way.
The bottom line is that most people have experienced social networking sites like LinkedIn and
MySpace and understand the value of having specific groups of people with similar interests, in this
case how others may be using the software, communicate with each other. The business relationship
expert, Bob (2008), stated that If these social networking concepts are not in your radar, you are
ignoring a dynamic trend that could have a profound impact on key areas of your business such as
profitable revenue growth, talent acquisition and development, and operational efficiency and
effectiveness".
Benefit of Social Network Tools for Corporate Businesses
One of the many ways to achieve industrial competitiveness is to manage and share efficiently
knowledge built inside an organization. In this context, social networks have shown signs of being an
efficient tool to proliferate individual and explicit knowledge. It can also improve tacit knowledge
dissemination, helping to capture organizational knowledge based on the knowledge of each of its
employees (Ricardo 2009).
A company that has a good early warning system would not miss opportunities or fail to meet
challenges quickly enough. Clueless organisations in many industries were surprised when PCs turned
out to be a big thing, because they had no way of absorbing that knowledge systemically, through
their own employees. Even though some people inside the organization undoubtedly knew the shape
of the future and were talking about it, they had no way to get it to decision-makers.
According to executive briefing from social networking for businesses & associations (2009), the top
10 ways businesses, associations and organizations can use social networking are highlighted. These
include:
I Customer and Member Relationship Development
Customer satisfaction is at an all-time low, perhaps as a result of reduced business focus on actual
relationships, and an increased focus on customer relationship management systems emphasizing
management of data rather than personal connections. Online social networks allow a prospective
customer or prospective member to easily facilitate a real, human level connection with individuals
within an organisation. This enables genuine business relationships to form and puts an authentic
human face on the interaction, changing the external perception of an organization from a sterile,
faceless behemoth into a collection of individuals who are ready to help.
II. Customer support (connecting the customer with the right resource)
Successful customer support achieves a number of goals. Basic customer service includes, of course,
assisting customers when they have problems or questions about an organizations products.
However, online networks enable exceptional customer support that goes beyond the basics, allowing
customers to connect with experts in an organization who have deep knowledge in a particular area.
III. Provide the whole product
728 Vol. 7, Issue 3, pp. 723-732

By creating a strong network of complementary providers with similar philosophies and business
practices, a single service provider can provide a much greater value proposition to a prospective
customer than an individual working without the benefit of the network.
IV. Supercharge Meeting Facilitation and Preparation
The unfortunate part of meetings and conferences is that it always seems that one do not connect with
the people he really want to meet until the final day of the event, when he meets them randomly in the
buffet line. A dedicated online social network created before the event enables attendees to use their
time at the event more efficiently, by determining with whom they want to connect before leaving
home.
V. Increasing the Value and Extend the Shelf Life Of Conferences
Similar to the above point, creating an online social network of event attendees extends the shelf
life of a conference, enabling the attendees to remain connected and take action on the items
discussed at the event. This can evolve a meeting, event or conference from a one time occurrence
into the catalyst of a community that more effectively achieves its goals.
VI. Share knowledge
By connecting a social network with basic subscription technologies (such as RSS, or Really Simple
Syndication), an individual can easily subscribe to updates from customers and colleagues. This
enables a straightforward way to stay abreast of the goings-on in projects of interest, as well as a way
to share knowledge within an organisation without additional effort.
IV. METHODOLOGY
Four international manufacturing and services companies were considered in the study. They include
Coca-cola, Ford, IBM and Sears. Five manufacturing and services companies in Malaysia were
targeted for the research. These include Celcom, Proton, J-Biotech, TM and Inter-Securities corporate
companies. The selection of manufacturing and services companies was done in such a way that it
will cut across major corporate companies that make use of social platform in their business activities.
For the international companies, the study was conducted by accessing the information regarding their
use of social networks as evidenced on their websites. For the local companies a combination of
questionnaires and interviews were adopted. Priority was also placed on the respondents that must
have been a part of the management in order to give the researchers a broader and better
understanding of the use of social networks in the companies.
V. DATA ANALYSIS AND FINDINGS
Social Network Tools for Corporate Businesses
The website analysis of the international corporate companies shows the kind of social networking
tools used for their business activities. Most of them use their own social networking sites in
promoting their business activities apart from the known social platforms. Ultimately, the selected
companies that employed successful use of social network in business activities introduced the
practice of publicity in marketing product on social platforms.
Meanwhile, the analysis of the local corporate companies obtained from questionnaire shows that all
of them employed only the use of some public social platforms. Nonetheless the local companies are
now motivated to use of their own social network for business activities.
Therefore, the study found out that, apart from the known social networking sites, all the international
and local corporate companies under study use the social platform in promoting their products and
services. The study established more extensive use of diverse social networks compared to the local
companies as shown in figure 1.

729 Vol. 7, Issue 3, pp. 723-732

Figure 1 frequency of social network tools

Internation
al
Local

Youtube
Linkedin
Flickr
Mysears
Mykmant
LotusConne
ct
Myspace
Twitter
Facebook
Youtube
Linkedin
Flickr
Mysears
Mykmant
LotusConne
ct
Myspace
Twitter
Facebook

4

1

2

1

1

1

2

3

4

1

1

4

0

0

0

0

1

4

730 Vol. 7, Issue 3, pp. 723-732

Social Network Marketing Strategy
Some social network strategies were brought and digested out by the researchers that could be used as
a challenge to some corporate companies. Many companies are now committed to improving the lives
of their customers by providing quality service products solutions that earn their trust and build
lifetime relationship. Ultimate, they are now strategizing to become an even bigger integral part of the
online community with new changes and developments that will likely place them as a front runner of
social networking and involvement. However, these companies according to research findings
developed a marketing-based strategy that gives them a competitive advantage. The social network
marketing strategies used by the companies are as indicated in Table 1.
Table 1: Social Network Market Strategies
CORPORATE COMPANIES
INTERNATIONAL LOCAL
Name
Social Network Strategy
Name
Social Network Strategy
Coca
Cola
Creates games and reinforce their
brand image in the target markets'
mind
J-Biotech
Organizing a sort of online quiz and
making product tag, Thus, the
customer contending with their
friend
Ford
Creates online quiz applications
making customers to challenge
their colleagues about companys
product and services
TM
Promoting products and then client
can recommend to their friends.

IBM
Create demand-pull video
advertising, web advertising &
viral marketing
Celcom
Sending out information to potential
customers through the network
Sears
Launch an open ID platform and
directly connect users
Inter-
pacific
Securities
Make clients to challenge their
friends about companys products
Proton
Making an application about
companys product to the clients
Social Network Usage Suitable For Corporate Business
As highlighted earlier, the use of social networking for corporate business indeed has tremendous
advantage. This may perhaps improve companys performance in terms of promoting product and
services. Most of the respondents were of the view that corporate companies need to be on social
platform (their owned or public) in order to have competitive advantage. This is a great challenge to
corporate companies.
Using social network, a company can benefit from the striking ways of doing business which could of
a great competitive advantage. Suitable social networks are the ones that enable corporate businesses
to achieve the following:
i. Finding buyers
ii. Find manufacturers
iii. Finding potential job candidates
iv. Create persona; friends or business connection group
v. Hire people
vi. Market product
VI. CONCLUSIONS
In this paper we compared the use of social networks between selected international and local
corporate businesses in Malaysia. By connecting on social network, individuals can easily get updates
from customers and colleagues regarding companys products and services. This perhaps enables a
731 Vol. 7, Issue 3, pp. 723-732

direct way to stay well-informed about what is going on in a particular company. Therefore, a
company can diversify the means of sharing knowledge within an organization without additional
effort especially company research related issues. Social media, if used effectively, can leads
corporate companies to find potential job candidates who are skilled social networkers. The
implication of this study is that the local companies can benefit more by adopting some of the
strategies used by the international corporate companies.The companies set up the skills they want on
social platforms, while the candidates had to be very active within social platform before they even
come across it. This helps in getting the right potential employees needed by the companies.
ACKNOWLEDGMENT
This work was sponsored by the Research Management Center, Universiti Teknologi Malaysia. My
thanks to the anonymous referees whose comments helped considerably in the preparation of this
paper for publication.
REFERENCES
[1]. Boyd, D., & Ellison, N., (2007), Social network sites: Definition, history, and scholarship, Journal of
Computer-Mediated Communication, 13(1),.Retrieved August 22, 2009 from
http://jcmc.indiana.edu/vol13/issue1/boyd.ellison.htm , pp.1-11
[2]. Christopher,C.(2008) Executive Briefing: Social network for business associations. Retrieved August
20, 2009 from http://haystack.cerado.com/html/haystack_directory.php. PP.1-8.
[3]. Dan, M. (2009). Social networking for dentists made easy. Retrieved August 22, 2009 from
http://www.dental- tribune.com/articles/content/id/315/scope/news/region/usa.
[4]. DiMicco. J., Millen. D., Geyer. W., Dugan, C., Brownholtz, B., and Muller, M. (2008) Motivations for
Social Networking At Work. In Proceedings of The 2008 ACM Conference on Computer Supported
Cooperative Work, San Diego, CA, USA, November 08 - 12, 2008, pp.711-720
[5]. David,R.(2007) YouTube for Your Business; Computerworld. Retrieved August 20, 2009 from
http://www.pcworld.com/article/133278/youtube_for_your_business.html.
[6]. Emin , D. & Cneyt , B. (2007) Web 2.0 - an Editors Perspective: New Media for Knowledge Co-
creation. International Conference on Web Based Communities (2007),pp 27-34.
[7]. Facebook Adds Marketplace of Classified Ads (2007-05-12) .Retrieved August 24, 2009 from
www.physorg.com/news98196557.html .
[8]. Faith, D. (2009) Writing Level Star: Promote your business using social networking tools.
[9]. Pagiamtzis, J. (2009 ) Social Networking Strategy: Communication is the key .Retrieved August 12,
2009 from: http://www.articlesbase.com/social-marketing-articles/social-networking-strategy-
communication-is-the-key-1112082.html
[10]. Kioskea.net (2009) Corporate Tools for Online Business, Retrieved August 24, 2009 from
http://en.kioskea.net/faq/sujet-1904-corporate-tools-for-onlinebusiness.
[11]. Krista, (2009) 18 Using Social Networking Sites: You are hired! CV. 2009.
[12]. Marshall Kirkpatrick Top Ten Social Networking Sites See 47% Growth, the socialsoftwareweblog
.http://socialsoftware.weblogsinc.com/2006/05/17/top-10-social-networking-sites-see-47-growtw.
[13]. Mayer. R., Davis. J., and Schoorman, F., (1995). An Integrative Model of Organizational Trust, The
Academy of Management Review (20) 3, pp. 709-734.
[14]. Metzger. M., (2004) Privacy, Trust, and Disclosure: Exploring Barriers to Electronic Commerce 9
(4), Journal of Computer-Mediated Communication (9) 4.
[15]. Nicole, S., (2008). The Valorization of Surveillance: Towards a Political Economy of Facebook.
[16]. Sonja, J., (2008) How to use LinkedIn to bring in business. Retrieved August 24, 2009 from
http://www.sonjajefferson.co.uk/.
[17]. Steven, C. (2008) Twitter as KM Technology.
[18]. Suzanne, V., (2007) Marketing Leadership Council 2008, Marketing Leadership Council 2008:
Retrieved August 14, 2009 from www.mlc.executiveboard.com.
[19]. Teodor, Z., (2009). Cultured Perl: Flickr, a business's bst frnd Create charts and upload them to Flickr
using CPAN modules, IBM DeveloperWorks
[20]. Directory Submissions (2009), Social Networking Definition, Retrieved May 01, 2011,
http://directorysubmissions.eu/news/2009/04/22/social-networking-definition/
[21]. Barker, Robert L. (Ed.). (1999). The social work dictionary (4th Ed.). Washington, DC: NASW Press.
[22]. Susan Gunelius (2011). Social Networking Retrieved May 01, 2011,
http://weblogs.about.com/od/bloggingglossary/g/SocialNetwork.htm
732 Vol. 7, Issue 3, pp. 723-732

ABOUT THE AUTHORS
Mohmed Y. Mohmed AL-SABAAWI is a Lecture of University of Mosul. in the College of Administration &
Economic at the Department of Management Information System .
Address University of Mosul. College of Administration & Economic

733 Vol. 7, Issue 3, pp. 733-742

PRECISE CALCULATION UNIT BASED ON A HARDWARE
IMPLEMENTATION OF A FORMAL NEURON IN A FPGA
PLATFORM
Mohamed ATIBI, Abdelattif BENNIS, Mohamed BOUSSAA
Hassan II - Mohammedia Casablanca University, Laboratory of Information Processing,
Cdt Driss El Harti, BP 7955 Sidi Othman Casablanca, 20702, Maroc

ABSTRACT
The formal neuron is a processing unit that performs a number of complex mathematical operations on real
format data. These calculation units require hardwares architectures capable providing extremely accurate
calculations treatments. To arrive upon more accurate hardware architecture in terms of the calculation, the
new proposed method uses data coding in single precision floating point. This allows handling of infinitely
small and infinitely large data and; consequently, a diverse field of application. The formal neuron
implementation requires an embedded platform whose implementation must be flexible, efficient and fast. This
article aims at presenting in detail a new precise method to implement this calculation unit. It uses a number of
specific blocks described in VHDL hardware description language in an embedded FPGA platform. The data
handled by these blocks are coded in 32-bit floating point. The implementation of this new method has been
developed and tested on an embedded FPGA platform of Altera DE2-70. The calculation results on the platform
and those obtained by simulation are very conclusive.
KEYWORDS: FPGA, precision, formal neuron, Floating point, HARDWARE implementation.
I. INTRODUCTION
The artificial neurons networks (ANN) present heuristic models whose role is to imitate two basic
skills of the human brain:
Learning from examples.
The Generalization of knowledge and skills learned through examples to others which are
unseen in the learning phase [1].
The ANN is configured through a learning process for a specific application, this process involves
adjusting the synaptic connections between neurons. These models are used in a wide range of
applications such as patterns recognition, the classification, robotics, signal and image processing...
etc. For example in the field of information processing, these models simulate the way biological
nervous systems process information.
The ANNs are networks based on a simplified model of neuron called formal neuron, this model can
perform a number of functions of the human brain, like the associative memory, supervised or
unsupervised learning, parallel functioningetc. Despite all these features, formal neuron is far from
having all the performances of biological neurons that human being possess like synapse sharing and
membrane activation [2].
A major problem of the use of formal neurons in the ANNs, is the lack of Hardware method to
implement in embedded platforms [3] [4]. The respect of, on the one hand, the neurons architecture,
and on the other hand, the format of neurons manipulated data which takes often the form of a real
number has a great impact on the calculation results of this neuron and their precision. This is
especially true in the case of an application requires an architecture consisting of a large number of
neurons.
734 Vol. 7, Issue 3, pp. 733-742

Several attempts have allowed implementing formal neurons as integrated circuits. The field-
programmable gate array (FPGA) is the preferred reconfigurable hardware platform. It has proven its
capacity through several applications in various fields [5]; implementing complex control algorithms
for high speed robot movements at [6], efficient production of multipoint random distributed variable
[7], the design of hardware platforms / software for the car industry [8], or in applications of energy
production [9]. However, the design of the neuron presents several challenges, the most important is
to choose the most effective format of arithmetic representation to ensure both good precision and
processing speed.
The article examines in detail the precision of formal neuron design with a sigmoid activation
function on FPGA, the architecture is tested with a floating point arithmetic format by using an
advanced hardware description language (VHDL).
The article is organized as follows. Section II provides a global overview on different Hardware
architectures. Section III presents a theoretical study of the formal neuron with its different activation
functions, and the existing data formats. Section IV dedicated to the details of the implementation of
formal neuron Hardware detail. Section V presents the tests of efficiency of this implementation.
Finally Section VI presents the conclusion.
II. RELATED WORK
Several approaches of architecture have been proposed for hardware implementation of the formal
Neuron in a platform such as FPGA, as shown in Figure 1. In 2007, Antony W. Savich has made a
detailed study on FXP and FLP representations and the effect of the accuracy on the implementation
of the multilayer perceptron. The obstacle found in this study was related to the implementation of
formal neuron with the sigmoid activation function which requires complex operations such as the
exponential and division [3].

Figure 1. Neuron structure
In 2011 (Horacio Rostro-Gonzalez) has presented [4] a numerical analysis of the role of asymptotic
dynamics in the design of hardware implementations of neural models like GIF (generalized integrate-
and-fire). The implementation of these models was carried out on an FPGA platform with fixed-point
representation (figure 2).
In 2012 (A. Tisan) has introduced an implementation method of the learning algorithm of networks
artificial neural on FPGA platform. The method aims at constructing a network of specific neurons
using generic blocks designed in the math Works Simulink environment. The main features of this
solution are mainly the implementation of the learning algorithm on high capacity chip of
reconfiguration and functioning of real time constraints [5].
In 2011 (Cheng-Jian Lin) has presented in his article the hardware implementation of neurons and
neural networks, with a representation of real numbers in fixed-point format, using the perturbation
method as a method of networks learning [2].

735 Vol. 7, Issue 3, pp. 733-742

Figure 2. Architecture for a single neuron
III. THE FORMAL NEURON THEORY
3.1. History
In 1943, McCulloch and Pitts have proposed a model that simulates the functioning of biological
neuron. This model is based on a neurobiological inspiration, which is a very rudimentary modeling
the neurons functioning, in which the accumulation the neuron synaptic activities are ensured by a
simple weighted summation [1]. The interconnections of a set of such units provide a connectionist
neural system, also referred to as neural network.
These networks can perform logical functions, complexes arithmetic and symbolic. Figure 3 shows
the schema of a formal neuron:

Figure 3. Schema of a formal neuron
With:
X1Xn: the neuron vector object.
W1Wn: synaptic weights contained in the neuron.
: a function which calculates the sum of the multiplication between the object vector and the
synaptic weights according to equation (1).
b: the bias of the summation function.
F(V): the neuron activation function.
Y: the formal neuron output.
A "formal neuron" (or simply "neuron") is a nonlinear algebraic and bounded function. In fact the
neuron receives at its input an object vector of which each object parameter is multiplied by a synaptic
weight. The sum of these multiplications and the bias constitute internal activation:
(1)
V will gets, at the end, to an output through an activation function. There are many activations
functions of a formal neuron, the most used are:
736 Vol. 7, Issue 3, pp. 733-742

The threshold function (Figure 4): In 1949, Mculloch and Pitts have used this function as an
activation function in their formal neuron models. Mathematically this model of neuron
generates an application from Rn to {0,1}.
1 si V(X)

Y(X)=f(V(X))= (2)
0 sinon

Figure 4. Threshold function
The sigmoid function (Figure 5): this is a function proposed by by Rosenblat in 1962, also
called logistic function, defined by:
(3)
It is a function with values in the interval [0,1], which allows to interpret the output of the neuron as a
probability. In addition, it is not polynomial and is infinitely continuously differentiable.

Figure 5. Sigmoid function
The Gaussian function (Figure 6): It is a function proposed in 1989 by MOODY and
DARKEN in order to be used in specific networks called radial based networks (RBF). It is
defined by:
Y(X)=f(V(X))= (4)
It is a function that depends on the center points of the input space and its width. In addition it is a
continuous and differentiable function.

Figure 6. Gaussian function
3.2. Data formats handled by a formal neuron
The formal neuron, in most cases, makes its calculations with real numbers. To represent a real
number, there are a finite number of bits and one can imagine different ways to represent it with this
bit set. The two well-known methods of representation are fixed-point representation and floating
point representation.
737 Vol. 7, Issue 3, pp. 733-742

3.2.1. Fixed-point representation
This is the usual representation as is done on paper except for the sign and the decimal point. For the
sign, we keep a bit for it; in general the left bit. Also the point is not represented, working with an
implicit point located at a definite place.
This representation is not widely used because it has a low accuracy because of the numbers lost to
the right of the decimal point. Another problem with this representation is that it does not represent
very large numbers.
3.2.2. The floating point representation
Inspired by the scientific notation (ex : +1,05 x 10-6 ) and in which a number is represented by the
product mantissa and a power of 10 (decimal) or a power of 2 (binary). To normalize the floating
point representation of real numbers [3], the IEEE-754 norm is recommended by the Institute of
Electrical and Electronics Engineers. It is widely used to represent real numbers. Each number is
represented by:
bit for the sign (s).
Ne bits for the signed exponent (E).
Nm bits for the absolute value of the mantissa (M).
Real numbers are represented either in 32 bits (simple precision), 64 bits (double precision) or 80 bits
(extended).
Example of a real number represented in 32 bits:
Table 1. Floating point representation
Bits
Bits (31 down to 0) 31 30 - 23 22 - 0
Contents (s, E, M) Sign (s) 0 = positive
1 = ngative
Exponent (E)
An 8 bits integer
Mantissa (M)
An 23 bits integer

X = (- 1)
S
x 2
E- 127
1, M with 0 < E < 255
IV. IMPLEMENTATION DETAILS
This section reviews the different steps and necessary modules for the design of formal neuron with
sigmoid activation function. The formal neuron module consists of many multipliers, Additionners,
and a block of sigmoid activation function.
Among the problems of the design of a formal neuron in a FPGA platform with the VHDL language,
is that the real numbers are not synthesized in this language, the solution which has proven this
implementation is to design a formal neuron that manipulates these data, representing them in
floating-point . This provides an efficiency in terms of the calculation precision.
To achieve this accuracy, blocks called mega functions , which are blocks offered by the constructors
of the FPGA, have been used . These blocks are written in VHDL to handle complex arithmetic
operations with floating point representation (32 or 64 bit), these blocks are useful for the calculation
accuracy in the formal neuron.
4.1. Megafunctions
As the design complexity increases in a fast manner, the use of specific blocks has become an
effective method of design to achieve complex applications in different domains such as robotics,
signal and image processing etc. The simulation software QUARTUS offers a number of IP (
Intellectual Properties ) synthesizing complex functions (memory, multipliers, comparators etc ...)
optimized for Altera circuits. These IP, designated by the term megafunction , are grouped into
libraries, including Library of Parameterized Modules (LPM), containing the most complex
functions that are useful for the design of formal neuron.
The use of megafunctions replacing in the coding of a new logic block saves precious time for design.
In addition to the functions provided by Altera allows us to offer more efficient logic synthesis for the
realization of the application. It also allows the opportunity to redimensionning the size of these
738 Vol. 7, Issue 3, pp. 733-742

megafunctions by adjusting the settings and to access specific functionality of the architecture in the
memory, DSP blocks, registers shift, and other simple and complex functions [10].
The formal design of the neuron was based on the sigmoid activation function; it's an indefinitely
differentiable function.
The design detail is divided into two parts (Figure 7):
The first part: the design of the internal activation.
The second part: the design of the sigmoid activation function.

Figure 7. Design of the formal neuron
4.2. Design detail of the internal activation
The formal neuron receives as input an object vector X=(X1,X2,..,Xn) which is represent forms to
recognize in the example of application of pattern recognition and a vector of synaptic weights
W=(W1,W2,,Wn) representing the connection between the neuron and one of its inputs (Figure 8).
The function of the neuron consists of calculating firstly the weighted sum of its inputs. The sum
output is called internal neuron activation (1).

Figure 8. Internal activation
This module implements this operation using megafunctions multiplication and addition according to
Equation (1).
4.2.1. Multiplication:
The multiplication block used is a megafunction block that implements the functions of the
multipliers. It follows the IEEE-754 norm for representations of floating point numbers in simple
precision, double precision and single extended precision. More, it allows the representation of special
values like zero and infinity.
The Representation followed in this paper is the representation of single precision 32 bits as follows;
this is a High Precision representation which consumes less space compared to 64 bits:
X = (-1)
S
x 2
E-127
x 1,M
The result (R) of the multiplication algorithm of two real inputs (A and B) represented in floating
point used by this megafunction is calculated as follows:
R= (Ma x 2
Ea
) x (Mb x 2
Eb
) = (Ma x
Mb
) x 2
Ea
+
Eb

Where:
R: multiplication result.
Ma: the mantissa of a number A.
Mb: the mantissa of a number B.
Ea: the exponent of a number A.
Eb: the exponent of a number B.
Sign: (sign of A) XOR (sign of B).
4.2.2. Addition:
739 Vol. 7, Issue 3, pp. 733-742

The addition block used is a megafunction block that implements the functions of addition and
subtraction; it follows the IEEE-754 norm for representations of floating point numbers in single
precision, double precision and single extended precision, with handling of selecting operation
between addition and subtraction.
The result (R) of the addition algorithm of two real inputs (A and B) represented in floating point used
by this megafunction is calculated as follows:
R = (-1)
Sa
x 2
Ea
x 1,Ma + (-1)
Sb
x 2
Eb
x 1,Mb
Where:
R: addition result.
Ma: the mantissa of a number A.
Mb: the mantissa of a number B.
Ea: the exponent of a number A.
Eb: the exponent of a number B.
Sa:the sign of number A.
Sb:the sign of number B.
These 2 blocks are the basis for designing the internal activation of the formal neuron. The following
Figure 9 shows an example of the implementation of this internal activation.

Figure 9. Design of internal function
4.3. Design detail of the sigmoid function
The second block is a transfer function called activation function. It limits the output of the neuron in
the range [0,1]. The most used function is the sigmoid function (Figure 10).

Figure 10. Sigmoid function
The implementation of this function requires a number of complex operations such as division and
exponential. It requires the use of the exponential and division megafunctions.
4.3.1. Exponential and division
The used blocks of the exponential and the division are megafunction blocks that implement the
functions of division and exponential. These blocks require a number of resources for their designs,
the following 2 tables show these resources:

740 Vol. 7, Issue 3, pp. 733-742

Table 2. Exponential resources
Precision Output
latency
Logic usage Fmax
(MHz)
ALUTs Registers 18 bit DSP memory
Single 17 527 900 19 0 274,07
Double 25 2905 2285 58 0 205,32
Table 3. Division resources
Precision Output
latency
Logic usage Fmax
(MHz)
ALUTs Registers 18 bit DSP memory
Single 20 314 1056 16 0 408,8
Double 27 799 2725 48 0 190,68

4.3.2. Sigmoid function implementation
The implementation of the sigmoid function in addition to these displays two blocks, the already
mentioned blocks of multiplication and addition, as it is shown in the following diagram (Figure 11):

Figure 11. Design of sigmoid function
V. TEST AND RESULT
This test is designed to evaluate the precision of calculation of the formal neuron with VHDL
language by comparing it with the software results. Before performing this test, it is necessary to
initialize the synaptic weights. The following table summarizes the initialization (the case of 4 inputs),
representing the floating point data:
Table 4. Values of the synaptic weight

value
Floating point
representation in 32
bits (hexadecimal)
W1 1 3F800000
W2 0.5 3F000000
W3 -0.5 BF000000
W4 -1 BF800000
These values of the synaptic weights will be used during the entire test phase for design.
The simulation of the complete implementation of a formal neuron using the sigmoid activation
function, in the FPGA platform of the family Cyclone II Version EP2C70F896C6. Requires a number
of steps in the simulation software Quartus II 9.1:
1. Create a new project by specifying the reference of the chosen platform.
2. Choose the Block Diagram/Schematic file.
3. Draw the model of the formal neuron by combining between required megafunctions blocks
(figure 12).
741 Vol. 7, Issue 3, pp. 733-742

Figure 12. Design of the complete formal neuron
4. Compile the project by clicking START COMPILER.
5. Choose VECTOR WAVEFORM FILE by specifying the values of the manipulated inputs
(table 4) and outputs.
6. Start SIMULATOR TOOL to simulate a module of the formal neuron.
7. View the simulation result (Figure 13).

Figure 13. Test result
The following table shows the obtained results of the formal neuron test based on the sigmoidal
activation function.
Table 5. Hardware and Software result
vector object Hardware
result
Software
result X1 X2 X3 X4
1 1 0.5 0.5 0.67917 0.679178
1 0.5 -0.5 -0.5 0.88 0.88
0.5 0.5 0.5 0.5 0.5 0.5
0.5 0.5 -0.5 -0.5 0.817 0.817574
4 test vectors are tested in this neuron. The table shows the output result of these 4 inputs with the
synaptic weights in the table (4). These data are represented in the FPGA with floating point format at
32 bits, leading to a good precision while doing the calculation in this neuron. Moreover, the table
shows also a comparison of the same neuron calculations carried in software. The result of these
calculations has shown great precision thanks to the representation of floating point data. This
precision is due to the use of megafunctions blocks (multiplier, additionner, exponential etc.).
VI. CONCLUSION AND FUTURE WORKS
This article has examined the technics of Hardware implementation of the formal neuron with
sigmoid activation function in the FPGA platform using a floating point 32-bit format, of neuron
processed data. The objective of this Hardware implementation is to materialize the formal neuron as
a specific component in the calculations and can, therefore, be added to the library of the Quartus
software.
742 Vol. 7, Issue 3, pp. 733-742

For future work, this module of a formal neuron will be the base for the design of architecture of
artificial neuron networks, as the architecture of the multilayer perceptron (MLP). And apply this
network to various applications such as image processing, signal processing, pattern recognition...
REFERENCES
[1]. Sibanda, W., & Pretorius, P. (2011). Novel Application of Multi-Layer Perceptrons (MLP) Neural
Networks to Model HIV in South Africa using Seroprevalence Data from Antenatal Clinics. International
Journal of Computer Applications, 35.
[2]. Lin, C. J., & Lee, C. Y. (2011). Implementation of a neuro-fuzzy network with on-chip learning and its
applications. Expert Systems with Applications, 38(1), 673-681.
[3]. Savich, A. W., Moussa, M., & Areibi, S. (2007). The impact of arithmetic representation on implementing
MLP-BP on FPGAs: A study. Neural Networks, IEEE Transactions on, 18(1), 240-252.
[4]. Rostro-Gonzalez, H., Cessac, B., Girau, B., & Torres-Huitzil, C. (2011). The role of the asymptotic
dynamics in the design of FPGA-based hardware implementations of gIF-type neural networks. Journal of
Physiology-Paris,105(1), 91-97.
[5]. Tisan, A., & Cirstea, M. (2013). SOM neural network designA new Simulink library based approach
targeting FPGA implementation. Mathematics and Computers in Simulation, 91, 134-149.
[6]. Shao, X., & Sun, D. (2007). Development of a new robot controller architecture with FPGA-based IC
design for improved high-speed performance. Industrial Informatics, IEEE Transactions on, 3(4), 312-321.
[7]. Bruti-Liberati, N., Martini, F., Piccardi, M., & Platen, E. (2008). A hardware generator of multi-point
distributed random numbers for Monte Carlo simulation.Mathematics and Computers in Simulation, 77(1),
45-56.
[8]. Salewski, F., & Kowalewski, S. (2008). Hardware/software design considerations for automotive embedded
systems. Industrial Informatics, IEEE Transactions on, 4(3), 156-163.
[9]. Bueno, E. J., Hernandez, A., Rodriguez, F. J., Girn, C., Mateos, R., & Cobreces, S. (2009). A DSP-and
FPGA-based industrial control with high-speed communication interfaces for grid converters applied to
distributed power generation systems. Industrial Electronics, IEEE Transactions on, 56(3), 654-669.
[10]. Online in: http://www.altera.com.
AUTHORS
ATIBI Mohamed received his master degree in information processing from Faculty of
Science Ben Msik, Hassan II University Mohammedia-Casablanca in 2013, he is
preparing his PhD thesis in the same university, and his area of interest includes:
application of image processing and artificial neuron networks in road safety.

BENNIS Abdellatif Is a professor of higher education at the Laboratory of Information
Processing, Faculty of Science ben m'sik, , Hassan II University Mohammedia-
Casablanca, and responsible for the software engineering and telecommunications team in
the same laboratory.

BOUSSAA Mohamed received his master degree in information processing from Faculty
of Science Ben Msik, Hassan II University Mohammedia-Casablanca in 2013, he is
preparing his PhD thesis in the same university, and his area of interest includes:
application of signal processing and artificial neuron networks in the cardiac signals.

743 Vol. 7, Issue 3, pp. 743-755

TEMPERATURE PROFILING AT SOUTHERN LATITUDES BY
DEPLOYING MICROWAVE RADIOMETER
A.K.Pradhan
1
, S. Mondal
2
, L.A.T.Machado
3
and P.K.Karmakar
4

1
Department of Electronic Science, AcharyaPrafullaChandraCollege, New Barrackpore,
North 24 Parganas, Kolkata 700 131, India.
2
Dumkal Institute of Engineering and Technology, P.O-Basantapur, Dist- Murshidabad, India
3
Instituto Nacional de Pesquisas Espaciais (INPE) Previso de Tempo e Estudos Climticos,
CPTEC Road, Cachoeira Paulista, SP, 12630000, Brazil.
4
Institute of Radiophysics and Electronics, University of Calcutta, Kolkata 700 009, India

ABSTRACT
Multifrequency Microwave Radiometer (MP3000A) from Radiometrics Corporation is deployed at different
places of Southern latitude for the profiling of one of thermodynamic variables like Temperature. The radiative
intensity down- welling from the atmosphere and is expressed in an equivalent brightness temperature . The
radiation is a nonlinear function of the required quantities and we linearise the expression around a suitably
chosen first guess, such as a climatological mean. We describe changes in the brightness temperature around
the first guess by means of a weighting function which expresses the sensitivity of to the variation of the
humidity or the temperature around their initial values. The variation of brightness temperature
with height occurs at 51 53 GHz is observed but on the other hand the constancy of brightness temperature
with height at 56 57 GHz is noticeable. This suggests that the measurement of temperature at a certain place
by a ground based radiometer may provide good result by exploiting the 56 57 GHz band. In this band we
have used four frequencies for the purpose. But to extend our study we have also made an attempt to retrieve the
temperature profiles in 51 53 GHz band. The retrieval process starts with the calculations of and for
two sub-ensembles of radiometric observations separately. Here and . The
measured brightness temperatures at eight specified channel frequencies at Fortaleza, Brazil on 11th April,
2011 at 05:39 and 17:36 UTC are as shown in table 2. The summery of measured brightness temperature in
oxygen band at Belem, Brazil on 26th June, 2011 at 05:30 UTC and on 17th June, 2011 at 17:33 UTC and also
at Alcantara, Brazil on 12th March, 2010 at 06:03 UTC and on 15th March, 2010 at 17:57 UTC.
KEYWORDS: Microwave Radiometer, Temperature profile, Optimal Estimation, Inversion, Brightness
Temperature.
I. INTRODUCTION
Till date the radiosonde observations (RAOBs) are the fundamental method for atmospheric
temperature, wind, and water vapour measurement, in spite of their inaccuracies, cost, sparse temporal
sampling and logistic difficulties [1]. A better technology has been sought for decades, but until now,
no accurate continuous all weather technology has been demonstrated. The highly stable multichannel
radiometer ( MP3000A :Radiometrics Corporation, USA) has the capability of producing temperature
and water vapour profiles ([2]; [3]; [4]; [1]; [5]; [6]; [7]; [8]) within the admissible accuracies.
Applications for this passive radiometric profiling include: weather forecasting and now casting;
detection of aircraft icing and other aviation related meteorological hazards; refractivity profiles;
corrections needed for radio-astronomical studies; satellite positioning and GPS measurements;
atmospheric radiation fluxes; measurement of water vapour density and temperature as they affect
hygroscopic aerosols and smokes.
744 Vol. 7, Issue 3, pp. 743-755

The present studies of temperature profiling are in the scope of CHUVA project (Cloud processes of
the main precipitation systems in Brazil: A contribution to cloud resolving modelling and to the
Global Precipitation Measurement). It aims to investigate the different precipitation regimes in Brazil
in order to improve remote sensing precipitation estimation, rainfall ground validation and
microphysical parameterizations of the tri-dimensional characteristics of the precipitating clouds.
Clouds play a critical role in Earth's weather and climate, but the lack of understanding of clouds has
been long limited scientists' ability of making accurate predictions about weather and climate change.
Climate simulations are sensitive to parameterizations of deep convective processes in the
atmosphere. The weather and climate modelling is improving space and time resolution and to
accomplish this it is necessary to move from cloud parameterization to explicit microphysical
description inside the cloud. Therefore, to improve space-time resolution reducing climate model
uncertainties is necessary to better understand the cloud processes. A3-D microphysical processes
description of the main precipitating system in Brazil can strongly contributes to this matter and thus
will be one of the tasks of the project. Keeping these in view the present authors are intending to get
the temperature profile at three different locations and they are: a) Fortaleza (3
0
S; 38
0
W),b)Belem
(1.46
0
S; 48.48
0
W) and c) Alcantara ( 2.4
0
S; 44.4
0
W) in Brazil by exploiting the ground based
radiometric brightness temperatures at desired frequencies. It is to be mentioned here that these
locations are urban coastal city in the north-east of Brazil characterized by tropical climate, which
according to Kppen classification is the type as Equatorial - summer dry. Rainfall and wind regime
are governed mainly by the meridional shift of Inter-tropical Convergence Zone (ITCZ). The ITCZ is
located in its northernmost position, normally from August to October, and intense south-easterly
winds and low rainfall dominate in the area (dry seasonal). On the other hand, when the ITCZ is in its
southernmost position, from March to April, weak south-easterly winds and high rainfall prevail (wet
season).
Temperature profiles can be obtained by measuring the radiometric brightness temperature around 60
GHz .Centering this frequency there lies the continuum which we call as oxygen complex band. The
opacity is larger near the oxygen feature center, limiting emission observation to several meters in
height. Away from the oxygen feature center the opacity is smaller and emission can be observed at
increasing height. Since local temperature contributes to emission intensity, temperature profiles can
be obtained. In this context it is to be mentioned that in this band the emission is almost dependent on
ambient pressure and temperature.
The section III contains the physical principle needed behind the profiling technique. This ultimately
culminates to temperature weighting functions at the said three locations. The next section i.e., section
IV summarises the basic technique for inverting temperature from the measured radiometric
brightness temperatures at or near the oxygen complex. Here, we have chosen the Optimal Estimation
Method, in-spite of the in-built facilities available in the radiometer. Incidentally, the inversion
depends on the historical background of the chosen parameter of the said location. As the campaign
has been performed at three different locations of Brazil it is reasonably taken granted to reconstruct
the new background taking help of data taken from BADC (British Atmospheric data Center). This is
elaborated in the section V. The results obtained are summarised in section VI and the present work
ends with discussions and conclusions in section VII.
II. INSTRUMENT
The The Radiometrics ground based WVP-3000A portable water vapour and temperature profiling
radiometer measures the calibrated brightness temperature from which one can derive profiles of
temperature, water vapor, and limited resolution profiles of cloud liquid water from the surface to
approximately 10 km. The detailed descriptions of the system are given by[1]. A short summary of
the instrument characteristics is given here. The noticeable characteristics of the system include a
very stable local oscillator, an economical way to generate multiple frequencies and the multi-
frequency scanning capability ([5]). The radiometer system consists of two separate subsystems in the
same cabinet which shares the same antenna and antenna pointing system. A highly stable synthesizer
act as local oscillator and allows tuning to a large number of frequencies within the receiver
bandwidth. The water vapour profiling subsystem receives thermal emission at five selected
frequencies within 22 30 GHz [9]. The temperature profiling subsystem measures sky brightness
745 Vol. 7, Issue 3, pp. 743-755

temperature within 51-59 GHz. An inbuilt infrared thermometer is provided with the said radiometer
to observe the presence of cloud and also measures the cloud base height [10]. Here in this paper, we
will restrict ourselves only to derive the temperature profiles at different places of Southern latitude as
mentioned earlier, under a collaborative CHUVA Project implemented at Instituto Nacional de
Pesquisas Espaciais - INPE, Brazil, at an initial stage. The salient characteristics of the radiometer are
shown in Table 1.
Table 1. Characteristics of the Radiometer
Frequencies(GHz) for water vapor and liquid
water profiling
22.234,23.035,23.835,26.235,30.00
Frequencies(GHz) for temperature profiling 51.248, 51.76, 52.280, 52.804, 56.02, 56.66, 57.288,
57.964
Absolute accuracy(K) 0.5
Sensitivity(K) 0.25
FWHP bandwidth(deg) 2.2 2.4
Gain (dB) 36 - 37
Side lobes(dB) < - 26
III. GENERAL PHYSICAL PRINCIPLE
The scalar form of Radiative Transfer Equation is remarkably simple in Rayleigh-Jeans limit and is
considered to be sufficient to the large majority of microwave applications. The radiative intensity
down- welling from the atmosphere and expressed in an equivalent brightness temperature can be
written as ([11])
(1)
Here, is the cosmic background radiation. The attenuation coefficient is a function of different
meteorological parameters.
The radiation is a nonlinear function of the required quantities and we linearise the expression around
a suitably chosen first guess, such as a climatological mean. We describe changes in the brightness
temperature around the first guess by means of a weighting function which expresses the sensitivity of
to the variation of the humidity or the temperature around their initial values
for a certain frequency and elevation angle . Here, stands for water vapour, for temperature,
for liquid water and for ambient atmospheric pressure respectively. However, the weighting
function analyses for humidity and temperature by ([12]) showed that the temperature
weighing function is given by
(3)
To explain more clearly about the weighting function, we take the temperature weighting function
(km
-1
). If we have a change in over a height interval (km), the brightness temperature
response to this change is where is called the height average of over the
height interval . The weighting functions are determined from the height profile of attenuation
coefficients at different frequencies. However equation (1) and its Rayleigh-Jeans approximation are
well discussed by [13] and its more general form including scattering is discussed by [14].
Information on meteorological variables may be obtained from measurements of radiometric
brightness temperature as a function of and/or . Equation (1) is used: a) in forward model
studies in which the relevant meteorological variables are obtained by radiosonde sounding, b) in
inverse problem and parameter retrieval applications in which meteorological information is inferred
from measurements from radiometric brightness temperature , c) in system modelling studies in
746 Vol. 7, Issue 3, pp. 743-755

determining the effects of instrument noise on retrieval and optimum measurement ordinates such as
and/or ([15]).
However, for the sake of clarity the temperature weighting function at Fortaleza, Belem, and
Aclantara, Brazil is shown in Figure 1a - 1c.

Figure 1a. Temperature weighting function at Fortaleza at 17.56UTC. The derived median values of daily
radiosonde data are taken into consideration during the month April, 2011, for the purpose.

Figure 1b. Temperature weighting function at Belem at 17.33 UTC. The derived median values of daily
radiosonde data are taken into consideration during the month June, 2011, for the purpose.

Figure 1c. Temperature weighting function at Alcantara at 17.57 UTC. The derived median values of daily
radiosonde data are taken into consideration during the month March, 2010, for the purpose.
These figure show that the variation of brightness temperature with height occurs at 51 53 GHz but
on the other hand the constancy of brightness temperature with height at 56 57 GHz is noticeable.
This suggests that the measurement of temperature at a certain place by a ground based radiometer
may provide good result by exploiting the 56 57 GHz band. In this band we have used four
frequencies for the purpose. But to extend our study we have also made an attempt to retrieve the
temperature profiles in 51 53 GHz band which will be presented in the subsequent sections.
IV. INVERSION TECHNIQUE
For the purpose of formulating the inverse problem in this particular context i.e., retrieving the
temperature profile at three specified locations of Southern latitude we have purposely set the
radiometer towards zenith, direction. Hence, equation (2) can be rewritten retaining only the
temperature term as
(4)
Here, (refer to equation 4) the integral equation is linear in assuming is independent
747 Vol. 7, Issue 3, pp. 743-755

of .The non-linear form is not commonly encountered in the microwave region as it is in the
infrared region. The infrared equivalent of the linear integral equation, for example, involves the
Planck brightness function, which is strongly nonlinear in Temperature in the infrared region but is
approximately linear for most of the microwave region. However, in practice is usually
measured at discrete number of frequencies, and the objective of the inversion technique is to find out
a function that, when substituted in equation (4) will give values of which might
be approximately equal to the measured values [16]. If the integral of equation (4) is approximated as
a summation over layers each of height , the radiometric temperature at frequencies can be
written as
(5)
Here, in our study, we have chosen four closely spaced frequencies 56 58 GHz in the lower shoulder
of the 60 GHz oxygen spectrum and thereafter frequencies 51 53 GHz, for the purpose. For the sake
of simplicity we write the equation (5) in a more compact form as
(6)
Where, and are the vectors of dimension and respectively and is an (n matrix.
The vector represents the observations; is a weighting matrix also presumed known and is
the unknown atmospheric temperature profile. To get the good vertical resolution we have
considered . But with this idea, this will produce infinite number of solutions. Now to get rid
of this issue i.e., to make the problem solvable we need a priori information about the character of the
atmosphere for a given geographic location and a given time of year. This information includes the
statistics about the temperature profile, constraints imposed by atmospheric physics, any other
information that, if integrated into the inversion algorithm, would narrow the range of values that
can have. However, the degree of accuracy to which this information is incorporated in the
inversion algorithm depends on the underlying structure of the variety of inversion method. Here, we
have chosen the Optimal Estimation Method purposely for the retrieval method.
The a- priori (statistical) information, in the present context, is comprising of monthly averages of the
vertical profiles of temperature and the constrains imposed by the atmospheres of particular places of
Argentina, Brazil, China, Newzeland, India for the months July through August, 2005.
V. OPTIMAL ESTIMATION METHOD
We consider the brightness temperature as measured by the radiometer is linearly related to the
unknown or sought function . Then equation (6) can be written as
(7)
where is the vector of order and is of order m and is (n matrix and is commonly
known as kernel or weighting function of the sought function ). In practice we cannot measure the
true exactly because of experimental error which may include both measurement error and
modeling error. However, our main purpose is to achieve successful retrieval of an unknown vector
using observations with . The key factor is to supplement the observations with sufficient
a priori information for regularizing the ill-posed problem. This a priori information is the mean
profile and its covariance matrix where
(8)
It is to be mentioned here that if represents the atmospheric temperature profile , then the
representative ensemble of radiosonde measured temperature profiles can provide and its
covariance . We also assume that the error vector has a zero mean and is statistically independent
of , but the error covariance matrix is known.
748 Vol. 7, Issue 3, pp. 743-755

This optimal estimation method has the advantage of finding the most likely value of , based on the
combination of a priori information and a real measurement through along with the
associated covariance matrix. The optimal estimate of may be obtained by generalizing the method
to vectors as and , where D is exact solution such that , the identity matrix ([17];
[18]; [19]; [20]).
The solution of equation (6) is given by ([16])
(9)
VI. ANALYSES AND RESULTS
The relationship between the measurements, represented by the - dimensional vector
measurement , and the quantities to be retrieved, represented by the dimensional profile vector
may be expressed as
(10)
This relationship is satisfied by infinite number of profile vectors for a finite particular set of y. Thus
by applying inversion method to obtain an exact solution x, it must be constrained which is the
statistical ensemble of a large number of historic radiosonde profiles as mentioned earlier. Now, the
modification of equation (9) is done to get a better approximation and minimization of experimental
error as well and is expressed as,

Here, contains both measurement errors and the errors originated due to the assumptions and
approximations associated with models. The n X 1 matrices , , and denote the a- priori
data set, retrieved temperature profile vector and radiosonde measurements respectively, being carried
out at Fortaleza, Brazil on 11
th
April, 2011 at 05:39 and 17:36 UTC. A similar approach (as discussed
earlier) being carried out using a- priori data set, radiometric measurements at specified frequencies
and radiosonde measurements respectively, at Belem, Brazil on 26th June, 2011 at 05:30 UTC and on
17
th
June, 2011 at 17:33 UTC. And, following the modified equation (equation 11) of optimal
estimation method, the entire measurements being repeated for Alcantara, Brazil on 12
th
March, 2010
at 06:03 UTC and on 15
th
March, 2010 at 17:57 UTC, so as to approximate and validate the model.
Here, is the mean and covariance matrix of . represents m X m error covariance matrix
associated with m-dimensional radiometric observations. The Lagrangian Multiplier , is basically a
positive real quantity to be determined empirically. Depending upon the band of frequencies and the
time at which the observations were carried out, it can take up the values ranging from 1 to 200.
The profile vector representing statistical ensemble of temperature is a 6 X 1 matrix having elements
exactly at the vertical coordinates of 0.351 km, 0.6866 km, 1.4327 km, 3.1005 km, 5.862 km and
7.613 km forming has 8 elements (1 x 8 matrix) each at the eight specified frequencies.
For faithful analysis this 1 x 8 matrix is subdivided into two sub-ensembles each of (1 x4). The
weighting functions associated with was calculated analytically. These weighting functions
are then normalized to unit maxima for different frequencies separately. The experimental errors
allied to radiometric observation forming 1 x 4 matrices for each segment, assuming zero mean error.
These errors can be assumed rather easily and correctly by calculating the elements of
corresponding to each radiometric observation and then finding their lowest possible
ratio. The range of these errors lies within -1 to +1. The retrieval process starts with the calculations
of and for two sub-ensembles of radiometric observations separately. Here and
. The measured brightness temperatures at eight specified channel frequencies at
Fortaleza, Brazil on 11
th
April, 2011 at 05:39 and 17:36 UTC are as shown in table 2. The summery of
measured brightness temperature in oxygen band at Belem, Brazil on 26
th
June, 2011 at 05:30 UTC
749 Vol. 7, Issue 3, pp. 743-755

and on 17
th
June, 2011 at 17:33 UTC and also at Alcantara, Brazil on 12
th
March, 2010 at 06:03 UTC
and on 15
th
March, 2010 at 17:57 UTC are tabulated in table 3 and 4 respectively. The approximated
results of temperature retrieval are shown from figure 2 to figure 7 separately for the three places of
choice.
Table 2. Summary of Brightness Temperature Measured in Oxygen Band at Fortaleza
Date &
Time
Brightness
Temperature
in K
Frequency in GHz
51.248 51.76 52.28 52.804 56.02 56.66 57.288 57.964
11/4/2011
5:39
124.092 140.774 166.392 199.965 293.843 294.943 295.669 296.053
11/4/2011
17:36
137.786 152.597 176.91 207.676 294.534 296.245 296.477 297.553
Table 3. Summary of Brightness Temperature Measured in Oxygen Band at Belem

Table 4. Summary of Brightness Temperature Measured in Oxygen Band at Alcantara
In Figures 2-7, we plot the statistical difference between retrieved air temperature profile using
brightness temperatures measured by scanning radiometers and RAOBs measured air temperature
profile, the mean value (BIAS), and the root mean square (RMS) of the difference between retrieved
estimation and RAOBs measurements at the specified dates and times. Figure 3 shows highest air
temperature profile retrieval accuracy comparable with the radiosonde observation being observed for
Fortaleza, Brazil on 11
th
April, 2011 at 17:36 UTC. The RMS is smaller than 1.0 K up to 8 km; the
only exception is around 6 km (1.5 K), while the BIAS does not exceed 0.7 K except around 6 km.
While at Belem, Brazil, figure 5 clearly suggest that the highest air temperature profile retrieval
accuracy comparable with the radiosonde observation found to be on 17
th
June, 2011 at 17:33 UTC.
The RMS is smaller than 1.0 K up to 5 km and beyond this it is around 1.6 K, while the BIAS does
not exceed 1.0 K, except above 5 km (1.3 K). A similar result is observed for Alcantara, Brazil on
15
th
March, 2010 at 17:57 UTC (Figure 7). The RMS is smaller than 1.0 K up to 8 km; the only
exception is around 6 km (1.2 K), while the BIAS does not exceed 0.6 K except around 6 km. On the
other hand, in case of Fortaleza particularly, retrieval accuracy slightly degraded with observable
RMS errors are varying with a maximum up to 2.5 K, while the BIAS does not exceed 2.5 K on 11
th

April, 2011 at 17:36 UTC for retrieved profile (being derived with BT of lower frequency sub-
ensemble). But it is effected by fairly high RMS (2.0 K to 6.0 K) and BIAS (-4.0-5.0 K) on 11
th
April,
2011 at 05:39 UTC (Figure 2). This is might be related to the temperature inversion around 0.7 km,
but also to the lower surface temperature during local dawn time. While at Belem, Brazil we observed
that the retrieval accuracy is affected by observably high RMS errors varies maximum up to 2.0 K and
BIAS does not exceed 1.6 K on 26
th
June, 2011 at 05:30 UTC (Figure 4). This is might be related to
the fairly lower temperature difference between two subsequent layers near surface, during morning
UTC. We noticed a similar result for Alcantara with retrieval accuracy is effected by fairly high RMS
(1.0 K to 2.0 K) and BIAS (-1.5-1.5 K) on 12
th
March, 2010 at 06:03 UTC (Figure 6). This
observation is also valid for the retrieved profile when derived using the brightness temperatures of
Date &
Time
Brightness
Temperature
in K
Frequency in GHz
51.248 51.76 52.28 52.804 56.02 56.66 57.288 57.964
26/6/2011
5:30
115.899 133.291 160.164 195.107 294.612 296.093 295.933 296.324
17/06/2011
17:33
124.305 141.799 167.661 200.014 296.673 297.739 299.324 299.718
Date &
Time
Brightness
Temperature
in K
Frequency in GHz
51.248 51.76 52.28 52.804 56.02 56.66 57.288 57.964
12/3/2010
6:03
123.084 136.897 162.706 198.142 296.006 296.065 296.852 297.793
15/3/2010
17:57
124.421 138.322 163.987 197.55 296.63 297.695 298.022 299.08
750 Vol. 7, Issue 3, pp. 743-755

lower frequency sub-ensemble. The following figures show the comprehensive inaccuracies happen at
05:39 UTC. Statistics confirm that at Fortaleza as well as Alcantara, the results are in good agreement
during afternoon when the retrieval method being carried out with the brightness temperatures for
upper frequency sub-ensembles. But this is partly supported by the results at Belem. Here also
statistically fairly high accuracy being observed during afternoon Universal Time Coordinates [UTC],
when the retrieval method is being carried out not only with the brightness temperatures for upper
frequency sub-ensemble but also with the BTs of lower frequency channels in the oxygen band
separately.

(a) (b)
Figure 2. Statistical difference between retrieved air temperature profiles when retrieved with BTs of two
frequency sub-ensembles separately and temperature profile measured by RAOBs for Fortaleza, Brazil on 11
th

April, 2011 at 05:39 UTC are shown in red (upper frequencies), blue (lower channels) and black respectively
(a). The corresponding measured BIAS is shown in black and blue; while the RMS is shown in red and magenta,
while retrieved using BTs measured for both upper and lower frequency sub-ensembles respectively (b).

(a) (b)
frequency sub-ensembles separately and temperature profile measured by RAOBs for Fortaleza, Brazil on 11
th

April, 2011 at 17:36 UTC are shown in red (upper frequencies), blue (lower channels) and black respectively

(a) (b)
frequency sub-ensembles separately and temperature profile measured by RAOBs for Belem, Brazil on 26
th

June, 2011 at 05:30 UTC are shown in red (upper frequencies), blue (lower channels) and black respectively (a).
The corresponding measured BIAS is shown in black and blue; while the RMS is shown in red and magenta,
751 Vol. 7, Issue 3, pp. 743-755

(a) (b)
frequency sub-ensembles separately and temperature profile measured by RAOBs for Belem, Brazil on 17
th

June, 2011 at 17:33 UTC are shown in red (upper frequencies), blue (lower channels) and black respectively (a).
The corresponding measured BIAS is shown in black and blue; while the RMS is shown in red and magenta,

(a) (b)
frequency sub-ensembles separately and temperature profile measured by RAOBs for Alcantara, Brazil on 12
th

March, 2010 at 06:03 UTC are shown in red (upper frequencies), blue (lower channels) and black respectively

(a) (b)
frequency sub-ensembles separately and temperature profile measured by RAOBs for Alcantara, Brazil on 15
th

March, 2010 at 17:57 UTC are shown in red (upper frequencies), blue (lower channels) and black respectively
VII. DISCUSSIONS AND CONCLUSIONS
The optimal estimation method is a kind of method that combines the observations with a background
taken from numerical weather prediction (NWP) model outputs. The assumed error characteristics of
both are taken into account ([21]). However, the 1DVAR approach (One- Dimensional Variation
Technique was demonstrated to be advantageous over methods using background from statistical
climatology ([22]). In fact, as background information, 1DVAR uses a forecast state vector, which is
usually more representative of the actual state than a climatologic mean. A comparative analysis
752 Vol. 7, Issue 3, pp. 743-755

between a variety of retrieval methods applied to ground-based observations from conventional
microwave radiometers (such as MWRP) indicated that the 1DVAR technique outperforms the other
considered methods, these being based on various kinds of multiple regression and neural network.
Thus, it seemed convenient to couple the sensitivity of millimetre-wave radiometry with the
advantages of the 1DVAR technique for the retrieval of temperature and humidity profiles [23]. While
developing this technique the standard notation is used as used by [24] and indicated with B and R the
error covariance matrices of the background and observation vector y, respectively. In addition, the
forward-model operator (i.e., radiative transfer model) with F(x) was used. Thus, the technique
adjusts the state vector x from the background state vector to minimize the following cost function
(12)
Here and represents the matrix transpose and inverse. The radiometric noise,
representativeness, and forward-model errors all contribute to the observation-error covariance R. The
minimization is achieved using the LevembergMarquardt method; this method was found to improve
the convergence rate with respect to the classic GaussNewton method ([21]) by introducing a factor
that is adjusted after each iteration depending on how the cost function J has changed; thus, calling
K the Jacobian matrix of the observation vector with respect to the state vector, the solution
(13)
is iterated until the following convergence criterion is satisfied
(Observations)
(14)
Here, and (Obs.) indicates the number of observations (i.e., the
dimension of y).
The GSR was first deployed during the Water Vapour Intensive Operational Period (WVIOP, March
April 2004) and, later, during the Radiative Heating in Underexplored Bands Campaign (RHUBC,
FebruaryMarch 2007), both held at the Atmospheric Radiation Measurement (ARM) Programs
North Slope of Alaska (NSA) site in Barrow, Alaska ([25] ).
The state vectors that used by [23] are profiles of temperature and total water (i.e., total of specific
humidity and condensed-water content ([26]). The choice of total water has the advantages of
reducing the dimension of the state vector, enforcing an implicit correlation between humidity and
condensed water, including a super-saturation constraint. Moreover, the introduction of natural
logarithm of total water creates error characteristics that are more closely Gaussian and prevents
unphysical retrieval of negative humidity. The background-error covariance matrices for both
temperature and humidity profiles may be computed from a set of simultaneous and co-located
forecast-RAOB data (both in clear and cloudy conditions). This calculation of inherently includes
forecast errors as well as instrumental and representativeness errors from the radiosondes. The
radiosonde instrumental error is assumed to be negligible compared with the representativeness error,
which consists of the error associated with the representation of volume data (model) with point
measurements (radiosondes). The matrix including these terms seems appropriate for the
radiometric retrieval minimization; since the grid cell of the NWP model is much larger than the
radiometer observation volume, the latter can be assumed as a point measurement compared with the
model cell, similar to radiosondes. It may be assumed that the matrix estimated for humidity to be
valid for control variable total water, since no information on the background cloud-water error
covariance was available. This assumption is strictly valid during clear sky conditions only, while it
underestimates the background error in cloudy conditions. The implications are that, under cloudy
conditions, humidity retrieval would rely more on the background and less on measurements than
would be possible by adopting a matrix that includes both humidity and liquid-water errors.
However, considering the infrequent and optically thin cloudy conditions encountered during
RHUBC, it is understood that this assumption does not affect results significantly.
The observation vector is defined as the vector of measured by GSR at a number of elevation
angles, plus the surface temperature and humidity given by the sensors mounted on the lowest level
753 Vol. 7, Issue 3, pp. 743-755

(2m) of the meteorological tower. The observation error covariance matrix may be estimated using
the GSR data taken from the WVIOP, following the approach by [21]. The forward model F( is
provided by the NOAA microwave radiative-transfer code ([27])which also provides the weighting
functions that were used to compute the Jacobians with respect to temperature, humidity, and liquid
water. The typical errors with respect to band-averaged are within 0.1 K and were accounted for in
the forward-modelling component of the observation error.
The 1DVAR retrieval technique and the settings described in the previous section were applied to
GSR data collected during the three-week duration of RHUBC. These observations were found to be
consistent with simultaneous and co-located observations from the other two independent 183-GHz
radiometers and with simulations obtained from RAOBOs ([28]), generally within the expected
accuracy. As a comparison, the background NWP profiles were shown that were used as a first guess
and the in situ observations from the radiosonde.
Concerning the temperature profile, it is noted that in this case, the NWP forecast is in good
agreement with the RAOB, particularly in the atmospheric layer from 0.5 to 3.0 km. Conversely, in
the upper part of the vertical domain (35 km), the NWP forecast shows about 12-K bias with
respect to the RAOB, while in the very first layer (00.5 km), it differs from RAOB by more than 10
K. Conversely, the 1DVAR retrieval agrees better with the RAOB in the lowest levels, while for the
upper levels, the retrieved temperature tends to lie over the NWP background.
As for the humidity, again it is noted that the NWP forecast captures well the vertical structure,
although with lower resolution, except for the first 500 m, where the 1DVAR retrieval shows a much
better agreement with the RAOB.
The analyses show that at Fortaleza, the upper frequency band i.e., 56-58 GHz consisting of four
frequency channels built in the said radiometer provides a good agreement regarding temperature
profiling, with the RAOBs. But on the other hand, at Belem the lower frequency band i.e., 51-53
GHz consisting of four frequency channels built in the said radiometer provides a good agreement
regarding temperature profiling, with the RAOBs. It may also be noted that at Alcantara, the lower
frequency band show good agreement, in this regard. All these agreement happen to be good during
afternoon and the situation is worst during midnight (UTC). It is also observed that as we move
towards the higher latitude the possibility of getting good agreement with RAOBs upper air data lies
in favor of using the higher frequency channels.
VIII. FUTURE SCOPE
In future we are very much interested to use others retrieval method such as Backus Gilbert Synthetic
Averaging Inversion method, neural network method etc. for retrieval of vertical profiles of
atmospheric temperature over the aforesaid three places choice and try to find out the most suitable
and simplified inversion method for retrieving of atmospheric temperature profile both in terms of
accuracy and resolution with minimal operating and instrumental limitations. In future, using the
potentials offered by ground-based multichannel microwave radiometry we are very much interested
to apply the Modified Optimal Estimation Method for continuous vertical profiling of humidity as
well as water vapour in order to validate the model.
REFERENCES
[1].Solheim, F., Godwin, J. R., Westwater, E.R., Han,Y., Keihm, S. J., Marsh, K., Ware, R., (1998)
Radiometric profiling of temperature, water vapor and cloud liquid water using various inversion methods,
Radio Science, Vol. 33, No. 2, pp 393-404.
[2].Westwater, E. R., Crewell, S., Matzler, C., (2004) A Review of Surface-based Microwave and Millimeter
wave Radiometric Remote Sensing of the Troposphere, URSI Radio Science Bulletin, No. 310, pp 59-80.
[3].Rocken,C., Johnson, J. M., Neilan, R. E., Cerezo, M., Jordan , J. R., Falls, M. J., Nelson , L. D., Ware, R. H.,
Hayes, M., (1991) The Measurement of Atmospheric Water Vapor: Radiometer Comparison and Spatial
Variations, IEEE Transactions on Geoscience and Remote Sensing, Vol. 29, Issue 1, pp 3-8.
[4].Westwater, E.R., (1993) Ground-based Microwave Remote Sensing of Meteorological Variables, in:
Michael A. Janssen (Eds.), Atmospheric Remote Sensing by Microwave Radiometry. J. Wiley & Sons, Inc.
(New York), pp.145-213.
754 Vol. 7, Issue 3, pp. 743-755

[5].Westwater, ED R., Han, Y., Solheim, F., (2000) Resolution and accuracy of a multi-frequency scanning
radiometer for temperature profiling, in: P. Pampaloni and S. Paloscia (Eds.), Microwave Radiometry and
Remote Sensing of the Earths Surface and Atmosphere. VSP 2000 (Netherland), pp. 129-135.
[6].Ware, R., Solheim , F., Carpenter, R., Gueldner, J., Liljegren, J., Nehrkorn, T., Vandenberghe, F., (2003)
A multichannel radiometric profiler of temperature, humidity and cloud liquid, Radio Science, Vol. 38, No. 4,
8079, pp 4401-4413.
[7].Karmakar, P. K., Maiti, M., Sett, S., Angelis, C. F., Machado, L.A.T., (2011a) Radiometric Estimation of
Water Vapor Content over Brazil, Advances in Space Research, Vol. 48, No. 9, pp 1506-1514.
[8].Karmakar, P. K., Maiti, M., Calheiros, A.P.J., Angelis, C.F., Machado, L.A.T., Da Costa, S.S., (2011 b)
Ground based single frequency microwave radiometric measurement of water vapour, International Journal
of Remote Sensing, Vol. 32, No. 23, pp 8629-8639.
[9].Mondal, S., Pradhan, A.K., Karmakar, P.K., (2014) water vapour profiling at southern latitudes by
deploying microwave radiometer, International Journal of Advances in Engineering & Technology, Vol. 6,
Issue 6, pp 2646-2656.
[10].Karmakar, P.K., (2013) Ground-Based microwave radiometry and remote sensing, CRC press, Boca Raton,
FL (USA), pp 224.
[11].Askne, J. I. H., Westwater, E. R., (1986) A Review of Ground-Based Remote Sensing of Temperature and
Moisture by Passive Microwave Radiometers, IEEE Transactions on Geoscience and Remote Sensing, GE-
24(3), pp 340-352.
[12].Canavero, F. G., Einaudi, F., Westwater , E. R., Falls, M. J., Schroeder, J. A., Bedard Jr, A. J., (1990)
Interpretation of ground-based radiometric observations in terms of a gravity wave model, Journal of
Geophysical Research, Vol. 95, No. D6, pp 7637-7652.
[13].Goody, R. M., Yung, Y. L., (1995). Atmospheric Radiation: Theoretical Basis, 2
nd
Edition, Oxford
University Press (USA).
[14].Gasiewski, A. J., (1993) Microwave Radiative Transfer in Hydrometeors, in: Michael A. Janssen (Eds.),
Atmospheric Remote Sensing by Microwave Radiometry. J. Wiley & Sons, Inc (New York), pp.1-36.
[15].Westwater, Ed. R., Crewell, S., Matzler, C., Cimini, D., (2005) Principles of Surface-based Microwave
and Millimeter wave Radiometric Remote Sensing of the Troposphere, Quaderni Delle Societa Italiana Di
Elettromagnetismo, Vol. 1, No. 3, pp 50-90.
[16].Ulaby, F. T., Moore, R. K., Fung, A. K., (1986) Microwave Remote Sensing-Active and Passive, vol. 3,
Artech House, Inc. (Norwood).
[17].Cimini, D., Shaw, J. A., Han, Y., Westwater, E. R., Irisovi, V., Leuski, V., Churnside, J. H., (2003) Air
Temperature Profile and Air-Sea Temperature Difference Measurements by Infared and Microwave Scanning
Radiometers, Radio Science, Vol. 38, No. 3, 8045, pp 1001-1019.
[18].Westwater, E. R., Snider, J. B., Carlson, A. C., (1975) Experimental Determination of Temperature
Profiles by Ground-Based Microwave Radiometry, Journal of Applied Meteorology, Vol. 14, No. 4, pp 524-
539.
[19].Rodgers, C. D., (1976) Retrieval of Atmospheric Temperature and Composition from Remote
Measurement of Thermal Radiation, Reviews of Geophysics, Vol. 14, No. 4, pp 609-624.
[20].Rodgers, C. D., (2000) Inverse Methods for Atmospheric Sounding: Theory and Practice, i-xvi. World
Scientific Publishing (Singapore).
[21].Hewison, T., (2007) 1D-VAR retrievals of temperature and humidity profiles from a ground-based
microwave radiometer, IEEE Transactions on Geoscience and Remote Sensing, Vol. 45, No. 7, pp 21632168.
[22].Cimini, D., Hewison, T. J., Martin, L., Gldner, J., Gaffard, C., Marzano, F. S., (2006) Temperature and
humidity profile retrievals from groundbased microwave radiometers during TUC, Meteorology Z., Vol. 15,
No. 5, pp 4556.
[23].Cimini, D., Westwater, Ed R., Gasiewski, A. J., (2010) Temperature and Humidity Profiling in the Arctic
Using Ground-Based Millimeter-Wave Radiometry and 1DVAR, IEEE Transactions on Geoscience and
Remote Sensing, Vol. 48, No. 3, pp 1381-1388.
[24].Ide, K., Courtier, P., Ghil, M., Lorenc, A. C., (1997) Unified notation for data assimilation: Operational,
sequential, and variational, Journal of Meteorological Society of Japan, Vol. 75, No. 1B, pp 181189.
[25].Ackerman, T. P., Stokes, G. M., (2003) The atmospheric radiation measurement program, Physics
Today, Vol. 56, No. 1, pp 3844.
[26].Deblonde, G., English, S., (2003) One-dimensional variational retrievals for SSMIS simulated
observations, Journal of Applied Meteorology, Vol. 42, No. 10, pp 14061420.
[27].Schroeder, J. A., Westwater, E. R., (1991) Users guide to WPL microwave radiative transfer software,
Nat. Ocean. Atmos. Admin., Boulder, CO, NOAA Technical Memorandum, ERL WPL-213.
[28].Cimini, D., Nasir, F., Westwater, E. R., Payne, V. H., Turner, D. D., Mlawer, E. J., Exner, M. L., Cadeddu,
M. P., (2009) Comparison of ground based millimeter-wave observations and simulations in the Arctic
winter, IEEE Transactions on Geoscience and Remote Sensing, Vol. 47, No. 9, pp 30983106.
755 Vol. 7, Issue 3, pp. 743-755

AUTHORS
Ayan Kanti Pradhan (b. 1980) is an M.Sc. of the Vidyasagar University, West Bengal,
India and currently pursuing his research work under the supervision of Dr. Pranab Kumar
Karmakar in the Department of Radiophysics & Electronics, University of Calcutta. His
field of research includes vertical profiling and estimation of atmospheric constituents
especially ambient temperature deploying ground based microwave radiometer as well as
employing Radiosonde data.
Since 2010, till date he acted as Assistant Professor in the Department of Electronics,
Acharya Prafulla Chandra College, West Bengal, India, where he is involved in teaching in both the Post-
graduate and under-graduate classes in areas such as microwave devices, communication, control system and
instrumentation.
Subrata Mondal (b. 1984) received the B-Tech degree in Electronics & Communication
Engineering from Murshidabad College of Engineering & Technology, W.B.U.T in 2007
and M-Tech degree in Radiophysics & Electronics from Institute of Radiophysics &
Electronics, C.U in 2009. He is currently working as an Assistant professor in the
department of Electronics & Communication Engineering at Dumkal Institute of
Engineering and Technology, W.B.U.T and pursuing research work leading to PhD degree
under the supervision of Dr. P.K. Karmakar at Institute of Radiophysics & Electronics,
C.U., India.
He taught courses on electromagnetics, remote sensing and wave propagation theory. His technical interest
include passive and active remote sensing, radiative transfer, microwave engineering.
Pranab Kumar Karmakar is currently pursuing his research work basically in the area of
modeling of integrated water vapour and liquid water in the ambient atmosphere. Ground
based microwave radiometric remote sensing is his special area of interest. This includes
the vertical profiling of thermodynamic variables. He is presently involved in the research
and teaching in the Post-graduate classes of Institute of Radiophysics and Electronics,
University of Calcutta.
Since his joining in 1988 in the University of Calcutta, India, Dr. Karmakar published his
noteworthy outcomes of researches over tropical locations in different international and national journals of
repute. All these are culminated into a book entitled Microwave Propagation and Remote Sensing: Atmospheric
Influences with models and Applications published by CRC Press in 2012.
He had been awarded the International Young Scientist award of URSI in 1990. Also, he had been awarded the
South-South Fellowship of TWAS in 1997. He acted as visiting scientist in Remote Sensing Laboratory,
University of Kansas, USA; Centre for Space Sciences, China, and National Institute for Space Sciences, Brazil.

International Journal of Advances in Engineering & Technology, July 2014.
756 Vol. 7, Issue 3, pp. 756-764

DEVELOPMENT AND EVALUATION OF TROLLEY-CUM-BATCH
DRYER FOR PADDY
Mohammed Shafiq Alam and V K Sehgal
Department of Processing and Food Engg.,
Punjab Agricultural University, Ludhiana, Punjab, India

ABSTRACT
The tractor trolley available with the farmer was converted into a batch dryer by making suitable modifications. The
trolley was modified in such a way that it can hold one tonne of paddy per batch. A set of three heaters (8 kW each)
was provided to heat the air for drying and a controllable system was provided to control the air temperature. The
performance of the dryer was evaluated by drying paddy and found that the system was capable of drying one tonne
of paddy in 100 minutes consuming 56 kW-hr of energy. The operating cost for moisture reduction of one tonne of
paddy per batch by 5 per cent i.e. 18 to 13% (w.b.) was approximately Rs. 250. Among the nine drying models fitted
to the experimental data the Verma model was found to be the best to represent the drying behaviour of paddy in the
trolley-cum-batch dryer.
KEYWORDS: Dryer, Energy, Moisture content, Paddy, Trolley.
I. INTRODUCTION
The drying of grains followed by scientific storage practice is universally accepted to be the safest and
most economical practice of preserving quality and quantity of grains. The high moisture content crop at
the time of harvest is liable to be infested with moulds, fungus and insect attack during a short period of
storage. The safe storage moisture content for most of the crops is between 8-10% in case of grains is 14
percent. Since most of the crops that need drying are harvested in October, the farmer is under great
pressure to sow the next crop in November. Under such circumstances, a suitable drying system is and 5-
6% in case of fruits and vegetables. For paddy milling, the appropriate moisture content an integral part of
farm mechanization. In order to store the produce, a farmer has to opt for safe drying methods. It is
neither desirable nor practically feasible for the farmer to dry his crop in the open sun which is a
highly time consuming, sunshine dependent and inefficient method of drying. A mechanical dryer helps
to reduce the losses and drying time as well as to improve and stabilize the quality of the produce. But in
the absence of a suitable and cheap mechanical dryer he has no option but to sell his crop and be deprived
to take the advantage of higher price of dried/dehydrated product during off-season.
Efforts have been made consistently by several researchers to develop a batch drying system that has a
low capital and operating cost but were of low capacities which do not fulfill the requirement of the
farmers[1][2][3][4][5]. Keeping in view the problem faced by the farmers regarding high moisture paddy
especially if rains persist an attempt has been made by the authors to modify a tractor trolley to a multi
crop batch dryer which runs on electricity and was evaluated for its performance for drying of paddy in
bulk and assessing its quality in terms of Head yield.

757 Vol. 7, Issue 3, pp. 756-764

II. MATERIALS AND METHODS
A tractor trolley available with the farmer has been converted to dryer. The various components of the
existing trolley dryer such as plenum chamber, holding bin, heat exchanger and hot air supply system
were modified /improved to make trolley dryer a multi crop dryer with temperature control system. The
schematic view of the developed trolley-cum-batch dryer is presented in Figure 1. The trolley was
equipped with hydraulic jack for easy unloading. The air, heat and energy requirements of the dryer have
been calculated as per the procedure suggested [6] [7]. A trolley has been constructed in such a manner
that the trolley can be used for conventional transport purposes when it is not being used as a dryer. The
dryer consists of a common trolley modified into a drying chamber by providing a perforated surface
above the trolley floor. The total size of the screen on which the material is to be placed for drying is
4267.2 x 2133.6 mm where as the perforated screen is of the dimension of 4054 x 2042 mm. An axial
flow blower powered by 3.73kW electric motor is attached in the center of the trolley floor along with
heating system (electric heaters), in such a way that it sucks the heated air and distribute uniformly
through the plenum chamber and the air velocity through the plenum chamber was kept nearly in the
range of 1 0.2 m/s. Commodity that needs drying is to be placed on top of the perforated surface of the
trolley. Electric heaters (a set of 3 heaters, each of 8 kW) heats up the air and thus the air is forced
through the drying bed. The total energy consumed was worked out by noting down the input power of
driving fan and thermal energy consumed based on the time required for the drying operation. The energy
meter was used for recording the energy consumption during drying of paddy and the economics was
calculated based on total energy consumed and labour engaged for the operation. Three replication of
each data were taken and average values were taken for the analysis.

Figure1. Schematic view of developed trolley-cum-batch dryer

Figure 1. Schematic view of developed trolley cum batch dryer

8
LEGEND

1. Perforated sheet
2. Blower (axial flow)
3. Electric heater
4. Control panel
5. Trolley with hydraulic
jack for unloading
6. Air heating unit
7. Drying chamber
8. Test bins

All dimensions are in mm.
A B
C
D
E
8
758 Vol. 6, Issue 3, pp. 1049-1054

2.1 Evaluation of Trolley-Cum-Batch Dryer
The modified / improved multi crop trolley dryer was tested for its performance and evaluated for paddy.
The moisture content and temperature variations were noted by using specially designed test bins which
were placed at five different locations i.e. one at center (E) and four at each corner (A, B, C, D) of
perforated platform of trolley drier (Figure 1). The observations were recorded from the top and bottom
layer of the grain bed. Temperature of air in the plenum chamber was also observed at each of the five
locations at every 15 minutes interval by a laser temperature thermometer having least count of 0.1C.
The moisture content of paddy was measured in the field by wile-35 moisture meter and was again
measured for accuracy by oven method [8]. In order to avoid non-uniform drying the paddy was stirred at
regular intervals of 30 minutes. The dryer was evaluated for drying capacity, moisture reduction, total
drying time required, specific energy consumption and economics[5][9]
The performance of dryer was evaluated on the basis of EHE, HUF and COP as suggested [6]
t1 t2
Effective heat efficiency (EHE) = --------------- ------ eq 1
t1 tw1
t1 t2
Heat utilization factor (HUF) = --------------- ------ eq 2
t1 t0
t2 t0
Co-efficient of performance (COP) = --------------- ------ eq 3
t1 t0
Where; t1 = drying air temperature, C
t2 = exhaust air temperature C
to = ambient air temperature, C
tw1 = wet bulb temperature of drying air, C
2.2 Modeling of Drying Curve of Paddy in Trolley Dryer
The moisture contents of paddy during thin-layer drying were expressed in dimensionless form as
moisture ratio, that is expressed as (M-Me)/(M0-Me) where M is the moisture content at any time t; M0 is
the initial value; and Me is the equilibrium moisture content. However, (M-Me)/(M0-Me) could be
simplified to M/M0[10][11] as the equilibrium moisture content is too small in comparison to M0.
For mathematical modeling, nine thin layer drying mathematical models were tested to select the best
model for describing the drying curve equation of paddy during drying process in a trolley dryer (Table
1). The non-linear regression analysis was performed using SPSS (Statistical Package for Social Sciences
Version 11.0). The coefficient of determination, r
2
was one of the primary and main criteria for selecting
the best equation to account for variation in the drying curves of dried samples [12]. In addition to
coefficient of determination, the goodness of fit was determined by various statistical parameters such as
reduced chi-square,
2
, root mean square error, RMSE and percent mean deviation modulus, P and was
calculated as suggested [13]. For quality fit, r
2
value should be higher and the error terms
2
, RMSE and P
values should be lower [12]
Table1. Mathematical models used for modeling of drying curve of paddy
Name of the Model Model equation References
Newton MR = Exp (-kt) [14]
Page
MR = Exp (-kt
n
) [15]
Logarithmic MR = a Exp (-kt) + c [16]
Two term MR = a Exp (-k1t) + b Exp (-k2t) [17]
Midilli
MR = a Exp (-kt
n
) + ct [18]
Two-term exponential
MR = a Exp (-kt)+ (1-a) Exp(-kat) [19]
Wang and Singh
MR = 1+at+bt
2
[20]
759 Vol. 6, Issue 3, pp. 1049-1054

Verma
MR = a Exp (-kt)+(1-a) Exp (-ct) [21]
Thomson
t = a ln(MR) + b (ln MR)
2
[22]
2.3 Milling Quality
III. RESULTS AND DISCUSSION
3.1 Drying Of Paddy
The trolley-cum-batch dryer was tested and evaluated at the farm of Punjab Agricultural University,
Ludhiana. The average initial moisture content of paddy was 18% (w.b.), which was reduced to 13%
(w.b.) in approximately 100 minutes. The dryer was evaluated for 18.5 mm bed thickness of paddy
resulting in one tonne/batch holding capacity for reducing the moisture content to safe moisture level for
milling of paddy. The dryer was tested at its full load and was observed that the airflow rate through the
plenum chamber was 496.8m
3
/min and to raise the temperature of drying air from 27C (ambient air) to
50C the dryer took approximately 30 minutes. During testing the maximum temperature attained by the
dryer were 54C and the drying air temperature varied from 50 to 54C throughout drying process.
During experimentation the ambient air conditions in respect of temperature and humidity varied between
27C to 29C and 35% to 40% respectively. The reduction in moisture content with time was measured
by taking samples at regular time intervals from five different locations of the trolley loading platform for
assessing the uniformity in drying (Table 2). The samples were taken from the top as well as bottom layer
to see the variation in temperature and moisture content with time (Figure 2 & Figure 3). It is very clear
from the curve that in the initial stage of drying the moisture reduction is fast in the bottom layer while
there is a little decrease in moisture content from the top layer, the difference of moisture content of
paddy between the bottom and the top layer is caused by the difference of drying temperature of air
flowing through the paddy bed, higher drying temperature at the bottom layer than at the top layer. Also
the relative humidity is another factor that affects the drying rate of paddy. The drying rate from both top
as well as bottom layer stabilized when the dryer temperature reaches approximately 50C. The total time
taken to reduce moisture content up to 13% (w.b.) was found to be 100 minutes while the total electrical
power consumed in reducing moisture content by 5 per cent i.e. from 18% to 13% (w.b.) was
approximately 56 kW-hr. One labour was required to operate the dryer.
3.2 Mathematical Models For Fitting Drying Curves
The average moisture content data (means of moisture content data from five test kits placed at different
locations of trolley dryer) obtained at various time intervals for paddy dried in trolley dryer were
converted to the more useful moisture ratio and then curve fitting computations with the drying time were
done by using the thin layer drying models (Newton, Page, Logarithmic, Two term, Midilli, Two term
exponential, Wang and Singh, Verma, Thomson) as shown in Table 1. The coefficient of correlation and
results of statistical analyses and model constants are depicted in Table 3. All the models showed higher
R
2
(>0.90) except Newton model. The examination of statistical terms showed that the Page, Midilli,
Two term and Verma model experienced maximum R
2
(> 0.98) and minimum error values for paddy
dried in trolley dryer, selecting them the best model representing the experimental data showing minimum
(<1.2) mean relative deviation modulus values (P).
Among the best four models fitted, Verma model noticed minimum error terms (0.009569 of RMSE;
0.00016 of
2
; 1.0 of P) and maximum R
2
(0.9861), supporting their superiority to fit the experimental
data. Thus, the Verma model was judged to be the best model for representing the convective drying
kinetics of paddy in a trolley dryer

760 Vol. 6, Issue 3, pp. 1049-1054

Figure 2. Temperature variation of drying air and paddy in trolley dryer

0
10
20
30
40
50
60
0 20 40 60 80 100 120
Drying time (minutes)
T
e
m
p
e
r
a
t
u
r
e

(
C
)

(
C
)

Ambient air
Plenum chamber
temp of bottom layer of
paddy
temp of top layer of
paddy
S.D varied between: 0.24 to 0.78, based on N=3 replications for
temperature recorded at each drying time

Figure 3. Variation of moisture content at various depth of
grain in trolley dryer
10
11
12
13
14
15
16
17
18
19
20
0 50 100 150
Drying time ( Minutes)

M
o
i
s
t
u
r
e

c
o
n
t
e
n
t

(
%

w
.
b
.
)
Top layer
Bottom layer
761 Vol. 6, Issue 3, pp. 1049-1054

Table2. Moisture content of paddy sample during drying at various testing positions of trolley dryer

Table 3. Statistical results obtained from different thin layer models

3.3 Performance Evaluation Of Trolley Dryer
The modified trolley dryer was evaluated for its efficiency factors i.e. Effective heat efficiency (EHE),
Heat utilization factor (HUF) and Co-efficient of performance (COP). The EHE, HUF and COP of the
trolley dryer evaluated were 40.4, 52 and 51.5 per cent respectively. The operating cost of drying one
tonne of paddy was about Rs. 250 by taking into account the cost of electricity (Rs 4/unit) and the labour
cost. Thus the paddy can be dried safely in a trolley dryer with an acceptable quantity and quality.

Moisture content (% w.b.)
Testing Bin
positions

15 min

30 min

45 min

60 min

75 min

100 min
A
16.4(19.6)
0.13
15.0(17.06)
0.34
14.6(17.0)
0.16
14.3(16.6)
0.26
13.9(16.3)
0.52
13.0 (15.0)
0.17
B
16.24(19.4)
0.18
15.2(17.9)
0.40
14.9(17.6)
0.63
14.7(17.2)
0.58
14.5(16.9)
0.24
13.1(15.1)
0.14
C
16.2(19.3)
0.14
14.7(17.2)
0.38
14.4(16.8)
0.44
14.1(16.4)
0.22
13.8(15.9)
0.28
12.7(14.6)
0.17
D
16.8(20.2)
0.28
15.5(18.3)
0.73
15.1(17.8)
0.45
14.7(17.2)
0.35
14.4(16.8)
0.09
13.1(15.1)
0.12
E
16.6(19.8)
0.22
15.5(18.4)
0.27
15.0(17.7)
0.30
14.7(17.2)
0.15
14.4(16.8)
0.20
13.2(15.2)
0.10
Average 16.5(19.7) 15.2 (17.9) 14.8(17.4) 14.5(16.9) 14.2(16.6) 13.0(15.0)

Initial moisture content of paddy: 18% wet basis (21.95 % dry basis);
The data in parenthesis are moisture content (dry basis)
* Mean of N=3 replications; Bold values are the standard deviation based on N=3 replications

Models Model constants Comparison
criteria

k n k1 k2 a b c R
2
RMSE
2
P
Newton 0.00352 - - - - - - 0.8498 0.031529 0.00116 3.15
Page 0.02220 0.56398 0.982 0.0109 0.000166 1.14
Logarithmic 0.02291 0.72416 0.9716 0.013671 0.000262 1.43
Two term 0.002312 5 0.92513 0.074872 0.9752 0.012769 0.00038 1.17
Midilli 0.98209 0.52252 1.00058 -0.00023 0.9821 0.010867 0.000276 1.14
Two term
exponential
0.05099 0.05422 0.9348 0.020725 0.000601 2.02
Wang & Singh

-
0.00502
0.000025 0.9454 0.019007 0.000506 1.96
Verma* 0.07447 0.10598 0.00187 0.9861 0.009569 0.00016 1.00
Thomson
model
-98.279 753.247 0.9654 6.002119 50.43561 9.08

* Drying model with minimum error terms and maximum R
2

762 Vol. 6, Issue 3, pp. 1049-1054

3.4 Milling Quality Characteristics
The test results indicated that trolley dried samples showed comparatively higher head recovery (62.04%)
than tray-dried samples (60.07%). The total yield obtained from trolley and tray dried samples was 73.43
and 72.47 percent respectively. In comparison to tray dried samples the trolley-dried samples showed
comparatively lower broken percentage (15.50%). This slight improvement in head yield in trolley dried
samples may be due to its proper churning during drying operation thus resulting in uniform drying of the
sample. The paddy samples dried in the trolley dryer and tray dryer experienced non-significant
difference in milling quality at 5% level of significance (Table 3).
IV. CONCLUSIONS
The developed trolley-cum-batch dryer was evaluated for the holding capacity of one tonne/ batch of
paddy and found that the dryer successfully reduced 5 per cent moisture i.e. 18 to 13% (w.b.) in 100
minutes. The operating cost and energy consumption of trolley dryer in drying one tonne of paddy was Rs.
250 and 56 kW-hr respectively. The model developed [22] showed good agreement with the experimental
data. The milling quality of samples dried in trolley dryer was at par to the samples dried in mechanical
tray dryer. Thus, high moisture paddy can be dried in a trolley drier at farm level/grain market to reduce
losses during various post harvest operations.
V. FUTURE RESEARCH DIRECTION
In future, the developed trolley-cum-batch dryer should be tested for drying of red chillies, turmeric
fingers etc. The trolley cum batch dryer should be modified for utilization of solar energy for heating the
drying air.
REFERENCES
[1]. Chancellor, W.J. (1968). A simple grain dryer using conducted heat. Trans of ASAE, 11, 857-62.
[2]. Schuler, R.T., Hirwing, H.J., Hofman, V.L. & Lundstrom, D.R. (1978). Harvesting, Handling and Storage of
Seeds. Inc. Publishers, Madison, Wisconsin, USA. Pp. 45-167.
[3]. Bose, A.S.C., Ojha, T.P. & Maheshwari, R.C. (1980). Drying of paddy in a batch dryer with solar-cum-husk
fired furnace. Proc. Solar Energy Convention, held at Annamalai University. Pp 14-20.
[4]. Sahay, K.M., Saxena, R.P. & Singh, B.P.N. (1981). Development and performance evaluation of husk fired
furnace with continuous dryer for small rice mill. Paper presented at the International Conference on Agric
Engng and Agro industries in Asia at AIT, Bangkok.
[5]. Arora, Sadhna; Sehgal, V.K. & Singh, R. (2000). Evaluation of a farm grain dryer. Intern. J Trop Agric 18(2),
159-163.
[6]. Chakraverty, A. (2000). Post Harvest Technology of Cereals, Pulses and Oilseeds. Oxford and IBH Publishing
Co.
Table 4. Milling quality of paddy

Milling parameters Drying source
Trolley dryer Tray dryer
Husk content (%) 21.91 21.67
Total yield (%) 73.43 72.47
Bran (%) 4.66 5.86
Degree of Polish (%) 5.97 7.48
Head yield (%) 62.04 60.07
Brokens (%) 15.50 17.11
CD at 5% NS
t stat -0.4056
t critical two tail 2.5705
763 Vol. 6, Issue 3, pp. 1049-1054

[7]. Xing, ZuoQun., Yin, XiaoHui., Gao, GuangZhi., Xiu, DeLong., Yin, SiWan. & Sun, PeiDong. (2009).
Calculation and discussion of condition correction coefficient for drying performance of grain dryer.
Transactions of the Chinese Society of Agricultural Engineering, 25 (2), 91-95.
[8]. AOAC (1965). Official Methods of Analysis. Association of Official Agricultural Chemists, Washington, USA.
[9]. Singh, R.P. (1978). Energy accounting in food process operations. Fd Technol, 31-40.
[10]. Maskan, A., Kaya, S., Maskan, M. (2002). Hot air and sun drying of grape leather (pestil). J Food Engng ,54,
81-88.
[11]. Goyal, R.K., Kingsly, A.R.P., Manikanthan, M.R. & Ilyas, S.M. (2007). Mathematical modeling of thin layer
drying kinetics of plum in a tunnel dryer. J Food Engng, 79, 176-180.
[12]. Togrul, I.T. & Pehlivan, D. (2002). Mathematical modeling of solar drying of apricots in thin layers. J. Food
Engng ,55, 209-216.
[13]. Gomez, K.A. & Gomez, A. A. (1983). Statistical Procedure for Agricultural Research. John Wiley and Sons,
New York.
[14]. Brooker, D.B., Bakker Arekma Fred, W. & Hall. C.W. (1997). Theory and simulation of grain drying. In:
Drying and storage of grains and oilseeds. CBS publishers and distributors, New Delhi, India.
[15]. Doymaz, I. & Pala, M. (2002). Hot-air drying characteristics of red pepper. J. Food Engng, 55, 31-335
[16]. Yaldiz, O., Ertekin, C. & Uzun, H.I. (2001). Mathematical modeling of thin layer solar drying of sultana
grapes. Energy Oxford. 26, 457-465.
[17]. Rahman, M.S., Perera, C.O. & Theband, C. (1998). Desorption isotherm and heat pump drying kinetics of
peas. Food Res Int ,30, 485-491.
[18]. Midilli, A., Kucuk, H. & Yapar, Z. (2002). A new model for single layer drying. Drying Technol., 20, 1503-
1513.
[19]. Sharaf-Eldeen, J.L., Blaisdel & Hamdy. M.Y. (1980). A model for ear corn drying. Trans ASAE. 28: 1261-
1265.
[20]. Wang, C.Y. & Singh, R.P. (1978). A single drying equation for rough rice. Trans ASAE ,11,668-672.
[21]. Verma, L.R., Bucklin, R.A., Endan, J.B. & Wraten, F.T. (1985). Effects of drying air parameters on rice drying
models. Trans ASAE, 28, 296-301.
[22]. Thompson, T.L., Peart, R.M., Foster, G.H. (1968). Mathematical simulation of corn drying-A new model.
Trans ASAE 11, (4), 582-586.
[23]. Bal, S. (1974). Measurement of milling quality of paddy. REPC, Indian Institute of Technology, Kharagpur,
Publication No.72.
AUTHORS
Mohammed Shafiq Alam born in the year 1973, graduated from JNKVV, Jabalpur in the
year 1994 and completed M.Tech in the year 1998 from Punjab Agricultural University,
Ludhiana, awarded Ph.D. (Processing and Food Engineering) from Punjab Agricultural
University, Ludhiana in the year 2008 as in-service candidate. Presently, he is working as
Processing Engineer, Department of Processing and Food Engineering, Punjab Agricultural
University, Ludhiana. His area of specialization is Food Engineering and Post Harvest
Technology. He has developed many post harvest machinery and process technologies for
honey, food grains, fruits, vegetables etc. He has published more than 100 papers in the national and international
journals, magazines etc. He has been awarded with many prestigious awards and recognitions of national and
international repute.

V K Sehgal born in the year 1952 graduated from College of Agricultural Engineering, PAU,
Ludhiana and post graduated from Asian Institute of Technology, Bangkok, Thailand and is a
Fellow of Indian Society of Agricultural Engineers and Institution of Engineers. He has
undergone three months Advance Training in Post Harvest Technology and Agricultural Waste
Management at Kansas State University, USA, one month Advance Training in Biomass Fuel
Conversion Technology at Fuel and Combustion Laboratory, Jyvaskyla, Finland and two
months Advance Training in Biomass Fuel Conversion Technology at University of California,
Davis, USA. He has developed many post harvest machinery and process technologies like PAU Moisture Meter,
PAU briquetting Machine, Trolley-cum-batch dryer, suspension type husk fired furnace, honey processing
equipments, fruit & vegetable washing machine, turmeric washing & polishing machine etc. He has published more
than 250 papers in the national and international journals, magazines etc. He has been awarded with many
764 Vol. 6, Issue 3, pp. 1049-1054

prestigious awards and recognitions of national and international repute. He has been graced with several positions
like Head, Department of Processing and Agricultural Structures, Coordinator of Research (Engineering), additional
charge of Dean, College of Agricultural Engineering, additional charge of Estate Officer-cum-Chief Engineer and
Director School of Energy studies in Agriculture. He also had additional charge of Head, Department of Mechanical
till his retirement on 31
st
Aug 2012.

765 Vol. 7, Issue 3, pp. 765-772

JOINT CHANGE DETECTION AND IMAGE REGISTRATION
METHOD FOR MULTITEMPORAL SAR IMAGES
1
Lijinu M Thankachan,
2
Jeny Jose
Department of Electronics and Communication
College of Applied Science, Konni, India

ABSTRACT
This paper presents a novel method for jointly change detection and image registration over multitemporal
Synthetic Aperture Radar (SAR) images. Image registration is performed based on histogram of unregistered
images. Detecting the changes occur between the multitemporal images is performed based on the evolution of
local statistics of the images. The local statistics are estimated using radon transform which removes the
speckle noise occurs in the SAR images. Histogram curve fitting is used to approximate the radon probability
distribution function (pdf) into Gaussian distribution which is on the assumption that radon pdf should obeys the
central limit theorem. In order to measure the variation between the two pairs of projection pdf, the algorithm
uses a local statistic similarity measure called Jeffrey divergence. Experiments results demonstrate that the
proposed method can perform change detection rapidly and automatically over unregistered remote sensing
images.
KEYWORDS Synthetic Aperture Radar (SAR) images, Change detection, Image registration, Radon
transform, Jeffrey divergence.
I. INTRODUCTION
Detecting temporal changes in the state of remotely sensed natural surfaces by observing them at
different times is one of the most important applications of Earth orbiting satellite sensors, because
they can provide multidate digital imagery with consistent image quality, at short intervals, on a
global scale, and during complete seasonal cycles. A lot of experience has already been accumulated
in exploring change detection techniques for visible and near infrared data collected by Satellites. The
changes that occurred in earth surface are identified by analyzing the multitemporal images acquired
at different time.
Automatic change detection [1] [2] [3] in images of a given scene acquired at different times is one of
the most interesting topics in image processing. Since the Synthetic Aperture Radar (SAR) sensors are
capable of acquiring data in all weather conditions and are not affected by cloud cover or different
sunlight conditions, registered SAR images [4] are mainly used for change detection. In recent years,
SAR change detection applications have been widely extended to environmental Earth observation,
national security, and other fields.
The great prospect of change detection has led to a rapid development of techniques, such as image
differencing/ratioing [5], vegetation index differencing, principal component analysis [6], and change
vector analysis [7]. However, less attention has been paid to change detection with SAR images. But
the change detection procedure using SAR images become difficult due to the presence of speckle
noise [8][9], which is a type of multiplicative noise that often severely degrades the visibility in an
image. Thus, the simple pixel-by-pixel comparison used in optical remote sensing images is
ineffective; instead, alternative approaches have been proposed, based on local-statistic similarity
measure.
Change detection can be performed by various methods like pixel intensity comparison, feature point
matching, probability density function (pdf) comparison etc. Unlike the classical detector which is
766 Vol. 7, Issue 3, pp. 765-772

based on ratio of local means, more information can be extracted from the comparison of the local
probability density functions (pdfs).
Higher order statistics include more information about the probability density function (pdf). For
example, the third central moment indicates the lopsidedness, and the fourth central moment is a
measure of whether the distribution is tall or short. Since the higher order statistical information has
proven to be helpful, here willing to compare the local pdf. Once the pdfs are estimated, their
comparison can also be performed using different criteria.
The paper is organized as follows. The overview of the proposed method is briefly presented in
section II-A. Section II-B presents the image registration process. Radon transform is discussed in
Section II-C. Histogram curve fitting method is described in section II-D and Jeffrey divergence is
analyzed in section II-E and then experimental results are presented and analyzed in section III.
Finally, conclusions and future works are drawn.
II. METHODOLOGY
A. Overview of the Proposed Method
Consider two coregistered SAR images Ix and IY acquired over the same geographical area at two
different dates. Our goal is to generate a change/no change map identifying the changes that
occurred during the two dates. The problem can be decomposed into two steps: the generation of a
change indicator and the thresholding of the change image. In this paper, only focus on the first step
of the procedure. The process is applied for each possible pixel position within the image area. Then,
at each pair of SAR images, the following processing stages are performed.
1) Pre-processing stage: Image registration procedure is performed if unregistered images are
selected from dataset.
2) Radon transform stage: Radon transform is applied to the co-registered SAR images respectively,
to generate a pair of projections namely horizontal projection and vertical projection.
3) Jeffrey divergence stage: The Jeffrey divergence is calculated as the distance measure between the
two pairs of pdf. The probability distribution functions (pdfs) used in Jeffrey divergence is
approximated to Gaussian pdf by Histogram curve fitting method.
B. Image registration
The backbone of any SAR image processing task requires initial image matching and registration
procedure. Intuitively, registration correctly aligns all or parts of two or more images containing the
same scene. The two matched images are used to extract temporal changes in a scene, to relate
differences in the appearance of a scene under differing imaging conditions, to detect parallaxes, to
mosaic the images, or to create a multidimensional data set or automated analysis. The many
important tasks that depend on precision registration make this a very significant problem of image
processing and the analysis of multiple SLR (Single Look Radar) images. An approach has been
described in the registration process of two images incorporating the difference in translation and
rotation.
The most common image matching method is dimension correlation, where image windows are found
that resemble one another in the two images to be matched. If the images are not matched then the
image registration is performed.
C. Radon Transform
The Radon transform is the projection of the image intensity along a radial line oriented at a specific
angle [10]. The resulting projection is the line integral of the pixel intensities in each direction. It is a
mapping from the Cartesian rectangular coordinates (x, y) to a distance and an angle (, ) also known
as polar coordinates.
For a given pixel, a new random variable is generated by averaging along a line oriented at an angle
with respect to the x axis, and it is named as projection variable which is denoted by . For any
and , the value of R (, ) is the amount of density that falls along the -line that passes within a
distance of the origin. For image processing, the 0
and 90
cases are chosen and these projections

are called v-projection or vertical projection and h-projection or horizontal projection respectively.
767 Vol. 7, Issue 3, pp. 765-772

Projection variable of an image f(x, y) oriented at an angle can be written mathematically as,
= x cos + y sin (1)
Radon transform of that image can be written as
(, ) = (, )( )
(2)
where (.) is the Dirac delta function.
Fig.1 shows the geometry of Radon transform of the image f(x,y). x and y is obtained by rotating
image axis around the centre of the image at an angle degrees. R (x) is the amount of distribution
that falls along the angle that passes within a distance x of the origin.
Fig.2 shows a single projection at a specified rotation angle. It is a parallel beam projection at rotation
angle . For the radon transform computation,source and sensor is rotated about the center of the
image having coodinates x and y. For each angle the distribution of the image is obtained by when
the rays from the source passes through the image and it is acccumulated at the sensor. This is
repeated for a given set of angles usually from [0:180]. The angle 180 is not included since the
result would be identical to the angle 0.

Fig. 1 Geometry of Radon transform.
In a way, by averaging the pixels, the Radon transform makes the speckle noise weak and shortens the
tail. The shape of horizontal radon pdf and vertical radon pdf is much closer to the Gaussian
distribution and the histogram curve fitting approximation method fits them better.

Fig.2 Parallel beam projection at angle .
D. Histogram curve fitting
768 Vol. 7, Issue 3, pp. 765-772

The method called Histogram curve fitting is used to approximate Radon pdf R(, ). Histogram
curve fitting succeeds based on the assumption that the density to be approximated is not too far from
a Gaussian pdf, which has been satisfied after Radon transform [12]. A histogram is a way to
graphically represent the distribution of data in a data set. Each data point is placed into a bin based
on its value. Bin means pixel intensity range.
Histogram curve fitting plots a histogram of the values in the data is plotted using a number of bins
equal to the square root of the number of elements in data, which will give a fitted normal
distribution. In order to obtain the normal pdf the mean and standard deviation are estimated from
the histogram data, and these values are substitute into the equation of the normal pdf.
Reason of using the curve fitting method is that the central limit theorem (CLT) says, the mean of a
sufficiently large number of independent random variables, each with finite mean and variance, will
be approximately normally distributed. Therefore, the pdf of Radon should not be too far from a
normal distribution.
E. Jeffrey Divergence
Let fx(x) and fy(x) be two probability distribution function of the random variables X and Y. The
Jeffrey divergence between the densities fx(x) and fy(x), is given by

(/) = (
()
()
() +
()
()
()) (3)

where m(x) = (fx(x) + fy(x))/2.
It is an information similarity measure that calculates the divergence from one pdf with other [13].
This measure is symmetric and nonnegative; therefore, it can be used as a statistical similarity
distance [14]. The Jeffrey divergence is numerically stable and robust with respect to noise and the
size of the bins than that of KL divergence. Since it is a modification of the KL divergence [15], it can
be written as
J(X, Y) = K (Z|X) + K (Z|Y) (4)
In order to estimate the Jeffrey distance, decompose it into two KL divergences.
III. RESULTS AND DISCUSSIONS
A pair of SAR images, acquired by the Space borne imaging radar sensor taken at the Manaus, Brazil
in April 1994 and October 1995, respectively is used in the experiment. In order to better understand
the behavior of projection based detection, simulations have been performed.

Fig 3 SAR images (a) Before change (SAR image 1)
769 Vol. 7, Issue 3, pp. 765-772

(b) After change (SAR image 2)
Fig 3(a) and (b) are the registered SAR images. After the pre-processing radon transform is applied to
both images. The resulting projection is the sum of the intensities of the pixel in each direction. The
Radon pdf obtained is in random manner. It is approximated into the Gaussian pdf for statistical
analysis by histogram curve fitting method. Fig 4(a) and Fig 4(b) shows the projection pdfs of SAR
image 1.

Fig 4(a) Vertical projection pdf of SAR image 1

Fig 4(b) Horizontal projection pdf of SAR image 1
Fig 5(a) and Fig 5(b) shows the Vertical and Horizontal projection pdf of SAR image 2. Gaussian
pdfs are obtained by the approximation of projection pdfs by using the method called Histogram
curve fitting. In order to detect the changes between the SAR images, comparison of the local
770 Vol. 7, Issue 3, pp. 765-772

probability density functions (pdfs) of the homologous pixels of the pair of images are performed. Fig
6(a) and (b) shows the variation between vertical Gaussian pdfs and horizontal Gaussian pdfs of SAR
images.

Fig 5 (a) Vertical projection pdf of SAR image 2

Fig 5 (b) Horizontal projection pdf of SAR image 2

Fig 6 (a) Comparison of horizontal pdfs
As the divergence value is not equal to zero, it shows that changes have occurred. If it is equal to zero,
there is no change between the images. The percentage deviation between the vertical pdfs and
horizontal pdfs of the images under consideration are 13.68 % and 21.69% respectively. From the
results obtained it is clear that as the value of Jeffrey divergence increased, the percentage of change
between the images also increased.

771 Vol. 7, Issue 3, pp. 765-772

(b) Comparison of vertical pdfs.
IV. CONCLUSION AND FUTURE WORKS
In this paper, a new similarity measure between images in the context of multitemporal SAR image
change detection is discussed. This measure is based on the Radon transform combined with Jeffrey
divergence, which removes the speckle noise generated by random fluctuations in SAR returning
signals.
A divergence value between the pdfs is calculated using Jeffrey divergence, which is the similarity
measure for change detection. From the results obtained it is clear that as the value of Jeffrey
divergence increased, the percentage of change between the images also increased.
Some questions still remain open about the use of projection. As stated in Section II, in theory, it can
detect the rotation change of the texture. It would be tested when there were appropriate data.
Some improvements could be done in order for it to be a more reliable approach by using more
information rather than only the pdf of the image ie,a change map has to be created from the
divergence value to visually analyze the changes between the images. All these aspects will be studied
as a future development of this paper.
REFERENCES
[1] W. Luo, H.Li and G.Liu, Joint change detection and image registration method for remote sensing
images, IEEE Trans. Geosci. Remote Sens., vol. 6, pp. 634640, Mar. 2013.
[2] M.V. Wyawahare and P. M.Patil ,Image registration techniques and change detection :An overview, Int.
J. Signal Proce., vol. 2, no. 3, Sep 2013.
[3] Jin Zheng and Hongjian You, A New Model-Independent Method for Change Detection in
Multitemporal SAR Images Based on Radon Transform and Jeffrey Divergence, IEEE Trans.Geosci.
Remote Sens., vol. 4, no. 2, pp. 278282, Jan.2012
[4] V. A. Krylov1 and G. Moser, Change detection with synthetic aperture radar images by wilcoxon statistic
likelihood ratio test, IEEE conf. image processing, Aug. 2011.
[5] N. Milisavljevic and D. Closson, Detecting human induced scene changes using coherent change
detection in SAR images, ISPRS. Remote sens, Vol.8, July 2010.
[6] C. Oliver and S. Quegan, Understanding Synthetic Aperture Radar Images, West Perth, WA, Australia:
SciTech, 2004.
[7] G. Mercier, G. Moser, and S. B. Serpico, Conditional copulas for change detection in heterogeneous
remote sensing images, IEEE Trans. Geosci.Remote Sens., vol. 46, no. 5, pp. 14281441, May 2008.
[8] J.Inglada and G. Mercier, A new statistical similarity measure for change detection in multitemporal SAR
images and its extension to multiscale change analysis, IEEE Trans. Geosci. Remote Sens., vol. 45, no. 5,
pp. 14321445, May 2007.
[9] Maryna Rymasheuskaya, Land cover change detection in northern Belarus using image differencing
technique, in Proc. ScanGIS, May 2007.
[10] S. Baronti, R. Carla, S. Sigismondi Principal Component Analysis for change detection on polarimetric
multitemporal SAR Data, IEEE Trans. Geosci. Remote Sens., vol. 46, no. 7, Nov 1994.
[11] J.Chen, P.Gong, Chunyang He, R. Pu, Land-Use/Land-Cover Change Detection Using Improved
Change-Vector Analysis, IEEE Trans. Photogrammetric Engg & Remote Sens., Vol. 69, No. 4, pp.
369379, April 2003.
772 Vol. 7, Issue 3, pp. 765-772

[12] E. J. M. Rignot and J. J. van Zyl, Change Detection Techniques for ERS-1 SAR Data, IEEE Trans.
Geosci. Remote Sensing, vol. 31, no. 4,pp. 896906, July 1993. V. Frost, J. Stiles, K. Shanmugan, and J.
Holtzman, A model for radar images and its application to adaptive digital filtering of multiplicative
noise, IEEE Trans. Pattern Anal. Machine Intell., vol. 4, no. 2, pp. 157165, Mar. 1982.
[13] V.G.G. Moser and S. Serpico, Unsupervised change detection from multichannel SAR images, IEEE
Trans. Geosci. Remote Sens., vol. 4, no. 2, pp. 278282, Apr. 2007.
[14] Y. Bazi, L. Bruzzone, and F. Melgani, An Unsupervised Approach based on the Generalized Gaussian
Model to Automatic Change Detection in Multitemporal SAR Images, IEEE Trans. Geosci.
RemoteSensing, vol. 43, no. 4, pp. 874887, Apr. 2005.
[15] J.-S. Lee, Digital Image Enhancement and Noise Filtering by Use of Local Statistics, IEEE Trans.
Pattern Anal. Machine Intell, vol. 2, no. 1,pp. 165168, 1980.
[16] C. L. Nikias and A. P. Petropulu, Higher-Order Spectra Analysis: a nonlinear signal processing
framework. Englewoods Cliff, NJ, PTR Prentice Hall, 1993.
[17] J. Inglada, Change detection on SAR images by using a parametric estimation of the KullbackLeibler
divergence, in Proc. IEEE IGARSS, Jul. 2003, vol. 6, pp. 41044106.

AUTHORS
Lijinu. M. Thankachan received the Electronics and Communication engineer degree in 2011
from Mahatma Gandhi University and the post-graduation degree in signal processing in 2013
from CUSAT University. She has been since working at IHRD College of Applied Science,
Konni as Asst. Professor in department of Electronics and Communication. Her research
interests include the development of image processing algorithms for the operational
exploitation of Earth Observation images, mainly in the fields of image registration, change
detection and object recognition.

Jeny Jose received the B.Tech degree in Electronics and Communication engineering and the
Masters (engineer) degree in VLSI and Embedded System engineering from Mahatma Gandhi
University, in 2011 and 2013, respectively. She is currently a Lecturer in the IHRD College of
Applied Science, MG University. Her research interests include VLSI design and
implementation for discrete wavelet transform and image coding.

773 Vol. 7, Issue 3, pp. 773-781

LOAD - SETTLEMENT BEHAVIOUR OF GRANULAR PILE IN
BLACK COTTON SOIL
Siddharth Arora
1
, Rakesh Kumar
2
and P.K. Jain
3

1
M. Tech. Student, M.A.N.I.T Bhopal, (M.P) India.
2
Assistant Professor, Department of Civil Engineering, M.A.N.I.T Bhopal, (M.P) India.
3
Professor, Department of Civil Engineering, M.A.N.I.T Bhopal, (M.P) India.

ABSTRACT
The paper discusses the results of study conducted on floating granular piles constructed in soft black cotton
soil. The soil beds were prepared in a model tank of diameter173 mm and height605 mm. The granular pile of
diameter55mm was constructed at the centre of soil bed using crushed stone chips as the pile material. The load
test was conducted on the granular pile. The length to diameter ratio (L/d ratio) of the pile was varied, from 1
to 11 with an interval of two, and the effect of pile length on load carrying capacity was studied. Further, to
observe the effect of encasement of pile material, it was wrapped by geogrid and the load test was conducted.
The test results show that the ultimate load carrying capacity, Qult of the granular pile increases as L/d ratio
increases in both the cases i.e. without and with geogrid encasement. The increasing trend of Qult with L/d is
observed continuing for the maximum L/d ratio i.e. 11, for the granular pile tested in the study. No well-defined
critical length of granular pile is observed in the present investigation of floating granular piles in soft black
cotton soil.
KEY WORDS: Granular Pile, crushed stone chips, Geogrid, load carrying capacity, critical length.
I. INTRODUCTION
Granular pile, sometimes referred as stone column technique, is the one of the most commonly used
soil improvement technique for soft soils. The technique has been utilized worldwide to increase the
load carrying capacity of poor soils and to reduce the settlement of superstructures constructed on
them. The technique is particularly suitable for flexible structures such as road embankments, oil
storage tanks, etc constructed on soft marine clays (Murugesan and Rajagopal, 2006). Different
aspects of soft soil improvement using granular piles have been reported in the literature. The
construction methods are discussed by Datye and Nagraju, 1981; Ranjan and Rao, 1983 and
Greenwood and Krisch, 1983. The mechanism of load transfer is described by Greenwood, 1970;
Hughes and Withers, 1974; Hughes et al., 1976; Madhav and Vitkar, 1978; Aboshi et al., 1979;
Barksdale and Bachus, 1983 and Black et al., 2007. Granular pile is reported to be most effective in
clayey soils with undrained shear strength ranging from 7-50 kPa (Barksdale and Bachus 1983, Juran
and Guermazi 1988, IS 15284(Part-I): 2003).
Black cotton soils also behave like soft soils on wetting. The structures, particularly the light one,
founded on such soils experience large settlement, heave and undulation due to alternate wetting and
drying of the soil. A large number of methods that are commonly suggested to improve the behavior
of such soils include the soil replacement, mechanical and chemical stabilization, thermal treatment
etc. A brief review of the common methods and their shortcomings has been discussed by Kumar and
Jain (2012). Recently, Kumar and Jain (2013) and Kumar (2014) show that the technique of
improving the soft soils by granular piles/ stone columns can also be used in soft expansive black
cotton soil. Kumar (2014) performed model test in the laboratory on granular piles of sand
constructed in soft expansive soil of different UCS values. End bearing granular Pile were casted in
774 Vol. 7, Issue 3, pp. 773-781

the soil and the load test were performed. The test results showed that load carrying capacity of a
footing Qult, defined as the load corresponding to 10% diameter of the footing ( or pile), resting on
granular pile is significantly more than the corresponding value for the footing placed directly on the
soft soil bed. The increase in Qult is observed for all soils i.e. with UCS 30, 40 and 50 KPa. Further,
by mixing nylon fibers and/ or encasing the pile by geogrid, the load carrying capacity of the granular
pile is found increasing. As all the test results reported by Kumar (2014) were performed on end
bearing granular piles, the effect of the ratio of length, L to diameter, d of the pile (L/d ratio) on load
carrying capacity of the pile could not be established, hence to fill this gap a test program has been
planned in which floating piles of different L/d ratio were casted in expansive soft soil prepared bed
and the load test were performed. The details of which are discussed in subsequent paragraphs. It
includes the experimental investigation, results, discussion and the conclusions.
II. EXPERIMENTAL INVESTIGATION
The experiments were carried out on a 55 mm diameter granular pile surrounded by black cotton soil
(i.e. soft clay) in cylindrical mould of 173 mm diameter and 605mm height. The soil bed was
prepared at a dry density of 14.32kN/m
3
and 32% water content at which the cohesion of soil was 36
kN/m
2
. Care was taken to ensure that no significant air voids were left out in the soil bed. The
granular pile was constructed at the centre of the soil bed by making hole with auger, removing the
soil and filling pile material in layers. The dry density of the granular material was 17.5 kN/m
3
. Three
Series of laboratory model tests were performed. First series of tests were performed on the soft soil
bed without any granular pile by placing a mild steel plate of 55mm diameter on the soil top. Second
series of tests were performed on ordinary granular piles (O.G.P) of length, L equal to 55mm, 165mm,
275mm, 385mm, 495mmand 605mm. Thus the L/d ratio was varied as 1, 3, 5, 7, 9, and 11. The third
series of tests were performed on encased granular piles (E.G.P) in which the nylon geogrid was
wrapped around the granular pile for the different cases of L/d ratio as mentioned in O.G.P test series.
III. TEST SETUP
A typical test arrangement for a single pile test is shown in Fig. 1. The load was applied on a circular
footing (mild steel plate) placed on the granular pile through a proving ring at a constant displacement
rate of 1.25 mm/min. The diameter of the footing was same as that of the granular pile. The load
corresponding to equal interval of settlement of the footing was noted.

Figure 1 Typical test arrangement for studying load-settlement behavior of floating granular pile in soft soil

775 Vol. 7, Issue 3, pp. 773-781

IV. MATERIAL PROPERTIES
Three basic materials used for this study are black cotton soil representing the soft soil to be
improved, crushed stone chips as granular pile forming material and the geogrid for encasing the
granular pile. The properties of each of these are as follows:
(i) Black cotton soil: The soil used in the investigation is black cotton soil taken from MANIT
Bhopal campus and its properties are given in table 1.
Table 1 - Properties of black cotton soil
Properties Values
Liquid limit (L.L.), % 53.6
Plastic limit (P.L.), % 28.95
Plasticity index (P.I.), % 24.65
MDD (kN/m
3
) 15.36
OMC (%) 23.5
DFS (%) 40
Specific gravity (G) 2.64
Clay and silt content (%) 95
Classification (IS:1498-1972) CH
Dry unit weight (d ), ( kN/m
3
) 14.32
Degree of saturation (Sr), (%) 100
Cohesion (c) or ( Undrained shear strength ), (kN/m
2
) at 32%
water content
36

(ii) Crushed stone chips: Properties of the crushed stone chips having size less than 6.35mm, used in
the granular pile are listed in table 2.
Table 2 - Properties of crushed stone chips

(iii)Geogrid: The properties of the geogrid (net) used are presented in table 3. Geogrid are stitched
to form the tube for encasing the granular pile.
Table 3 - Properties of geogrid used for encasement of soil
Aperture size (mm) 0.25
Stiffness (KN/m) 116
Weight (gm/m
2
) 255.85
V. RESULTS AND DISCUSSIONS
The load settlement graphs for O.G.P and E.G.P series of test for different L/d ratios are presented in
figure 2 to 7. For the purpose of comparison the load settlement curve for the footing placed on the
top of the soft black cotton soil alone is also shown in these figures. The ultimate load, Q for the
footing placed on soil is found as 0.42kN corresponding to settlement value as 10% size of the
footing.
Properties Values
D10 (mm) 0.95
D20 (mm) 1.40
D30 (mm) 1.75
D50 (mm) 2.70
D60 (mm) 3.25
Cu 3.42
Cc 0.992
Minimum dry unit weight ( d min.), ( kN/m
3
) 16.4
Maximum dry unit weight ( d max.), ( kN/m
3
) 17.5
at ( d.= 17.21 kN/m
3
) 44
0

Suitability number 2.42
776 Vol. 7, Issue 3, pp. 773-781

From figure 2 to 7 it is noted that for all L/d ratios the Qult of footing on granular pile (O.G.P series)
is significantly more in comparison to the footing resting on the soil alone. Further, appreciable
increment in Qult is seen for E.G.P cases too. In order to show the effect of L/d ratio on load carrying
capacity of granular pile, the ratio of Qult for pile to Q of soil is calculated for both the test series and
is given in Table 4. The graphical representation of Table 4 is shown in Figure 8.
Table 4 Effect of L/d ratio on pile load carrying capacity
L/d ratio Qult.pile/Q (O.G.P test series) Qult.pile/Q (E.G.P test series)
1 1.93 3.15
3 4.95 6.24
5 6.24 7.54
7 6.99 8.69
9 7.83 9.99
11 8.77 11.49
From Table 4 and Figure 8 it is noted that the critical length of granular pile as suggested by some of
the researchers in soft soils other than black cotton as 4-6 times the pile diameter (Hughes and
Withers, 1974; Mitra and Cahttopadhyay, 1999; IS 15284(Part-I): 2003; McKelvey et al., 2004; Black
et al., 2007; Samadhiya et al., 2008 and Najjar et al., 2010) is not observed in the present case of soft
soil as the expansive black cotton soil.

Figure 2:Load-settlement curve for L/d ratio 1
0
2
4
6
8
10
12
14
0 1 2 3 4 5 6 7 8 9 10
s
e
t
t
l
e
m
e
n
t

(
m
m
)
Load (KN)
untreated clay
O.G.P
E.G.P
777 Vol. 7, Issue 3, pp. 773-781

Figure 3: Load-settlement curve for L/d ratio 3

0
2
4
6
8
10
12
14
0 1 2 3 4 5 6 7 8 9 10
S
e
t
t
l
e
m
e
n
t

(
m
m
)
Load (KN)
untreated clay
O.G.P
E.G.P
0
2
4
6
8
10
12
14
0 1 2 3 4 5 6 7 8 9 10
S
e
t
t
l
e
m
e
n
t

(
m
m
)
Load (KN)
untreated clay
O.G.P
E.G.P
778 Vol. 7, Issue 3, pp. 773-781


0
2
4
6
8
10
12
14
0 1 2 3 4 5 6 7 8 9 10
S
e
t
t
l
e
m
e
n
t

(
m
m
)
Load (KN)
untreated clay
O.G.P
E.G.P
0
2
4
6
8
10
12
14
0 1 2 3 4 5 6 7 8 9 10
S
e
t
t
l
e
m
e
n
t

(
m
m
)
Load (KN)
untreated clay
O.G.P
E.G.P
779 Vol. 7, Issue 3, pp. 773-781


Figure 8 - Graph between L/d ratio and load at settlement equal to 10% of pile diameter
VI. CONCLUSIONS
From above, following conclusions are drawn
The load carrying capacity of a footing on granular pile is found more in comparison to
that resting on soil alone for all L/d ratio of the pile.
In case of a geogrid encased granular pile of particular L/d ratio, the increase in load
carrying capacity was observed to be higher than the corresponding value in ordinary
granular pile.
The critical length of granular pile is not observed in the present investigation. As the L/d
ratio of the pile increases, the ratio Qult.pile/Q is found increasing.
VII. FUTURE WORK
Similar work can be conducted by changing the shear strength of the black cotton soil and by
using different types of granular materials as pile forming material.
0
2
4
6
8
10
12
14
0 1 2 3 4 5 6 7 8 9 10
S
e
t
t
l
e
m
e
n
t

(
m
m
)
Load (KN)
untreated clay
O.G.P
E.G.P
0
2
4
6
8
10
12
14
0 1 2 3 4 5 6 7 8 9 10 11 12
Q
u
l
t
.
p
i
l
e
/
Q
L/d ratio
O.G.P
E.G.P
780 Vol. 7, Issue 3, pp. 773-781

The swelling and shrinkage behavior of black cotton soil reinforced with granular pile also
needs to be studied.
Single granular pile was used in the present study. The behavior of group of granular piles in
black cotton soil may also be studied.
In the present investigation, the load was applied only on the granular pile whereas it may be
applied on entire area of soil and pile so as to study the behavior of the composite material.
REFERENCES
[1]Aboshi, H., Ichimoto, E., Harada, K., and Emoki, M. (1979). The composer-A method to improve the
characteristics of soft clays by inclusion of large diameter sand columns. Proc., Int. Conf. on Soil
Reinforcement., E. N. P. C., 1, Paris, 211-216
[2]Barksdale, R.D., and Bachus, R.C. (1983). Design and construction of stone columns. Federal
Highway Administration, RD-83/026
[3]Black, J.A., Sivakumar, V., Madhav M.R. and Hamill, G.A. (2007). Reinforced stone columns in weak
deposits: laboratory model study. Proc., J. Geotech. Geoenviron. Eng., ASCE 133(9), 1154-1161.
[4]Datye, K. R., and Nagaraju, S. S. (1981). Design approach and field control for stone columns. Proc.,
10th Int. conf. on Soil Mech. and Found. Eng., Stockholm, Vol. 3, 637-640.
[5]Greenwood, D. A. (1970). Mechanical improvement of soils below ground surfaces. Ground Eng.
Conf., Institution of Civil Engineers, London, 11-22.
[6]Greenwood, D. A. and Kirsch, K. (1983). Specialist ground treatment by vibratory and dynamics
methods. Proc., Int. Conf. on Piling and Ground Treatment, Thomas Telford, London, 17-45.
[7]Hughes, J. M. O., and Withers, N. J. (1974). Reinforcing of soft cohesive soils with stone columns.
Ground Eng., 7(3), 42-49.
[8]Hughes, J. M. O., Withers, N. J., and Greenwood, D.A. (1976). A field trial of reinforcing effect of
stone column in soil. Geotechnique, Vol. 25, No. 1, 32- 44.
[9]Indian Standards (IS). (2003). Indian standard code of practice for design and construction for ground
improvement-guidelines. Part 1: Stone columns. IS 15284 (Part 1), New Delhi, India
[10]Juran I. and Guermazi A. (1988). Settlement response of soft soils reinforced by compacted sand
columns. J. Geotech. Geoenviron. Eng., ASCE 114(8), 903-943.
[11]Kumar, R. and Jain P. K. (2012). Prospect of using granular piles for improvement of expansive
soil. Int. J. of Advanced Eng. Tech., Vol.III/ Issue III/July-Sept, 2012, 79-84.
[12]Kumar, R. and Jain P. K. (2013). Expansive Soft Soil Improvement by Geogrid Encased Granular
Pile. Int. J. on Emerging Technologies, 4(1): 55-61(2013).
[13]Kumar, R. (2014). A Study on Soft Ground Improvement Using Fiber- Reinforced Granular Piles.
Ph. D. thesis submitted to MANIT, Bhopal, India
[14]Madhav, M. R., and Vitkar, P. P. (1978). Strip footing on weak clay stabilized with a granular trench
or pile. Can. Geotech. J., 15, 605-609.
[15]McKelvey, D., Sivakumar, V., Bell, A., Graham, J. (2004). Modeling vibrated stone columns in soft
clay. Proc., Institute of Civil Engineers Geotechnical Eng., Vol. 157, Issue GE3, 137-149.
[16]Mitra, S., and Chattopadhyay, B.C. (1999). Stone columns and design limitations. Proc., Indian
Geotech. Conf., Calcutta, India, 201-205.
[17]Murugesan, S. and Rajagopal, K. (2006). Geosynthetic-encased stone columns: Numerical
evaluation. Geotextiles and Geomembranes, J. Vol. 24, 349-358.
[18]Najjar, S. S., Sadek, S., and Maakaroun, T. (2010). Effect of sand columns on the undrained load
response of soft clays. Proc., J. Geotech. Geoenviron. Eng, ASCE 136(9):1263-1277.
[19]Ranjan, G. and Rao, B. G. (1983). Skirted granular piles for ground improvement. Proc., VIII
European Conf. on Soil Mech. and Found. Eng., Halainki.
[20]Samadhiya, N. K., Maheshwari, P., Basu, P., and Kumar, M. B. (2008). Load settlement
characteristics of granular pile with randomly mixed fibers. Indian Geotech. J., 38(3), 345-354.
AUTHORS
Siddharth arora was born in Haridwar (U.K), India, in 1991. He received the Bachelor
degree in civil engineering from govind ballabh pant engineering college, pauri-garhwal, in
2012 and He is currently pursuing the the Master degree in geotechnical engineering from
maulana azad national institute of technology, Bhopal, which will complete in july, 2014.
His research interests include geotechnical engineering.

781 Vol. 7, Issue 3, pp. 773-781

Rakesh kumar was born in Gajrola (U.P), India, in 1977. He received the Bachelor degree
in civil engineering from the MMMEC, Gorakhpur, in 1999 and Master degree in
geotechnical engineering from IIT Roorkee, Haridwar, in 2001. He has completed his PhD
from maulana azad national institute of technology, Bhopal, in 2014. His research interests
include geotechnical engineering.

Pradeep Kumar Jain was born in Jhansi (U.P), India, in 1964. He received the Bachelor in
civil engineering degree from the MITS, Gwalior, in 1986 and Master degree in construction
technology and management from MITS, Gwalior, in 1988. He has completed his PhD from
IIT Roorkee, Roorkee, in 1996. His research interests include geotechnical engineering.

782 Vol. 7, Issue 3, pp. 782-789

HARMONIC STUDY OF VFDS AND FILTER DESIGN: A CASE
STUDY FOR SUGAR INDUSTRY WITH COGENERATION
V. P. Gosavi
1
and S. M. Shinde
2

1
P. G. Student, Govt. College of engineering, Aurangabad, India
2
Asst. Professor, Electrical Department
Govt. College of Engineering, Aurangabad, India

ABSTRACT
In majority of Sugar Industries, to achieve better efficiency of power, VFDs are extensively used. Initially D.C.
Drives were used, but now these are slowly getting phased out and are replaced by A.C. Drives .In a case study
of Harmonic Analysis of SAMARTH CO-OPERATIVE SUGAR FACTORY, ANKUSH NAGAR, MAHAKALA,
JALNA it is commonly seen perception that, latest technology A.C. drives are having in built harmonic filters
and hence do not produce harmonics / produce harmonics within prescribed tolerance limit. Because of
variation in load pattern these drives still produce harmonics, which must be suppressed by using appropriate
Harmonic Filter, otherwise it may result into malfunction of sensitive devices. It is more essential to suppress
these harmonics, for SAMARTH CO-OPERATIVE SUGAR FACTORY, ANKUSH NAGAR, MAHAKALA,
JALNA, as Sugar Industry has grid connected Cogeneration plant. The suggested method of filter design gives
better result, reduces harmonic level (5th harmonics) to almost 10A from 154A. The values of load Kw, P.F.,
Voltage, current load inductance, capacitance, resistance are derived from average of readings actually
available from site.
KEYWORDS: Passive Harmonic Filter, Tuned Frequency, Cable Impedance.
I. INTRODUCTION
In Co-generation Sugar Mills, Reduction in, specific power consumption increases power revenue by
about Rs. 3 per ton cane [1]. In a sugar mill, one of the method of increasing power efficiency is to
replace smaller low efficiency (25-30 %) mill turbines by better efficiency drives such as AC motors.
Multistage steam turbines can operate at efficiency of 65-70 %. Hence equivalent quantity of steam
saved by installation of AC motors, DC motor, Hydraulic drives can be passed through power turbines
to generate additional power [2].
Modern Drive manufacturer proved the line reactor with the A.C. Drives only, but these line reactors
produce high Voltage drop and does not provide the expected filtrations of harmonics. It is common
perception among industries that these line reactors, use the harmonic filter and hence these Drives do
not produce or produce harmonics within to tolerance limit. Because of variation harmonics, which
must be supposed by Harmonic Filters otherwise it may result into malfunction of sensitive
equipments. It is more essential to suppress these harmonics in case the Sugar Industry has the grid
interconnected cogeneration power plant.
This paper discuss the case study of Harmonic Analysis of SAMARTH CO-OPERATIVE SUGAR
FACTORY, ANKUSH NAGAR, MAHAKALA, JALNA, which confirms the presence of very high
amount of harmonic current and the necessity of proper harmonic filter.
The organization of this paper is as follows
In section II The methodology of the case study is discussed
In section III The actual readings taken during the case study are presented
In section IV The filter design as per recommended method is explained
783 Vol. 7, Issue 3, pp. 782-789

In section V The result of the installation of filter simulated with the help of matlab
program is discussed.
In section VI The interpretation of the case study and the results derived is discussed as
Conclusion
In section VII The future scope in the same research area of harmonic filter design for the
Sugar Industry based co-generation plant is presented.
II. METHODOLOGY OF CASE STUDY
For the case study of harmonic analysis of SAMARTH CO-OPERATIVE SUGAR FACTORY,
ANKUSH NAGAR, MAHAKALA, JALNA the methodology adopted is as follows
Initially the walk in audit of the entire plant was done.
The probable sources of generation of harmonics were identified, VFDs were came out as most
probable sources of harmonics generation.
To confirm the current harmonics content generated by the VFDs, reading were taken for two VFDs
for a time cycle of ten minutes each with the help of power quality analyser. [3]
Out of this two, the one which was generating more harmonic current was taken for the case study and
case study is presented as below
III. CASE STUDY (ACTUAL READING)
An A.C. DRIVE OF 300 kW 3- INDUCTION MILL DRIVE MOTOR
SAMARTH CO-OPERATIVE SUGAR FACTORY, ANKUSH NAGAR, MAHAKALA, JALNA
The Actual readings are kW consumed 261 kW.
Avg. Line Current = 406 A
P.F. = 0.88
Current Harmonics
I (h1) 368.88 A I (h2) 2.06 A
I (h3) 28.363 A I (h4) - 0 A
I (h5) 154.03 A I (h6) 0 A
I (h7) - 59.543 A I (h8) 0 A
I (h9) 4.483 A I (h10) 0 A
I (h11) 26.33 A

Table 1 : R Current Harmonics
TIME I1(01) I1(02) I1(03) I1(04) I1(05) I1(07) I1(09) I1(11) THDF_I1(%)
15:54:40 351.00 0.00 19.00 0.00 148.00 47.00 4.00 30.00 45.60
15:55:40 329.00 0.00 21.00 0.00 137.00 51.00 0.00 24.00 46.00
15:56:40 282.00 0.00 17.00 0.00 136.00 61.00 0.00 25.00 54.30
15:57:40 480.00 5.00 20.00 0.00 183.00 49.00 3.00 31.00 40.80
15:58:40 338.00 5.00 19.00 0.00 149.00 57.00 0.00 27.00 48.60
15:59:40 397.00 4.00 17.00 0.00 165.00 57.00 0.00 29.00 45.10
16:00:40 393.00 5.00 22.00 0.00 173.00 67.00 0.00 33.00 48.50
16:01:40 373.00 0.00 21.00 0.00 161.00 62.00 4.00 33.00 48.10
16:02:40 340.00 0.00 17.00 0.00 153.00 62.00 0.00 24.00 49.80
16:03:40 202.00 0.00 23.00 0.00 117.00 63.00 5.00 19.00 68.20
16:04:40 330.00 4.00 21.00 0.00 152.00 62.00 4.00 26.00 51.10

784 Vol. 7, Issue 3, pp. 782-789

Table 2 : Y Current Harmonics
15:54:40 402.00 0.00 31.00 0.00 144.00 51.00 4.00 26.00 39.80
15:55:40 378.00 0.00 27.00 0.00 137.00 55.00 4.00 20.00 40.70
15:56:40 325.00 0.00 24.00 0.00 134.00 63.00 4.00 21.00 47.10
15:57:40 532.00 5.00 30.00 0.00 177.00 52.00 6.00 26.00 36.00
15:58:40 386.00 4.00 28.00 0.00 148.00 62.00 6.00 24.00 42.80
15:59:40 442.00 6.00 25.00 0.00 162.00 59.00 5.00 25.00 40.20
16:00:40 441.00 4.00 26.00 0.00 168.00 70.00 5.00 28.00 42.50
16:01:40 416.00 4.00 23.00 0.00 160.00 64.00 4.00 30.00 42.80
16:02:40 384.00 0.00 26.00 0.00 152.00 64.00 4.00 21.00 44.20
16:03:40 244.00 4.00 19.00 0.00 121.00 68.00 0.00 18.00 59.00
16:04:40 378.00 4.00 26.00 0.00 151.00 65.00 4.00 23.00 44.70

Table 3 : B Current Harmonics
15:54:40 377.00 0.00 41.00 0.00 155.00 49.00 8.00 30.00 45.80
15:55:40 349.00 0.00 42.00 0.00 146.00 54.00 7.00 25.00 47.30
15:56:40 301.00 0.00 35.00 0.00 144.00 63.00 7.00 26.00 54.60
15:57:40 503.00 0.00 44.00 0.00 190.00 49.00 9.00 32.00 41.00
15:58:40 361.00 0.00 40.00 0.00 157.00 59.00 8.00 28.00 48.80
15:59:40 416.00 4.00 38.00 0.00 172.00 57.00 8.00 29.00 45.40
16:00:40 412.00 0.00 40.00 0.00 178.00 68.00 7.00 32.00 48.20
16:01:40 387.00 5.00 38.00 0.00 168.00 63.00 7.00 33.00 48.60
16:02:40 361.00 0.00 39.00 0.00 160.00 63.00 7.00 24.00 49.70
16:03:40 213.00 0.00 36.00 0.00 126.00 66.00 6.00 21.00 70.50
16:04:40 350.00 5.00 41.00 0.00 159.00 63.00 8.00 26.00 51.30
Table 4 : The impedance from Generator to Mill Motor
Sr. Cable size R + j X L ( Length Impedance / Length No. of
Runs
Total
1 120 sq mm
(L.T.)
0.3050 + j 0.0744 25 0.00762 + j 0.00185 4 0.0019 + j 0.00046
2 300 sq mm
(L.T.)
0.1220 + j 0.0732 15 0.00183 + j 0.00109 10 0.00018 + j 0.00019
3 400 sq mm
(H.C.)
0.10 + j 0.099 25 0.0025 + j 0.00247 1 0.0025 + j 0.00247
4 Transformer 0 + j 0.00338 0 + j 0.00338
5 Turbine j 0.5377 0 + j 0 . 5377

785 Vol. 7, Issue 3, pp. 782-789

Figure 1 Current Harmonic Spectrum
IV. FILTER DESIGN AS PER RECOMMENDED METHOD
The load is 261 Kw with 0.88 logging P.F. The fifth harmonic current produced by load is maximum
that is 41.75 % of fundamental current. The system voltage is 422 V.
The fifth harmonic background voltage distortion on utility side of transformer is 0.5 %
Step 1 : Select a turned frequency for filter
The filter will be slightly tuned below the 5
th
harmonic frequency to allow for tolerances in the filter
components and variations in system impedance. The filter is designed to be tuned to the 4.7
th

Step 2 : Compute capacitor Bank size and Resonant Frequency.
Here it is assumed that P.F. is to be improved from 0.89 logging to 0.96 logging.
Reactive power demand for 0.88 logging PF
1
296.59 sin [cos (0.88)] 140.87 K
var

.. (1)

Reactive power demand for a 0.96 logging PF
1
Var
296.59 sin [cos (0.96)] 83.04 K

.. (2)

Required Compensation from filter
140.87 83.04 = @60
var
K

For a nominal 422 V system, the net wye equivalent filter reactance (capacitive)
Filt
X is determined
by
2
(1000) (0.422) (1000)
= = = 2.9680
60
2
Filt
Var
KV
X
K
,
.. (3)

Filt Cap L
X = X - X
.. (4)

For tuning at 4.7
th
harmonic
2 2
Cap L L
X =h X =(4.7) X
.. (5)

Thus, desired capacitive reactance can be determined by
2
2
2.968(4.7)
3.1087
1 (4.7) 1
2
Filt
Cap 2
X h
X
h

.. (6)

To achieve this reactance at a 422 V rating the capacitor would have to be rated
2
(1000) (0.422) 1000
57.28
3.1087
2
var var
Cap
KV
K K
X

var
@50 K
.. (7)

Now the filter will be designed using 422 V capacitor rated
var
60K . For this capacitor rating.
2.968
Cap
X
Step 3 : Compute Filter Reactor Size
The filter reactor size is computed from the wye equivalent capacitive reactance as follow
0
50
100
150
200
250
300
350
400
I1 I2 I3 I4 I5 I6 I7 I8 I9 I10 I11
Series1
786 Vol. 7, Issue 3, pp. 782-789

2
2.968
0.134
(4.7)
Cap(wye)
L 2
X
X (Fund)
h

.. (8)

0.134
0.427 mH
2 50 2 50
LFund
X
L

.. (9)

Step 4 : Computation of Fundamental duty requirements
a) The apparent reactance of combined capacitor and reactor at fundamental frequency is
0.134 2.968
Fund L Cap(wye)
X X - X

= 2.834
.. (10)

b) The fundamental frequency filter current is
/ 3 422/ 3
85.97 A
2.834
actual
Fund
fund
KV
I
X

.. (11)

c) The fundamental frequency operating voltage across capacitor bank is
3
L-L (Fund) fund Cap (wye)
V Cap L X = 3 85.97 2.968

= 441.95 V
.. (12)

This is nominal fundamental voltage across the capacitor. It should be adjusted for any contingency
conditions and it should be less than 110 percent of capacitor rated voltage.
d) actual reactive power produced is
var
3 62.83 K
var Fund fund actual
K I KV
.. (13)

Step 5 : Computation of harmonic duty requirements
a) The 5
th
harmonic current produced by load is 154.03 A
(amps)
Ih
b) Fundamental frequency impedance of service transformer
2
(0.422)
(%) 0.06 0.00534
2
2
actual
T (Fund) T
KV
X Z
MVA(Xmer)

.. (14)

T(harm) T
X hX (Fund)

= 5 0.00534 = 0.0267
.. (15)

The harmonic impedance of capacitor bank is
XCap (wye), harm = 0.5936
The harmonic impedance of reactor is
XL(harm) = hXLFund = 5 0.134 = 0.67
The fifth harmonic current contributed to filter from source side would be
3 ( )
h(utility)(pu) actual
h(utility)
T (harm) Cap (wye) harm L
V KV
I
X X X (harm)

.. (16)

0.005 422
11.81 A
3 (0.0267 0.5936 0.67)

c) The maximum harmonic total current 154.03 11.81 165.84
h(total)
I
d) The harmonic voltage across capacitor
3
Cap (wye)
Cap (L-L rms harm) h(total)
X
V I
h

2.968
3 165.84
5

= 170.50
.. (17)

Step 6 : Evaluate total rms current and peak voltage requirement
a)
2 2
(85.97) (170.50)
2
rms total Fund hutility
I I I
= 190.94A
.. (18)

b) The maximum peak voltage across capacitor
Cap
LL (max. peak) LL (Fund)+ V
V Cap =V Cap
.. (19)

= 442 + 170.5 = 612.5
c) The rms voltage across capacitor
2 2
(442) (170.5) 473.74V
.. (20)

d) The total
Var
K seen by the capacitor is
3 190.94 0.47374
Var (wye)
K Cap total
.. (21)
= 156.67
var
K

787 Vol. 7, Issue 3, pp. 782-789

Capacitor Rated for high voltage must be
2
var 2
(600)
60 121.29 K
(422)
.
Table : 5 Comparison able for Evaluating Filter Duty Limit
Duty Definatio Limit % Actual
Values
Actual
Value %
Peak
voltage
KV rated
LL (max. peak)
V Cap

120 612.5 /
600
102
RMS
voltage
KV rated
LL (rms total)
V Cap

110 473.74 /
600
80
RMS
current
. total
I caprated
rms
I

180 190.94/
116.71
164
Kvar
Kvar rated
Kvar Cap (wye total)

135 156.67/
121.29
129
V. RESULT

Figure 2 Waveform of 5th Harmonic Current Figure 3 Waveform of 5th Harmonic Current
(w/o filter) (with recommended filter)

Figure 4 Initial distorted Waveform Figure 5 Distorted Waveform
(w/o filter) (with recommended filter)
Table : 6 Result Of Performance Of Filters Designed By Recommended Method.
Parameter Without Filter With Filter
Power Factor 0.88 0.968
2nd Harmonics 2.06 A 2.14 A
3rd Harmonics 28.363 A 14.237 A
4th Harmonics 0 0
5th Harmonics 154.03 A 10.3128 A
788 Vol. 7, Issue 3, pp. 782-789

6th Harmonics 0 0
8th Harmonics 0 0
10th Harmonics 0 0
Finding of results
1) Figure 1 and 3 represent the current wave form of fifth harmonic current and the total rms current
generated by the VFD before installation of the proposed harmonic filter.
2) Figure 2 and 4 represent the current wave form of fifth harmonic current and the total rms current
generated by the VFD after installation of the proposed harmonic filter, simulated with the help of
MATLAB program. It is seen that the originally distorted wave forms as observed in figure 1 and 3
are smoothened in figure 2 and 4 respectively, which clearly indicates the effective performance of
proposed passive harmonic filter
3) The table 5 reveals that the power factor of load is improved, third, fifth, seventh and ninth
harmonic current generated by the load are minimized with the incorporation of the proposed passive
harmonic filter. It is seen that the fifth harmonic current for which the filter is designed, as it is the
major harmonic contributor current, is drastically reduced from 154.03 A to 10.3128 A.
4) It is also seen from table 5 that eleventh harmonic current is increased marginally due to
incorporation of the proposed filter.
5)The even harmonic currents that is second, fourth, sixth, eighth and tenth harmonics are absent in
both the cases with filter and without filter
VI. CONCLUSIONS
This paper explains the design of harmonic filter by a method which is generally not followed by
industry. The recommended method designs a filter which caters all requirement of the filter as shown
in table 5.
i) The recommended filter suppresses 5
th
harmonic current from 154.03 A to 10.3128 A.
ii) A Filter tuned for frequency 235 Hz is designed by considering 0.96 power factor gives all duty
requirements within acceptable limits as per IEEE Standard18-1992, IEEE Standard for Shunt
Power Capacitors
VII. FUTURE WORK
By this method filters tuned for different frequencies can be designed but for every frequency the cost
of filter as well reduction in %THD of current changes. Filters can be designed for different power
factors, but for different power factors the cost of filter as well reduction in %THD of current
changes. An optimized approach is necessary for the optimum design of the filter to give the most
cost effectiveness subjected to the better % THD of current and power factor of the system. Research
is necessary in this area. A program can be designed to get this optimum result.
ACKNOWLEDGEMENTS
Authors are thankfull to General Manager Shri S.N.Surwase, Manager Co-generation Shri
Vijaykumar Hulgiri & all supporting staff of Samarth Co-operative Sugar Factory
Ankushnagar,Mahakala,Jalna, Shri.V.P.Sonawane(A.E.)&Shri.G.H.Chavan(A.E.)M.S.E.T.C.L, Shri.
R.V. Kulkarni (Govt. College of Engg. Abad), Shri. Padhye of Ambik Harmonic Filters Pvt Ltd.
Pune for their support and help during measurement of data.
REFERENCES
[1] Efficient Cogeneration Scheme For Sugar Industry. M.Premalata, S.Shanmuga Priya, V. Sivaramakrishnan
Journal of Scientific and Industrial Research. Vol 67,March 2008,PP 239-242
789 Vol. 7, Issue 3, pp. 782-789

[2] Energy Conservatin in Sugar Mills.All India Seminar on Ethanol & Cogeneration
2003,Nainital,Uttaranchal
[3] IEEE Recommended Practice for Monitoring Electric Power Quality, 1159.
[4] Electrical Power Systems Quality, Roger C. Dugan, Mark F. McGranaghan, Surya Santoso, H. Wayne
Beaty, Tata McGraw-Hill Publication, Second Edition, pp-264-270.
[5] Harmonic Minimization In Sugar Industry Based Cogeneration Plant.S.M.Shinde ,W.Z.Gandhare IEEMA
Journal December 2013 pp-122-125
[6] Modeling Variable AC Drives And Harmonic Distortion Erika Twinning, Ian Cochrane Department Of
Elctrical and Electronics Engineering The University Of Melbourne ,Australia
[7] IEEE Standard18-1992, IEEE Standard for Shunt Power Capacitors

BIOGRAPHY

Vivek P. Gosavi was born in Aurangabad, Maharashtra, India in 1972. He received the
Bachelor in Electrical Engineering degree from the Marathawada (Dr.B.A.M.U.)
University, Aurangabad in 1993. He is currently pursuing the Master of Engineering degree
with the Department of Electrical Engineering of Govt College of Engineering,
Aurangabad. His research interests include Power quality issues and design of harmonic
filters.

Sanjay M. Shinde was born in Manwat, Maharashtra, India in 1966. He received the
Bachelor in Electrical Engineer degree from the Marathwada (Dr.B.A.M.U.) University,
Aurangabad in 1990 and the Master in 1998 degree from the Dr. B.A.M.U. of Aurangabad
in Year, both in Electrical engineering. He is currently pursuing the Ph.D. degree with the
Department of Electrical Engineering, Govt. College of Aurangabad. His research interests
include Power quality study in co-generation systems and renewable energy sources.

790 Vol. 7, Issue 3, pp. 790-797

PRECIPITATION AND KINETICS OF FERROUS CARBONATE
IN SIMULATED BRINE SOLUTION AND ITS IMPACT ON CO
2

CORROSION OF STEEL
G. S. Das
Material Science and Metallurgical Engineering
National Institute of Foundry and Forge Technology, Hatia, Ranchi, India

ABSTRACT
The aim of present study was to study the stability of iron carbonate films formed on the surface of low alloy
carbon steel used in pipeline applications in brine solution. All the tests were carried out in a stirred autoclave
at various temperatures and low partial pressure of CO2. Usually, American Petroleum Institute (API)
specification steels are used for carrying the crude oil and gas from the offshore to refining platforms and
ultimately to the customers through pipeline routes. Developed and modified alloy steel of API grade such as
API X-52, API X-56, API X-60 and L-80 are used in this study that are rapidly using in petroleum industries to
transport the oil and gas. The first three grade of steel are used for transportation while L-80 grade is used for
tubing and drilling the wells for recovery of oil and gas from offshore. The comparative study of their
susceptibility of corrosion in severe corrosive condition has been made by measuring the corrosion rate of steel
using weight loss method in simulated 3.5% NaCl test solution. The fluid velocity of simulated brine solution
was maintained at a rate of 1.5 m/s for each test. The exposed samples were characterized by using Scanning
Electron Microscope and X-ray diffraction techniques.
KEY WORDS: API grade steel, Partial pressure of carbon dioxide, Autoclave, Brine solution
I. INTRODUCTION
The CO2 corrosion of carbon steel is one of the most important problems in the petrochemical
industries since 1940, because of their localized attack and severe corrosion [1-3]. These steels have
wide application in just about every sphere of oil and gas industries and it requires critically
assessment of corrosion severity to ensure safe utilization of these steels. An increased number of
offshore developments are based on transporting the unprocessed or partially processed multiphase
well streams from wells to onshore processing platforms and from there to ultimate consumers. In
such circumstances carbon dioxide often present in the fluids and is known for its high corrosion
potential when dissolved in water. These aspects are responsible to adequately assess the integrity of
the steels used in the oil and gas industry for production and transport equipment. Many parameters
are involved in the process and even small differences in chemical composition between one steel and
another can strongly influence scale formation during CO2 corrosion [4]. Scale formation and salt
accumulation are another analogous rigorous problem for the operation of multiphase pipelines [4-5].
Usually, well fluids and other products are transported through pipelines in the oil and gas industry
and these fluids contain aqueous phases which are considered to be inherently corrosive. CO2 is a gas
that reacts with water to form carbonic acid which lowers the pH of the solution and thus responsible
for increased corrosion rate. Despite, of its many studies frequent questions are still raised regarding
the mechanism of CO2 corrosion responsible for its occurrence. Many developed corrosion resistant
alloys such as 13% Cr steel and duplex stainless steel are used for the downhole drilling and also for
short flow lines [7-9]. But for long distance and large diameter pipelines, carbon steels are the only
economically feasible alternatives [9]. However, these steels are susceptible to corrosion in both either
791 Vol. 7, Issue 3, pp. 790-797

CO2 and H2S or combination of these two. The susceptibility of their corrosion depends on various
parameters such as temperature, CO2 partial pressure, H2S concentration, pH, oxygen solubility,
chloride concentration, flow rate and characteristics of the materials [1-3]. The presence of acidic gas
like CO2, H2S and free water can cause severe corrosion problem in the oil and gas industries [5]. The
small change in one of these parameters can change the corrosion rate drastically due to change in the
properties of the layers of corrosion products that accumulates on the steel surface. The selection of
materials that transport oil and gas is not always made with sufficient emphasis on corrosion
resistance, but rather on good mechanical properties, ease of fabrication and low cost. Due to the
material loss rates resulting from internal corrosion, it becomes necessary to thoroughly characterize
the behavior of these high strength steels used for oil and gas pipelines. Many authors

emphasized on
the importance of the evaluation of the specific nature of the film formed under a given set of
conditions, as these conditions can affect the type of film formed on the steel [10]. In general, the
precipitation of an iron carbonate film on the steel surface significantly affects the rate of corrosion.
The corrosion rate may increase or decrease depends on the nature of the film that forms on the metal
surface. Usually the carbonate film reduces the corrosion rate however in some certain cases it can
contribute to localized corrosion due to its non-uniform nature [11]. Also, the corrosion rate is closely
dependent on a series of parameters such as temperature, pressure, pH and steel composition. Nesic et.
al proposed a theoretical model of carbon dioxide corrosion in which the main focus was on the
factors which influence FeCO3 film formation [6]. They suggested that corrosion rates are not, in fact,
closely associated simply with the thickness of the protective film, but rather with its degree of
coverage and homogeneity on the metal surface. These films can partially or completely cover the
metal surface and consequently block only a portion of the exposed metal or, in some cases, it can
homogeneously cover the entire surface of steel. By covering entire surface it can prevent the further
corrosion due to restrict in further dissolution. The voids and holes may form at the underneath of
corrosion films that interface with the metal. The rate of void formation is important for determining
the type of film that formed on the metal surface. The protectiveness of film depends on many factors
such as film density, degree of protectiveness, porosity, and thickness. With the aid of their model,
they also confirmed that high levels of pH, temperature, partial pressure of CO2 and Fe
2+

concentration associated with a low formation rate of the above mentioned voids can favour the
formation of a protective iron carbonate film [10-12].
II. MECHANISM AND KINETICS OF CO2 CORROSION
Iron carbonate is the main corrosion product in carbon dioxide corrosion is also known as sweet
corrosion since 1940 and it is considered as one of the most important problems in the oil and gas
industry due to its presence in the well fluids [10-11]. The iron carbonate scale formed on the metal
surface may be protective or non protective depends on the conditions at which they are formed and
influenced by many factors such as the concentration of ions present in the solution, pH, temperature,
pressure and velocity of the solution. The formation of protective films of iron carbonate called
siderite is possible when the concentration of bicarbonate ions increases rather than carbonic acid.
Since carbon dioxide dissolve in water and form carbonic acid which is mild acidic in nature and
corrosive. Among the most popular mechanisms postulated for this type of corrosion are proposed by
De Waard and Milliams, Schmitt and Rothmann and George and Nesic et. all which involve carbonic
acid and bicarbonate ions formation during dissolution of CO2 in water [1-6]. Both chemical and
electrochemical reactions are important factors in CO2 corrosion and one of the most common
chemical reactions that occur in the system is as follows;

This is followed by the carbonic acid dissociation:
792 Vol. 7, Issue 3, pp. 790-797

On the surface of the steel, the electrochemical reactions which occur include an anodic reaction and
three cathodic reactions. The anodic reaction is:

The following cathodic reactions takes place are as follows;

In CO2 corrosion of carbon steel, when the concentrations of Fe
2+
and CO3
2
ions exceed the solubility
limit, they can precipitate to form solid iron carbonate film according to the reaction as below;

Fe
2+
+ CO3
2
= FeCO3 (s) (9)
The properties of these films depend on precipitated iron carbonate (FeCO3) film called siderite on the
metal surface. It may act either in a protective or non-protective manner, depending on the specific
conditions under which it forms. At pH < 7, the carbonate ion (CO3
2
) is a minority species, and the
direct reduction of bicarbonate ions (HCO3
-
) is an important factor in the formation of the FeCO3
films on the steel surface. The FeCO3 forms by the following equation:
Fe
2+
+ HCO3
= FeCO3 + H
+
(10)
The overall reaction can be expressed as:

Thus dissolved CO2 have the tendency to form carbonic acid (H2CO3) in the solution which increases
the cathodic reaction kinetics by dissociation to bicarbonate and hydrogen ions. Under stagnant
condition, dissolved ferrous ions combine with the H2CO3 to form ferrous carbonate (FeCO3).
However under flow condition some part of the iron carbonate dissolved and thus the reaction rate
become faster due to removal of corrosion scale from the metal surface. De Waard and Milliams
developed a semiempirical correlation between corrosion rate and CO2 partial pressure. This
relationship later simplified in the form of a nomogram and given as below;

Log C.R. (mm/y) = 5.8-1780/ T +0.67 Log (pCO2) (12)

The above equation provide a conservative estimate of corrosion rate under flowing condition because
it does not account for the effect of non-ideality of the gas phase and scale formation. In equation 4,
the hydrogen ion is produced through the bicarbonate ions near the surface rather than direct H2CO3.
In this sense, the carbonic acid acts as catalysts for the evolution of hydrogen. The high concentration
of H2CO3 or its precursor CO2 relative to H
+
ion in the bulk solution increases the rate of overall mass
transfer and depolarized the cathodic reaction. The ability of carbonic acid to sustain high cathodic
reduction rate at relatively high pH indirectly aids the anodic reaction as per equation (5), which is
faster at higher pH. Also, the value of pH has the tendency to both either increase or decrease the CO2
793 Vol. 7, Issue 3, pp. 790-797

based corrosion. The determination of pH is a function of partial pressure of mild acidic gas,
concentration of bicarbonate ions and temperature. Practically, the contribution of mild carbonic acid
(H2CO3) and temperature in pH determination is another way of representing effective level of partial
pressure of carbon dioxide (CO2) in the solution. As the solution pH decreases the localized attack
involved in the steel surface. The temperature of the solution has a significance influence on the CO2
corrosion. An Arrehenius type of relationship has been reported as below;

Log (Icorr) = 7.58-10,700/2.303 RT (13)

The temperature effect on film formation has been summarized by Burke in 1984 and reported that
below 60
o
C the corrosion product formed on the metal surface is mainly consists of siderite and
magnetite which is non-protective in nature. However, at higher temperature range (80-120
o
C) this
scale becomes dense in association with Fe2O3 and Fe3O4 and protects the metal for further
degradation.
III. EXPERIMENTAL
The materials used in the current study were obtained in pipe form. The as-received materials were
cut into the rectangular specimen of dimension 20 mm x 12 mm x 2.5 mm with a center hole of
diameter 1.5 mm at the top edge of each specimen to facilitate the suspension inside of the autoclave
of capacity 2.2 liter. The faces of the samples were initially coarse grounded on a SiC belt grinder
machine then consequently machine polished in the successive grade of emery papers (220, 400, 600
800 and 1000). The initial weight and area of each sample was measured using digital weighing
machine upto an accuracy of four digits. Four different experiments were carried out at different
temperature (30
o
C, 60
o
C, 90
o
C and 120
o
C) for each measured sample at a constant partial pressure
of 50 psi. The velocity of fluid was maintained at a constant speed of 1.5 m/s for 96 hours. The
oxygen solubility of the system was maintained below 40 ppb using the Ar gas purging in the system.
In the beginning of the test, the actual temperature of the machine was set as 30
o
C, 60
o
C, 90
o
C and
120
o
C separately and respectively and then created the actual partial pressure of carbon dioxide by
releasing the carbon dioxide gas inside the autoclave. The exposed samples were taken out from the
system washed in distilled water, rinsed in acetone and then dried in air. After cleaning, the coupon
weighed again and finally the corrosion rates of the samples were measured in mills per year using the
formula;
Corrosion rate (mpy) = Weight loss (g) X Constant (K)/Metal density (gm/cm
3
) X Metal
Area (A) X Exposure time (hrs.)
Where K is 5.34 x 10
5
. Also, the exposed samples were characterized by using different techniques
such as environmental scanning electron microscope (ESEM) and X-ray diffraction (XRD)
Table 1: Chemical composition of the as received materials (in wt %)
Elements Sample 1
(L-80)
(Wt %)
Sample 2
(API X-52)
(Wt %)
Sample 3
(API X-56)
(Wt %)
Sample 4
(API X-60)
(Wt %)
C 0.19 0.17 0.14 0.12
Mn 1.22 1.23 1.27 1.25
Si 0.36 0.37 0.035 0.037
S 0.003 0.003 0.0031 0.003
P 0.004 0.004 0.0039 0.0038
Cr 0.07 0.080 0.010 0.015
Mo - 0.0001 0.00025 0.00034
Cu 0.16 0.09 0.22 0.23
794 Vol. 7, Issue 3, pp. 790-797

50 Psi
0
2
4
6
8
10
30 60 90 120
Temperature (90
o
C)
C
o
r
r
o
s
i
o
n

r
a
t
e

(
m
p
y
)
API X-60
API X-56
API X-52
L-80

Fig. 1: Corrosion rate patterns of exposed samples in CO2 environment
50 psi
0
5
10
15
20
0 30 60 90 120 150
Temperature (
0
C)
C
o
r
r
o
s
i
o
n

R
a
t
e

(
m
p
y
)
API X-52
API X-56
L-80
API X-60

Fig. 2: Weight loss graph patterns of exposed samples in CO2 environment

795 Vol. 7, Issue 3, pp. 790-797

Fig. 3: XRD Patterns of all the four exposed samples exposed at 90
o
C and in CO2 environment

Fig. 4: ESEM micrographs of exposed samples (a) API X-52 (b) API X-56 (c) API X-60 and (d) L-80 at 90
o
C
and in 50 psi CO2
(a) (b)
(c) (d)
796 Vol. 7, Issue 3, pp. 790-797

IV. RESULTS AND DISCUSSION
The corrosion rate patterns of exposed samples in CO2 environment are shown in Fig. 1. The results
indicated that all the steel exhibit corrosion in mildly acidic condition, however, the severity of
corrosion is different in different compositional steel. All exposed samples shown higher corrosion
rate at the 90
o
C due to formation of porous layer of siderite and their spallation from the metal
surface. Also, the results indicated that the corrosion rate decreases at higher temperature near about
100-120
o
C due to formation of adherent and dense layer of protective iron carbonate films that protect
the metal from further reaction of corrosion. The formations of iron carbonate scales on the metal
surface were confirmed by using X-ray diffraction analysis and are shown in Fig.3. The other phases
observed are Fe3O4, Fe2O3 and Fe3C respectively are also indicated in Fig. 3. The micrographs of
corroded samples were captured in environmental scanning electron microscope and presented in
Fig.4.
V. CONCLUSIONS
The corrosion rate patterns of all the steels increases with increasing temperature upto 90
o
C due to
formation of porous layer on the surface of the metal. The weight loss data as a function of
temperature at a constant partial pressure of carbon dioxide are shown in Fig. 2. Beyond this
temperature the corrosion rate again fall down due to formation of dense and adhered layer of siderite
with magnetite films on the metal surface which acts as protective in nature. At aggressive
temperature the layer formed on the surface dissolving continuously and the reaction rate becomes
faster and shown higher corrosion rate while at higher temperature the iron carbonate films formed are
very dense in nature and compact on the surface results in slower corrosion rate.
REFERENCES
[1] C. De Ward, D.E. Milliams, Prediction of carbonic acid in natural gas pipelines, First International
Conference on the Internal and External Protection of Pipes paper F-1, University of Durham, September
1975.
[2] C. De Waard, U. Lotz and D.E. Milliams, Predictive model for CO2 corrosion engineering in wet natural
gas pipelines. Corrosion 47 (1991), pp. 976985.
[3] C. DeWaard and U. Lotz Prediction of CO2 corrosion of carbon steel in the Oil and Gas Industry, Institute
of Materials Publisher, UK (1994), pp. 3049.
[4] C.A. Palacios and J.R. Shadley , Characteristics of corrosion scales on steel in a CO2-saturated NaCl brine.
Corrosion 47 (1991), pp. 122127.
[5] C. De Waard and D.E. Milliams, Carbonic acid corrosion of steel. Corrosion 31 (1975), pp. 177181.
[6] S. Nesic, N. Thevenot, J.L. Crolet, D.M. Drazic, Electrochemical properties of iron dissolution in the
presence of CO2 Corrosion'96 NACE, USA, paper 3, 1996.
[7] K.D. Efird, E.J. Wright, J.A. Boros, T.G. Hailey, Wall shear stress and flow accelerated corrosion of carbon
steel in sweet production Proceedings of the 12th International Corrosion Congress, Houston 1993, pp.
26622679.
[8] G.I. Ogundele and W.E. White, Some observations on corrosion of carbon steel in aqueous environments
containing carbon dioxide. Corrosion 42 (1986), pp. 7178.
[9] K. Videm and A. Dugstad, Corrosion of carbon steel in an aqueous carbon dioxide environment. Part 2.
Film formation. Mats. Perf. 28 (1989), pp. 4650.
[10] A. Dugstad, The importance of FeCO3 supersaturation of carbon steel Corrosion'92, paper no. 14, USA,
1992.
[11] M.L. Johnson, M.B. Tomson, Ferrous carbonate precipitation kinetics and its impact CO2 corrosion,
Corrosion'91, NACE, USA, paper 268 1991.
[12] Videm K. and Kvarekval J., Corrosion of Carbon Steel in CO2 Saturated Aqueous Solutions Containing
Small Amounts of H2S, NACE CORROSION/94, paper No.12, 1994
[13] Srinivasan S. and Kane R.D., Prediction of Corrosion Rates of Steel in CO2/H2S Production
Environments, Prevention of Pipeline Corrosion Conference, Houston, TX, 1995.
[14] Ikeda A., Ueda M. and Mukai S., Influence of Environmental Factors on Corrosion in CO2 Source Well,
Advances in CO2 Corrosion, Vol. 2, 1985
[15] Kvarekval J., The Influence of Small Amounts of H2S on CO2 Corrosion of Iron and Carbon Steel,
EUROCORR 97, Trondheim, Norway.
797 Vol. 7, Issue 3, pp. 790-797

[16] Valdes, A., Case, R, Ramirez, M., and Ruiz, A., The Effect of Small Amounts of H2S on CO2 Corrosion
of a Carbon Steel, Paper No. 22, CORROSION/98.

AUTHOR
Ghanshyam Das, Asst. Professor in Materials and Metallurgical Engineering Department
at National Institute of Foundry and Forge Technology, Hatia, Ranchi-834003. He
graduated in Metallurgical Engineering from B.I.T. Sindri in 1996 and Doctor in
philosophy in 2007 from Materials Science and Metallurgical Engineering Department, IIT
Bombay, Powai, Mumbai. His research interest is degradation of ferrous materials and
published more than 20 papers in international and national journals and conference
proceedings. He has attended many international and national conferences in India and
abroad. He received best technical paper award in Nov. 2012 at International conferences (M3) in Singapore and
in Jan. 2014 at Raigarh, Chattishgarh, India.

798 Vol. 7, Issue 3, pp. 798-806

PERFORMANCE COMPARISON OF POWER SYSTEM
STABILIZER WITH AND WITHOUT FACTS DEVICE
Amit Kumar Vidyarthi
1
, Subrahmanyam Tanala
2
, Ashish Dhar Diwan
1

1
M.Tech Scholar,
2
Asst. Prof.
Dept. of Electrical Engg., Lovely Professional University, Phagwara, Punjab, India

ABSTRACT
Power system damping through coordinated design of power system stabilizer (pss) & different types of FACTS
devices (SVC, SSSC & UPPC) is presented in this paper. In the condition of single phase fault power system
stabilizer is enough to damp the oscillations for the two area system but the settling time increases for the whole
system parameter, but in the case of three phase faults the whole system parameter goes into unbalanced
condition and the machine losses its synchronism even in the presence of power system stabilizer, in that case
PSS is not enough to suppress the faults, so the other compensator must put into action for the better stability of
the two area system. PSS including with FACTS devices reduce the settling time and enhance the response time
of the system for single phase faults and damp the oscillation for three phase fault and make the system stable.
This paper contains coordinated simulink model of PSS & different types of FACTS devices for two area systems
& checked for single phase fault and three phase fault condition with & without FACTS devices.
KEYWORDS- PSS, SVC, SSSC, UPFC, FACTS.
I. INTRODUCTION
For the better operation of power system without scarcity in system security and power quality
furthermore in the case of contingency conditions namely transmission lines and generating units
losses, advance and modern control strategies need to be implemented. The FACTS is a power
electronic based controller which improve the power transfer capability of the transmission network.
Since these controllers operate very fast so they outbid the safe operating limits of transmission
system without taking a chance on stability. When the large power system is interconnected by weak
tie lines then the low frequency oscillation has been observed. Low frequency oscillation may
continue and rise above to cause system separation if perfect damping is not available. In these days
the conventional power system stabilizer is used by the power system utilities all over the world. The
FACTS devices play an important role in the operation and control of power system such as
scheduling power flow; reducing net losses; providing voltage support; limiting short circuit currents;
mitigating sub synchronous resonance; damping power system oscillation and improving transient
stability. Power system is nothing but a business of generation, transmissions & distribution of
electrical energy. Since power system is in normal operating condition works as a synchronous system
i.e. all the machines in power system are synchronized with each other i.e. they run at a common
frequency. During short circuit condition the very heavy current flow in the circuit as well as the
voltages at different part of the system go to very low value & this condition is known as abnormal
operating condition. So the fault does create a disturbance in the power system operation & due to
these disturbances will the power system remain in synchronism or not is a big problem for the power
system. Due to the fault there is an imbalance between the electrical & mechanical power & this
affects the rotor speed variations & this may lead the step of falls. Power system stabilizer improves
the damping oscillations of power system during the time of electromechanical transients [1].
799 Vol. 7, Issue 3, pp. 798-806

Sometimes power system stabilizer is not enough to damp out the oscillation. So we use the
coordinated model of PSS & FACTS device that has a less settling time & more efficient.
This paper is organized as follows: In section 2 single machine infinite bus systems is introduced. In
section 3 power system stabilizer is introduced. In section 4 SVC, SSSC & UPFC are introduced. In
section 5 simulation result is shown & in the last section conclusion and prospects for future work is
given.
II. SINGLE MACHINE INFINITE BUS SYSTEM
Figure.1 shows the basic building block of single machine infinite bus system with exciter & AVR [2]
this is the basic block for better understanding of the machine & infinite bus which is connected to the
complex power system. The complete power system state space model is given in the following form.
/ [

1
]=[
11 12
21 0
13 0
0 0
0 32
0 42
33 34
43 44
] [

1
]+ [
1
0
0
0
]
The heffron phillips model of single machine infinite bus system is given below[9].

Fig: 1 Single machine infinite bus system
III. POWER SYSTEM STABILIZER (PSS)
Power system stabilizer is generally adding extra damping to the rotor oscillation of synchronous
machine by controlling its excitation. The block diagram is given below.

Fig: 2 Power system stabilizers
The above block model contains a low pass filter, a gain, a washout high filter, a phase compensator
system & limiter. The input for the PSS may be speed deviation, frequency deviation or accelerating
power & its output is stabilizing voltage. The working of PSS is very simple & straight forward.
When the fault occur in power system it generally introduce electromechanical oscillation in the
synchronous generator & to stabilize the system this oscillation must be damp out. The PSS take any
of the three signal which is illustrated above & supply an additional output stabilize voltage signal to
the excitation system so the overall system may remain in a stable state. The block diagram of pss
with single machine infinite bus system is given below. [3]
800 Vol. 7, Issue 3, pp. 798-806

Figure-3 Single machine infinite bus system with PSS
IV. FACTS DEVICE
FACTS are defined by the IEEE as a power electronic based system & other static equipment that
provide control of one or more AC transmission system parameters to enhance controllability &
increase power transfer capability.[4]
4.1 static var compensator (SVC)
Static var compensator is mainly a shunt facts device use to control power flow & improve transient
stability of the system. It mainly injects or absorbs reactive power in the bus when the voltage of the
bus is low or high respectively [5] .It is mainly connected mid- point of the transmission line because
at the mid -point static var compensator gives its maximum efficiency .So static var compensator is
use for mid- point voltage regulation of line segmentation, end of line voltage support to prevent
voltage instability, improvement of transient stability, power oscillation damping etc. Figure-4 which
is given below is the conventional figure of static var compensator and mainly contains thyristor
control reactor, thyristor switch capacitor, fixed capacitor and switch resistor. The inductor is used in
this circuit specially to prevent inrush current.

Figure-4 Static var compensator
4.2 static synchronous series compensator (SSSC)
Static synchronous series compensator is series facts device which is use to control power flow &
improve power oscillation damping of system. [6]. the schematic of static synchronous series
compensator is given in the figure below.
801 Vol. 7, Issue 3, pp. 798-806

Figure-5 Static synchronous series compensator with line
4.3 Unified power flow controller (UPFC)
Unified power flow control is a device nothing but a combination of series & shunt facts device & it
obviously do the same work what is done by the series & shunt fact device alone. It is the most
powerful facts device [7]. UPFC is mainly a combination of SSSC & STATCOM. Use to improve the
transient stability of the power system [8].The schematic figure of unified power flow controller is
given below.

Figure-6 Unified power flow controller
V. SIMULATION
We have taken two generating unit 1000MW and 5000MW respectively, connected with 500KM
long transmission line which is given below.

Figure.7 Simulation of Two area system with power system stabilizer only
The initial power output of the SM1 and SM2 are 0.95 and 0.8 respectively. In the above model a
three phase fault occurs at sending end bus at time t=0.2s.
Action of Power system stabilizer on single phase fault is given below.
Voltage and power waveform of positive sequences is shown below.
802 Vol. 7, Issue 3, pp. 798-806

Figure.8 Positive sequence voltage of power system

Figure.9 Waveform of Power
Now the machines rotor angle, speed and voltage waveform for single phase faults is given below.

Figure.10 Rotor angle difference between two machines

Figure.11 Speed of Machine (PU)
803 Vol. 7, Issue 3, pp. 798-806

Figure.12 Terminal voltages waveform of two machines
Action of Power system stabilizer on three phase faults is given below.

Figure.13 Positive sequence voltages waveform

Figure.14 Line power of a system

Figure15.Rotor angle difference between two machines

Figure16.Machine speeds (PU)
804 Vol. 7, Issue 3, pp. 798-806

Figure17.Terminal voltages of two machines
The simulink model of two area power system with power system stabilizer and static var
compensator is given below.

Figure18.Two area power system with PSS and SVC
For a single phase faults

Figure19.Positive sequence voltages and line power waveform

Figure20.Rotor angle, Machine speed, Terminal voltages of machine

For three phase faults
805 Vol. 7, Issue 3, pp. 798-806

Figure21.positive sequence voltages and Line power

Figure22. Rotor angle, Speed and Terminal voltage of Machine
The simulink model result of two areas with power system stabilizer and static synchronous series
compensator is given below.

Figure23.Sequence voltage, Line power, Reactive power waveform
The simulink model result of power system stabilizer with unified power flow controller is shown
below.

Figure24.Sequence voltages, Line power, Reactive power waveform
VI. CONCLUSION AND PROSPECTS FOR FUTURE WORK
The objective of this study was to analyze power system stabilizer and coordinated model of PSS and
FACTS devices performance under several perturbations. Different types of simulations are carried
out in simulink environment. The FACTS devices are simulated for the transient stability on a two
area power system. The system is simulated by initiating a single phase fault and three phase fault
near the first machine in the absence of FACTS devices. In this case the difference between the rotor
806 Vol. 7, Issue 3, pp. 798-806

angles of two machines is increased enormously and eventually loses its synchronism. But, when the
same faults are simulated in the presence of FACTS devices the system becomes stable. From the
simulink result it is shown that UPFC is the best one to suppress the three phase fault so the best
combination for suppressing the three phase fault in two area power system is power system stabilizer
and unified power flow controller when the tuning is in appropriate manner. In the future we will do
the same with the help of different artificial intelligent techniques such as fuzzy logic, artificial neural
network and compare the result for the optimal one.
REFERENCES
[1] Saidi Amara and Hadj Abdallah , Power system stability improvement by FACTS devices: a comparison
between STATCOM, SSSC and UPFC, First international conference on renewable energies and vehicular
technology, 2012.
[2] Kamalesh Chandra Rout and Dr.P.C. Panda, Performance Analysis of Power System Dynamic Stability
Using Fuzzy Logic Based PSS for Positive and Negative Value of K5 Constant, International conference on
computer, communication and Electrical Technology ICCCET2011, 18th and 19
th
March2011.
[3] P. Kundur, Power system stability and control, Mc Grew Hill, U.S.A 1994
[4] Proposed terms and definitions for flexible AC transmission system(FACTS) , IEEE Transactions on power
Delivery, Volume 12, Issue 4, October1997,pp1848-1853.
[5] Pasala Gopi, Dr. I. Prabhakar Reddy and P.Sri Hari , Shunt FACTS Devices for First Swing Stability
Enhancement in Inter-area Power System, Third International Conference on Sustainable Energy and
Intelligent system (SEISCON2012).
[6] N.G.Hingorani, L Gyugyi, Understanding FACTS; Concepts and Technology of Flexible AC Transmission
System,IEEE Press book, 2000.
[7] Mehrdad AhmadiKamarposhti, MostafaAlinezhadAlinezhad, Effects of STATCOM, TCSC, SSSC and
UPFC on voltage stability, International Review of Electrical Engineering, vol.4, no.6, pp.1376-1382.2009.
[8] Eskandar Gholipour and Shahrokh Saadate, Improving of Transient Stability of Power Systems Using
UPFC, IEEE TRANSACTIONS ON POWER DELIVERY vol.20 (2), pp.1677-1682.2005.
[9] D.K. Sambariya and Rajendra Prasad , Robust Power System Stabilizer
Design for Single Machine Infinite Bus System with Different Membership functions for Fuzzy Logic
Controller, 7
th
International Conference on Intelligent system and Control (ISCO 2013).
BIOGRAPHIES
Amit Kumar Vidyarthi was born in Dhanbad, India on 8
th
Dec 1986. He is pursuing his
M. Tech from Lovely Professional University, Punjab. His research area interests are in the
areas of power system operation and control and Flexible AC Transmission systems
(FACTS).

Subrahmanyam Tanala is an Assistant Professor in the Department of Electrical
Engineering at Lovely Professional University, Phagwara, Punjab. His area of research and
interest includes Power electronics, power systems, FACTS devices and renewable energy.

Ashish Dhar Diwan is pursuing his M.Tech in Electrical Engineering (power system) from
Lovely Professional University. His area of interest includes Power systems, FACTS,
Transmission systems, etc. and has published papers in various journals subjecting the
same.

807 Vol. 7, Issue 3, pp. 807-817

HYDROLOGICAL STUDY OF MAN (CHANDRABHAGA)
RIVER
Shirgire Anil Vasant
1
, Talegaokar S.D.
2

1
Student M.Tech,
2
Professor,
Department of Civil Engineering, Bharati Vidyapeeth Deemed University
College of Engineering, Pune, India

ABSTRACT
Hydrological study of Man (Chandrabhaga) river basin were done from the available rainfall and stream flow
records of years 1996-2010.Data was taken from HYDROLOGY PROJECT DIVISION NASHIK FOR CASE
STUDY. The mean annual rainfall of Sidhewadi station is about 627.64 mm. Cumulative mass inflow curve was
drawn to estimate the maximum reservoir capacity of small dam. In this study, a flood frequency analysis of
Man River basin is carried out by Log-Pearson Type-III probability distribution method and Gumbels
distribution. From the flood frequency analysis maximum flood discharge is calculated for the recurrence
interval of 10, 100, 200, 1000 years. From the study duration of flood was calculated. A small dam can be
constructed of storage capacity of 4.60 Mm3.Study was helpful to increase area under irrigation and design and
construction of hydraulic structure on Man river.
KEYWORDS: stream flow, rainfall, Log Person Type III, flood frequency, flow duration.
I. INTRODUCTION
Planning, design, and management of water resources systems often require knowledge of flood
characteristics, such as peak, volume, and duration. Fundamental importance in many design
problems is the determination of the probability distribution of maximum annual discharge (7). Based
on an assumed probability distribution, one can compute statistics of flows of various magnitudes,
which can then be used for planning, design, and management of water resources projects. (11)
The river system in any region depends upon the slope of land surface and velocity of stream. Rivers
of Indian plateau are shorter in length than the Himalaya region. These rivers have ample amount of
water in rainy season and in summer most of the rivers are dry (8). These rivers have greater
importance in western Maharashtra region. In the western Maharashtra region one of eastward
flowing river is Bhima it has tributary Man River.
The present study was taken to development of mass curve, flow duration curve and carry out flood
frequency analysis from the available records of the rainfall and stream flow. This study is a
prerequisite study in decision making policies for the design of any hydraulic structure.
The data is taken from hydrology project division Nashik for study.
Study Area
River Gauge Station: Sidhewadi River Gauge station is selected for hydrological study. This station is
65 km away from Solapur. The hydrological data of this station is taken from Hydrology project
division Nasik (Government of Maharashtra). Sidhewadi station is on Mangalvedha -Pandharpur road,
& 17 km away from Pandharpur.
Objectives of Research-
Keeping the above perspectives in view, the proposed study is undertaken with the following
objectives:
808 Vol. 7, Issue 3, pp. 807-817

1. To calculate the storage capacity of dam if small dam is constructed by use of mass curve to
increase area under irrigation, water for drinking etc.
2. To develop flow duration curve from which assessments of hydropower design schemes, water
supply, planning and design of irrigation systems is possible.
3. To carry out comparative flood frequency analysis between Gumbels and Log Pearson type III
distribution from the available hydrological data for calculation of maximum annual discharge.

Fig 1. Diagram showing Man River basin
II. REVIEW OF LITERATURE
L.R. Singo et al., 2011; Annual maximum flow data from 8 stations with 50 years hydrological data
were used to analyze flood frequencies in the catchment. To derive the probability of occurrence of
flood events, the frequency distributions which could best describe the past characteristics and
magnitudes of such floods were tested. This involved the determination of the best flood frequency
models, which could be fitted to the available historical recorded data. The distribution models used
included the Generalized Extreme Value, Gumbel or Extreme Value type 1, Log-Normal and the Log
Pearson type III distributions. The extreme value analysis showed that the Gumbel and Log Pearson
type III distributions provided the best fit.
T. A. Ewemoje et al., 2011; The paper discusses how Normal, Lognormal, and log-Pearson type 3
distributions were investigated as distributions for annual maximum flood flows using the Hazen,
Weibull, and California plotting positions at Ogun-Oshun river basin in Nigeria. All the probability
distributions when matched with Weibull plotting position gave similar values near the center of the
distribution but varied considerably in the tails. The Weibull plotting position when matched with
Normal, Log-normal and Log Pearson Type III probability distributions gave the highest Coefficient
of determinations of 0.967, 0.987, and 0.986 respectively. Hazen plotting position gave minimal
errors with the RMSE of 6.988, 6.390, and 6.011 for Normal, Log-normal, and Log-Pearson Type III
probability distributions respectively. This implies that, predicting statistically using Hazen plotting
position, the central tendency of predicted values to deviate from observed flows will be minimal for
the period under consideration. Minimum absolute differences of 2.3516 and 0.5763 at 25- and 50-
year return periods were obtained under the Log-Pearson Type III distribution when matched with
Weibull plotting position, while an absolute difference of 0.2338 at 100-year return period was
obtained under the Log-Pearson Type III distribution when matched with California plotting position.
Comparing the probability distributions, Log-Pearson Type III distribution with the least absolute
differences for all the plotting positions is the best distribution among the three for Ona River under
Ogun-osun river basin study location.
809 Vol. 7, Issue 3, pp. 807-817

Tefaruk Haktanir et al., 1993 ; A statistical model comprising nine different probability distributions
used especially for flood frequency analysis was applied to annual flood peak series with at least 30
observations for 11 unregulated streams in the Rhine Basin in Germany and two streams in Scotland.
The parameters of most of those distributions were estimated by the methods of maximum likelihood
and probability-weighted moments. The distributions were first compared by classical goodness-of-fit
tests on the observed series. Next, the goodness of predictions of the extreme right-tail events by all
the models were evaluated through detailed analyses of long synthetically generated series. The
general extreme value and 3-parameter lognormal distributions were found to predict the rare floods
of return periods of 100 years or more better than the other distributions used. The general extreme
value type 2 and log-Pearson type 3 (when skewness is positive) would usually yield slightly
conservative peaks. The Wakeby distribution also gave peaks mostly on the conservative side. The
log-logistic distribution with the method of maximum likelihood was found to overestimate greatly
high return period floods.
III. MATERIALS AND METHODS
To carry out hydrological studies of Man river basin, the rainfall data and discharge data of years
1995-2010 were collected from Sidhewadi station. The stream flow data recorded at Sidhewadi
stream flow gauging station. The data was taken from Hydrology project division Nasik utilized in
these studies.
3.1 Mean Annual Rainfall
Daily rainfall data is utilized for calculating mean rainfall of station Sidhewadi. Avarage annual
rainfall is calculated from daily data of station.
Table 1; Annual Rainfall data of siddhewadi station
Sr. No Year Rainfall (mm)
1 1995 527.4
2 1996 1071
3 1997 684
4 1998 1161
5 1999 552
6 2000 440
7 2001 454
8 2002 732
9 2003 423
10 2004 726
11 2005 671
12 2006 560
13 2007 532
14 2008 484
15 2009 522
16 2010 505
Avg Rainfall 627.64
3.2 Point Rainfall
The point rainfall is also known as station rainfall refers to the rainfall data of a station. Depending
upon the need, data can be listed as daily, weekly, monthly, seasonal or annual values for various
periods. Graphically these data are represented as plots of magnitude versus chronological time in the
form of a bar diagram. Such a plot however is not convenient for discerning a trend in the rainfall as
there will be considerable variations in the rainfall values leading to rapid changes in the plot. The
trend is often discerned by the method of moving averages, also known as moving means.
3.3 Moving Average
Moving average is a technique for smoothening out the high frequency fluctuations of a time series
and to enable the trend, if any, to be noticed. The basic principle is that a window of time range m
810 Vol. 7, Issue 3, pp. 807-817

years is selected. Starting from first set of m years of data, the average of of the data for m years is
calculated and placed in the middle year of the range m. The window is next moved sequentially one
time unit (year) at a time and the mean of the m terms in the window is determined at each window
location. The value of m can be 3 or more years; usually an odd value. Generally, the larger the size of
range of m, the greater is the smoothening .There are many ways of averging (and consequently the
plotting position of the mean) and the method described above is called central simple moving
average.
3.4 Mass curve analysis
The flowmass curve is a plot of the cumulative discharge volume against time plotted in
chronological order. The ordinate of the mass curve, V at any time t is thus
V= (1.1)
Where t0 is the time at the beginning of the curve and Q is the discharge rate. Since the hydrograph is
a plot of Q vs t, it is to see that the flow-mass curve is an integral curve (summation curve) of the
hydrograph. The maximum vertical ordinate was measured which gives the maximum storage
capacity of the reservoir. Fundamentally, a reservoir serves to store water and the size of the reservoir
is governed by the volume of the water that must be stored, which in turn is affected by the variability
of the inflow available for the reservoir.
3.5 Flood frequency study
The procedure for estimating the frequency of occurrence (return period) of a hydrological event such
as flood is known as (flood) frequency analysis. Two methods were used for analysis.
3.5.1 Log-Pearson Type III distribution
The Log-Pearson Type III distribution tells you the likely values of discharges to expect in the river at
various recurrence intervals based on the available historical record. This is helpful when designing
structures in or near the river that may be affected by floods.
The Log-Pearson Type III distribution is calculated using the general equation:
If X is the variate of a random hydrologic series, then the series of Z variates where
Z = Log x (1.2)
are first obtained. For this Z series, for any recurrence interval T gives
(1.3)
Where K is a frequency factor which a function of recurrence interval T and the coefficient of skew
Cs,
z = standard deviation of the Z variate sample
(1.4)
Next, the skewness coefficient Cs of Z variate can be calculated as follows:

(1.5)
, N= sample size = number of years of record,
XT = antilog (ZT) (1.6)
3.5.2 Gumbels Equation for Practical Use
Equation (1.3) giving the values of the variate X with a recurrence interval T is used as
xT = x
+ Kn-1 (1.7)
Where n-1= standard deviation of the sample size N, K= frequency factor expressed as
(1.8)

in which yT = reduced variate, a function of T and is given by
(1.9)
or
(1.10)
yn = reduced mean, a function of sample size N.
811 Vol. 7, Issue 3, pp. 807-817

Sn = reduced standard deviation, a function of sample size N
Table 02- Reduced Mean (yn ) and Reduced Standard Deviation ( Sn )
N 10 15 20 25 30 40 50
yn 0.4952 0.5128 0.5128 0.5309 0.5236 0.5436 0.5309
Sn 0.9457 1.0206 1.0206 1.0915 1.0628 1.1413 1.1607
N 60 70 80 90 100 200 500
yn 0.5521 0.5548 0.5569 0.5586 0.5600 0.5672 0.5724
Sn 1.1747 1.1854 1.1938 1.2007 1.2065 1.2360 1.2588
IV. RESULTS AND DISCUSSIONS
4.1 Mean Annual Rainfall

Graph 1: Annual Rainfall Vs time in years
4.2 Moving Average Method
Annual rainfall data is used for moving average method.
Addition of three consecutive years is done as shown in following table.
Table 03 -Computation of three year moving mean
1 2 3 4
Year
Annual
Rainfall (mm)
Three consecutive year total
for moving mean (Pi-1 +Pi +Pi+1 )
3-year moving
Mean (Col 3/3)
1995 527.4
1996 1071 527.4+1071+684=2282.4 760.8
1997 684 1071+684+1161=2916 972
1998 1161 684+1161+552=2397 799
1999 552 1161+552+440=2153 717.7
2000 440 552+440+454=1446 482
2001 454 440+454+732=1626 542
2002 732 454+732+423=1609 536.34
2003 423 732+423+726=1881 627
2004 726 423+726+671=1820 606.67
2005 671 726+671+560=1957 652.33
2006 560 671+560+532=1763 587.7
2007 532 560+532+484=1576 525.3
812 Vol. 7, Issue 3, pp. 807-817

2008 484 532+484+522=1538 512.7
2009 522 484+522+505=1511 503.7
2010 505
627.64 mm

Graph 2: Annual rainfall Vs year for Three year moving mean
Moving mean calculation are shown in table 04 .Three year moving mean curve is shown plotted in
graph:02 with the moving mean value as the ordinate and the time in chronological order as abscissa.
Note that the curve starts from 1996 and ends in the year 2010. No apparent trend is indicated in this
plot.
4.3 Mass inflow curve
The cumulative stream flow data were used to derive the mass inflow curve for calculating the
reservoir capacity corresponding to specific yield. The mass curve is shown in graph 03 .The average
cumulative inflow is obtained as 2592.22 m
3
/s and total cumulative inflow into the reservoir for a
period of the 13 years is 38884.242 m
3
/s. From the mass curve, it was estimated that total quantity of
water available for utilization is about 8000 m
3
/sec.
Water available for storage is 8000 m
3
/sec for Five month .Therefore mean daily discharge available
is 8000/150 = 20 m
3
/s. i.e. (= 53.338.64 10
4
=4.60 Mm
3
).
Table 04: Inflow cumulative discharge
Year Inflow (m
3
/s) Cumulative Inflow (m
3
/s)
1991 3364.19 3364.19
1993 764.76 4128.95
1996 837.52 4966.47
1997 63.97 5030.44
1998 14975 20005.44
1999 1387.319 21392.75
2000 559.489 21952.239
2001 1223.816 23176.055
2002 130.63 23306.685
2004 302.27 23608.955
2005 236.037 23844.992
2007 2320.97 26165.962
2008 532.39 26698.352
2009 9630 36328.352
2010 2555.89 38884.242
Total 38884.242
Avg 2592.22
813 Vol. 7, Issue 3, pp. 807-817

Graph 3: Cumulated volume Vs Time in years for Storage capacity of reservoir from mass curve
4.4 Flow duration curve
The annual discharge data was used for drawing the flow duration curve. From the data it is found that
maximum and minimum discharges of 14975 m
3
/sec, 63.97

m
3
/sec are recorded in the year 1998 and
1997 respectively .The average discharge for the period of the 13 years was estimated as 2592

m
3
/sec.
out of which only 2 years, the river flow is above average flow and in remaining 11 years the river
flow is below the average discharge rate. A plot of annual discharge Q verses the plotting position
(Pp) is shown in fig from this graph flood discharge for different probabilities of time can be
predicted.
Table 05: Annual discharge in descending order
Year
Inflow
(m
3
/s)
Discharge
(Descending order ) (m
3
/s)
M
Plotting position

1996 837.52 14975 1 7.14
1997 63.97 9630 2 14.28
1998 14975 2555 3 21.42
1999 1387.319 2320.97 4 28.57
2000 559.489 1387.319 5 35.71
2001 1223.816 1223.816 6 42.85
2002 130.63 837.52 7 50
2004 302.27 559.489 8 57.14
2005 236.037 532.39 9 64.28
2007 2320.97 302.27 10 71.42
2008 532.39 236.037 11 78.57
2009 9630 130.63 12 85.71
2010 2555 63.97 13 92.85
32.497 %
814 Vol. 7, Issue 3, pp. 807-817

Graph 4: Accumulated volume Vs Percent time
for Flow duration Curve or Man basin Gauged at Sidhewadi station
4.5 Flood frequency study
Comparative flood frequency study was done by following two methods
4.5.1 Log Pearson Type III distribution
The Log Pearson Type III distribution was employed for flood frequency studies. The annual
maximum one day flood data was taken for the period of 13 years. From the data it was found that the
highest maximum one day flood of 2416.143 m
3
/s were recorded in the year 2009, whereas a
minimum flood of 16.92904 m
3
/s were recorded in 2005.The average maximum one day flood for the
man river basin over a period of 13 years was estimated as 449.31 m
3
/s with standard deviation of
0.5977. The maximum one day flood for recurrence interval of 100,200 and 1000 years were
determined as 4207.87m
3
/s, 5611.77 m
3
/s and 10071.6 m
3
/s respectively.
Table 06: Flood frequency studies of Man River Basin by log Pearson type III Distribution
Year Inflow
Max Flood
discharge
X
Z=Log x

1996 837.52 209.5074 2.321 0.006 0.000036 0.000000216
1997 63.97 63.97 1.805 -0.5101 0.2602 -0.1327
1998 14975 897.0745 2.952 0.6369 0.4056 0.2583
1999 1387.319 460.8193 2.6635 0.3479 0.1210 0.04210
2000 559.489 263.6401 2.4210 0.1059 0.01121 0.0011
2001 1223.816 252.2801 2.4018 0.0867 0.0075 0.00065
2002 130.63 30.49228 1.4841 -0.831 0.6905 -0.5738
2004 302.27 75.92681 1.8803 -0.4348 0.1890 -0.0821
2005 236.037 16.92904 1.2286 -1.0865 1.1804 -1.2825
2007 2320.97 698.1013 2.843 0.5279 0.2786 0.1471
2008 532.39 219.781 2.341 0.0259 0.0006 0.000017
2009 9630 2416.143 3.383 1.0679 1.1404 1.2178
2010 2555.89 236.4403 2.373 0.0579 0.0033 0.00019

Sum=
4.28834
Sum=
-0.40384

815 Vol. 7, Issue 3, pp. 807-817

Table 07: Result Table of log Pearson Type III distribution
T (Years)
z= 0.5977 = 2.3151 Cs = -0.186
Xt = antilog Zt
kz
(For Cs= -0.1860)
z kz

100 2.190 1.308 3.6231 4207.87
200 2.400 1.434 3.7491 5611.77
1000 2.825 1.688 4.0031 10071.6
Table: 08 Frequency Factors K for Gamma and log-Pearson Type III Distributions (Haan, 1977)
Skew Coefficient
Cs
Recurrence Interval In Years
1.0101 2 5 10 25 50 100 200
99 50 20 10 4 2 1 0.5
3 -0.667 -0.396 0.420 1.180 2.278 3.152 4.051 4.970
2.9 -0.690 -0.390 0.440 1.195 2.277 3.134 4.013 4.904
2.8 -0.714 -0.384 0.460 1.210 2.275 3.114 3.973 4.847
2.7 -0.740 -0.376 0.479 1.224 2.272 3.093 3.932 4.783
2.6 -0.769 -0.368 0.499 1.238 2.267 3.071 3.889 4.718
2.5 -0.799 -0.360 0.518 1.250 2.262 3.048 3.845 4.652
2.4 -0.832 -0.351 0.537 1.262 2.256 3.023 3.800 4.584
2.3 -0.867 -0.341 0.555 1.274 2.248 2.997 3.753 4.515
2.2 -0.905 -0.330 0.574 1.284 2.240 2.970 3.705 4.444
2.1 -0.946 -0.319 0.592 1.294 2.230 2.942 3.656 4.372
2 -0.990 -0.307 0.609 1.302 2.219 2.912 3.605 4.298
1.9 -1.037 -0.294 0.627 1.310 2.207 2.881 3.553 4.223
1.8 -1.087 -0.282 0.643 1.318 2.193 2.848 3.499 4.147
1.7 -1.140 -0.268 0.660 1.324 2.179 2.815 3.444 4.069
1.6 -1.197 -0.254 0.675 1.329 2.163 2.780 3.388 3.990
1.5 -1.256 -0.240 0.690 1.333 2.146 2.743 3.330 3.910
1.4 -1.318 -0.225 0.705 1.337 2.128 2.706 3.271 3.828
1.3 -1.383 -0.210 0.719 1.339 2.108 2.666 3.211 3.745
1.2 -1.449 -0.195 0.732 1.340 2.087 2.626 3.149 3.661
1.1 -1.518 -0.180 0.745 1.341 2.066 2.585 3.087 3.575
1 -1.588 -0.164 0.758 1.340 2.043 2.542 3.022 3.489
0.9 -1.660 -0.148 0.769 1.339 2.018 2.498 2.957 3.401
0.8 -1.733 -0.132 0.780 1.336 1.993 2.453 2.891 3.312
0.7 -1.806 -0.116 0.790 1.333 1.967 2.407 2.824 3.223
0.6 -1.880 -0.099 0.800 1.328 1.939 2.359 2.755 3.132
0.5 -1.955 -0.083 0.808 1.323 1.910 2.311 2.686 3.041
0.4 -2.029 -0.066 0.816 1.317 1.880 2.261 2.615 2.949
0.3 -2.104 -0.050 0.824 1.309 1.849 2.211 2.544 2.856
0.2 -2.178 -0.033 0.830 1.301 1.818 2.159 2.472 2.763
0.1 -2.252 -0.017 0.836 1.292 1.785 2.107 2.400 2.67
0 -2.326 0.000 0.842 1.282 1.751 2.054 2.326 2.576
-0.1 -2.4 0.017 0.846 1.27 1.716 2.000 2.252 2.482
-0.2 -2.472 0.033 0.850 1.258 1.680 1.945 2.178 2.388
-0.3 -2.544 0.050 0.853 1.245 1.643 1.890 2.104 2.294
-0.4 -2.615 0.066 0.855 1.231 1.606 1.834 2.029 2.201
-0.5 -2.686 0.083 0.856 1.216 1.567 1.777 1.955 2.108
-0.6 -2.755 0.099 0.857 1.200 1.528 1.720 1.880 2.016
-0.7 -2.824 0.116 0.857 1.183 1.488 1.663 1.806 1.926
-0.8 -2.891 0.132 0.856 1.166 1.448 1.606 1.733 1.837
-0.9 -2.957 0.148 0.854 1.147 1.407 1.549 1.660 1.749
-1 -3.022 0.164 0.852 1.128 1.366 1.492 1.588 1.664
-1.1 -3.087 0.180 0.848 1.107 1.324 1.435 1.518 1.581
-1.2 -3.149 0.195 0.844 1.086 1.282 1.379 1.449 1.501
-1.3 -3.211 0.210 0.838 1.064 1.240 1.324 1.383 1.424
-1.4 -3.271 0.225 0.832 1.041 1.198 1.270 1.318 1.351
816 Vol. 7, Issue 3, pp. 807-817

-1.5 -3.33 0.240 0.825 1.018 1.157 1.217 1.256 1.282
-1.6 -3.880 0.254 0.817 0.994 1.116 1.166 1.197 1.216
-1.7 -3.444 0.268 0.808 0.970 1.075 1.116 1.140 1.155
-1.8 -3.499 0.282 0.799 0.945 1.035 1.069 1.087 1.097
-1.9 -3.553 0.294 0.788 0.920 0.996 1.023 1.037 1.044
-2 -3.605 0.307 0.777 0.895 0.959 0.980 0.990 0.995
-2.1 -3.656 0.319 0.765 0.869 0.923 0.939 0.946 0.949
-2.2 -3.705 0.330 0.752 0.844 0.888 0.900 0.905 0.907
-2.3 -3.753 0.341 0.739 0.819 0.855 0.864 0.867 0.869
-2.4 -3.800 0.351 0.725 0.795 0.823 0.830 0.832 0.833
-2.5 -3.845 0.360 0.711 0.711 0.793 0.798 0.799 0.800
-2.6 -3.899 0.368 0.696 0.747 0.764 0.768 0.769 0.769
-2.7 -3.932 0.376 0.681 0.724 0.738 0.740 0.740 0.741
-2.8 -3.973 0.384 0.666 0.702 0.712 0.714 0.714 0.714
-2.9 -4.013 0.390 0.651 0.681 0.683 0.689 0.690 0.690
-3.0 -4.051 0.396 0.636 0.660 0.666 0.666 0.667
4.5.2 Gumbel distribution:
The annual maximum one day flood data were arranged in descending order. The recurrence interval
(T) and percent probability P was calculated. From the data it was found that the highest maximum
one day flood of 2416.143 m
3
/s were recorded in the year 2009, whereas a minimum flood of
16.92904 m
3
/s were recorded in 2005.The average maximum one day flood for the man river basin
over a period of 13 years was estimated as 449.31 m
3
/s with standard deviation of 646.14 m
3
/s. The
maximum one day flood for recurrence interval of 10,100 and 200 years were determined as 1538.17
m
3
/s, 3101.05 m
3
/s and 3553.96 m
3
/s respectively.
Table 09 Gumbels distribution
Sr
No Year
Max Flood
discharge
Max Flood
discharge
X

(

Order M Tp
1 1996 209.5074 2416.143 1966.84 3868471.3 1 14
2 1997 63.97 897.0745 447.77 200501.5 2 7
3 1998 897.0745 698.1013 248.801 61901.93 3 2.8
4 1999 460.8193 460.8193 11.51 132.68 4 0.7035
5 2000 263.6401 263.6401 -185.6 34469.6 5 2.8
6 2001 252.2801 252.2801 -197.02 38816.8 6 2.33
7 2002 30.49228 236.4403 -212.86 45309.3 7 2
8 2004 75.92681 219.781 -229.781 52678.9 8 1.75
9 2005 16.92904 209.5074 -239.79 57500.68 9 1.55
10 2007 698.1013 75.92681 -373.37 139408.14 10 1.4
11 2008 219.781 63.97 -385.37 148513.11 11 1.27
12 2009 2416.143 30.49228 -418.80 175400.14 12 1.16
13 2010 236.4403 16.92904 -432.37 186944.68 13 1.07

Mean=449.31 x
= 449.3
(x- )
2
=
5010048.76

Table 10: Gumbels distribution Result
T (Years)
n-1= 646.14
= 449.3

K k n-1
xt = + k n-1
10 1.6852 1088.87 1538.17 m
3
/s
100 4.104 2651.75 3101.05 m
3
/s
200 4.804 3104.05 3553.96 m
3
/s
1000 6.4186 4147.35 4596.62 m
3
/s
817 Vol. 7, Issue 3, pp. 807-817

V. CONCLUSIONS
-From the mass inflow curve, it was estimated that a quantity of 4.60 Mm
3
water is available for
storage if small dam is constructed.
- From the hydrological studies it was estimated that the annual rainfall for Sidhewadi station is about
627.64 mm
-From the available records of annual discharges, flow duration curve was established.
- Log Pearson type III distribution and Gumbles method were used to calculate flood frequency study.
- It was found that by Log Pearson type III distribution the maximum one day flood for recurrence
interval of 100, 200 and 1000 years were determined as 4207.87 m
3
/s, 5611.77 m
3
/s and 10071.6 m
3
/s
respectively.
-Also It was found that by Gumbels distribution for return period of period of 10, 100 , 200 and 1000
years, the maximum one day flood discharges were 1538.17 m
3
/s , 3101.05 m
3
/s , 3553.96 m
3
/s and
4596.62 m
3
/s respectively.
-These are the prerequisite studies for the construction of any hydraulic structure across the Man river
basin.
REFERENCES
[1] Arora, K.R. 2007. Irrigation, Water Power and Water resources engineering. Standard Publishers
Distributors, New Delhi
[2] Asawa, G.L. 2005. Irrigation and Water resources engineering. New Age International Ltd. Publishers, New
Delhi
[3] Chow, V. T. (1964). Handbook of Applied Hydrology. McGraw-Hill, New York.
[4] Gumbel E.J.(1958) Statistics of extremes Columbia University ,New York.
[5] Gumbel E.J.(1954) Statistical theory of droughts Proc. American Society of civil engineers,Volume 80 no
439 pp 1-19
[6] Haan, C. T. (1994). Statistical Methods in Hydrology. Iowa State University Press. Ames.
[7] Ibrahim M.H , and E.A. Isiguzo, 2009. Flood Frequency Analysis of Guara River Catchment at Jere, Kaduna
State, Nigeria,. Scientific Research and Essay Vol. 4 (6), pp. 636 646.
[8] M.V.Manjunatha and G.S.Somnatha (2001): Hydrological studies of Netravathy river basin, Indian Journal
of Power and river valley development,Vol.,1,Jan-Feb 2001 pp 22-28.
[9] Sathe B. K. (2012) ; Flood Frequency Analysis of Upper Krishna River Basin catchment area using Log
Pearson Type III Distribution IOSR Journal of Engineering (IOSRJEN) ISSN: 2250-3021 Volume 2, Issue 8
(August 2012), PP 68-77.
[10] Subramanya k.(1984):Engineering Hydrology; Tata McGraw hill publishing company Ltd, New Delhi
[11] Weibull W (1939); A statistical theory of the strength of Materials Ing.Ventenskps Akod Handi Vol 151
pp15
[12] Todkari G.U. (2012) ; Impact of irrigation on agriculture productivity in solapur district of Maharashtra
state.
AUTHORS BIOGRAPHY
Shirgire Anil Vasant is a student of M.Tech (Civil Engg) course in Bharti Vidyapeeth
College of Engineering, Pune. He has CGPA-8.34 upto sem III for M.tech-Civil course.

S. D. Talegaonkar, is working as an Assistant Professor in Bharti Vidyapeeth College of Engineering Pune. He
has total 10 years of teaching work experience. He has guided so many student in UG level for projects.
818 Vol. 7, Issue 3, pp. 818-826

CROP DETECTION BY MACHINE VISION FOR WEED
MANAGEMENT
Ashitosh K Shinde and Mrudang Y Shukla
Dept. of Electronics and Telecommunication
Symbiosis Institute of Technology, Pune, Maharashtra, India

ABSTRACT
Weed management is one of the costliest input to the agriculture and it is one of the un-mechanised area. To
bring mechanization in this area the most important step is the detection of weed in agricultural field. Weed can
be detected by using machine vision techniques. Machine vision uses special image processing techniques.
Weeds in agricultural field can be detected by its properties such as Size, Shape, Spectral Reflectance, Texture
features. In this paper we are demonstrating weed detection by its Size features. After the image acquisition
Excessive green algorithm is developed to remove soil and other unnecessary objects from the image. Image
enhancement techniques are used to remove Noise from the images, By using Labelling algorithm each
components in the Image were extracted, then size based features like Area, Perimeter, longest chord and
longest perpendicular chord are calculated for each label and by selecting appropriate threshold value Weed
and Crop segmentation is done . Result of all features is compared to get the best result.
KEYWORDS: Machine vision, Camera, Area, Perimeter, Longest chord, longest perpendicular chord.
I. INTRODUCTION
Weeds have been existing on earth since men started cultivating. Every vegetation present in the
agricultural field which is unwanted is called as weed. Weeds compete with crop for Sunlight, Space,
Water and Nutrients in the soil. Weeds are the most underestimated crop pests in tropical agriculture
although they cause maximum reduction/loss in the yields of crops than other pests and diseases. The
total annual loss of agricultural produce from various pests in India, weeds roughly account for 37%
[15]. They decrease quantity and quality crop yield and cause health hazards for humans and animals.
Thus weed management is most important in every crop production system. Weeds are one of the
major constraints in agricultural production. As per the available estimates, weeds cause up to one-
third of the total losses in yield, besides impairing produce quality and various kinds of health and
environmental hazards. Despite development and adoption of weed management technologies [14],
the weed problems are virtually increasing. This is due to intercropping, mulching and crop rotations
involving shift in weed flora, due to adoption of fixed cropping systems and management practices
including herbicides development of herbicide resistance in weeds e.g. Phalaris minor in the 1990s
growing menace of wild rice in many states and Orobanche in mustard growing areas invasion by
alien weeds like Parthenium, Lantana, Ageratum, Chromolaena, Mikania and Mimosa[14]in many
parts of the country impending climate change favoring more aggressive growth of weed species, and
herbicide residue problems. This suggests that weeds problems are dynamic in nature, requiring
continuous monitoring and refinement of management strategies for minimizing their effects on
agricultural productivity and environmental health. A number of factors affects the quality and
quantity of yield such as competitiveness of crop and weed present, density of crop and weed present,
time of emergence of the weed relative to the crop, duration of weed present. The paper is organised
as follows, In Section 1 literature survey, Filed survey is discussed, Section 2 Image acquisition,
819 Vol. 7, Issue 3, pp. 818-826

Excessive green algorithm, Image Enhancement, Size based feature extraction and Crop masking is
given, Conclusion and future work is discussed in section 3
1.1: Related Work
In biological morphology based techniques shape and size recognition can be conducted. Most
machine vision research on plant species identification has been done at leaf geometry level with
some at whole plant level. Biological morphology can be defined as the shape and structure of an
organism or any of its part. A wide range of machine vision shape features for leaf or plant detection
is used such as area, length, width, perimeter etc. and generally achieve high recognition rates under
ideal conditions.
Weed detection processes included were normalized excessive green conversion, statistical threshold
value estimation, adaptive image segmentation, median filter, morphological feature calculation and
Artificial Neural Network (ANN) by Hong Y. Jeon Lei F. Tian. With an accuracy of 72.6%[2]. For
crop and weed detection the seven shape features extracted from the images, four were selected by
discriminate analysis which was able to classify the two groups with 98.9% accuracy. This method is
developed to detect only corn crop in the field. By S. Kiani, A. Jafari .[3][15][16].
1.2: Field Survey
To get familiar with Indian agricultural field condition and to gather information about farmers
expectation from this project field survey is carried out in Loni kalbhor, Pune, Maharashtra, India.
According to farmer Onion and Sugar Cane is the main crop taken by maximum farmer in that area.
One of the costliest inputs to the agriculture is the De-Weeding which is 21.15%, 31.81% and 21.87%
of total expenses for onion, Sugar Cane and corn respectively as shown in table 1. Currently the
farmers are using expensive herbicides to kill the weed or the weeds can be removed manually which
is labor dependent task.
Table 1 De Weeding Expenses for 1 Acre
Crop Onion Sugar Cane Corn
Time Period 4 Month 1 Year 2 Month
Total Yield 15,000KG 60,000KG 3000KG
Market Price Rs 10/KG Rs 2/KG Rs 13/KG
Total Earning Rs 150000 Rs 1,20000 Rs 39000
Expenses
Labor Charges Rs 10000 Rs 3000 Rs 3000
Fertilizers Rs 3000 Rs 3000 Rs 3000
Transport Rs 6000 NA Rs 5000
Pesticides Rs 1500 Rs 1500 Rs 1500
Total Rs 20500 Rs 7500 Rs 12500
De-Weeding
After 10 Days NA NA Rs 2000
After 25 Days Rs 2000 Rs 2000 Rs 1500
After 2 Month Rs 2000 Rs 1500 NA
After 3 Month Rs 1500 NA NA
Total Rs 5500 Rs 3500 Rs 3500
II. MATERIALS AND METHODS
2.1 Image Acquisition
The digital images were captured under perspective projection and stored as 24-bit color images with
resolutions of 5MP saved in RGB (red, green and blue) color space in the JPG format. The images
were processed with MATLAB R2010 under Windows 7 and Intel Core i3-2370M CPU, 2.4 GHz, 2
GB RAM. The images were taken in month of March, April 2013. With different angles. The result of
a project of this type relies heavily on the quality of the photo material that is used as input. Ideally we
820 Vol. 7, Issue 3, pp. 818-826

want an image acquisition system that is robust in different lighting and weather conditions. But it is
also important to keep the photo acquisition process uncomplicated since the system should be easy to
use. All photographs used in this project were taken under natural lighting conditions. During all
image acquisition the camera was pointing directly towards the ground.
2.2 Excessive Green
After the image acquisition it is necessary to remove unwanted information from the image. It is
necessary to segment image pixels into vegetation and Non vegetation. For this excessive green color
extraction algorithm is developed.
Outimage (x,y,z) = inimage (x,y,z) if
{
inimage (x, y, r) < (x, y, g)
inimage(x, y, b) < (x, y, g)

}

outimage(x,y,z) = 0 otherwise
where outimage (x,y,z) is the output image after excessive green segmentation saved in jpg format,
inimage(x,y,z) is the image acquired by an camera, x is the no of pixels in each row, y is the no of
pixels in each column and z is the primary color plane for red the z is equal to 1, for green the z is 2
and for blue the z is 3. Input image is shown in figure 1 and output image is shown in figure 2[5-8]

Fig 1. To be processed Image

Fig 2. Excessive Green
821 Vol. 7, Issue 3, pp. 818-826

2.3 Image Enhancement
The aim of image enhancement is to improve the interpretability or perception of information in
images to provide better input for automated image processing techniques. In the proposed system
spatial domain image enhancement techniques are used.
4.6.1 RGB to gray image conversion:
When converting an RGB image to grayscale, we have to take the RGB values for each pixel and
make as output a single value reflecting the brightness of that pixel. One such approach is to take the
average of the contribution from each channel: (R+B+C)/3. However, since the perceived brightness
is often dominated by the green component, a different, more "human-oriented", method is to take a
weighted average, e.g.: 0.3R + 0.59G + 0.11B.
2.3.1 Median Filtering
Median filtering is a nonlinear method used to remove noise from images. It is widely used as it is
very effective at removing noise while preserving edges. It is particularly effective at removing salt
and pepper type noise. The median filter works by moving through the image pixel by pixel,
replacing each value with the median value of neighboring pixels. The pattern of neighbors is called
the "window", which slides, pixel by pixel over the entire image pixel, over the entire image. The
median is calculated by first sorting all the pixel values from the window into numerical order, and
then replacing the pixel being considered with the middle (median) pixel value.
2.3.2 Intensity Adjustment
Intensity adjustment is an image enhancement technique that maps an image's intensity values to a
new range.
2.4 Labeling Algorithm
Connected component labeling works by scanning an image, pixel-by-pixel (from top to bottom and
left to right) in order to identify connected pixel regions, i.e. regions of adjacent pixels which share
the same set of intensity values V. Connected component labeling works on binary or gray level
images and different measures of connectivity are possible. However, for the following we assume
binary input images and 8-connectivity. The connected components labeling operator scans the image
by moving along a row until it comes to a point p (where p denotes the pixel to be labeled at any stage
in the scanning process) for which V={1}. When this is true, it examines the four neighbors of p
which have already been encountered in the scan (i.e. the neighbors (i) to the left of p, (ii) above it,
and (iii and iv) the two upper diagonal terms). Based on this information, the labeling of p occurs as
follows:
If all four neighbors are 0, assign a new label to p, else
if only one neighbor has V={1}, assign its label to p, else
if more than one of the neighbors have V={1}, assign one of the labels to p and make a note
of the equivalences.
After completing the scan, the equivalent label pairs are sorted into equivalence classes and a unique
label is assigned to each class. As a final step, a second scan is made through the image, during which
each label is replaced by the label assigned to its equivalence classes. For display, the labels might be
different gray levels or colors.
2.5 Size based feature Extraction
Size based features can be extracted by using Mathematical morphology. Morphology is an approach
to image analysis which is based on the assumption that an image consists of structures which may be
handled by set theory. This is unlike most of the rest of techniques. As it can be seen in Figure 3, there
is a significant difference between the sizes of corn leaves and the leaves of the weeds.[4]

822 Vol. 7, Issue 3, pp. 818-826

Fig. 3 Corn crop with Weed

2.5.1 Area
Area of any object in an image can be defined as number of pixels in that region. It possible to
Differentiate Weed from Crop By analysing area features of each object in an image which are
detected after performing the Labelling algorithm. From figure 4 it is clearly seen that the area
mapped by corn crop is higher than that mapped by a weeds. By selecting appropriate area of an
object weed and crop can be easily identified. In Figure 4. The crop of corn is identified.[4]

Fig.4 Area Based Crop detection
2.5.2: Perimeter
Perimeter of any object in an image can be defined as number of pixels in object boundary of that
region. It possible to Differentiate Weed from Crop By analysing Perimeter features of each object in
an image which is detected after performing the Labelling algorithm. From figure 5 it is clearly seen
that the Perimeter of corn crop is higher than that mapped by a weeds. By selecting appropriate
Perimeter of an object weed and crop can be easily identified. In Figure 5 the crop of corn is
identified.[4]

823 Vol. 7, Issue 3, pp. 818-826

Fig.5 Perimeter Based Crop detection
2.5.3: Longest Chord
A line connecting two pixels on the object boundary is called a chord. For many shapes, the length
and orientation of the longest possible chord give us an indication of size and orientation of the object.
If the chord runs between the boundary pixels (x1, y1) and (x2, y2), then its length lc and orientation
are given by equation 2 . It possible to Differentiate Weed from Crop By analysing Longest Chord
features of each object in an image which is detected after performing the Labelling algorithm. From
figure 6 it is clearly seen that the significant difference between Longest Chord of corn crop and
weeds. By selecting appropriate Longest Chord Length of an object weed and crop can be easily
identified. In Figure 6 the crop of corn is identified.
= (2 1)
2
+(2 1)
2
(2)

Fig.6 Longest Chord Based Crop detection
2.5.4: Longest perpendicular chord
Longest perpendicular chord of any object in an image can be defined as the maximum length of all
chords that are perpendicular to the longest chord can give us additional shape information. It
possible to Differentiate Weed from Crop By analyzing longest perpendicular chord features of each
object in an image which is detected after performing the Labelling algorithm. From figure 7 it is
clearly seen that the Longest perpendicular chord of corn crop is higher than that mapped by a weeds.
By selecting appropriate Longest perpendicular chord Length of an object weed and crop can be
easily identified. In Figure 7 the crop of corn is identified.[4]
824 Vol. 7, Issue 3, pp. 818-826

Fig. 7 Minor Axis
2.6 Crop Masking
After Successful detection of weed by different techniques it is necessary to find the weed in an image
for that crop detected is masked with black color by finding the origin, length, width of each bounding
box the result is shown in figure 8.

Fig. 8 Crop Masking
2.7 Weed Detection
After the crop masking weed is detected by applying excessive green algorithm the weed is detected.
The bounding box algorithm is performed on the image to map the weeds. Once the weed is mapped
by bounding box. Algorithm to find the co-ordinate of the each bounding box is developed and
performed and the co-ordinate of each detected weed is printed on the Image as shown in figure 9.
825 Vol. 7, Issue 3, pp. 818-826

Fig. 8. Weed Detection
III. CONCLUSION & FUTURE WORK
Image processing algorithm for detection of weed in the indian agriculuteral field for the management
of weed is successfully developed and tested. For each stages of De weeding the thresholding
parameter for each feature is calculated for Suger cane and corn crop. Detected Weed coordinates
can be used for calculation of actuation parameters. Through the serial communication calculted
Coordinates can be given to the robot controller. The developed algorithm is simple and faster than
the algorithms which uses artifisial inteligence techniques as the number of calculation in this
algorithm is much lesser than that of AI Algorithms as their is no requirement training algorithms and
Huge Database. Accuracy of the algorithm can be increase by using the more features and localized
image processing techniques. In the future Accuracy of the algorithm can be incresed by using
spectral reflectance features based weed detection and texture features based weed detection, It is
poosible to develop Robotic machine which will Run through Agriculutral field and By using the
weed Co-ordinates it can Spray herbisides on perticuler weed plant precisely or by using mechanical
tool it Can Up-Root the Weeds.
REFERENCE
1.] D.C. Slaughter, D.K. Giles, D. Downey Autonomous robotic weed control systems: A review. computers
and electronics in agriculture 6 1 ( 2 0 0 8 ) 6378
2] Hong Y. Jeon ,, Lei F. Tian and Heping Zhu, Robust Crop and Weed Segmentation under Uncontrolled
Outdoor Illumination, Sensors 2011, 11, 6270-6283; doi:10.3390/s110606270
3] S. Kiani, and A. Jafari, Crop Detection and Positioning in the Field Using Discriminant Analysis and Neural
Networks Based on Shape Features, J. Agr. Sci. Tech. (2012) Vol. 14: 755-765
4] Kamal N. Agrawal, Karan Singh, Ganesh C. Bora and Dongqing Lin, Weed Recognition Using Image-
Processing Technique Based on Leaf Parameters, Journal of Agricultural Science and Technology B 2 (2012)
899-908
5] Xavier P. Burgos-Artizzu , Angela Ribeiro , Alberto Tellaeche , Gonzalo Pajares , Cesar Fernndez-
Quintanilla Analysis of natural images processing for the extraction of agricultural elements. Image and Vision
Computing 28 (2010) 138149
6] J. Romeo, G. Pajares, M.Montalvo, J.M. Guerrero, M. Guijarro, and A. Ribeiro, Crop Row Detection
inMaize Fields Inspired on the Human Visual Perception, The ScientificWorld Journal Volume 2012, Article ID
484390
7] Muhammad Asif, Samreen Amir, Amber Israr and Muhammad Faraz A Vision System for Autonomous
Weed Detection Robot, International Journal of Computer and Electrical Engineering, Vol. 2, No. 3, June, 2010
1793-8163
826 Vol. 7, Issue 3, pp. 818-826

8] Xavier P. Burgos-Artizzu,, Angela Ribeiroa, Alberto Tellaecheb, Gonzalo Pajaresc, Cesar Fernndez-
QuintanilladImproving weed pressure assessment using digital images from an experience-based reasoning
approach. computers and electronics in agriculture 6 5 ( 2 0 0 9 ) 176185
9] J. Bossua, Ch. Gea,, G. Jones, F. TruchetetbWavelet transform to discriminate between crop and weed in
perspective agronomic images. computers and electronics in agriculture 6 5 ( 2 0 0 9 ) 133143
10.] Sajad KAN Crop-Weed Discrimination Via Wavelet-Based Texture Analysis Internatonal Journal of
Natural and Engineering Sciences 6 (2) : 7-11 , 2012
11.] Alberto Tellaeche, Gonzalo Pajares A computer vision approach for weeds identification through Support
Vector Machines Applied Soft Computing 11 (2011) 908915
12.] Lanlan Wu, Youxian Wen, Xiaoyan Deng and Hui Peng Identification of weed/corn using BP network
based on wavelet features and fractal dimension. Scientific Research and Essay Vol.4 (11), pp. 1194-1200,
November, 2009
13] Anup Vibhute, S K Bodhe Applications of Image Processing in Agriculture: A Survey. International Journal
of Computer Applications (0975 8887)
14] Imran Ahmed, Awais Adnan, Muhammad Islam,Salim Gul, Edge based Real-Time Weed Recognition
System for Selective Herbicides. IMECS 2008, 19-21 March, 2008, Hong Kong
15] T. K. Das, Weeds and their Control Methods Division of Agronomy, Indian Agricultural Research Institute,
New Delhi 110 012
16] Dr. A.R. Sharma, vision 2050 ,the Director, Directorate of Weed Science Research Jabalpur-482 004 (M.P.),
India
AUTHORS BIOGRAPHY
Ashitosh Shinde pursuing the M.tech degree in Electronics & Telecommunication
Engineering from Symbiosis International University, Pune. He has Bachelor of engineering
degree from University of Pune, India. He has Diploma degree from Cusrow Wadia Institute
of Technology. His research interests include Technology in agriculture, Image processing,
Robotics, Embedded systems.

Mrudang Shukla is assistant professor at Symbiosis Institute of Technology in Electronics
and Telecommunication department. His research interest is in image processing and defining
vision and path for automated vehicle in agricultural field. He has M.Tech in Electronics and
Communication System DDU(Dharmsinh Desai University, Nadiad, Gujarat and BE in
Electronics and Telecommunication D N Patel COE Shahada, North Maharatra University
NMU, Jalgaon)

827 Vol. 7, Issue 3, pp. 827-837

DETECTION & CONTROL OF DOWNEY MILDEW DISEASE IN
GRAPE FIELD
Vikramsinh Kadam
1
, Mrudang Shukla
2

1
Mtech (E&TC),
2
Assisstant Professor
Symbiosis institute of Technology, Pune, Maharashtra, India

ABSTRACT
Grape field is greatly affected by Downey mildew disease. The development of the disease is faster, within six
hour the spread of the disease get multiplied by twice. Once it is affected it will diminish the quantity & quality
of Grapes, it reduce the Photosynthesis process. For this, disease detection is an important part. By using
detection technique we can prevent the disease. Detection & control strategies are carried over in order to
control Downey mildew disease. The electromechanical system we have used. In this system we have used
Raspberry pi module for disease detection. The raspberry pi is the module which has image processing
capabilities. Traditionally farmer visually checks the disease. If disease is there, he applies pesticide spray
manually. In our system disease is detected by raspberry pi module, which is installed on the robo car. As soon
as disease is detected, detected signal is transmitted to another electromechanical module. Then pesticide is
sprayed automatically on infected area by using electromechanical system. The farmers need not to go in the
farm to check each leaf. Prevention is always better than cure. Instead of waiting for disease development, we
can prevent this disease on Grape field. If disease is prevented then Export quality grapes can be produced &
Farmer can have more profit from Grape production.
KEYWORDS: Image Processing, Robotics, Control of Downey mildew disease, Raspberry pi module
I. INTRODUCTION
In the Grape field Downey mildew disease is the biggest threat to the plant. Native of Downey
mildew disease is North America. Downey mildew disease caused by fungus Plasmopara Viticola.
Vinifera cultivars is the most susceptible for this disease, Wild species are more resistant. Downey
mildew comes naturally in the rainy season when humidity of environment is high. In the first 40 to
65 days after the cutting of plants for grape production, the leaves of Grapes are delicate & immature.
At that time this disease comes. It can reduce profitability by 50%. The correct identification should
be done in time. Little delay in identification can harm plant.
There are two favorable conditions for Downey mildew disease development as 10 to 23 degree
temperature or 23 to 27 degree temperature with relative humidity greater than 80%.then destruction
to the grape starts. Downey Causes deformed shoot, cluster growth reduction, premature defoliation
causes delayed ripening of fruit, young berries will turn light brown, becomes soft then fall off the
cluster easily. Downey mildew disease comes of fungus growth on the back side of the leaf. Thats
why its name is Downey mildew. Before proceeding towards detection we should know how disease
comes. If temperature remains in between 10 to 23 degree. This plasmopara viticola pathogen grows
rapidly. Another suitable temperature for this is from 23 to 27 degree with greater than 80% relative
humidity.
Once the fungus grows on back side of leaf. It finds stomata to enter in to the leaf tissue. If they dont
get sufficient amount of stomata they will break 3 layers of leaf & get entry in to the leaf tissue. These
three layers are cutin, pectin & cellulose. Once it enters in to the leaf tissue. Its effect comes in the
form of yellowish leaves in the first 40 to 65 days. As mentioned earlier, if pathogen doesnt get
828 Vol. 7, Issue 3, pp. 827-837

enough stomatas to enter in to leaf, it will enter by breaking 3 layers. Because of all these leaves are
delicate in these first 40 to 65 days, it is easy to break three layers. Once the leaves are getting older,
then there is no threat of Downey mildew disease as its immune system increased to resist that
disease. The fungus creates plasmopara viticola pathogen, which creates oogonium, oospores,
sporangia, zoospores & it creates infection to leaves. In another cycle it creates sporangia, zoospores
& it creates infection to leaves.
a) Need of detection & control:
Two cycles of disease development:
First cycle: oogonium->oospores->sporangia->zoospores
Second cycle: sporangia->zoospores
1) Disease is identified Visually by farmers. But when they identify this disease, the disease gets well
developed. To avoid this we can use image processing technique to identify the disease in early stage.
2) The duration of first cycle is 2 days & duration of second cycle is 6 hours. If we failed to detect this
disease for first 2 days then this disease pathogen production get multiplied by every 6 hours. It will
create huge destruction to grape field & ultimately it reduces profitability. So detection of disease is
more important
b)Statistics about grape field of farmer at Location Phaltan:
We visited the farmer named as Sampatro kadam At Post: Girvi, Taluka: Phaltan, District: Satara,
State: Maharashtra. They are grape producer since 1999. They have used grape variety named as
Tashi ganesh. The first cutting of grape field is done in the month of February. Once the cutting is
done they grow & nourish the stake. The duration of this process is 150 days. In these 150 days grape
stake get developed. After 150 days once again cutting is done at the top side. This is the starting of
grape production development process. This process is carried over for 105 days. In these days there
is great threat to grape field by Downey Mildew disease. In these 105 days total pesticide cost goes up
to 60000 to 70000 rupees.
In one acre the production of grape field is up to 8 to 10 tonne. The harvesting is done in October. In
these days grapes get exported. Farmer gets 80 to 100 rupees per kilogram of grape. Thus in one acre
of grape field, farmer gets 8 to 10 lakh rupees. Total expenditure is 1.5 to 2 lakh. Farmer has profit of
7 to 9 lakh rupees in one acre of farm.
When plasmopara viticola get sufficient amount of atmosphere (10 to 23 degree Celsius temperature
& greater than 80% relative humidity). It starts affecting grape leaves, berries& twigs. It forms
whitish growth on the back side of leaf. These spot kills the plant tissue. Thus photosynthesis process
gets stopped at there. Let us see how disease gets created. [41]
1) There are organs (Antheridium) [41] Producing male gamets & immature ovarian egg (Oogonium)
within which fetus is developing. The fusion (Karyogami) of these two nuclei is done. It produces
fertilized female zygote (oospores). Germination of oospores leads to sporangium. Sporangium is
nothing but container in which Sporangias are stored. In one sporangium contains near about 40000
to 50000 sporangia. Germination of sporangia leads to zoospores. Zoospore again infects leaves,
debris & twigs. This process takes 3 to 6 days.
2) The dormant twigs [41] get affected due to vegetative part of fungus (mycelium).it again forms
sporangia & then zoospores are created. This process takes 6 to 8 hours.
An infected leaf gives sporangia. It again forms zoospores. This process takes 6 to 8 hours. Therefore
detection is required for first step.
II. DEVELOPMENT OF DOWNEY MILDEW DISEASE
Leaf color changes due to disease:
a) Stage 1: Creation of white spots on backside of leaf
Plasmopara viticola is the pathogen responsible for Downey mildew disease. It creates the white spots
on the back side of leaves. It is shown in figure no.1.
829 Vol. 7, Issue 3, pp. 827-837

Figure 1: White spot of Downey mildew disease on backside of leaf
b) Stage2: creation of yellow spots on upper side of leaf
As seen in first stage white spots are present on backside of leaf. Exactly on the opposite side (upper
side) of white spot (downside).yellow spot comes. Primary stage of yellow spot development is
shown in figure no.2.

Figure 2: Yellow spot of Downey mildew disease on upper side of leaf
Full development of yellow spots on leaf (on upper side) is shown in figure.no.3

Figure 3: Dark Yellow spot created
c) Stage 3: Leaf get damaged at areas where white spots come
This Disease kills the leaf tissues. It is shown in figure no.4. Its color turns to brownish. It stops the
photosynthesis process. Then it affects the production of Grape
830 Vol. 7, Issue 3, pp. 827-837

Figure 4: Yellow spot kills the tissue of leaf
III. ACTUAL DETECTION FOR DOWNEY MILDEW DISEASE
a) White spot detection
Detection of white spots is very important. Because this is a primary stage of disease. So with the help
of image processing toolbox of Matlab, We can detect it. First we have converted this binary image to
double precision intensity image. Then green color thresholding is done. That image is converted to
gray. To reduce salt & pepper noise we used median filtering. After that we complement the image. In
morphological operation, we removed connected components less than 400. Then we find perimeter
for less than 800 & then locate that point. We applied Excessive white condition on infected leaf. fig.
No.5 shows Excessive white condition. fig. No. 6 shows thresholding & fig.no.7 shows white spot
detection.

831 Vol. 7, Issue 3, pp. 827-837

Flow chart no.1: White spot detection

Fig. no.5: Gray image Fig. no.6: Morphological operation Fig. no.7: White spot detected
b) Yellow spot detection:
First we have converted this binary image to double precision intensity image. Then green color
thresholding is done. That image is converted to gray image. To reduce salt & pepper noise we used
median filtering. After that we complement the image. In morphological operation remove
connected component less than 400.then we find perimeter for less than 800 & then locate that point.
Gray image is shown in fig.no.8. Binary image is shown in fig.no.9.then yellow spot detection is
done, as shown in fig.no.10
832 Vol. 7, Issue 3, pp. 827-837

Flow chart no.2: Yellow spot detection

Fig. no.8: Gray image Fig. no.9: Morphological operation Fig. no.10: Yellow spot detected
IV. CONTROL OF DOWNEY MILDEW DISEASE
a) Raspberry pi module
The Raspberry pi module, Robocar, pesticide sprinkling system is used to control the disease.
833 Vol. 7, Issue 3, pp. 827-837

Fig.no.11: Raspberry pi module
Raspberry pi consist of following parts:
1) ARM 11 with 700 MHz frequency
2) Ethernet port
3) Two USB port
4) SD card slot
5) Micro USB power
6) HDMI port for communication
7) GPIO ports
The Raspberry Pi is a credit-card sized computer that plugs into your TV and a keyboard. It is shown
in fig no.11.It is a capable little computer which can be used in electronics projects, and for many of
the things that your desktop PC does, like spreadsheets, word-processing and games. It also plays
high-definition video. The Raspberry Pi measures 85.60mm x 56mm x 21mm (or roughly 3.37 x
2.21 x 0.83), with a little overlap for the SD card and connectors which project over the edges. It
weighs 45g. The Soc is a Broadcom BCM2835. This contains an ARM1176JZFS, with floating point,
running at 700 MHz; and a Video core 4 GPU. The GPU is capable of Blue Ray quality playback,
using H.264 at 40MBits/s. The camera module is an Omni vision 5647. It is comparable to cameras
used in mobile phones.
b) Detection System

834 Vol. 7, Issue 3, pp. 827-837

Fig.no.12: Block diagram of Detection system Fig.no.13: Hardware of detection system
At first we downloaded the Raspbian operating system. The file size is 577 MB. After extraction of
that file, it becomes 2.75 GB. This operating system is taken in to 4 GB memory card. Then it is
inserted in to raspberry pi module with help of SD card reader. Raspberry pi uses Pithon
programming. Open CV is a library from which we fetch the function in pithon programming. In fig.
No. 11. We have shown Pibot. Pibot is nothing but a robo car which has four wheels. We used two
DC motors, These DC motors require 12 volt for operation. We selected its rpm is 30.Also to run this
pibot we use 12 volt battery. On that pibot we install Raspberry pi module & camera is connected to
the Raspberry pi module. Zigbee module we have used. It is used for the wireless transmission.
Raspberry pi module & Zigbee module is on the pibot. This pibot will run in the grape field. Camera
face is in the upward direction. As soon as disease is detected by Raspberry pi module, That detected
signal is transmitted through Zigbee module. The range of Zigbee module is 100 meters. The grape
field consists of number of rows, our robot will move in between these rows. Robot can run in
between these rows because there are no ups & downs in this area.
c) Control System
\
Fig.no.14: Block diagram of control system Fig.no.15: Microcontroller based hardware
The system shown in fig.no.13 indicates control of Downey mildew Disease. The block diagram &
Hardware design is shown in fig.no.13. The detected signal is received by Zibee module. As earlier
said its range is 100 meters. This zigbee module connected to the microcontroller AT89C51 through
MAX 232 IC. The 12 volt relay we can use. But it is also depends on which kind of pesticide pump
835 Vol. 7, Issue 3, pp. 827-837

we are using. If it is three phase then we can use contractor for switching on & off the pesticide pump.
These pesticide pumps are connected to the pesticide sprinklers. There are number of pesticide
sprinklers we can use in entire grape field. It is distributed in all rows of grape field through pesticide
pipes.
Suppose our Robo car is going through a first row & it has detected the disease in first row. Then
detection system will send signal to control system that disease is detected through Zigbee module.
Then control system will on the particular sprinklers in first row. Therefore we have used valves to on
sprinklers in first row. Solenoid valve is normally closed. But when we energize the solenoid coil,
solenoid valve becomes open. With the help of solenoid valve, we can switch on particular set of
sprinklers. If we want to on all the sprinklers at a time, then it will create low pressure at the output of
sprinklers. Thus by using number of solenoid valves, we can spray pesticide in to whole Grape field
V. FUTURE WORK
In our system we have detected & control of downey mildew disease. We can use GSM based system
to operate robo. If a farmer is far away from a Grape field & he wants to check for disease, then he
can operate that robo from a long distance. The system used in our paper is only for one disease in
Grape field. We can integrate all Grape field diseases. If we integrate all the diseases then it will give
complete solution for every disease in Grape field. Apart from Grape field, We can use this
technology for identifying diseases of the crops who has canopy like structure.
VI. CONCLUSION
As plasmopara viticola is dangerous pathogen causing Downey Mildew Disease. it reduces
Profitability of farmer. With the help of cultural practices we can reduce the disease up to some
extent. Therefore we can use novel approach to reduce this disease by detection. Prevention is better
than controlling the disease. As detection will give precautionary message, that will lead to prevent
the disease. Detection & control technology is reliable, accurate. It will reduce labor cost, cost on
pesticides, and destruction due to Downey mildew diseases & finally it will lead to increase
production & profit.
REFERENCES
[1] Sindhuja sankaran, Ashish Mishra, Reza Ehsani, Cristina Davis, 2010, a review of advanced techniques for
detecting plant diseases, Computers and Electronics in Agriculture 72
[2] Cesare Gessler, Ilaria, Pertot & Michele Perazzolli, 2011, plasmopara viticola: a review knowledge on
Downey mildew of grapevine& effective disease management.
[3] Veronica Saiz-rubio, Francisco Rovira Mas, 2013, proximal sensing mapping method to generate field maps
in vineyards
[4] Jaime lloret, Ignacio Bosch, Sandra Sendra, Arturo Serrano, 2011, wireless sensor network for vineyard
monitoring that uses image processing, ISSN 1424-8220.
[5] Anushka Srivastava, 2010, Robo kisan-a helping hand to the farmer.
[6] Anushka Srivastava & Swapnil Kumar Sharma, 2010, development of a robotic navigator to assist the
farmer in field.
[7] Federico Hahn, 2009, actual pathogen detection : sensor & algorithm, ISSN 1999-4893.
[8] R.C.Seem, P.A.Magarey, P.I.Mccloud & M.F, Wachlet, 1985, a sampling procedure to detect grapevine
Downey mildew.
[9] N. Lalancette, L. V. Madden, and M. A. Ellis, 1988, A Quantitative Model for Describing the Sporulation of
Plasmopara viticola on Grape Leaves.
[10] G.Staudt and H.H.Kassemeyer, 1995, Evaluation of Downy mildew resistance in various accessions of wild
Vitis species.
[11]Stuart.P.Falk,Roger.C.Pearson,David.M.Gadoury, Robert, C.Seem & Abraham Sztejnberg, 1996, Fusarium
proliferation as a Biocontrol agent against grape Downey mildew
[12] A. Kortekamp, 1997, Epicoccum Nigrum Link: A biological control agent of Plasmopara viticola
[13] Maurus V. Brown1 and James N. Moore2, Patrick Fenn3, Ronald W. McNew4,1999, Comparison of Leaf
Disk, Greenhouse and Field Screening Procedures for Evaluation of Grape Seedlings for Downy Mildew
Resistance
836 Vol. 7, Issue 3, pp. 827-837

[14] R. Beresford, H. Paki, G. Brown, G. Follas and G. Hagerty,1999, strategies to avoid resistance
development to Strobilurin and related fungicides in Newziland
[15] S. M. Liu, S. R. Sykes and P. R. clingeleffer, 2003, a method using leafed single-node cuttings to evaluate
downy mildew resistance in grapevine
[16] A. Calonneca, P. Cartolaroa, C. Poupotb, D. Dubourdieub and P. Darrietb, 2004, Effects of Uncinula
necator on the yield and quality of grapes
[17] Thomas M. Perring, 2004, Epidemiological analysis of glassy winged sharpshooter and pierces disease
data.
[18] Ken Shackel, John Labavitch, 2004, magnetic resonance imaging: A non destructive approach for detection
of xylem blockages in xylella fasttidiosa-infected grapevines
[19] Mark A. Matthews, Thomas L. Rost, 2004, mechanisms of pierces disease in grape vine: the xylem
pathways of Xylella fastidious a progress report: comparison with symptoms of water deficiated the impact of
water stress.
[20] D. Gobbin, M. Jermini,B. Loskill, I. Pertot,M. Raynal and C. Gessler,2005, Importance of secondary
noculums of Plasmopara viticola to epidemics of grapevine downy mildew
[21] S.Bosco, M.C.Martinez, S.Anger and H.H.Kassemeyer, 2006, Evaluation of foliar resistance to downy
mildew in different cv. Albario clones
[22] Wei-SenChen, Francois Delmotte, Sylvie Richard-cervera, lissette douence, Charles Greif & Marie France
Corio-Cortest, 2007, at least two origins of fungicides resistance in grapevine Downey mildew populations
[23] Sotol R, 2007, comparison of grape seedling population against Downey mildew by using different
provocation methods.
[24] Franco mannini, 2007, Hot water treatment and field coverage of mother plant vineyards to prevent
propagation material from Phytoplasma infections.
[25] Y.cohen, Erubin, T.Hadad, D.gotlieb, u.gisi, 2007, sensitivity of phytothora infestans to mandipropamid &
the effect of enforced selection pressure in the field
[26] Santell Burrano, Antonto Alfonzo, 2008, Interaction between Acremonium byssoides & plasmopara
viticola in vitis vinifera
[27] Lance Cadle-Davidson, 2008, Variation Within and Between Vitis spp. for Foliar Resistance to the Downy
Mildew Pathogen Plasmopara viticola.
[28] M.Jermini, P.Blaise, C.Gessler, 2010, Quantitative effect of leaf damage, caused by Downey mildew on
growth & yield quality of grape vine morlot.
[29] W.S.Lee, V.Alchantis, C. Yang, M. Hirafuji, D. Moshou, 2010, sensing technologies for precision
speciality crop production
[30] Jayamala, K. Patil, raj Kumar, 2011, advances in image processing for detection of plant diseases
[31] Jan-Cor brink, 2012, Optimization of fungicide Spray coverage on grapevine & the incidence of botrytis
Cineria (book).
[32] Kanji Fatema Aleya, Debabrata Samantha, 2013, automated damaged flower detection using image
processing
[33] Paolo Tirelli, Massimo Marchi, Aldo Calante, Sara vitalini, Marcello iriti, n.alberto Borghese, Roberto
oberti,2013,multispectral image analysis for grapevine diseases automatic detection in field conditions
[34] Dan Egel, Downey mildew of pumpkin
[35] T.J.Wicks, B.H.hall & A. Somers, first report of matalaxyl resistance of grapevine Downey mildew in
Australia.
[36] Andrew Taylor, Farm note-how to bag test for Downey mildew of grapes
[37] Joost H.M. Stassen, identification & functional analysis of Downey mildew effectors in lettuce &
Arabidopsis
[38] Ron Becker, Sally miller, fact sheet: managing Downey mildew in organic & conventional vine crops.
[39] Uncorking the grape, Genome
[40] Jenna Burrell, Tim Brooke & Richard Beckwith, vineyard computing: sensor networks in agricultural
production.
[41] Plant pathology Book by George Agrios.
[42] Diseases of fruit crops Book by R.S.Singh

837 Vol. 7, Issue 3, pp. 827-837

AUTHORS BIOGRAPHY
Vikramsinh Kadam pursuing the Mtech degree in Electronics & Telecommunication
Engineering from Symbiosis International university, Pune. His research interests include
Technology in agriculture, Image processing, Robotics, Embedded systems.

and Telecommunication department. His research interest is in image processing and
defining vision and path for automated vehicle in agricultural field.

838 Vol. 7, Issue 3, pp. 838-844

GROUND WATER STATUS- A CASE STUDY OF ALLAHABAD,
UP, INDIA
Ayush Mittal, Munesh Kumar
PG Student, Department of Civil Engineering, MNNIT Allahabad, Uttar Pradesh, India

ABSTRACT
The ground water study is essential for assessing the ground water quality. These studies are of vital importance
from the point of view of development of water-logging. Vast area of state of Uttar Pradesh is already water-
logged, and any addition to such area is not in the interest of the state. It therefore, becomes imperative to have
a careful watch over the behaviour of ground water table since rise of ground water can be permitted only to a
certain extent. More than 90% of rural population uses ground water for domestic purpose. It is therefore
extremely important to have detailed information and knowledge about the quantity and quality of ground
water. In the present research work we have discussed about the groundwater scenario which includes the data
of groundwater availability in different blocks, groundwater quality of Allahabad district.
KEYWORDS: Alluvial, Chaka, Ganga, Groundwater, Vindhyan.
I. INTRODUCTION
Ground water is an important component of the water system for domestic, industrial and agricultural
purpose. It is commonly used source for drinking water for urban and rural sectors in India. Ground
water is a renewable natural resource with a relatively short and shallow circulation with close
dependence on precipitation and surface water. Ground water once was supposed to be the hygienic,
secure and safe for human consumption. Groundwater is now being gradually polluted by human
being because of the intense industrial activities. The quality of ground water depends upon the
characteristics and type of the subsurface soil and nature of recharge water.(Srinivasa CH, 2000)
Today the accelerated pace of development, rapid industrialization and population density have
increased demand of water resources. Ground water, a gift of nature, is about 210billion m
3
including
recharge through infiltration, seepage and evaporation. Out of this nearly one third is extracted for
irrigation, industrial and domestic use, while most of the water is regenerated into rivers. Over 98% of
the fresh water on the earth lies below its surface. The remaining 2% is what we see in lakes, rivers,
streams and reservoirs. Of the fresh water below the surface, about 90% satisfies the description of
ground water, that is, water which occurs in saturated materials below the water table. About 2%
water occurs as soil moisture in the unsaturated zone above the water table and is essential for plant
growth. The crucial role that the ground water plays as a source of drinking water for millions of rural
and urban families cannot be neglected. According to some estimates it accounts for nearly 80 percent
of the rural domestic water needs and 50 percent of urban domestic water needs(Kumar M. Dinesh,
2004)India receives annual precipitation of about 4000 km
3
, including snowfall rain. India is gifted
with a river system comprising of more than 20 major rivers with several tributaries. Many of these
rivers are perennial and some of them are seasonal. India occupies 3.29 million km
2
geographical area,
which forms 2.4 percent of the worlds land area and having 4 percent of worlds fresh water
resources. Monsoon rain is the main source of fresh water, with 76 percent of the rain fall occurring
between June to September. The Precipitation in volumetric terms is 4000 Billion Cubic meters
(B.C.M.).The average annual flow out of this is 1869 B.C.M. The rest of water is lost in infiltration
and evaporation. Due to topographical and other constraints only 600 B.C.M., can be utilized.
839 Vol. 7, Issue 3, pp. 838-844

Ground water is generally less susceptible to contamination and pollution when compared to surface
water bodies. Also the natural impurities in rain water which replenishes ground water system get
removed while infiltrating through soil strata. But in India, where ground water is used intensively for
irrigation and industrial purposes, a variety of land and water based human activities are causing
pollution of this precious resource.
In many parts of country available water is rendered not-potable because of the presence of toxic
material in excess. The situation gets worsened during the summer due to water scarcity and rain
water discharge. Contamination of water resources available for household and drinking purpose with
toxic element, metal ions and harmful microorganisms is one of the serious major health problems. As
a result huge amount of money is spent in India for chemical treatment of contaminated water to make
it potable.
Ground water contamination includes spillage or disposal of pesticides, fertilizers, petroleum by
carbon industrial chemicals, and waste products. Contamination can also respect from change in land
use pattern. The contaminated ground water system can pose a threat to other connected ecosystem.
The contamination can be described as coming from either point sources or diffuse sources. Point
source contamination may range from land fill sites on the other hand diffuse source contamination
includes the spreading of fertilizer to agriculture land, urban runoff, and the fallout from industrial
smoke stacks. When ground water becomes polluted it is difficult or impossible to clean it up
completely. The slow rates of ground water flow and low microbial activity limits any self-
purification process which takes place in days or weeks in surface water system may take decades to
occur in ground water. In addition the coasts of remediating ground water system are very high. It is
therefore better to prevent or reduce the risk of ground water contamination than to deal with its
consequences. This large use and dependency upon ground water dictates that these resources are
valuable and must be protected for both present day and future use (Donald K., Keech, 1979). The
ground water of the region has also been classified and characterized on the basis of hydro chemical
facies and their quality for agricultural use. [Jain C.K. et al (2000)].
II. DESCRIPTION OF STUDY AREA
The district of Allahabad lies at tail end of Allahabad division to the south between latitude 24 47
and 25 47 N and longitude 81 19 and 82 29 E. On the North it is bounded by the districts of
Pratapgarh and Jaunpur, the former being separated by the river Ganga for about one third of its
boundary. On the East is the district of Varanasi and the district of Mirzapur on the Southeast. The
southern boundary is formed by the State of Madhya Pradesh, and the district of Banda and Fatehpur
bound it on the south-west and West. The length of the district from east to west is about 117 km and
the breadth from north to south is about 101 km while the total area 7261 sq. kms. The district
headquarters is located at Allahabad which is also known as Prayag, situated at the confluence of the
great rivers the Ganga, the Yamuna and the mythical Saraswati. Allahabad is one of the most
important towns which are situated along the river Ganga. This great city is famous for the annual
MaghMela and for MahaKumbh, which is held at every twelve years interval, the biggest Mela in the
World. The main town is bounded by river Ganga on Northern and Eastern sides, the river Yamuna
and Doab plain forms its Southern and Western boundaries respectively. The Kanpur- Varanasi Road,
in most of its length runs on the ridge line dividing the town in two parts. The area on the north of this
road slopes towards Ganga whereas the area on the south side slopes towards river Yamuna. The
general information about Allahabad is as given below:
2.1 Information Data
Table-1: Information
S.No Parameter Value
1. Population 5,959,798 As Per 2011 Census
2. Area 63.07 Sq.Kms
3. Altitude 98 Meters Above Sea Level
4. Temperature Summer 26.6 To 41.0 C
5. Temperature Winter 9.1 To 29.0 C
6. Rainfall 102.8 Cms
7. Language Hindi, Urdu & English
840 Vol. 7, Issue 3, pp. 838-844

III. PHYSIOGRAPHY
The district is drained by river Ganga and its right bank tributary Yamuna and Tons broadly
represents following geomorphic units:-
1. Ganga Alluvial plain
2. Yamuna Alluvial plain
3. Vindhyan Plateau
Topographically Allahabad can be divided into three parts- the trans-Ganga tract or the Gangapar
plain, the doab and the trans-Yamuna tract. These are formed by the two main rivers, the Ganga and
the Yamuna. The trans Ganga part consists of the Soran, Phulpur and Handia tehsils. It is plain area
but there are long belts of Khadar land. The high banks of the Ganga are covered with poor sandy soil.
Belts of loam and usar lands also exist in this part. The doab tract comprises the Chail, Manjhanpur
and Sirathu tehsils and lies between the Ganga on the north and the Yamuna on the south. It is rich
and fertile. The land is plain and it consists of alluvial and light loam soils. In the south west the soil is
dark and it resembles the mark of the adjoining parts of Madhya Pradesh. The trans Yamuna tract lies
to the south of the Yamuna and comprises the Karchhana and Meja tehsils. It forms a part of
Bundelkhand region. The ridge formed by the Ganga and the Yamuna which lies in the north of
Karchhana is crowned with light sandy soil. The Kachhar land lies near the confluence of the Ganga
and the Tons. The central parts of karchanna tehsil and some parts old meja tehsil consist of upland.
The ranges of the vindhyan series of the Deccan plateau also lies in this tract. The Panna range lies for
about 16 km along the southern boundary of the district.

Figure 1. Allahabad district
IV. CLIMATE & RAINFALL
Allahabad district is continental. The climate of Allahabad is tropical with moderate winter and severe
extended summer. The Allahabad experiences both very dry hot summer and very cold winter every
year. During winter the temperature ranges from 9.5C to 26.2C whereas in summer it ranges from
29.5C to 48.0C. The average normal maximum temperature has been observed as 47.8C during
June and minimum 3.5C during January. The district receives rainfall from south west monsoon from
June to September. The average rainfall being 973.8 mm. Total evaporation of Allahabad district is
1476.9 mm. The maximum is observed during May and June which are the peak summer months.
841 Vol. 7, Issue 3, pp. 838-844

V. HYDROGEOLOGY
Ground water in the district occurs both in alluvium and the weathered and joined sandstones in areas
which are underlain by the hard rocks. Two broad based hydrogeological units, namely,
unconsolidated (Alluvium) and consolidated (hard rock) are the major components. The Alluvial
formations occur in the Trans-Ganga and Doab region. Localized patches of Trans-Yamuna region are
also covered by unconsolidated formations. Occurrence of consolidated formations is restricted
primarily to Trans Yamuna tract.
5.1 Alluvium Area
Field observations by government agencies indicates that depth of water table is less than 15 m during
pre-monsoon in the Trans- Ganga region whereas in Doab it stands in the depth range between 5.5
and 20 m. During post monsoon period, however depth of water table in the Trans-Ganga region
ranges between 0.65 and 12.0 mbgl. The Doab region indicates the depth of water table ranging
between 4.2and 10.0 mbgl.
5.2 Hard Rock Area
The ground water in the widely covered Vindhyan Plateau region is primarily under unconfined
condition. Exploratory data indicates that kaimur sandstones found at depths do have enough
potentiality at favourable locations. These sandstones after leaching of cementing materials get
disintegrated and reduced to silica sands which are loose and act as promising repository of ground
water. The lithological characteristics of bore holes have clearly indicated the presence of loose silica
sands.

Figure 2 . Hydrogeological map of Allahabad, UP, India
842 Vol. 7, Issue 3, pp. 838-844

VI. WATER EXTRACTION SOURCE
Table-2: Overall Status of Water Supply

(Source: U.P Jal Nigam)
Overall Status of Water Supply:
The total length of distribution pipeline is 1055 Kms. There are three zonal service reservoirs which
are located at Daraganj, Bhardwaj Ashram and Mayo Hall having capacity of 1.8 ML, 1.35 ML and
2.7 ML respectively. In addition to these three service reservoirs, there are 34 Over Head Tanks
(OHTs) which store the water before supply. Unaccounted flow of water is about 30% of total water
supply.
VII. GROUND WATER RESOURCES
To facilitate the ground water development the ground water resources of the district have been
worked out and are as follows
Table-3: Block wise ground water resource of Allahabad district
Sl. No. Assessment
unit (Blocks)
Ground
Water
Availability
(Ham)
Ground
Water Draft
(Ham)
Level of
development
(%}
Category
as on
05/2013
Balance Ground
Water (Ham)
1. Bahadurpur 6269.09 552.88 82.64 Safe 5212.89
2. Baharia 8923.83 2509.56 81.44 Safe 4778.45
3. Chaka 2545.76 403.23 75.68 Safe 2123.22
4. Dhanupur 4743.23 854.79 87.12 Safe 3387.86
5. Handia 4834.02 1025.95 71.22 Safe 3232.22
6. Holagarh 7023.11 2544.43 69.82 Safe 2389.99
7. Jasara 5904.15 1634.21 57.00 Safe 2756.66
8. Kaundhiara 6952.68 1822.22 84.21 Safe 3154.32
9. Karchhana 6300.47 1263.39 89.79 Safe 3989.25
10. Kaurihar 5933.43 1309.67 75.12 Safe 3909.24
11. Koraon 8512.02 2605.88 34.54 Safe 5124.78
12. Manda 3633.87 764.60 69.89 Safe 2467.55
13. Mauiama 5754.46 1588.20 82.13 Safe 2971.35
14. Meja 4533.22 987.44 53.33 Safe 3103.38
15. Phulpur 5998.23 1102.93 63.33 Safe 4455.33
16. Pratapur 5331.22 1122.74 75.08 Safe 3487.90
17. Saidabad 5334.67 1205.76 73.05 Safe 3212.21
18. Soraon 4654.23 1190.98 67.19 Safe 2690.09
19. Shankargarh 3278.34 671.56 33.03 Safe 2012.08
20. Urva 5278.55 744.22 69.17 Safe 4232.33
Total 111738.19 24775.88 68682.32
(Source: U.P Jal Nigam)
VIII. GROUND WATER QUALITY
8.1 Quality of shallow ground water
The chemical analysis of shallow ground water consists of pH, E.C., Na, K, Ca, Mg, HCO
3
, CL, SO
4
,
NO
3
, F and TH as CaCO
3
reflects that there is no contamination of the shallow ground water in the
843 Vol. 7, Issue 3, pp. 838-844

district and all the constituents are well within the range. The chemical data of shallow aquifers
reveals that the ground water quality is more deteriorated in canal command area. The map of E.C.
and Chloride show that in most of the area E.C. varies from 180-1780 siemens/cm at 25C. It is
interesting to find that different radicals in the shallow ground water have not changed over the year's
in spite of upcoming canal irrigation and use of fertilizers.
8.2 Quality of Deeper Aquifers
Data of water samples from deeper aquifers are few but there analysis reveals that the water is safe
and potable. It is observed that the E.C. and other salts are in higher concentration in alluvial area than
hard rock area. The quality in hard rock area is inferior near the stream than away from the stream.
IX. RECOMMENDATIONS
1. Variation in water quality was observed during both the periods of the study i.e. per-monsoon
and post-monsoon period.
2. Ground water quality varies from place to place with the depth of water table which is
reflected from the values obtained at same locations with different sources.
3. Water source should be thoroughly investigated before recommending it for use, whether it is
private or government boring.
4. Periodical investigation should be conducted every two to three years on quarterly basis to
evaluate the level of ground water contamination.
5. Ground water withdrawal should be minimized in Bahadurpur and Chaka blocks
immediately.
6. Alternative drinking water source may be provided along the river bank because people
residing nearby are using hand pump water for drinking and other domestic purposes.
7. Public awareness should be created among the masses particularly for the people residing
along the bank of the river Yamuna for consumption of safe drinking water.
8. It is suggested that some low cost and easy to implement technique may be provided to the
consumers for removing hardness, total dissolved solids and chloride in water where the
values exceed the permissible limit.
X. CONCLUSIONS
The stage of groundwater development in the district is 69.73%. Maximum groundwater development
in Karchhana block (89.79%) and minimum is in Shankargarh i.e. 33.03%. In five blocks viz:
Bahadurpur, Baharia, Dhanupur, Kaundhiara, Mauiama the stage of groundwater development is 80%
to 90%. In five blocks viz: Chaka, Handia, Kaurihar, Pratappur and Saidabad, the stage of
groundwater development is 70% to 80%. All the blocks fall under "SAFE" category. Construction of
canals or strengthening of the existing canal system should be emphasized in four blocks viz:
Bahadurpur, Chaka, Kaundhiara, and Meja. In rest of the blocks, emphasis may be given to irrigation
through groundwater development either by medium to shallow or deep tubewells. There is no block
in the district identified under polluted area but localized area like Chand Khamria (Meja), Naini
Industrial area and Shankargarh (part) where E.C., NO
3
and Fe has increased the permissible limit.
Ground water quality in general is fresh and potable except few pockets. Deeper aquifer also reveals
that there is no contamination or pollution of groundwater.
REFERENCES
[1] Kumar M. Dinesh and Shah Tushaar,(2004)Ground Water pollution and contamination In India- The
emerging challenge. pp 1-6.
[2] Kumar. Rakesh, Singh R.D., and Sharma K.P,(2005)Water resources of India current science, Vol. 89,
No.5, pp 794-811.
[3] Central Pollution Control Board (Ministry Of Environment And Forests),(2008)Status Of Groundwater
Quality In India -Part-II.
[4] Uttar Pradesh Development Report Vol-2.
844 Vol. 7, Issue 3, pp. 838-844

[5] cpcb.nic.in/upload/NewItems/NewItem_50_notification.pdf.
[6] Uttar Pradesh Jal Nigam, Lucknow, India.
[7] Donald, K., Keech Inc. N.Y., (1979), Ground Water Assessment of Surface and subsurface, Ground Water
Hydrology journal vol. 25.
[8[ Jain CK, Bhatia KKS, Kumar Vijay (Natl Inst Hydro, Roorkee 247667). Ground water quality in Sagar
district, Madhya Pradesh. Indian J Environ Health, 42(4) (2000), 151-158 [13 Ref].
[9] Jain CK, Sharma MK (Natl Inst Hydro, Roorkee, 247667, UP). Regression analysis of ground water quality
data of Sagar district, Madhya Pradesh. Indian J Environ Health, 42(4) (2000), 159-168 [8 Ref].
[10] Jain CK, Sharma MK, Bhatia KKS, Seth SM (Natl Inst Hydro, Roorkee 247667, UP). Ground water
pollution endemic of fluorosis. Polln Res, 19(4) (2000), 505-509 [2 Ref].
[11] Srinivasa CH, Piska Ravi Shankar, Venkateshwar C, Satyanarayana Rao MS, Ravinder Reddy R (Dept
Zoo, PG Coll Sci, Saifabad, Osmania Univ, Hyderabad 500004). Studies on ground water quality of Hyderabad.
Polln Res, 19(2) (2000), 285-289 [15 Ref]

AUTHORS BIOGRAPHIES
Ayush Mittal presently a postgraduate student in department of civil Engineering MNNIT
Allahabad, India. His area of research is Environmental geo technology. He obtained his
Bachelor of technology from United College of Engineering and Research, Naini,
Allahabad, India. His Interests are Geotechnical Engineering, Groundwater Management,
Rock Mechanics and Foundation Engineering.

Munesh Kumar presently a postgraduate student in department of civil Engineering
MNNIT Allahabad, India. His area of research is Geotechnical Engineering. He obtained his
Bachelor of technology from GLA University, Mathura, India. His interests are Geo
environmental Engineering, Water Conservation and Remediation, Soil Dynamics and
Earthquake Engineering.

845 Vol. 7, Issue 3, pp. 845-855

CLUSTERING AND NOISE DETECTION FOR GEOGRAPHIC
KNOWLEDGE DISCOVERY
Sneha N S
1
and Pushpa
2

1
PG Student, Department of Computer Science and Engineering,
Adichunchanagiri Institute of Technology, Chikmagalur, Karnataka, India
2
Associate Professor & Head, Department of Computer Science and Engineering,

ABSTRACT
Ample amount of geographic data has been collected with modern data acquisition techniques such as a Global
Positioning System, high resolution remote sensing and internet based volunteered geographic information.
Spatial datasets are large in size, multidimensional and have high complexity measures. To address these
challenges Spatial Data Mining (SDM) for Geographic Knowledge Discovery (GKD) are the emerging fields
for extraction of useful information and knowledge mining for many applications. This paper addresses the
clustering and noise detection technique for spatial data. We considered multidimensional spatial data to
provide feasible environment to place sensitive devices in a laboratory by using the data collected from the
sensors. Various sensors were used to collect the spatial and temporal data. The GDBSCAN algorithm is used
for clustering, which relies on density based notation of clustering and is designed to discover clusters of
arbitrary shape and distinguish noise. The proposed work reduces the computation cost and increase the
performance.
KEYWORDS: Spatial Data, Temporal Data, Spatial Clustering
I. INTRODUCTION
Due to the development of information technology, a vast volume of data is accumulated on many
fields. Since automated methods for filtering/analyzing the data and also explaining the results are
required, a variety of data mining techniques finding new knowledge by discovering hidden rules
from vast amount of data are developed. In the field of geography, due to the development of
technology for remote sensing, monitoring, geographical information systems, and global positioning
systems, a vast volume of spatial data is accumulated. An automated discovery of spatial knowledge
is required because of the fast expansion of spatial data and extensive use of spatial databases.
Nowadays, the spatial data mining turn out to be more eminent and stimulating for the reason that
abundant spatial data have been stored in spatial databases. The mining of meaningful patterns from
spatial datasets is more knotty than mining the analogous patterns from conservative numeric and
categorical data, due to the difficulty of spatial data types, spatial relationships and spatial
autocorrelation. In various applications, spatial patterns have excessive demand. Since the spatial data
has its own characteristics different from the non-spatial data, direct using of general data mining
techniques incurs many difficulties .So there have been many studies of spatial data mining
techniques considering the characteristics of the spatial data [1].
Spatial data are the data related to objects that occupy space. A spatial database stores spatial objects
represented by spatial data types and spatial relationships among such objects. Spatial data carries
topological and/or distance information and it is often organized by spatial indexing structures and
accessed by spatial access methods. These distinct features of a spatial database pose challenges and
bring opportunities for mining information from spatial data. Spatial Data mining or knowledge
discovery in spatial database refers to the extraction of implicit knowledge, spatial relations, or other
846 Vol. 7, Issue 3, pp. 845-855

patterns not explicitly stored in spatial databases. Spatial data mining refers to the extraction of
knowledge, spatial relationships, or other interesting patterns not explicitly stored in spatial data sets.
It is expected to have wide applications in geographic information systems, geo-marketing, remote
sensing, image database exploration, medical imaging, navigation, traffic control, environmental
studies, and many other application areas where spatial data are used. A crucial challenge to spatial
data mining is the exploration of efficient spatial data mining techniques due to the huge amount of
spatial data, and the complexity of spatial data types and spatial access methods. The extra features
that distinguish spatial data from other forms of data are spatial co-ordinates, topological distance, and
direction information. By inclusion of many features, query language has become too complicated. In
contrast to the mining in relational databases, spatial data mining algorithms need to consider the
objects that are near-by in order to extract useful knowledge as there is influence of one object on the
neighboring object.
In Data analysis, Cluster analysis is very frequently used, which organizes a set of data items into
groups (or clusters) so that items in the same group are similar to each other and different from those
in other groups. Clustering methods can be broadly classified into Five groups they are Partitioning
algorithms, Density based clustering, Hierarchical Algorithms, Grid-Based Methods and Model-
Based Clustering Methods. Example algorithms of the above classification are K-Means, K-medoids,
Density-based spatial clustering of applications with noise (DBSCAN) and Generalized Density-
based spatial clustering of applications with noise (GDBSCAN), Chameleon. To consider spatial
information in clustering, three types of clustering analysis are existing; they are spatial clustering,
regionalization, and point pattern analysis. In this work only Density-Based clustering methods are
considered.
In section 2, we discuss the related literature with respect to various clustering algorithms for
geographic knowledge discovery using spatial data mining. The section 3 discusses data collection,
GDBSCAN algorithm, spatial clustering and noise detection methods. Section 4 presents the results
related to clustering and noise detection system.
II. LITERATURE SURVEY
N.Santhosh Kumar, V. Sitha Ramulu, K.Sudheer Reddy, Suresh Kotha, Mohan Kumar [2], presented
how spatial data mining is achieved using clustering. Spatial data is a highly demanding field because
huge amounts of spatial data have been collected in various applications, ranging from remote
sensing, to geographical information systems (GIS), computer cartography, environmental assessment
and planning, etc. Spatial data mining tasks include: spatial classification, spatial association rule
mining, spatial clustering, characteristic rules, discriminant rules, trend detection. Cluster analysis
groups objects (observations, events) based on the information found in the data describing the objects
or their relationships.
All the members of the cluster have similar features. Members belong to different clusters has
dissimilar features. Several clustering methods for spatial data mining include; Partitioning Around
Medoid (PAM), Clustering LARge Applications(CLARA), Clustering LARge Applications based
upon RANdomized Search (CLARANS), Spatial Dominant approach SD(CLARANS), Non Spatial
Dominant approach NSD(CLARANS).
Ester M., Kriegel H.-P., Sander J. and Xu X.[3] in their paper provided a Density-Based Algorithm
for Discovering Clusters in Large Spatial Databases with Noise. They presented the clustering
algorithm DBSCAN which relies on a density-based notion of clusters. It requires only one input
parameter and supports the user in determining an appropriate value for it. They also performed a
performance evaluation on synthetic data and on real data of the SEQUOIA 2000 benchmark. The
results of these experiments demonstrated that DBSCAN is significantly more effective in discovering
clusters of arbitrary shape than the well-known algorithm CLARANS. Furthermore, the experiments
have shown that DBSCAN outperforms CLARANS by a factor of at least 100 in terms of efficiency.
Ng R.T., and Han J.[4] developed and Efficient and Effective Clustering Methods for Spatial Data
Mining. They developed a new clustering method called CLAHANS which is based on randomized
search. We also develop two spatial data mining algorithms that use CLAHANS. Their analysis and
experiments show that with the assistance of CLAHANS, these two algorithms are very effective and
can lead to discoveries that are difficult to find with current spatial data mining algorithms.
847 Vol. 7, Issue 3, pp. 845-855

Furthermore, experiments conducted to compare the performance of CLAHANS with that of existing
clustering methods show that CLAHANS is the most efficient.
III. METHODOLOGY
3.1. Geographic Knowledge Discovery System
Geographic knowledge discovery (GKD) is the process of extracting information and knowledge from
massive geo-referenced databases. Spatial objects by definition are embedded in a continuous space
that serves as a measurement framework for all other attributes. This framework generates a wide
spectrum of implicit distance, directional, and topological relationships, particularly if the objects are
greater than one dimension. Figure 1 gives the System Architecture of Spatial Data Mining System
for Geographic Knowledge Discovery (GKD). The architecture is divided into three parts; they are,
the Data Collection from various databases, the Processing stage which consists of spatial clustering
method and noise detection and the analysis phase where the discovered patterns are analysed for
Equipment feasibility.

Figure 1: System Architecture of GKD
3.2 Data Collection
Spatial data has positional and topological data that do not exist in general data, and its structure is
different according to the kinds of spatial data. Temporal data are the data which explicitly refer to
time. The spatial data consisting of Temperature, Light, Humidity, Voltage, Location information,the
Temporal data consists of date, the non-spatial data consists of Sensor ID are collected.The spatial
dataset used in this work consists of multidimensional data of size 145.133 MB. The dataset includes
23,03,450 Records. The Table 1 gives ten sample records of the spatial data used in the proposed
work. These data are stored in the database in a well-defined format in form of table. The database
consists of set of similar tables to store data pertaining to location of the sensors and the spatial data
specifications of the equipments to be placed.
The attributes used for storing spatial data as shown in Table 1 are Date, Sens ID, Temp, Humid,
Light,Volt. Date gives the date on which these data are entered, Sens ID, gives the ID of the sensors
for which the data is entered, Temp, consists the temperature of the particular sensor defined by the
Data Collection phase
Data
Collection
Processing phase Pattern Analysis Phase
Pattern Discovery
and
Analysis
Concept
Hierarchies
Spatial
Data
Non-Spatial
Data
Temporal
Data
Spatial
Clustering
Noise
Detection
848 Vol. 7, Issue 3, pp. 845-855

sensor ID. Similary Humid, Light, Volt gives the humidity, light and voltage information of the
sensors selected. All these data are collected and stored.
Table 1: Spatial data Table of SDM for GKD
Date Sens ID Temp Humid Light Volt
18/04/14 S1 10.25 20.12 50.15 1.256
18/04/14 S18 24.26 14.26 78.24 6.456
18/04/14
S19
41.25 31.29 136.1 5.698
18/04/14
S2
8.25 16.25 36.24 2.020
18/04/14 S20 30.24 10.24 99.12 5.240
18/04/14 S21 45.78 35.65 187.2 4.250
18/04/14
S22
49.25 39.45 146.2 3.450
18/04/14 S25 14.36 26.24 151.4 2.456
18/04/14 S28 47.54 36.24 100.5 4.500
18/04/14 S30 34.25 61.24 200.1 1.458
3.3. Spatial Clustering and Noise Detection
Spatial Clustering is interpreted as the task of collecting the objects of a spatial database into
meaningful detectable subclasses (i.e. clusters) so that the members of a cluster are as similar as
possible whereas the members of different clusters differ as much as possible from each other.
3.3.1 Density Based Spatial Clustering
Density based algorithms typically regard clusters as dense regions of objects in the data space that
are separated by regions of low density. The main idea of density-based approach is to find regions of
high density and low density, with high-density regions being separated from low-density regions.
These approaches can make it easy to discover arbitrary clusters. A common way is to divide the
high-dimensional space into density-based grid units. Units containing relatively high densities are the
cluster centers and the boundaries between clusters fall in the regions of low-density units. In density-
based clustering, clusters are defined as areas of higher density than the remainder of the data set.
Objects in these sparse areas - that are required to separate clusters - are usually considered to be
noise and border points.
The most popular density based clustering method is DBSCAN. In contrast to many newer methods, it
features a well-defined cluster model called "density-reachability". Similar to linkage based
clustering; it is based on connecting points within certain distance thresholds. However, it only
connects points that satisfy a density criterion, in the original variant defined as a minimum number of
other objects within this radius. A cluster consists of all density-connected objects (which can form a
cluster of an arbitrary shape, in contrast to many other methods) plus all objects that are within these
objects' range. Another interesting property of DBSCAN is that its complexity is fairly low - it
requires a linear number of range queries on the database - and that it will discover essentially the
same results (it is deterministic for core and noise points, but not for border points) in each run,
therefore there is no need to run it multiple times. The key idea of density-based clustering is that for
each object of a cluster the neighborhood of a given radius has to contain atleast a minimum number
of objects, i.e. the cardinality of the neighborhood have to exceed a given threshold.
3.3.2 Generalized Density-Based Spatial Clustering of Applications with Noise (GDBSCAN)
The clustering algorithm DBSCAN relies on a density-based notion of clusters and is designed to
discover clusters of arbitrary shape as well as to distinguish noise. In this work generalized version of
this algorithm is used. The generalized algorithm - called GDBSCAN - can cluster point objects as
well as spatially extended objects according to both, their spatial and their non-spatial attributes.
GDBSCAN algorithm is based on center-based approach. In the center-based approach, density is
estimated for a particular point in the dataset by counting the number of points within a specified
radius, eps of that point. This includes the point itself. The center-based approach to density allows
us to classify a point as a core point, a noise or border point. A point is core point if the number of
points within eps , a user-specified parameter, exceeds a certain threshold, MinPts , which is also a
user-specified parameter taken as 3 in this work. Any two core points that are close enough within a
849 Vol. 7, Issue 3, pp. 845-855

distance eps of one another are put in the same cluster. Likewise, any border point that is close
enough to a core point is put in the same cluster as the core point. Noise points are discarded.
To find a density-connected set, GDBSCAN starts with an arbitrary object p and retrieves all objects
density-reachable from p with respect to Eps and MinPts. If p is a core object, this procedure yields a
density-connected set with respect to Eps and MinPts. If p is not a core object, no objects are density-
reachable from p and p is assigned to NOISE. This procedure is iteratively applied to each object p
which has not yet been classified. Thus, clusters are formed and the noises are detected.
The Algorithm of the function is[5],

SetofPoints is either the whole database or a discovered cluster from a previous run. Eps and MinPts
are the global density parameters whose parameters are considered as 3 and 115.0 respectively.
ClusterIds are from an ordered and countable datatype where UNCLASSIFIED < NOISE < other
Ids, and each object is marked with a clusterId Sens. The function nextId (clusterId) returns the
successor of clusterId in the ordering of the datatype. The function SetofPoints.get (i) returns the i-th
element of SetofPoints.
A call of SetofPoints.Region(Sens, Eps) returns the Eps-neighborhood of Point in SetOfPoints as a list
of objects. Obviously the efficiency of the above algorithm depends on the efficiency of the
neighborhood query because such a query is performed exactly once for each object in SetofPoints
which satisfies the selection condition.
The clusterId of some objects p which are marked to be NOISE because Eps (p)) < MinPts may be
changed later if they are density-reachable from some other object of the database. This may happen
only for border objects of a cluster. Those objects are then not added to the seeds-list because we
already know that an object with a ClusterId of NOISE is not a core object, i.e., no other objects are
density-reachable from them.
In algorithm, function Expand-Cluster constructing a density-connected set for a core object Object is
presented in more detail next[5] :

GDBSCAN (SetofPoints, Eps, MinPts,)
// SetofPoints is UNCLASSIFIED
ClusterId:= 1;
FOR i FROM 1 TO SetofPoints.size DO
Sens: = SetofPoints.get (i);
IF Sens.ClId = UNCLASSIFIED THEN
IF ExpandCluster (SetofPoints, Sens, ClusterId,Eps, MinPts) THEN
ClusterId: =nextId (ClusterId)
END IF
END IF
END FOR
END; // GDBSCAN

ExpandCluster (SetofPoints, Sens, ClId, Eps,MinPts): Boolean;
seeds:=SetofPoints.Region(Sens,Eps);
IF Count (seeds) < MinPts THEN // no core point
SetofPoints.changeClId(Sens, NOISE);
RETURN False;
END IF
// still here? sens is a core object
SetofPoints.changeClIds (seeds,ClId);
seeds.Remove(Sens);
WHILE seeds > 0 DO
currentSens := seeds.first();
result := SetofPoints.Region(currentSens, Eps);

850 Vol. 7, Issue 3, pp. 845-855

4.1 Experimental setup
For the proposed work, the data is collected from various data sources such as sensors which are
deployed in a research laboratory. These sensors are required to collect time stamped topology
information, along with humidity, temperature, light and voltage values. The x and y coordinates of
the sensors are considered in meters. The data collected includes more than 2.3 million records.
Sensor ids range from 1-54; Temperature is in degrees Celsius. Humidity is temperature corrected
relative humidity ranging from 0-100%. Light is Lux, voltage is expressed in terms of volts.
4.2. Data Collection Form
The data collection is the main part of data mining. This section deals with the various data collection
forms used in our project.
4.2.1 Data Entry Forms
Figure 2 shows the form used for obtaining the geographic details pertaining to sensors such as co-
ordinate location. The data entered are displayed in the grid format. The density based clustering is
done based on these location details. The sensor Id takes the Ids of each sensors and the location x,
location y takes the x and y coordinate values of the location of sensor. On clicking the Add button
these data are added to the database and these sensors entered are displayed in the grid-format. The
Sensor details can be edited or deleted. The Reset button is used to clear all the textbox values.
IF Count(result) MinPts THEN
FOR i FROM 1 TO result.size DO
P: = result.get (i);
IF P.ClId IN {UNCLASSIFIED,NOISE} THEN
IF P.ClId = UNCLASSIFIED THEN
seeds.Add(P);
END IF;
SetofPts.changeClId(P,ClId);
END IF; // UNCLASSIFIED or NOISE
END FOR;
END IF; // MinPts
seeds.Remove(currentSens);
END WHILE; // seeds Empty
RETURN True;
END; // ExpandCluster

851 Vol. 7, Issue 3, pp. 845-855

Figure 2. Sensor Location Entry
Figure 3 shows the form used to obtain the spatial and temporal details of the sensors such as the date
of entry, temperature, humidity, light and voltage. These data entered are represented in the grid form,
which can be updated if necessary. The data must be entered within a specified threshold. The sensor
Id lists the Ids of the sensors entered in the previous page and the required sensor to which the data
must be entered should be selected. The date for which the data needs to be entered must also be
selected. By default the current date will be considered. The spatial data will be entered by the
administrator. On clicking the Add button, these data entered will be stored into data base if they are
within the specified threshold. The entered data will be displayed in the grid format in the page. These
data can be edited or deleted.

852 Vol. 7, Issue 3, pp. 845-855

Figure 3. Sensor Spatial Data Entry
Figure 4, shows the equipment data form. This form is used to obtain the spatial data of the equipment
such as temperature, humidity, light and voltage specifications of the equipment to be placed. These
data are added to the database only if they do not exceed a specified threshold. These data are
represented in a grid format. These data can be updated if required.

Figure 4. Equipment Data Entry
4.3. Clustering and Noise Detection Forms
The clustering and noise detection is done through the density based spatial clustering algorithm with
noise called GDBSCAN. Figure 5. This page gives a grid representation of various clusters formed
853 Vol. 7, Issue 3, pp. 845-855

along with the sensor IDs of those sensors that fall under each of the clusters. The sensors which
constitute noise are also detected and are displayed in the noise grid. The clustering is done based
only on the location information of the sensors entered through the sensor location entry page.

Figure 5. Cluster and Noise Grid
Figure 6 shows the graphical representation of the sensors and their respective clusters. The noises are
those sensors which are not included in any cluster formations. The purple dots represent the sensors;
the red lines are used to group the sensors of a particular cluster together. On clicking the show button
present in the top of the screen we get the sensors which form the clusters and those sensors which
form noise.

Figure 6. Cluster Formation and Noise Detection Representation

854 Vol. 7, Issue 3, pp. 845-855

V. CONCLUSION
We considered 54 sensors to capture the spatial data, out of which 6 sensors do not belong to any of
the clusters, so that are treated as outliers. Hence 11% of the noise is detected. This method is found
to be feasible for Spatial Data Mining (SDM) in Geographic Knowledge Discovery (GKD). The
density based GDBSCAN algorithm is used for clustering. It estimates the density for a particular
point in the dataset by counting the number of points within a specified radius. There are 6 clusters
formed for 54 sensors, named as C1, C2, C3, C4, C5, C6. The number of sensors in each cluster, after
using GDBSCAN algorithm are such that, C1 consists of 17 sensors, C2 consists of 6 sensors, C3
consists of 5 sensors, C4 consists of 5 sensors, C5 consists of 8 sensors and C6 consists of 7 sensors.
By this analysis cluster C1 has more number of sensors and it senses more spatial data and coverage
of the location is also high so that we can place more sensitive devices in C1 compared to other
clusters.
VI. FUTURE WORK
Clustering and Noise Detection method for the GKD will help in placing Laboratory devices in a
feasible location, so that the damage to the devices due to high temperature, high humidity and high
voltage can be avoided. By using this technique maintenance cost can be reduced and increase the
lifetime of the devices. Further this work can be enhanced for pattern analysis, so that computed
scoring for the temperature, light, humidity and voltage values can increase the speed of analysis
process.
REFERENCES
[1]. Duck-Ho Bae, Ji-Haeng Baek, Hyun-Kyo Oh, Ju-Won Song, Sang-Wook Kim SD- Miner: A Spatial Data
Mining System, Proceedings of IC-NIDC 2009.
[2]. N.Santhosh Kumar, V. Sitha Ramulu, K.Sudheer Reddy, Suresh Kotha, Mohan Kumar, Spatial Data
Mining using Cluster Analysis, International Journal of Computer Science & Information Technology
(IJCSIT) Vol 4, No 4, August 2012.
[3]. Ester M., Kriegel H.-P., Sander J. and Xu X. 1996. A Density-Based Algorithm for Discovering Clusters
in Large Spatial Databases with Noise. Proc. 2nd Int. Conf. on Knowledge Discovery and Data Mining.
Portland, OR, 226-231.
[4]. Ng R.T., and Han J. 1994. Efficient and Effective Clustering Methods for Spatial Data Mining. Proc.
20th Int. Conf. on Very Large Data Bases. Santiago, Chile, 144-155.
[5]. Jrg Sander, Martin Ester, Hans-Peter Kriegel, Xiaowei Xu, Density- Based Clustering in Spatial
Databases: The Algorithm GDBSCAN and its Applications, Data mining and knowledge discovery 2,169-
194(1998).
[6]. Ertz L., Steinbach M., Kumar V.: "Finding Clusters of Different Sizes, Shapes, and Densities in Noisy,
High Dimensional Data", SIAM International Conference on Data Mining (2003)
[7]. Sayal M., Scheuermann P.: "A Distributed Clustering Algorithm for Web-Based Access Patterns", in
Proceedings of the 2nd ACM-SIGMOD Workshop on Distributed and Parallel Knowledge Discovery,
Boston, August 2000
[8]. Fayyad U., Piatetsky-Shapiro G., and Smyth P. 1996. Knowledge Discovery and Data Mining: Towards a
Unifying Framework. Proc. 2nd Int. Conf. on Knowledge Discovery and Data Mining, Portland, OR, 82-
88.
[9]. Ester M., Kriegel H.-P., and Xu X. 1995. A Database Interface for Clustering in Large Spatial Databases,
Proc. 1st Int. Conf. on Knowledge Discovery and Data Mining, Montreal, Canada, 1995, AAAI Press,
1995.
[10]. Jain A. K., Dubes R.C.: "Algorithms for Clustering Data", Prentice-Hall
[11]. Dr. Mohammed Otair, Approximate K-Nearest Neighbour Based Spatial Clustering Using K-D Tree,
IJDMS Vol.5, No.1, February 2013.
[12]. Lovely Sharma1, K. Ramya, A Review on Density based Clustering Algorithms for Very Large
Datasets, IJETEA, Volume 3, Issue 12, December 2013
[13]. Richa Sharma Bhawna Malik Anant Ram, Local Density Differ Spatial Clustering in Data Mining,
IJARCSSE, Volume 3, Issue 3, March 2013

855 Vol. 7, Issue 3, pp. 845-855

AUTHORS
Sneha N S, is currently Pursuing 4
th
Semester, Master of Technology in Computer Science
and Engineering at AIT, Chickmagalur. She has completed her Bachelor of Engineering
from Srinivas Institute of Technology, Mangalore. She had published a paper. Her areas of
interests include data mining and Information Security.

Pushpa, Associate Professor in the Department of Computer Science and Engineering,
AIT, Chickmagalur and Research Scholar at R.V.College of Engineering, Bangalore. She
had 12 years of teaching experience. She has completed her Master of Technology from
Visvesvaraya Technological University. She had publishes research papers in international
journals. Her research interests are in the area of data mining and Computer Networks.

856 Vol. 7, Issue 3, pp. 856-862

PROXY DRIVEN FP GROWTH BASED PREFETCHING
Devender Banga
1
and Sunitha Cheepurisetti
2
1,2
Department of Computer Science Engineering,
SGT Institute of Engineering and Technology, Gurgaon, India

ABSTRACT
Now days, number of people are connected to the web to get results for their various queries from the servers.
Servers are also heavily loaded to respond to large number of users. At this moment of time when millions of
users are connected to the web, results for their queries should be as fast as possible. There are number of
schemes by which we can reduce the response time of a query to a server, such as caching scheme, prefetching
scheme. This paper proposes a framework for prefetching in the proxy server. For the reduction in the web
latency and improving user satisfaction, prefetching is important. The proposed framework uses FP growth
algorithm for determining the users frequent sets of patterns. User data will be collected from the web log
history. Now, depending upon the threshold and patterns generated by the FP Growth, list of patterns to be
prefetched will be generated and passed to the predictor module. Predictor module will prefetch the web
objects by creating session with the main server. Using FP Growth algorithm of association rule mining
frequent patterns can be determined without candidate key generation. Working of FP Growth is shown using
Data-applied. The proposed framework improves the efficiency of the existing network.
KEYWORDS: Prefetching, Proxy server, cache, prediction.
I. INTRODUCTION
Ones dependency on network is increasing day by day, in developing countries where mostly every
person is addicted to web, as we can get result to any query of any subject we heard even once. The
moment we start searching with a statement or a keyword, that keyword is stored by browser cache
for further providing fast results. In todays network where speed is of major concern we need
schemes and algorithms to reduce the response time from server to a client, where using only a cache
is not sufficient to reduce the latency. As todays network is full of congestion and getting a response
faster is a major issue of concern. Many researchers worked in this area and also there are schemes
and frameworks available which reduces the response time. One of the schemes available is
prefetching with web caching. Web mining is used to mine the data received from the proxy server in
three tier web architecture and then results are fed back to proxy cache which then responds to user
queries on demand. One of the example of mining is, if we visited a shop to buy bread slices, then the
milk can, butter bar and cheese slices which are placed nearby bread can be mined by using
association rule based mining, in this example we have only one query named bread but other results
such as slices, milk can are not required but these are the expected results of query for persons which
may looked for milk can and cheese slices along with bread at some earlier point of time. So by
applying mining rules we can get results to expected queries from the cache rather than asking the
results for various pre-questioned queries again and again from the server. This greatly reduces the
latency of a network. There are three types of caches; browser cache, proxy cache and server cache.
Applying a cache with a browser gives faster results but while using three tier architecture where
proxy server is the mediocre; much of the burden is barred by proxy server instead of main server. So,
by applying web caching rules on proxy server instead of main server will help in providing the
results much faster. Proxy servers are used not only to reduce web latency but for achieving various
tasks like: NAT (Network Address Translation), Firewall and for security of web servers so that
unauthorized access could be stopped [1]. Proxy servers also provide a way where we can grant
857 Vol. 7, Issue 3, pp. 856-862

access to any of the link on the web by hiding the actual details of the users like IP (Internet Protocol)
address.
Paper is divided into different sections for the ease of understanding. Section II focuses on the
primary work done in this field. Section III depicts the proposed framework and section IV provides
the experimental work done to prove the proposed framework. At the end, conclusion of the work has
been provided.
Pallis et. al[2] addressed the short-term prefetching problem on a web cache environment using an
algorithm for clustering inter-site web pages. The proposed scheme efficiently integrates web caching
and prefetching. According to this scheme, each time a user requests an object, the proxy fetches all
the objects which are in the same cluster with the requested object.
Sharma and Dubey proposed a framework for web traffic reduction [3]. The approach first extracts
data from proxy server web log, and then the extracted data is preprocessed. The preprocessed data is
then mined using clustering, and sequence analysis is being done to know the patterns to be pre-
fetched.
Divya and Kumar presented a survey on three different association rule mining algorithms AIS,
Apriori and FP-tree algorithm and their drawbacks which would be helpful to find new solution for
the problems found in these algorithms [4].
Kosala et.al surveyed the research in the area of web mining [5]. The paper explores the connection
between web mining categories and the related agent paradigm. This paper focuses on representation
issues, on the process, and on the learning algorithm, and the application of the recent works as the
criteria.
Chang et.al[6] describes that Information extraction (IE) from semi-structured web documents is a
critical issue for information integration systems on the Internet. The discovery of repeated patterns is
realized through a data structure call PAT tree. The paper also focuses that incomplete patterns are
further revised by pattern alignment to comprehend all pattern instances.
Sharma and Dubey provided the literature survey in the area of web mining [7]. The paper basically
focuses on the methodologies, techniques and tools of the web mining. The basic emphasis is given
on the three categories of the web mining and different techniques incorporated in web mining.
Dominic and Abdullah dealt with FP-growths Variation algorithms [8]. The paper considers to the
Classic frequent itemsets problem, which is the mining of all frequent itemsets that exist in market
basket-like data with respect to support thresholds. The execution time and the memory usage were
recorded to see which algorithm is the best. For the time consumption AFOPT(A Frequent Pattern
Tree) algorithm took advantage for most of the data set even though it suffers from segmentation fault
in the low support values on connect4 data set.
Venketesh et. al[9] presented a prediction model that built a Precedence Graph by considering the
characteristics of current websites in order to predict the future user requests. The algorithm
differentiated the relationship between the primary objects (HTML) and the secondary objects (e.g.,
images) when creating the prediction model.
Kasthuri et. al[10] shows that deduction of future references on the basis of predictive Prefetching,
can be implemented by based on past references. The prediction engine can be residing either in the
client/ server side. But in our context, prediction engine resides at client side. It uses the set of past
references to find correlation and initiates Prefetching that is driving users future requests for web
documents based on previous requests.
Sharma and Dubey proposed a framework for web traffic reduction [11]. The paper presented a
framework for the prefetching and prediction in web. According to the framework, previous web
requests of the user will be extracted from the proxy web log. From this web log, strong rules will be
generated using FP Growth algorithm. These rules will be used to prefetch the upcoming requests of
the current user.
Singh et. al[12] proposed a framework for prediction of web requests of users and accordingly,
prefetching the content from the server. The proposed framework improved performance of web
proxy server using web usage mining and prefetching scheme. They have clustered the users
according to their access pattern and usage behaviour with the help of K-Means algorithm and then
858 Vol. 7, Issue 3, pp. 856-862

Apriori algorithm is applied to generate rules for prefetching pages. This cluster based approach was
applied on proxy server web log data to test the results using LRU and LFU prefetching schemes.
Ford Lumban Gaol proposed web log sequential pattern mining using Apriori-all algorithm. The
experiment will be conducted base on the idea of Apriori-all algorithm, which first stores the original
web access sequence database for storing non-sequential data. An important application of sequential
mining techniques is web usage mining, for mining web log accesses, where the sequences of web
page accesses made by different web users over a period of time, through a server, are recorded [13].
III. THE PROPOSED FRAMEWORK
Prefetching and caching scheme is used to improve the performance of web even when more number
of user are connected and demanding access to the content over the web. In our proposed framework a
browser sends a request to the listener (forwarding function) to setup a connection between client and
proxy server. Listener then forwards the request for connection establishment via TCP/IP
Handshaking Protocol to the proxy server. Once the connection is established then proxy server sends
a hint list to the predictor. Predictor maintains a web log history which tells us about the pages and
content demanded with the number of hits of a particular web page with the IP address of the user.
Hints from the predictor are then applied with FP (Frequent Pattern) growth algorithm. FP Growth
algorithm gives the compressed data objects as an output, applying association rule mining approach.
The proposed framework is as shown in Figure 1.

Figure 1: Proposed Framework for web prefetching
2.1. Predictor
Predictor collect hint list from the proxy server with the help of keywords from the user query. Hint
list is very much essential to trace the type of data for which user is demanding, which is again
forwarded to web log history and to the block where FP growth algorithm is applied.
2.2. Web Log History
Web log history is very important aspect in prefetching. Web logs redirect the prefetch engine to
collect and cache particular type of data according to the demand of the user. Web log maintains the
session of the user, IP address of the user, for how long a user was connected, type of information for
which a user demanded, most frequently visited pages etc.
2.3. FP Growth
FP Growth approach is used for producing frequent item data sets based on divide and conquer
method. FP Growth is mainly used for mining frequent item sets without candidate generation. Major
step in FP-growth is to compress the database showing frequent item set in to FP-tree. To build a FP
859 Vol. 7, Issue 3, pp. 856-862

tree further we need to apply 2 steps on the dataset. Further FP tree is divided into a set of more
number of conditional databases in which we applied association rule mining for mining of frequent
item datasets from FP tree [8].FP growth algorithm pseudo code is as shown in Figure 2 [14].

Figure 2:FP Tree algorithm pseudo code [14]
2.4. Proxy Cache
Proxy cache is used with proxy server to save certain type of web data. The data after prefetching and
mining is fed to proxy cache. A proxy cache works very similar to that of cache in operating system.
When certain type information is demanded by a user / client through web browser passes through a
proxy server than proxy server look for the response of that query in its cache memory. If the result of
query for which the user is asking for found in cache memory of proxy server than it is called a hit
else it is a miss. If no result is found in server cache then the query is forwarded to main server to get
the result of the query.
2.4. Handshaking Process
As we know that TCP is a connection oriented protocol, but TCP does not establish connection on its
own TCP uses handshaking protocol for establishment of connection. Connection establishment is
very essential part in TCP so that further session of query and response could be maintained.
Handshaking protocol is of two types:
2.4.1. Two way handshaking
In this type of handshaking protocol there are 2 types of messages passed between client and server
for connection establishment and they are: Request message from host to server, accept/reject
message from server to host.
2.4.2. Three way handshaking
In this type of handshaking protocol there are 3 types of messages passed between client and server
for connection establishment and they are: Syc message from host to server, Syc_Ack from server to
host, Ack from host to server.
IV. EXPERIMENTAL WORK
The result analysis is done using online tool data-applied. Demo data from the tool is being used and
FP Growth is applied on that pre-processed data set. The results found shows that pattern generation is
fast and more accurate as compared to existing methodologies. Since web log data is huge and
FP_GROWTH

Input: FP-tree
Method: Call FP-growth (FP-tree, null).
procedure FP-growth (Tree, )
{
1) if (single path P) do
2) for each combination
= minimum support of nodes in .
3) Else For each header ai do
{

5) Construct .s conditional pattern base and then .s
conditional FP-tree Tree
6) If Tree = null
7) Then call FP-growth (Tree , )}

}
Output: complete set of frequent patterns

860 Vol. 7, Issue 3, pp. 856-862

existing works like as Apriori is slow because it generates candidate set of keys, whereas FP Growth
doesnt use this intermediate step. Hence, results in fast pattern generation process. Figure 3: shows
the online web data which is pre-processed using FP Growth algorithm on the data set named Web
Clickstream. It shows a log of data having various entities like; date, city, visits, page views, visit
duration, origin etc. Figure 4 shows the Data set when association rule is applied on Clickstream data;
it shows the association between various data elements. Figure 5 shows the descriptive information
about Web Clickstream data in text format. It also shows association rule that how data is gathered
and the percentage of data which is a hit.

Figure 3: Online pre-processed web using FP Growth algorithm.

Figure 4: Data set after applying association
861 Vol. 7, Issue 3, pp. 856-862

Figure 5: Descriptive data of dataset Web Clickstream after applying association
V. CONCLUSION
By the use of FP Growth algorithm in association rule mining we reduced the response time of most
of the user queries thereby increasing proxy server hit ratio, it improves the overall performance of
three tier web architecture. The user query is forwarded to proxy server through the listener and then a
predictor maintains web log history and passes the hints to FP Growth block where association is
applied on the loaded data set using association rule mining. This proposed framework helps in
reducing network congestion and it also helps in providing security to the main server as by
increasing the hit ratio in proxy server somehow we kept the user away from main server. By using
this framework we can also reduce DOS (Denial of service) attacks up to some extent. This
framework will also reduces the time for which a network channel is allocated to a user in TCP, as
soon as a user gets results to its queries, the allocated channel will become free and network traffic
can be reduced. The merits of this framework includes: increase in proxy server cache hit ratio. The
simulation of this framework is performed using Data-Applied tool on data set named Web
Clickstream collected from data-applied.com. Actual realization of this framework results in
improved network architecture.
VI. FUTURE WORK
In this framework, change in hit ratio will result thereby changing the size of proxy server cache and
threshold value. We are now focusing on realization of this framework under various circumstances
and varying network conditions mostly when network channels are heavily loaded. We are also
planning to investigate the performance of this framework in different networks like mobile network
where mobility can affect the performance of a network.

862 Vol. 7, Issue 3, pp. 856-862

REFERENCES
[1]. R.Manjusha, Dr.R.Ramachandran, (2011) Web Mining Framework for Security in E-commerce
IEEE-International Conference on Recent Trends in Information Technology, ICRTIT 2011. MIT,
Anna University, Chennai. June 3-5, 2011.
[2]. Pallis G., A. Vakali and J. Pokorny, (2008) A clustering-based prefetching scheme on a Web cache
environment, Computers and Electrical Engineering 34, Elsevier, pg 309323.
[3]. N. Sharma and S.K. Dubey, (2013) Fuzzy C-means clustering based prefetching to reduce web
traffic, International Journal of Advances in Engineering & Technology, vol. 6, Iissue 1, pp. 426-435,
ISSN: 2231-1963,2013.
[4]. R. Divya and V. S. Kumar, (2012) Survey on ais, apriori and fp-tree Algorithms,International Journal
of Computer Science and Management Research, vol. 1, issue 2, pp 194-200,ISSN 2278-733X,2012.
[5]. Kosala R. and Blockeel H, (2000) Web Mining Research: A Survey, ACM SIGKDD Explorations
Newsletter, June 2000, Volume 2 Issue 1.
[6]. Chang C., Lui S., Wu Y., (2001) Applying Pattern Mining to Web Information Extraction, Advances
in Knowledge Discovery and Data,2001 Springer.
[7]. N. Sharma and S.K. Dubey, (2012) A Hand to Hand Taxonomical Survey on Web Mining,
International Journal of Computer Applications, vol. 60, issue 3, ISSN 0975 8887,2012.
[8]. A.M. Said, P.D.D. Dominic and A. B. Abdullah ,(2009) A Comparative Study of FP-growth
Variations, International Journal of Computer Science and Network Security(IJCSNS), vol.9,
issue.5,pp 266-272,2009.
[9]. P. Venketesh and R.Venkatesan, (2011) Graph based Prediction Model to Improve Web
Prefetching, International Journal of Computer Applications (0975 8887)Volume 36 No.10.
[10]. I. Kasthuri, M.A. Ranjit Kumar , K. SudheerBabu, and Dr. S. S. S. Reddy, (2012) An Advance
Testimony for Weblog Prefetching Data Mining, International Journal of Advanced Research in
Computer Science and Software Engineering, Volume 2, Issue 4.
[11]. N. Sharma and S.K. Dubey, (2013) FP tree use in prefetching, Proc. of Int. Conf. on Advances in
Computer Science, AETACS, Elsevier, 2nd December, 2013, pp 555-561.
[12]. N. Singh, A. Panwar and R. S. Raw, (2013) Enhancing the Performance of Web Proxy Server
through Cluster Based Prefetching Techniques, International Conference on Advances in Computing,
Communications and Informatics (ICACCI), 2013, ISBN 978-1-4799-2432-5.
[13]. Ford Lumban Gaol, (2010) Exploring The Pattern of Habits of Users Using WebLog Squential
Pattern Second International Conference on Advances in Computing, Control, and
Telecommunication Technologies. 978-0-7695-4269-0/10 $26.00 2010 IEEE.
[14]. Kuldeep Malik, NeerajRaheja and PuneetGarg, (2011) Enhanced FP-Growth Algorithm IJCEM
Vol. 12, April 2011.

AUTHORS
Devender Banga is Pursuing M.Tech. from SGT Institute of Engineering and
Technology, Gurgaon, India. His current research area is Web Usage Mining.

Sunitha Cheepurisetti is Assistant Professor in Department of Computer Science
Engineering, SGT Institute of Engineering and Technology, Gurgaon, India. Her
research area is Operating System, Networks and Security. She has more than 6
publications in various National/International Journals.
863 Vol. 7, Issue 3, pp. 863-867

SEARCH NETWORK FUTURE GENERATION NETWORK FOR
INFORMATION INTERCHANGE
G. S. Satisha
IMPACT College of Engineering & Applied Sciences, Banglore, India

ABSTRACT
With the turn of the century, these communication technologies started to evolve in ways that forever changed
how users connect with the world. Broadband quickly replaced slower dial-up connections. In Wireless
technology, Ten years later, usersfrom consumers to enterprisessee broadband and wireless technologies as
integral to what they do every day. Consumers and workers expect an always on connection to their world,
whether thats the ability to instantly share images with different network across the country or analyzing
supply chain issues in near real-time from virtually anywhere around the world. Businesses and governments
realize the value of networks to provide solutions to many of their greatest challenges, especially integrating
systems. Fundamental to innovation in communications are networks. The past decade saw significant
advancements; innovation in the next decade promises to be just as impactful.
I. INTRODUCTION
Over the next decade, communication between networks will pay close attention to consumer
Communication and network technology trends as this shape the workplace .Users will expect the
same experience wherever they are, with whatever they are doing. This expectation will drive the
need for new solutions, creating new opportunities for organizations that respond. Broadband,
wireless, and global IP technologies will be at the heart of global economic growth to meet future
needs, todays networks are already undergoing major changes. Advancements in the way technology
components relate, including moving toward a more service-oriented architecture, provide increased
bandwidth flexibility, more rapid provisioning of network services, and put much more control in the
hands of the enterprise. These network changes, along with user demand, produce trends that reduce
the challenges of today and introduce new solutions for tomorrow.
To enable major, new communication capabilities, intelligence will be embedded in networks.
Embedded intelligence will make network control more automated, dynamic, and flexible. In the
coming decade, communication between network components wont be tightly tied to specific
hardware, but will instead reside in an interoperable standards-based control layer so that newly-
added network components, like switches and multiplexers, can signal themselves to the other
network components, allowing dynamic configuration. Networks are moving towards being more
effectively application aware. This means they will provide different service levels, depending on the
application. Application-aware search network enable organizations to achieve
Prioritized levels of performance for Private IP network applications such as VoIP, enterprise
resource planning (ERP), and video. The proper network assessment, reporting, dynamic bandwidth,
and packet-marking tools let organizations closely monitor performance, make adjustments, and
achieve cost efficiencies. This paper gives information on the future network expectation and how the
Search tree network can incorporate the demand for the future traffic and for the future needs.
The paper is organized into pervasive bandwidth pervasive IP connectivity, Purpose built solution.
Amd future work.
864 Vol. 7, Issue 3, pp. 863-867

II. PERVASIVE BANDWIDTH
In the years ahead, multimedia applications will continue to increase. The surge includes 3-D video,
video sharing, Biometric analysis, conferencing, and streamingall in higher definition than is
possible today. As an example of the trend, a 2010 volcanic eruption in Iceland caused a sharp rise in
video conferencing. Globally, mobile data traffic will double every year through 2014, and will grow
to 3.6 Exabytes per month.
2
The increased volume will cause data traffic to skyrocket and require
capacity not currently available. In the past, bandwidth available to a user at a given site was
available only in coarse increments, selected at the time of provisioning. Enterprises upgrading to
higher capacity faced costly projects that could take months to plan and implement. Increasingly,
network technologies such as bandwidth deployment over fiber optic cables and via Ethernet access
networks will allow rapid upgrades often without the need for long projects, costly site visits, and
physical construction projects. On-demand bandwidth capability is available today, though not yet
widely used. Over the next decade, use will increase dramatically. Organizations will still need to be
conscious of bandwidth needs when designing networks, but will find that the flexibility after the
initial deployment is far greater. Greater flexibility allows easy adjustments that more readily meet
changing business needs.
The growth of bandwidth-intensive applications will make dynamic and on-demand bandwidth
capabilities routine. In many cases, businesses will be able to upgrade bandwidth almost instantly,
without human intervention, triggered by consumption patterns and parameters set in advance.
Organizations may pay for underlying capacity and then peak bandwidth and data transfer utilization,
in a model similar to todays commercial electricity suppliers. Wireless will realize the most
impactful bandwidth milestone. Fourth-generation (4G) search networks will bring broadband
capacity to mobile devices at rates approaching, and potentially surpassing, 10 times the current
capacity. The first commercial 4G wireless network was deployed in Scandinavia in late 2009 using
Long Term Evolution (LTE) technology, and large scale deployments began in the U.S. in 2010. LTE
not only provides customers with true broadband speeds, it will also embed wireless connections in
cars, buildings, machines, and appliances enabling what some people call the Internet of things.
Verizon Wireless was among the first carriers in the U.S. to launch LTE
9
. Trials in Boston and
Seattle demonstrate the network is capable of peak download speeds of 40 to 50 mega bits per
second (Mbps) and peak upload speeds of 20 to 25 Mbps, though average, real-world rates are
expected to be 5 to 12 Mbps on the downlink and 2 to 5 Mbps on the uplink. Based on internal
estimates, aggregate growth in wireless data carried by Verizon will skyrocket after LTE is deployed,
with an increase of more than 2,000 percent between now and the end of 2017.Search network allows
user to transmit and receive information via various evolutions like 3G, 4G, 5G.
III. PERVASIVE IP CONNECTIVITY
As technology platforms continue to evolve, the barriers between wireless and wired networks and
devices around the world will eventually disappear. Consumers and businesses will expect their
applications to move seamlessly between platforms, no matter which network theyre connected to at
the moment. And theyll demand access to all of their contentregardless of where its stored
anytime, anywhere, and on any device. This trend is already underway with fixed mobile convergence
(FMC), where mobile phones transparently switch a voice call in progress between the cellular
network and VoIP. The goal of Search network is to provide a seamless transition of voice, data, and
even video communications between different types of networks, no matter the location or what
device is used, providing the user with an optimized, always available experience. Soon it will be
commonplace to continually watch a television show or video presentation while moving between
devices.
The move toward 4G technology is pushing networks closer toward FMC. In the near term, LTE will
enable billions, perhaps trillions, of devices to connect. Wireless sensors will then integrate everyday
items, such as household appliances and medical monitoring equipment and businesses will widely
deploy Machine to Machine (M2M) wireless communications solutions to track assets, remotely
monitor inventory, and ensure that distant equipment is operating properly. These types of sensors
865 Vol. 7, Issue 3, pp. 863-867

will provide data that suggest the need for proactive maintenance, or instantly report service
interruptions. LTE will also enable a new generation of broadband to wireless applications.
IV. PURPOSE-BUILT SOLUTIONS
A trend in the next decade will be purpose-built networks that solve particular business requirements.
By separating the network functions or services from the technology, businesses can specify a custom
network to suit particular business needs. In the past, a physical network would have been built to
accomplish this. Virtualization of these services will make it possible to create a logical network
without building a physical network.
The industry is moving away from the slow and awkward methods of adding point-to-point links that
result in a tangle of lines and excess equipment. Many are choosing VPNs that run over the public
Internet or private clouds based on Multi-Protocol Label Switching (MPLS) network technologies.
VPNs and private clouds let organizations customize their network solution by specifying the type
and level of security required to meet the business and regulatory needs, the bandwidth required, and
the data storage features.
Through 2022, the VPN and private cloud trends will continue. While lingering concerns about
availability and security will encourage many enterprises to participate in private cloud-based
interconnections for key relationships, the flexibility of Internet-based connections have enormous
and growing appeal. However, even in industries like financial services, where the largest players
tend to be late adopters of many technologies, a marked shift away from physical private networks
and toward MPLS-based private cloud networks exists.
Its not too difficult to imagine different types of virtual industry markets and exchanges developing
as a result of the expected changes.
9
On the energy horizon is smart grid. By providing two-way
communication between the user and the utility company, and between utility companies, an energy
market is taking shape that allows more fluid pricing, encouraging high value conservation at times of
peak demand, and will let users sell electricity to other users. In the healthcare field, health
information exchanges are forming as a way to securely share patient information across different
medical facilities. These exchanges help doctors avoid repeat patient tests, reduce
Missing patient data, and increase the quality of care. Early in 2009, network selected to move to an
open connectivity model for its Nordic markets by replacing its proprietary participant network with
the Verizon financial Network (VFN). The VFN is a dedicated and purpose-built business
infrastructure specifically designed to share market data and execute timely trades. VFN offers
financial services customers a fully end-to-end managed and supported, highly scalable and low
latency interface to the financial services ecosystem
V. THE FUTURE WORK
By the year 2022, communications networks will be an even more integral part of everyday life than
they are today, both at home and in the workplace. Network-driven technology will be a key enabler
of daily activities, yet it will become more transparent to the user. No longer will the user care about
how it works just that it works. The network of tomorrow
8
will produce a hyper-connected
environment. Intelligence will be built into the fabric of everything imaginable and some things not
yet imaginedall enabled by pervasive communications technologies. While many of the
advancements that will be commonplace in 2022 are already taking shape, a few advancements seem
to be straight out of a science fiction novel. These innovations will impact everyone and everything.
For example
Apparel.
Various wearable devices such as glasses or visors with built-in cameras and video displays will both
record and transmit information. Inconspicuous displays will send streaming information to the user,
such as a restaurant menu as a diner walks by a restaurant or tech support through a virtual reality
demonstration. Gaming vests will provide forced feedback as part of an augmented reality
experience.
Home.
866 Vol. 7, Issue 3, pp. 863-867

The fiber-enabled smart home will be a platform for managing every function of the digital
ecosystem, from home security and energy management to medical monitoring, telework and distance
learning. Refrigerators will have a touch surface that displays grocery lists and coupons, and the
ability o track contents for real time inventory control. Even carpets will be smarter, tracking the
health of the elderly by sensing erratic movements that may predict a fall.
Energy.
Household appliances can already be remotely controlled to run at off-peak times. San Diego Gas and
Electric found that if 80 percent of its customers used their washers and dryers at off-peak times, it
could eliminate two power plants. Energy pricing will become dynamic, changing in real-time,
motivating users to be more energy efficient.
Healthcare.
A patients wireless device will receive reminders about medication and therapy, in-home devices
will provide daily monitoring of vital statistics for preventative care, and patients will consult with
out-of-town doctors and specialists over high-definition 3-D video connections. Many people will
have body sensors, tiny wireless devices intended to track their vital signs
Government.
City-centric applications will report traffic and parking conditions using GPS-enabled sensors that
provide real-time notifications for public transit, and even monitor the citys air and water quality
Crime detection will be aided with context-aware video surveillance that reports unusual activity.
Civil services will be tailored to the individual, supported by the full integration of government
systems.
Enterprise.
Using the cloud model, traditional businesses will sell their internal capabilities as services that are
separate and distinct from their regular business offering, just as Amazon does today with their Web-
store infrastructure. RFID tags will become multifunctional sensors that not only provide item
location but also item health, which is useful for tracking food shipment. The percentage of tele
workers with no fixed regular work location will grow significantly, with the ability to work and
video conference on one device that can connect anywhere. The combination of increasingly
powerful and intelligent networks and innovative applications and devices will create a whole new
way to run a home, an enterprise, a community, or an economy
VI. CONCLUSION
Users already have high expectations of their communication technology. This will continue to the
point of dependence. Users will come to expect always-on access to the Internet that supports their
lifestyles in every way. None of this would be possible without the foundation of solid and advanced
communication networks. In the future, network providers will continue to drive open, IP-based
technical standards that allow new technologies to work together.. To drive the solutions and services
of 2022, network providers will form alliances with other providers and partnerships with application
developers and device makers.. Unprecedented access to real-time data, combined with
communication platforms that are available anywhere and anytime, will not only increase the rate of
change for existing models, but will substantially increase the pressures of global competition.
Having identified potential opportunities, the team should then conduct forward-looking pilot
projects. While some opportunities may not prove fruitful, the ones that do will create a new
competitive advantage. Broadband, wireless, and global IP technologies will be the heart of coming
economic growth. The evolution of the network over the next decade will not only enable new
products and services, Its also so important because of the vital role broadband must play in
advancing key societal goals in areas like education, health care, energy, public safety, democracy,
and small-business opportunity. Future search tree network should be able to large traffic and can
transmit and receive data of images from any mobile network and it should be compatible with the
high speed traffic data. It should take care of traffic from different generation and should pass data
from different network layers.
867 Vol. 7, Issue 3, pp. 863-867

REFERENCES
[1]. N. R. French and J. C. Steinberg, Factors Governing the Intelligibility of speech Sounds, The Acoustical
Society of America Journal,19 (1) (1947): 90.
[2]. A. H. Inglis, Transmission Features of the New Telephone Sets, Bell System Technical Journal 17 (1938):
358-380.
[3]. Anthony Rix and Mike Hollier, Perpceptual speech quality assessment from narrowband telephony to
wideband audio, AES 107 th Convention, New York: 24-27 September 1999
[4]. Forge, Simon, Colin Blackman and Erik Bohlin (2005) The Demand for Future Mobile Communications
Markets and services in Europe, Technical Report EUR 21673 EN, April .
[5]. Hazlett, Thomas W. (2006) An economic evaluation of spectrum allocation policy," in Richards, Foster,
and Kiedrowski, pp. 249258.
[6]. Evolution of Mobile Wireless Communication Networks-1G to 5G as well as Future Prospective of Next
Generation Communication Network IJCSMC, Vol. 2, Issue. 8, August 2013, pg.47 53
[7]. Naik, G., Aigal, V., Sehgal, P. and Poojary, J. (2012). Challenges in the implementation of Fourth
Generation Wireless Systems. International Journal of Engineering Research and Applications (IJERA) 2 (2)L:
1353-1355.
[8]. Singh, S. and Singh, P. (2012). Key Concepts and Network Architecture for 5G Mobile Technology.
International Journal of Scientific Research Engineering & Technology (IJSRET) 1 (5): 165-170.
[9]. Hrikhande, S. and Mulky, E. (2012). Fourth Generation wireless Technology. Proceedings of the National
Conference "NCNTE-2012" at Fr. C.R.I.T., Vashi, Navi Mumbai: 24-25.
[10]. UMTS World (2009). UMTS/3G History and Future Milestones, [Online] Available:
http://www.umtsworld.com/umts/history.htm

AUTHORS BIOGRAPHY
The Author is currently working in the Electronics and Communication Engineering
Department as a HOD and Professor.
868 Vol. 7, Issue 3, pp. 868-878

A BRIEF SURVEY ON BIO INSPIRED OPTIMIZATION
ALGORITHMS FOR MOLECULAR DOCKING
Mayukh Mukhopadhyay
Tata Consultancy Services, Kolkata-700091, India

ABSTRACT
The idea in molecular docking is to design pharmaceuticals computationally by identifying potential drug
candidates targeted against proteins. The candidates can be found using a docking algorithm that tries to
identify the bound conformation of a small molecule ligand to a macromolecular target at its active site, which
is the region of an enzyme where natural substrate binds. Mathematically, molecular docking can be formulated
as an optimization problem in which the objective is to minimize the intermolecular bound conformational
energy of two interacting molecules. In this brief survey, after a concise introduction on nature inspired
computing, the application of bio inspired optimization algorithms in molecular docking has been studied.
KEYWORDS: Bio-Inspired Algorithm, Differential Evolution, Protein-Ligand Docking Problem
I. NATURE INSPIRED COMPUTING
1.1 Introduction
Nature Inspired Computing (NIC) is one that aims to develop new computing techniques after getting
ideas by observing how nature behaves in various situations to solve complex problems. A typical
NIC system is based on self-organization and complex systems. It is a computing system operated by
population of autonomous entities surrounded by environment.
Autonomous entity of the NIC system consists of two devices: detectors and effectors [25]. There
may be one or more detector that receives information related to its neighbours and its environment.
Obviously the information depends on the system to be modeled or problem to be solved. There may
be multiple effectors that make changes to the internal state, exhibits certain behaviors and make
changes to the environment. Basically the effector facilitates sharing of information among
autonomous entities. NIC system has a repository of local behavior rules. The rules of behavior are
crucial to autonomous entity. They are used to decide how autonomous entity must act or react to the
information collected by the detector from the environment and neighbors. Autonomous entity should
be capable of learning. It must respond to local changing conditions by modifying its rules of
behaviour over time.
Autonomous entities have ADEAS (Autonomous, Distributed, Emergent, Adaptive, Self-organized)
characteristics [25]. Autonomous entities are independent and rational. They use formal computing
methods to describe how entities acquire and improve reactive behavior. They have decision making
capabilities and are distributed in the environment. They follow predefined protocols and interact with
each other to exchange their state information. New complex behaviors emerge when the entities act
collectively. The entities respond to changes in the environment by changing their behavior. By
interacting with each other the entities self-organize to fine tune their behaviour.
Environment may be static or dynamic. In static environment, autonomous entities are free to roam.
Dynamic environment acts as notice board allowing autonomous entities post and read local
information. The central clock helps synchronize the actions of the autonomous entities.
By far the majority of nature-inspired algorithms are based on some successful characteristics of
biological system [22]. Therefore, the largest fraction of nature-inspired algorithms is biology-
inspired, or bio-inspired for short.
869 Vol. 7, Issue 3, pp. 868-878

Among bio-inspired algorithms, a special class of algorithms has been developed by drawing
inspiration from swarm intelligence. Therefore, some of the bio-inspired algorithms can be called
swarm-intelligence based. In fact, algorithms based on swarm intelligence are among the most
popular. Good examples are ant colony optimization, particle swarm optimization, cuckoo search, bat
algorithm, and firefly algorithm.
Obviously, not all algorithms were based on biological systems. Many algorithms have been
developed by using inspiration from physical and chemical systems. Some may even be based on
music. In the next section, we will briefly divide all algorithms into different categories, and we do
not claim that this categorization is unique. This is a good attempt to provide sufficiently detailed
references.
1.2 Classification
We can divide all existing nature inspired algorithms into four major categories: swarm intelligence
(SI) based, bio-inspired (but not SI-based), physics/chemistry-based, and others. We will summarize
them briefly in the rest of this section. However, we will focus here on the relatively new algorithms.
Well established algorithms such as genetic algorithms are so well known that there is no need to
introduce them in this brief paper. It is worth pointing out the classifications here are not unique as
some algorithms can be classified into different categories at the same time.
Loosely speaking, classifications depend largely on what the focus or emphasis and the perspective
may be. For example, if the focus and perspective are about the trajectory of the search path,
algorithms can be classified as trajectory based and population-based. Simulated annealing is a good
example of trajectory-based algorithms, while particle swarm optimization and firefly algorithms are
population-based algorithms.
If our emphasis is placed on the interaction of the multiple agents, algorithms can be classified as
attraction-based or non-attraction-based. Firefly algorithm (FA) is a good example of attraction based
algorithms because FA uses the attraction of light and attractiveness of fireflies, while genetic
algorithms are non-attraction-based since there is no explicit attraction used.
On the other hand, if the emphasis is placed on the updating equations, algorithms can be divided into
rule-based and equation-based. For example, particle swarm optimization and cuckoo search are
equation based algorithms because both use explicit updating equations, while genetic algorithms do
not have explicit equations for crossover and mutation. However, in this case, the classifications are
not unique. For example, firefly algorithm uses three explicit rules and these three rules can be
converted explicitly into a single updating equation which is nonlinear.
This clearly shows that classifications depend on the actual perspective and motivations. Therefore,
the classifications here are just one possible attempt, though the emphasis is placed on the sources of
inspiration.
1.2.1 Swarm intelligence based
Swarm intelligence (SI) concerns the collective, emerging behaviour of multiple, interacting agents
who follow some simple rules [22]. While each agent may be considered as unintelligent, the whole
system of multiple agents may show some self-organization behaviour and thus can behave like some
sort of collective intelligence. Many algorithms have been developed by drawing inspiration from
swarm-intelligence systems in nature.
All SI-based algorithms use multi-agents, inspired by the collective behaviour of social insects, like
ants, termites, bees, and wasps, as well as from other animal societies like flocks of birds or fish. A
list of swarm intelligence algorithms is presented in Table 1.
The classical particle swarm optimization (PSO) uses the swarming behaviour of fish and birds, while
firefly algorithm (FA) uses the flashing behaviour of swarming fireflies. Cuckoo search (CS) is based
on the brooding parasitism of some cuckoo species, while bat algorithm uses the echolocation of
foraging bats. Ant colony optimization uses the interaction of social insects (e.g., ants), while the class
of bee algorithms are all based on the foraging behaviour of honey bees.
SI-based algorithms are among the most popular and widely used. There are many reasons for such
popularity; one of the reasons is that SI-based algorithms usually sharing information among multiple
agents, so that self-organization, co-evolution and learning during iterations may help to provide the
high efficiency of most SI-based algorithms. Another reason is that multiple agent can be parallelized
easily so that large-scale optimization becomes more practical from the implementation point of view.
870 Vol. 7, Issue 3, pp. 868-878

1.2.2 Bio-inspired but not SI-based
Obviously, SI-based algorithms belong to a wider class of algorithms, called bio-inspired algorithms.
In fact, bio-inspired algorithms form a majority of all nature-inspired algorithms. From the set theory
point of view, SI-based algorithms are a subset of bio-inspired algorithms, while bio-inspired
algorithms are a subset of nature-inspired algorithms [22]. That is

SI-based bio-inspired nature-inspired (1)

Conversely, not all nature-inspired algorithms are bio-inspired, and some are purely physics and
chemistry based algorithms as we will see below. Many bio-inspired algorithms do not use directly
the swarming behaviour. Therefore, it is better to call them bio-inspired, but not SI-based. For
example, genetic algorithms are bio-inspired, but not SI-based. However, it is not easy to classify
certain algorithms such as differential evolution (DE). Strictly speaking, DE is not bio-inspired
because there is no direct link to any biological behaviour. However, as it has some similarity to
genetic algorithms and also has a key word evolution, we tentatively put it in the category of bio
inspired algorithms. These relevant algorithms are listed in Table 1.
For example, the flower algorithm or flower pollination algorithm developed by Xin-She Yang in
2012 is a bio-inspired algorithm, but it is not a SI-based algorithm because flower algorithm tries to
mimic the pollination characteristics of flowering plants and the associated flower consistency of
some pollinating insects.
1.2.3 Physics and Chemistry based
Not all metaheuristic algorithms are bio-inspired, because their sources of inspiration often come from
physics and chemistry. For the algorithms that are not bio-inspired, most have been developed by
mimicking certain physical and/or chemical laws, including electrical charges, gravity, river systems,
etc. As different natural systems are relevant to this category, we can even subdivide these into many
subcategories which are not necessary. A list of these algorithms is given in Table 1.
Schematically, we can represent the relationship of physics and chemistry based algorithms as the
follows [22]:
(2)
Though physics and chemistry are two different subjects, however, it is not useful to subdivide this
subcategory further into physics-based and chemistry. After all, many fundamental laws are the same.
So we simply group them as physics and chemistry based algorithms.
1.2.4 Other algorithms
When researchers develop new algorithms, some may look for inspiration away from nature.
Consequently, some algorithms are not bio-inspired or physics/chemistry-based, it is sometimes
difficult to put some algorithms in the above three categories, because these algorithms have been
developed by using various characteristics from different sources, such as social, emotional, etc. In
this case, it is better to put them in the other category, listed in Table 1 [22].
1.2.5 Some Remarks
Though the sources of inspiration are very diverse, the algorithm designed from such inspiration may
be equally diverse. However, care should be taken, as true novelty is a rare thing. For example, there
are about 28,000 living species of fish; this cannot mean that researchers should develop 28000
different algorithms based on fish. Therefore, one cannot call their algorithms trout algorithm, squid
algorithm or shark algorithm[22].
It is worth pointing out that studies show that some algorithms are better than others which are still
not quite understood why. However, if one looks at the intrinsic part of algorithm design closely,
some algorithms are badly designed, which lack certain basic capabilities such as the mixing and
diversity among the solutions. In contrast, good algorithms have both mixing and diversity control so
that the algorithm can explore the vast search space efficiently, while converge relatively quickly
when necessary. Good algorithms such as particle swarm optimization, differential evolution, cuckoo
search and firefly algorithms all have both global search and intensive local search capabilities, which
may be partly why they are so efficient.
871 Vol. 7, Issue 3, pp. 868-878

Table 1. A list of algorithms
Swarm intelligence based algorithms Bio-inspired (not SI-based) algorithms
Algorithm Author Algorithm Author
Accelerated PSO Yang et al. Atmosphere clouds model Yan and Hao
Ant colony optimization Dorigo Biogeography-based optimization Simon
Artificial bee colony Karaboga and Basturk Brain Storm Optimization Shi
Bacterial foraging Passino Differential evolution Storn and Price
Bacterial-GA Foraging Chen et al. Dolphin echolocation Kaveh and Farhoudi
Bat algorithm Yang Japanese tree frogs calling Hernandez and Blum
Bee colony optimization Teodorovic and Dell'Orco Eco-inspired evolutionary algorithm Parpinelli and Lopes
Bee system Lucic and Teodorovic Egyptian Vulture Sur et al.
BeeHive Wedde et al. Fish-school Search Lima et al.
Wolf search Tang et al. Flower pollination algorithm Yang
Bees algorithms Pham et al. Gene expression Ferreira
Bees swarm optimization Drias et al. Great salmon run Mozaffari
Bumblebees Comellas and Martinez Group search optimizer He et al.
Cat swarm Chu et al. Human-Inspired Algorithm Zhang et al.
Consultant-guided search Iordache Invasive weed optimization Mehrabian and Lucas
Cuckoo search Yang and Deb Marriage in honey bees Abbass
Eagle strategy Yang and Deb OptBees Maia et al.
Fast bacterial swarming algorithm Chu et al. Paddy Field Algorithm Premaratne et al.
Firefly algorithm Yang Roach infestation algorithm Havens
Fish swarm/school Li et al. Queen-bee evolution Jung
Good lattice swarm optimization Su et al. Shuffled frog leaping algorithm Eusuff and Lansey
Glowworm swarm optimization Krishnanand and Ghose Termite colony optimization Hedayatzadeh et al.
Hierarchical swarm model Chen et al. Physics and Chemistry based algorithms
Krill Herd

Gandomi and Alavi

Big bang-big Crunch

Zandi et al.

Monkey search Mucherino and Seref Black hole Hatamlou
Particle swarm algorithm Kennedy and Eberhart Central force optimization Formato
Virtual ant algorithm Yang Charged system search Kaveh and Talatahari
Virtual bees Yang Electro-magnetism optimization Cuevas et al.
Weightless Swarm Algorithm Ting et al. Galaxy-based search algorithm Shah-Hosseini
Other algorithms Gravitational search Rashedi et al.
Anarchic society optimization

Shayeghi and Dadashpour

Harmony search

Geem et al.

Artificial cooperative search Civicioglu Intelligent water drop Shah-Hosseini
Backtracking optimization search Civicioglu River formation dynamics Rabanal et al.
Differential search algorithm Civicioglu Self-propelled particles Vicsek
Grammatical evolution Ryan et al. Simulated annealing Kirkpatrick et al.
Imperialist competitive algorithm Atashpaz-Gargari and Lucas Stochastic difusion search Bishop
League championship algorithm Kashan Spiral optimization Tamura and Yasuda
Social emotional optimization Xu et al. Water cycle algorithm Eskandar et al.

1.3 Applications
Nature Inspired Computing techniques are so flexible that they can be applied to wide range of
problems, so adaptable that they can deal with unseen data and capable of learning, so robust that they
can handle incomplete data. They have decentralized control of computational activities.
Biological inspired computing is a subset of nature inspired computing. There are three key
differences between traditional computing systems and biological information processing systems:
components of biological systems respond slowly but implement much higher-level operations. The
ability of biological systems to assemble and grow on their own enables much higher interconnection
densities. The implementation of biological systems is not a planned one.
One who expects solutions from nature for complex problems has to first observe the natures
behaviour carefully. The next step is to use models and list all the behaviours observed so far. The
872 Vol. 7, Issue 3, pp. 868-878

above steps should be repeated till a near perfect working model is obtained. As a by-product some
unknown mechanisms may be found. Based on the observation from nature a problem-solving
strategy is formulated. Two main computational applications of NIC are Clustering and Optimization.
1.3.1 Clustering
Clustering is the unsupervised classification of patterns (observations, data items or feature vectors)
into groups (clusters). A loose definition of clustering could be the process of organizing objects into
groups whose members are similar in some way.
A cluster is therefore a collection of objects which are similar between them and are dissimilar to
the objects belonging to other clusters.
The goal of clustering is to determine the intrinsic grouping in a set of unlabeled data. It can be shown
that there is no absolute best criterion which would be independent of the final aim of the
clustering. Consequently, it is the user which must supply this criterion, in such a way that the result
of the clustering will suit their needs.
For instance, we could be interested in finding representatives for homogeneous groups (data
reduction), in finding natural clusters and describe their unknown properties (natural data types),
in finding useful and suitable groupings (useful data classes) or in finding unusual data objects
(outlier detection).
1.3.2 Optimization
An optimization problem is the problem of finding the best solution from all feasible solutions.
Optimization problems can be divided into two categories depending on whether the variables are
continuous or discrete.
Classification of optimization algorithm can be carried out in many ways. A simple way is to look at
the nature of the algorithm, and this divides the algorithm into two categories: deterministic
algorithms, and stochastic algorithms.
Deterministic algorithms follow a rigorous procedure, and its path and values of both design variables
and the function are repeatable. On the other hand, stochastic algorithms always have some
randomness and each individual path towards a feasible solution is not exactly repeatable.
II. MOLECULAR DOCKING
2.1 Purpose and Motivation
Proteins found in nature have evolved by natural selection to perform specific functions. The
biological function of a protein is linked to its three-dimensional structure or conformation.
Consequently, protein functionality can be altered by changing its structure.
The idea in molecular docking is to design pharmaceuticals computationally by identifying potential
drug candidates targeted against proteins. The candidates can be found using a docking algorithm that
tries to identify the bound conformation of a small molecule, also referred to as ligand, to a
macromolecular target at its active site, which is the region of an enzyme (large proteins that catalyse
chemical reactions) where the natural substrate (specific molecule an enzyme acts upon) binds.
Often, the active site is located in a cleft or pocket in the proteins tertiary structure. The structure and
stereochemistry (spatial arrangement of the atoms) at the active site complements the shape and
physical/chemical properties of the substrate so as to catalyse a particular reaction. The purpose of
drug discovery is thus to derive drugs or ligands that bind stronger to a given protein target than the
natural substrate. By doing this, the bio-chemical reaction that enzyme catalyses can be altered or
prevented.
Until recently, drugs were discovered by chance via a trial-and-error manner using a high-throughput
screening methods that experimentally test a large number of compounds for activity against the
target in question. This process is very expensive and time consuming. If a three- dimensional
structure of the target exits, simulated molecular docking can be a useful tool in the drug discovery
process because it allows for many possible lead candidates to be tested before committing expensive
resources for wet lab experiments (synthesis), toxicological testing, bioavailability and clinical trials.
The focus of the below sections will be on realizing the complexity of simulating such a docking of
ligand into the active site of target protein and a brief review of the most representative bio-inspired
metaheuristic algorithms for doing the same.
873 Vol. 7, Issue 3, pp. 868-878

2.2. Problem Definition
Molecular docking may be defined as an optimization problem, which would describe the best-fit
orientation of a ligand that binds to a particular protein of interest. Figure 1 below illustrates docking
of a small molecule into a proteins active site where solvent molecules are indicated by filled circles
[23]:

Figure 1: Protein Ligand docking in solvent
Protein-ligand docking can also be defined as a search problem where the task is to find the best
ligand conformation (low energy binding mode) within the active site of protein. Since the relative
orientation and conformation of both the protein and the ligand is taken into account it is not an easy
problem. Usually, the protein is treated as fixed in three-dimensional coordinate system while the
ligand is allowed to be flexible. The flexibility of the ligand is measured in terms of the number of
rotatable bonds in the ligand. Following are the four types of Docking Problem based on increasing
complexity:
(i) Protein-Rigid & Ligand-Rigid
(ii) Protein-Rigid & Ligand-Flexible
(iii) Protein-Flexible & Ligand-Rigid
(iv) Protein-Flexible & Ligand-Flexible
Category (iv) can be further divided in two cases:
(iv.a) Rigid protein backbone, selected side chains are flexible
(iv.b) Fully flexible protein backbone and side chains.
It can be thought of as a problem of lock-and-key, where one is interested in finding the correct
relative orientation of the key which will open up the lock (where on the surface of the lock is the
key hole, which direction to turn the key after it is inserted, etc.). Here, the protein can be thought of
as the lock and the ligand can be thought of as a key. Molecular docking may be defined as an
optimization problem, which would describe the best-fit orientation of a ligand that binds to a
particular protein of interest. However, since both the ligand and the protein are flexible, a hand-in-
glove analogy is more appropriate than lock-and-key. During the course of the process, the ligand
and the protein adjust their conformation to achieve an overall best-fit and this kind of
conformational adjustments resulting in the overall binding is referred to as induced-fit as illustrated
in figure 2 below:

Figure 2: Analogy for docking problem
2.2.1 The Docking Problem
Let A and B be a ligand and protein molecule, respectively. Further, let f be a scoring function(energy
function) that ranks solutions with respect to binding energy and let C be the conformational search
874 Vol. 7, Issue 3, pp. 868-878

space of all possible conformations(docking solutions) between A and B. Then we can define docking
problem as an optimization (minimization) problem in which the task is to find the following [23]:

(3)

A docking problem requires [24]:
1) A Scoring function to score protein-ligand complex by measuring the binding strength of the ligand
at a specific position of the protein.
2) A search algorithm to search for the protein-ligand combination that corresponds to the lowest
energy conformation.
2.2.2 Complexity of Docking Problem
Docking problems have huge search spaces, and the number of possible conformations to consider
increases dramatically when flexibility is taken into account. Unfortunately, the search space is not
unimodal but rather highly multimodal due to the numerous local optima caused by the energy
function used. Moreover, the complexity of the docking problem is influenced by the size, shape and
bonding topology of the actual ligand being docked.
Despite great improvements in computational power, docking remains a very challenging problem,
generally considered to be NP-hard (although no formal proof exists) [23]. Thus, a brute-force
approach looking at all possible docking conformations is impossible for all but the simplest docking
problems. To handle docking flexibility in an efficient manner, search heuristics that sample the
search space of possible conformations are required.
2.3. Survey of Bio Inspired Algorithms on Docking Problem
In most of the protein- ligand docking problems in literature, which has been solved by evolutionary
algorithms, the protein is kept rigid and ligands 3 translational, 3 rotational and r torsional degrees of
freedom is optimized. Thus the total number of variables, the dimension of the optimization problem
equals 6+r. A wide variety of optimization strategies is used to find the global minimum
corresponding to a complex structure.
One of the first applications of EAs to molecular docking was introduced by Dixon [3] .He used a
simple GA with binary encoding representing a matching of ligand atoms and spheres placed inside
the active site similar to DOCK (Kuntz et al., 1982. He used incremental construction algorithms.)
Only limited experimental results were presented by him. Later on, Oshiro, Kuntz and Dixon [16]
introduced a similar GA using a DOCK scoring function. This time the experiments resulted in the
low root mean square deviation (RMSD) values.
Genetic Algorithm for Minimization of Energy (GAME) by Xioa and Williams [18] was another early
attempt for molecular docking. GAME was designed to dock one or more rigid ligand to rigid protein.
Solutions were encoded as binary strings representing three rotational angles and three translational
coordinates. The algorithm was tested on one protein ligand complex. It was done with different
search strategies firstly; ligand was allowed to move with fixed rotation. Secondly, ligand was
allowed to rotate with fixed translation. At last all six degrees of freedom have been taken into
account.
Clark and Ajay [1] in 1995 introduced DIVALI. It used Gray-coded binary representation, bit-flip
mutation, and an elitism scheme. Two point crossover was used as it was believed to perform better.
The protein remained rigid during docking process whereas the ligand was flexible. Its key feature
was masking operator. It was evaluated on four protein ligand complexes using an AMBER-like force
field as scoring function and obtained good docking results.
Genetic Optimization for Ligand Docking, GOLD was developed by Jones et al [9].They used a
mixed encoding combining binary string and real valued numbers. Binary strings were used to
represent torsion angles of the ligand and selected side chain angles of the protein thus giving partial
flexibility to protein. Integer strings were used to represent mappings of hydrogen bonding sites
between ligand and protein. Both number and strength of hydrogen bonds and Van der Waals energy
contributed in the fitness function there. Standard one point crossover and bit-flip mutation were used.
Levine et al [12] developed Stalk system in which they applied new technologies to molecular
docking prediction. They combined the concepts of parallel and distributed computing, high speed
875 Vol. 7, Issue 3, pp. 868-878

networking, genetic algorithms and virtual reality for the study of molecular interactions. In Stalk they
have used CAVE (CAVE Automatic Virtual Environment) as the virtual reality system. Stalk
considered a rigid body docking in which it is assumed that no modification of proteins backbone or
side chain occurs. Each string of GA contained six parameters three translational and three rotational.
The fitness function included non-bonded interactions only. Stalk was run on one protein-ligand
complex and showed good docking results.
AutoDock is another popular docking program that originally used Simulated annealing which was
later replaced by Genetic Algorithm to search for promising docking conformations. GA was then
replaced by LGA, Lamarckian Genetic Algorithm (Morris et al [14] which was a hybrid GA
combining a genetic algorithm and local search. LGA used a real valued representation, two point
crossover and Cauchy mutation with fixed value of, defining the mean and spread of Cauchy
distribution. In their representations each chromosome (solution) was composed of a string of real
valued genes: three Cartesian coordinates for the ligand translation; four variables defining a
quaternion specifying the ligand orientation; and one for flexible torsion angle. Further, local search
was applied on each generation on a user defined proportion of population. When local search is
applied the genotype of the individual is replaced with the new best solution found by evaluating its
fitness, which is the sum of intermolecular interaction energy between the ligand and the protein and
the intramolecular interaction energy of the ligand. The implementations of AutoDock on various
protein-ligand complexes and corresponding numerical results have been presented (Morris et al,
1998). There they concluded that out of three search algorithms (SA, GA, LGA) which they have
applied in AutoDock, the most efficient, reliable and successful is Lamarckian Genetic Algorithm
LGA.
Vieth et al.[17] compared the efficiency of molecular dynamics (MD), Monte Carlo (MC) and
Genetic Algorithms (GA) for docking five representative ligand- receptor complexes. All the three
algorithms employed CHARMM-based energy function. The results were also compared with
AutoDock. The receptor was kept rigid while flexibility of ligand was permitted. They concluded that
MD was most efficient in case of large search spaces while GA outperformed others in small search
spaces.
In 1999, Wang [7] et al. applied GA combined with random search. Steric complementarity and
energetic complementarity of ligand with its receptor have been separately considered in two stage
automated docking. Eight complexes have been randomly selected. For most of the cases the root
mean square (RMS) of the GA solutions was smaller than 1.0. In the first stage rough searching of a
set of bound sites based on steric complementarity was done whereas in the second stage detailed
searching of the locally associated sites based on energetic complementarity was performed.
With the passage of time several other algorithms have been developed, changes in variation operators
have been made to improve the efficiency of docking algorithms. Multi-Objective, multi-population
genetic algorithms came into picture. Of recent improvements C-l Li et al. [13] proposed an
information entropy based evolution model for molecular docking. The model was based on a multi-
population genetic algorithm. The GA proposed to solve this was binary coded. In each generation
there were three operators: selection, crossover and mutation. Selection was performed by an integer
decimal method, crossover was two-point and mutation used was uniform. An elitist maintaining
mechanism was designed in the proposed GA.
Krob et al.[15] applied Ant Colony Optimization to structure based drug design. They introduced a
new docking algorithm PLANTS (Protein-Ligand Ant System) based on ACO metaheuristic. An
artificial Ant Colony was employed to find a minimum energy conformation of the ligand in the
proteins active site. They presented the effectiveness of PLANTS for several parameter settings as
well as a direct comparison to GOLD, which is based on a GA. PLANTS, treated the ligand as
flexible while almost whole receptor except rotatable hydrogen bond donors were kept rigid. The
continuous variables have been discretized so that ACO can directly be applied, as it is designed to
tackle combinatorial optimization problems.
Most recently Janson [8] et al. have done work on molecular docking with multi-objective particle
swarm optimization (PSO). A new algorithm ClustMPSO was proposed which is based on PSO and
follows a multi-objective approach for comparing the quality of solutions. Energy evaluations were
done using AutoDock tools. In the second part of the paper a new approach for predicting the docking
trajectory was proposed. It was shown empirically that ClustMPSO finds solution for docking
876 Vol. 7, Issue 3, pp. 868-878

problem with lower energy than LGA and SA that are incorporated in AutoDock. ClustMPSO is also
faster with respect to the solution evaluations.
Differential evolution (DE) was introduced by Storn and Price[19]. Compared to more widely known
EA-based techniques (e.g. genetic algorithms, evolutionary programming and evolution strategies),
DE uses a different approach to select and modify candidate solutions. The main innovative idea in
DE is to create offspring from a weighted difference of parent solutions. The DE algorithm has
recently demonstrated superior performance, outperforming related heuristics, such as GAs and PSO
on both numerical benchmark functions and in several real world applications.
Based on the great success of DE, a novel application of DE to flexible ligand docking (termed
DockDE) was introduced by Thomsen [20]. The comparison was performed on a suite of six
commonly used docking benchmark problems where DockDE obtained the best energy for all of them
compared to the DockEA and LGA algorithms as depicted in the below figure 3 extracted from [20].

Figure 3. Energy Graph of DockDE against LGA and DockEA
In a recent study, a new docking algorithm called MolDock was introduced by Thomsen and
Christensen [21] which is based on a hybrid search algorithm that combines the DE optimization
technique with a cavity prediction algorithm.

Figure 4. Comparative Survey of MolDock docking accuracy
The docking accuracy of MolDock on selected 77 complexes is 87%, outperforming the other docking
algorithms on the benchmark data set which is depicted in above Figure 4 extracted from [21]. In
conclusion, the results strongly suggest that the Differential Evolution algorithm has a great potential
for protein-ligand docking.
877 Vol. 7, Issue 3, pp. 868-878

III. CONCLUSIONS
Molecular docking is a valuable tool for pharmaceutical companies because they can be used to
suggest promising drug candidates at a low cost compared to making real-world experiments. As a
result of the need for methods that can predict the binding of two molecules when complexed,
numerous algorithms have been suggested during the last three decades. In this survey, a review of the
most representative bio-inspired optimization algorithms for protein-ligand docking was studied. In
particular, docking methods utilizing differential evolution have been surveyed to have superior
performance to other well studied approaches regarding accuracy and speed, issues that are important
when docking hundreds of thousands of ligands in virtual screening experiments.
IV. FUTURE/PROPOSED WORK
Cancer is a serious and often fatal disease, widespread in the developed world. One of the most
common methods of treating cancer is chemotherapy with cytotoxic drugs. These drugs, around 30
used regularly in chemotherapy, themselves can often have debilitating or life threatening effects on
patients undergoing chemotherapy. Hence chemotherapy can be perceived as a complex control
problem of how to balance the need for tumour treatment against the undesirable and dangerous side
effects that the treatment may cause. Proposed work can be subdivided into two phases:
A) Literature Survey and Virtual screening for drug delivery
Here we survey current scientific literature to track potential cytotoxic Ligands. Then we use
molecular docking tools to simulate novel protein-ligand complexes using these potential candidates
for virtual drug on the basis of scoring function and optimized conformation.
B) Drug Administrating Optimization
Here we design a system, specifically a workbench, which supports the design of novel multidrug
cancer chemotherapy treatments. The aim of the system is to allow clinicians to interact with models
of tumour response and to use nature-inspired optimization algorithm to search for improved
treatments. In simple words, the system should take in effective input of patient current status of
tumour growth and provide output of administered dose of a combination of cytotoxic drug for
optimal control.
ACKNOWLEDGEMENTS
The author would like to thank Department of Information Technology, Jadavpur University, Salt lake
Campus for providing necessary access to scientific literature while conducting this survey. The
author is grateful to Dr Parama Bhaumik, for guidance and valuable technical inputs. Finally, the
author is indebted to constant motivation and grammatical editing from his better-half Mrittika
Chatterjee.
REFERENCES
[1]. Clark, K.P. and A.N. Jain (1995) Flexible ligand docking without parameter adjustment across four
ligand receptor complexes, J. Comput. Chem., Vol. 16, pp. 1210-1226.
[2]. Deb, K.(2005). Optimization For Engineering Design, Prentice-Hall of India Pvt. Limited, New Delhi.
[3]. Dixon, J.S.(1993). Flexible docking of ligands to receptor sites using genetic algorithms, Proceedings
of the 9
th
European Symposium on Structure-Activity Relationships: QSAR and Molecular Modelling,
pp 412-413.
[4]. Dorigo, M., T. Stiitzle: Ant Colony Optimization.MIT Press, Cambridge, MA, USA 2004.
[5]. Goldberg, D.E. (1989). Genetic algorithms in search, optimization and machine learning, Addision
Wesley, New York.
[6]. Hart, W.E., C. Rosin, R.K. Belew, and G.M. Morris (2000). Improved evolutionary hybrids for flexible
ligand docking in AutoDock, Optim. Comput. Chem. Mol. Biol., pp. 209-230.
[7]. Hou, T., J. Wang, L. Chen, and X. Xu (1999) Automated docking of peptides and proteins using a
genetic algorithm combined with a tabu search, Pro. Engi., Vol. 12 No.8, pp. 639-647.
[8]. Janson, S., D. Merkle, and M. Middendorf (2008) Molecular docking with multi-objective Particle
Swarm Optimization, App. Soft Computing Vol. 8, pp. 666-675.
[9]. Jones, G., P. Willett, and R.C. Glen (1995). Molecular recognition of receptor sites using a genetic
878 Vol. 7, Issue 3, pp. 868-878

algorithm with a description of desolvation. J. Mol. Biol., Vol. 245, pp. 43-53.
[10]. Kennedy, J., R. Eberhart, Particle Swarm Optimization, IEEE International Conference on Neural
Networks (ICNN95), Vol. 4, 1995.
[11]. Kuntz, I.D., J.M. Blaney, S.J. Oatley, R. Langridge and T.E. Ferrin (1982). A geometric approach to
macromolecule-ligand interactions, J. Mol. Biol., Vol. 161, pp. 269-288.
[12]. Levine, D., M. Facello, P. Hallstrom, G. Reeder, B. Walenz, and F. Stevens (1997). Stalk: An
interactive system for virtual molecular docking, IEEE Comput. Sci. & Engi., Vol. 4, Issue 2, pp. 55-
65.
[13]. Li, Chun-lian, Y. Sun, D. Long, and X. Wang (2005) A genetic algorithm based method for molecular
docking, ICNC: International conference on advances in natural computation. LNCS 3611, pp. 1159-
1163.
[14]. Morris, G.M., D.S. Goodsell, R.S. Hallyday, R. Huey, W.E. Hart, R.K. Belew, and A.J. Olson (1998).
Automated docking using Lamarckian genetic algorithm and an empirical binding free energy
function, J. Comput. Chem., Vol. 19, pp. 1639-1662.
[15]. Oliver, K., T. Stiitzle, and T.E. Exner (2006) PLANTS: Application of Ant Colony Optimization to
Structure-Based Drug Design, 5th International workshop on Ants (proceedings), vol. 4150, pp. 247-
258.
[16]. Oshiro, C.M., Kuntz, and J.S. Dixon (1995). Flexible ligand docking using a genetic algorithm, J.
Comput. Aid. Mol. Des., Vol. 9, pp. 113-130.
[17]. Vieth, M., J.D. Hirst, B.N. Dominy, H. Daigler, and C.L. Brooks III (1998). Assessing search
strategies for flexible docking, J. Comput. Chem.,Vol. 19, pp. 1623-1631.
[18]. Xiao, Y.L. and D.E. Williams (1994), Proceedings of the 1994 ACM symposium on Applied
computing. pp. 196-200.
[19]. Rainer Storn, Kenneth Price (1997), Differential Evolution A Simple and Efficient Heuristic for
global Optimization over Continuous Spaces,December 1997, Volume 11, Issue 4, pp 341-359
[20]. Thomsen, R.(2003), Flexible ligand docking using differential evolution, Evolutionary Computation,
2003. CEC '03. The 2003 Congress on (Volume:4)
[21]. Thomsen R1, Christensen MH.(2006), MolDock: a new technique for high-accuracy molecular
docking,J Med Chem. 2006 Jun 1;49(11):3315-21.
[22]. Iztok Fister Jr., Xin-She Yang, Iztok Fister, Janez Brest, Duan FisterBbb(2013), A Brief Review of
Nature-Inspired Algorithms for Optimization, Elektrotehni\v{s}ki vestnik, 80(3), 2013
[23]. Thomsen, R.(2007), Protein-Ligand Docking with Evolutionary Algorithms, Computational
Intelligence in Bioiformatics, IEEE press 2007, pp. 169-196
[24]. Shashi, Kusum Deep, V.K Katiyar, C.K Katiyar(2009),A state of art review on application of nature
inspired optimization algorithms in protein-ligand docking, Indian Journal of Biomechanics: Special
Issue (NCBM 7-8 March 2009)
[25]. Jiming Liu, K.C. Tsui,Toward nature-inspired computing, Communications of the ACM, Vol. 49 No.
10, Pages 59-64

AUTHOR

The Author is presently working as a BI developer for British Telecommunications Retail
Team in DSS platform and pursuing Jadavpur University-TCS collaborative Masters of
Engineering in Software Engineering.

879 Vol. 7, Issue 3, pp. 879-886

HEAT TRANSFER ANALYSIS OF COLD STORAGE
Upamanyu Bangale and Samir Deshmukh
Department of Mechanical Engineering, SGBA University, Amravati, India

ABSTRACT
India is largest producer of fruit and vegetable in world scenario but availability of fruit and vegetable per
capita is significantly low because of post-harvest loses which is about 25% to 30% of production so the
requirement of refrigeration and air conditioning has been increased. Building or group of buildings with
thermal insulation and refrigeration system in which perishable food or products can be stored for various
length of time to slow down deterioration with necessary temperature and humidity is called as cold storage. In
cold storage refrigeration system brings down the temperature initially during start up but thermal insulation
maintains the temperature later on continuously. In this view, the simple methodology is presented to calculate
heat transfer by analytical method also attempt has been made to minimize the energy consumption by replacing
150 mm Expanded polystyrene (EPS) by 100 mm Poly Urethane foam (PUF) insulation. The methodology is
validated against actual data obtained from Penguin cold storage situated in Pune, India.
KEYWORDS: Cold storage, Steady state analysis, insulation, EPS, PUF
I. COLD STORAGE DESCRIPTION
Cold storage has floor plan area of 400 m
2
and sidewall length of 14 m. The overall dimensions of
cold storage are 17m x 14m x 12m. The cold storage building is of three floors with each floor having
4 cold chambers of 8m*5m sizes operating at different temperature as per the requirements of
commodities. Ground and First floor called as Defrost zone where -20
0
C temperatures is maintained
while second floor is called as Cold storage where 0
0
C is maintained. The height of zone wall is taken
to be 4 m, leading to a total storage volume of 160 m
3
. Entire cold storage building is considered for
study of simulation where the actual product is stored (for either long or short term storage
periods) and is referred to as the Defrost Zone or Cold Storage Zone. Defrost zone and Cold
Storage zone having area of 8x5 m has with the height floor of 4m with area of 40 m
2
and a volume of
160 m
3
.
The Dock is available immediate entrance of cold storage building which is a separate zone that
serves as the staging area for incoming and outgoing products. The office of a cold storage and ante
room for loading and unloading of the commodities is as shown in fig. 1.1. The condensing unit i.e.
evaporative cooled condenser with two fans and a pump is located outside the zone in still
atmosphere.
880 Vol. 7, Issue 3, pp. 879-886

Fig 1.1 Cold storage floor plan
Under normal operating conditions, the Zone is at a lower temperature than the dock. The dock
temperature is maintained around 10
0
C, while different set-point temperatures have been investigated
for the Zone depending on the optimum storage temperature for the various food products considered
during the course of study.
An important concept to be understood in the operation of cold storage warehouses is the way the
temperature of the products and the zone air is lowered. Since, the refrigeration equipment directly
cools the air in the zone; it is the zone air temperature that reduces first. As the stored product has a
higher thermal mass than the zone air, there is a time lag between when the air temperature is reduced
and when the product temperature is reduced over time as heat is extracted. Alternatively, when the
refrigeration equipment is shut-off, the air temperature shows a greater risk than the product
temperature in the same time interval. Adequate precaution and care still has to be taken in the design
of cold storage to make them energy efficient and to reduce the operational costs as much as possible.
1.1 Cold storage construction
In this simulation we assume that on an average, 60% of the cold storage floor area is covered with
pallets of stored product. The walls of the cold storage zones are modeled as a layer of solid concrete
block followed by three layers of EPS and a layer of Aluminum from outside to inside respectively.
The insulation thickness is different for the zone and the dock. For the zone walls, three different
values of insulation thicknesses were studied for the insulating material of EPS, resulting in different
thermal resistance values (R values) for each wall. Then 150 mm EPS insulation is compared with
100 mm insulation of PUF. While Surface Area of wall is shown in Table 1.2
Table-1.1 Surface Area of wall
NAME AREA(m
2
)
Ceiling 40
Wall between DF and Surrounding 32
Wall between DF and Loading Unloading 20
Wall between DF and Dock 32
Wall between DF and DF 20
Floor 40
The wall of cold chamber as shown in fig.1.2 has a layer of solid concrete block on external side with
20 cm of thickness facing towards the external surroundings, and 15 cm of expanded polystyrene with
1 mm layer of aluminum foil facing into the conditioned space or the cooler.
881 Vol. 7, Issue 3, pp. 879-886

Fig.1.2 Wall of cold chamber
The roof of the dock and the zone is assumed to have the same construction as the zone walls. The
floor construction for the warehouse is common for both the dock and the zone space.
Table 1.2 Applied insulation thickness details
Application Area Thickness of Insulation (mm)
U value
(W/m
2
K)
R value
(m
2
K/W)
Exposed walls 150 0.27 3.70
Intermediate walls and
Inter-floors
50 0.58 1.72
Exposed roofs 150 0.24 4.17
Floors 150 0.29 3.45
The floor has two layers, from outside to inside: 200 mm of concrete, and 150 mm of EPS insulation,
resulting in a floor with a total thickness of 350 mm , and a U-value of 0.686 W/m2-K. The
commonly used insulating materials for cold storage walls, floors and roofs and their details are given
in Table No. 1.3
1.2 Thermal insulation details
Table 1.4 Cold storage insulation characteristics
[1]
Types of Insulation
Material
Density(Kg/m
3
) K value (W/m
0
C)
EPS 16 0.036
PUF 32 0.023
XPS 30-35 0.025
Phenolic foam 50 0.026
Mineral wool 48 0.033
Bonded fiber glass 32 0.033
II. HEAT TRANSFER CALCULATION
2.1 Average temperature calculation
Average Temperature is taken as 31
0
C because on date 12 May 2014
Table 2.1 Average Temperature [2]

So average temperature on dated 12 May 2014 is
Time Temp
23 to 5 27
0
C
5 to 11 24
0
C
11 to 17 37
0
C
17 to 23 36
0
C
Total 124
0
C
882 Vol. 7, Issue 3, pp. 879-886

Total Temp./ 4=
124
4
= 31
0
C
Cold storage capacity 1400 MT, Plant Size 17 m (L) x 14 m (W) x 12m (H),Average Outside
Temperature 31
0
C, Operating Temperature -20
0
C
In Defrost and 0
0
C in Cold Storage
Heat transfer is calculated by using Fouriers law
Q=A (T1-T3) / (L1/k1)
Q=Heat transfer in Watt, A=Area (m2),(T1-T3)= Temperature difference (
0
C or K),
L1=Thickness of insulation (m), K1=Thermal Conductivity (W/mk)
Q=
(13)
1
1
=
32(31(20)
0.15
1.14
= 9302.4W
While similar calculations are carried out by using above formula [3]
2.2 When existing model with EPS Insulation
Table 2.2 Heat transfer through existing Model

883 Vol. 7, Issue 3, pp. 879-886

884 Vol. 7, Issue 3, pp. 879-886

2.3 Recommended model
Table 2.3 Heat transfer through recommended Model

885 Vol. 7, Issue 3, pp. 879-886

Total amount of heat transfer from existing model when EPS insulation of 150mm QEPS = 14928 Watt
Total amount of heat transfer from Recommended model when PUF insulation is of 100 mm
QPUF=12644 Watt
If we use PUF as Insulation then 14928-12644= 2283 Watt of heat transfer is restricted (i.e. 2.283
kW).
If plant runs for 24 hours, units saved will be 2.283 x 24 = 54.792 Units.
If cost of a unit of electricity is Rs 9.00, then saving per day =54.792 x 9 = Rs 493.128. Saving in
month = 493.128 x 30 = Rs 14,794
Saving in a year = Rs 14,794 x 12 = Rs 1,77,526.08
Table 2.4 Cost for EPS and PUF

(Cost values are given in Ref.4)
Pay Back Period = 9, 76,320 / 1, 77,526.08 = 5.49 years
886 Vol. 7, Issue 3, pp. 879-886

Thus to recover the additional cost invested for PUF insulation, the plant has to operate 5.49
years, without any extra profit but with the replacement of EPS by PUF insulation the cold
storage plant can saves Rs 1,77,526.08.
III. CONCLUSION
In existing system of cold storage for external wall, insulation is 150 mm,for wall between two zone
and zone and dock Insulation thickness is 50 mm while in between zone and loading and unloading
thickness of insulation is 150 mm so the total amount of heat transfer is 14927.9W, but for 100 mm
PUF 12644.4W of heat transfer which is 2283.46W lesser than EPS. So saving in RS is 1,77,526.08
and payback period is Extra Investment/Saving=9,76,320/1,77,526.08=5.49years.
Hence it is concluded that the PUF is best suitable insulation material for a Cold storage.
(Cost values are given in Ref.4)
REFERENCES
[1]. Technical Standards and Protocol for cold chain in India,2010 pp-1-61.
[2]. www.yr. no/place/India/maharashtra/pune/long.html
[3]. Refrigeration and Air Conditioning, Third Edition by C.P. Arora, MC Graw Hill Education pp. 1-848.
[4]. Quotation of New Star Engineers

AUTHORS BIOGRAPHY
Upamanyu Bangale, Graduate in Mechanical engineering from SGBAU, Year 2010 and
Pursuing ME CAD/CAM from same University.

Samir Deshmukh, Doctorate from Amravati University, ME Thermal Coordinator and
Senior Lecturer in P.R.M.I.T & R, Amravati
887 Vol. 7, Issue 3, pp. 887-895

LOCALIZED RGB COLOR HISTOGRAM FEATURE
DESCRIPTOR FOR IMAGE RETRIEVAL
K. Prasanthi Jasmine
1
; P. Rajesh Kumar
2

1
Research Scholar,
2
Professor & Head
Department of Electronics and Communication Engineering
Andhra University, Visakhapatnam, Andhra Pradesh, India

ABSTRACT
This paper proposes a new feature descriptor, localized color descriptor for content based image retrieval
(CBIR). The proposed method collects the local histograms from red (R), green (G) and blue (B) color spaces.
These local histograms are collected by dividing the images into subblocks (regions). The performance of the
proposed method is tested by conducting experiments on Corel-1000, natural benchmark database. The
performance of the proposed method is evaluated in terms of precision, recall, average retrieval precision
(ARP) and average retrieval rate (ARR) as compared to the global RGB histograms, global HSV histograms
and other existing features for image retrieval. The performance of the proposed method also tests with different
distance measures. The results after being investigated the proposed method shows a significant improvement as
compared to the other existing methods in terms of precision, recall, ARP and ARR on Corel-1000 database.
KEYWORDS: Color Features; Feature Extraction; Histogram; Image Retrieval
I. INTRODUCTION
A. Motivation
Recent years have seen a rapid increase in the size of digital image collections. Every day, both
military and civilian equipment generates gigabytes of images. A huge amount of information is out
there. However, we cannot access or make use of the information unless it is organized so as to allow
efficient browsing, searching, and retrieval. Image retrieval has been a very active research area since
the 1970s, with the thrust from two major research communities, database management and computer
vision. These two research communities study image retrieval from different angles, one being text-
based and the other visual-based.
The text-based image retrieval can be traced back to the late 1970s. A very popular framework of
image retrieval then was to first annotate the images by text and then use text-based database
management systems (DBMS) to perform image retrieval. Many advances, such as data modelling,
multidimensional indexing, and query evaluation, have been made along this research direction.
However, there exist two major difficulties, especially when the size of image collections is large
(tens or hundreds of thousands). One is the vast amount of labour required in manual image
annotation. The other difficulty, which is more essential, results from the rich content in the images
and the subjectivity of human perception. That is, for the same image content different people may
perceive it differently. The perception subjectivity and annotation impreciseness may cause
unrecoverable mismatches in later retrieval processes. Comprehensive and extensive literature survey
on CBIR is presented in [1][4].
Two major approaches, including spatial and transform domain-based methods can be identified in
CBIR systems. The first approach usually uses pixel (or a group of adjacent pixels) features like color
and shape. Among all these features, color is the most used signature for indexing. Color histogram
[5] and its variations [6] were the first algorithms introduced in the pixel domain. Despite its
888 Vol. 7, Issue 3, pp. 887-895

efficiency, color histogram is unable to carry local spatial information of pixels. Therefore, in such
systems retrieved images may have many inaccuracies, especially in large image databases. For these
reasons, two variations called image partitioning and regional color histogram were proposed to
improve the effectiveness of such systems [7, 8]. Color Correlogram is a different pixel domain
approach which incorporates spatial information with color [9, 10]. Color spatial schemes like color
Correlogram offer more effectiveness in comparison with color histogram methods with little
efficiency reduction. Despite the achieved advantage, they suffer from sensitivity to scale, color and
illumination changes. Edge orientation Auto-Correlogram [11] was an effort to reduce the sensitivity
of color Correlogram method to color and illumination variations. Shape-based and color shape-
based systems using snakes [12], contour images, and other boundary detection methods [11, 13],
were also proposed as pixel domain methods.
B. Related Work
Second feature, texture features (transform domain) currently in use are mainly derived from multi-
scale approach. Liu and Picard [14] have used Wold features for image modeling and retrieval. In
SaFe project, Smith and Chang [15] used discrete wavelet transform (DWT) based features for image
retrieval. Ahmadian et al. used the wavelet transform for texture classification [16]. Do et al. proposed
the wavelet transform (DWT) based texture image retrieval using generalized Gaussian density and
Kullback-Leibler distance (GGD &KLD) [17]. Unser used the wavelet frames for texture calsification
and segmentation [18]. Manjunath et al. [19] proposed the Gabor transform (GT) for image retrieval
on Bordatz texture database. They have used the mean and standard deviation features from four
scales and six directions of Gabor transform. Kokare et al. used the rotated wavelet filters [20], dual
tree complex wavelet filters (DT-CWF), dual tree rotated complex wavelet filters (DT-RCWF) [21],
rotational invariant complex wavelet filters [22] for texture image retrieval. They have calculated the
characteristics of image in different directions using rotated complex wavelet filters. Birgale et al. [23]
and Subrahmanyam et al. [24] combined the color (color histogram) and texture (wavelet transform)
features for CBIR. Jhanwar et al. [25] have proposed the motif co-occurrence matrix (MCM) for
content based image retrieval. The MCM is derived from the motif transformed image which is
calculated by dividing the whole image into non-overlapping 22 pixel patterns. They also proposed
the color MCM which is calculated by applying MCM on individual red (R), green (G), and blue (B)
color planes.
The main contributions of this paper are summarized as follows. (a) This paper proposes a localized
RGB histogram based feature descriptor. These features are collected by dividing the image into
subblocks. (b) The performance of the proposed method is analyzed with different distance measures
on Core-1000 database in terms of precision, recall, average precision (ARP) and average retrieval
rate (ARR).
C. Organization of Manuscript
The organization of the paper as follows: In section I, a brief review of image retrieval and related
work is given. Section II, presents a concise review of Color spaces. Section III, presents the feature
extraction and similarity measure. Experimental results and discussions are given in section IV. Based
on above work conclusions are derived in section V.
II. COLOR SPACES
A. RGB Color Space
The predominant color representation used in Image retrieval is the RGB [13] color space
representation. In this representation, the values of the red, green, and blue color channels are stored
separately. They can range from 0 to 255, with 0 being not present, and 255 being maximal. A fourth
channel, alpha, also provides a measure of transparency for the pixel. The distance between two pixels
is measured using Eq. (1).
3
(255 Re ).(255 ).(255 )
255
d Green Blue A A A
(1)
889 Vol. 7, Issue 3, pp. 887-895

Where Red is the red channel difference, Green is the green channel difference, and Blue is the
blue channel difference between the two pixels being compared. This distance metric has an output in
the range of [0-1]. When used to compare pixels, a cut-off, usually between 0.6 and 0.8, is used to
decide whether the two pixels are similar to each other. Because the distance measure is based on the
absolute difference between pixel color values, the size of the neighbourhood of similarity for a given
cut-off value is the same regardless of the location of the pixel value being compared to in the RGB
color space.
Although this color space description does not always match the intuition for color similarity of
people, it does provide an easy method for pixel value comparison. In most cases it performs
agreeably well, as it can differentiate gross color differences. It is also computationally cheap as
compared to any other color space representations; because the Red-Green-Blue color channel
differentiation is used in most images formats and therefore is readily available.
B. HSV Color Space
The alternative to the RGB color space is the Hue-Saturation-Value (HSV) color space. Instead of
looking at each value of red, green and blue individually, a metric is defined which creates a different
continuum of colors, in terms of the different hues each color possesses. The hues are then
differentiated based on the amount of saturation they have, that is, in terms of how little white they
have mixed in, as well as on the magnitude, or value, of the hue. In the value range, large numbers
denote bright colorations, and low numbers denote dim colorations.
This space is usually depicted as a cylinder, with hue denoting the location along the circumference of
the cylinder, value denoting depth within the cylinder along the central axis, and saturation denoting
the distance from the central axis to the outer shell of the cylinder. This is depicted in Fig. 1(a) on the
left. This description, however, fails to agree with the common observation that very dark colors are
hard to distinguish, whereas very bright colors are easily distinguished into multiple hues. A system
more accurate in representing this is shown in Fig. 1(b) on the right, which is the model used for the
HSV color space representation in the Image retrieval system.
The computation of hue, saturation, and value can be performed as detailed below. In the text below,
as in the figures, h stands for hue, v stands for value, and s stands for saturation. The RGB colors are
similarly represented, with r for red, g for green, and b for blue. The remainder of the variables are
temporary computation results. All of these computations have to be computed on a pixel-by-pixel
basis.

(a) (b)
Fig. 1: Two HSV color space representations.
The cylindrical representation is a good approximation, but the conical representation results in better
performance.
max( , , ). v r g b =
(2)

(3)

Here, max denotes a function returning the maximum value among its arguments, and where min
denotes a function returning the minimum value among its arguments.
min( , , ) v r g b
s
v
=
890 Vol. 7, Issue 3, pp. 887-895

( ) ( )
( ) ( )
( ) ( )
max , , min , , 5 int
max( , , ) min( , , ) 3- int
max
if r r g b and g r g b then bpo
if r r g b and b r g b then gpo
if g r g b and b r g b then rpo
h
if g r g b and r r g b then bpo
if b
= = +
= = +
= = +
=
= =
= ( ) ( )
( ) ( )
, , min , 3 int
r g b and r r g b then gpo
if b r g b and g r g b then rpo
= +
= =
(4)
Where
int .
min( , , )
v r
rpo
v r g b
(5)
int .
min( , , )
v g
gpo
v r g b
(6)
int .
min( , , )
v b
bpo
v r g b
(7)
The value computation, as defined in Eq. (2), maps to the range of values which r, g, and b can take.
III. FEATURE EXTRACTION
Each image from the database is analyzed using localized RGB histograms. Firstly, the RGB image is
divided into non-overlapped subblocks. Then the individual R, G and B histograms are calculated for
each subblock. Finally, the feature vector is constructed by concatenating all the histograms. The
algorithm of the proposed system is given bellow.

Algorithm:
Input: Query Image/Database Image; Output: Feature Vector
1. Load the color image.
2. Separate the R, G and B color spaces.
3. Quantize the each color space.
4. Divide the each color space into subblocks.
5. Calculate the histograms for each subblock.
6. Form a feature vector by concatenating the histograms.

A. Similarity Distance Measure
In the presented work four types of similarity distance metric ares used as shown follows.
Manhattan or L1 or city-block Distance
This distance function is computationally less expensive than Euclidean distance because only the
absolute differences in each feature are considered. This distance is sometimes called the city block
distance or L1 distance and defined as
( , ) ( ) ( )
i j
i
D Q T f Q f T =
(8)
Euclidean or L2 Distance
The Euclidean distance is defined as:
( )
1 2
2
( , ) ( ) ( )
i j
i
D Q T f Q f T =
(9)
The most expensive operation is the computation of square root.
D1 Distance
, ,
1 , ,
( , )
1
Lg
T i Q i
i T i Q i
f f
D Q T
f f
=
=
+ +
(10)
Canberra Distance
891 Vol. 7, Issue 3, pp. 887-895

, ,
1
, ,
( , )
Lg
T i Q i
i
T i Q i
f f
D Q T
f f
=
=
+
(11)
where Q is query image, Lg is feature vector length, T is image in database;
, I i
f is
th
i feature of image
I in the database,
, Q i
f is
th
i feature of query image Q.
IV. EXPERIMENTAL RESULTS AND DISCUSSIONS
In this paper, retrieval tests are conducted on Corel-1K database and results are presented as follows.
Corel database [26] consists of large number of images of various contents ranging from animals to
outdoor sports to natural images. These images have been pre-classified into different categories each
of size 100 by domain professionals. Some researchers think that Corel database meets all the
requirements to evaluate an image retrieval system, due its large size and heterogeneous content. We
have collected 1000 images to form database Corel1K. These images are collected from 10 different
domains namely Africans, beaches, buildings, buses, dinosaurs, elephants, flowers, horses, mountains
and food. Each category has NG (100) images with resolution of either 256384 or 384256. Fig. 5
shows the sample images of Corel1K database (one image from each category).
In all experiments, each image in the database is used as the query image. For each query, the system
collects n database images X=(x1, x2, , xn) with the shortest image matching distance computed
using (12). If the retrieved image xi=1, 2, ., n belongs to same category as that of the query image
then we say the system has appropriately identified the expected image else the system fails to find
the expected image.
The performance of the proposed method is measured in terms of precision, recall, average retrieval
precision (ARP) and average retrieval rate using Eq. (12) to Eq. (17).
.
( ) 100
.
No of Relevant Images Retrieved
Precision P
Total No of Images Retrieved
=

(12)
1
1 1
1
( )
N
i
Group Precision GP P
N
=
=
(13)
1
1 1
1
( )
j
AverageRetrieval Precision ARP GP
I
=
=
I
(14)
( )
Numberof relevant images retrieved
Recall R
Total Numberof relevant images
=

(15)
1
1 1
1
( )
N
i
Group Recall GR R
N
=
=
(16)
1
1 1
1
( )
j
AverageRetrieval Rate ARR GR
I
=
=
I

(17)
where
1
N is number of relevant images and
1
I is number of groups.
892 Vol. 7, Issue 3, pp. 887-895

Fig. 2: Comparison of the proposed method with other existing methods in terms of: (a) ARP and (b) ARR on
Corel-1000 database.
Table I & Fig. 2 (a) and Table II & Fig. 2 (b) summarize the retrieval results of various methods in
terms of average retrieval precision and average retrieval rate respectively. Fig. 3 summarize the
performance of proposed method with different distance measures in terms of average retrieval rate.
From the Tables I to II and Fig. 6 the following points can be observed:
1. The average retrieval precision of proposed method (64.1% to 44.29%) is more as compared to
Jhanwar et al. (58.7% to 39.0%), RGB Hist. (62.14% to 41.76%) and HSV Hist. (63.7% to
42.6%).
2. The performance of the proposed method with d1 distance is more as compared to Canberra,
Euclidean and Manhattan distances in terms of average retrieval precision.
From Tables I to II, Fig. 2, Fig. 3 and above observations, it is clear that the proposed method is
outperforming the other existing techniques in terms of ARR and ARP. Fig. 4 illustrates the query
results of the proposed method on Corel-1000 database (top left image is the query image).
893 Vol. 7, Issue 3, pp. 887-895

Fig. 3: Comparison of proposed method with various distance measures on Corel-1000 image database.
Table i: performance of various methods in terms of group precision on corel-1000 database.
Precision (%)
Jhanwar et al. RGB Hist. HSV Hist. PM
53.15 84.1 66.7 75.0
43.85 52.7 32.2 49.6
48.7 52.5 56.8 44.8
82.8 52 83.4 65.6
95 100 99.8 100.0
34.85 69 50.65 70.1
88.35 82.6 66.2 87.4
59.35 91.3 87.25 94.6
30.8 33.1 24.75 48.3
50.4 77 69.95 70.8
58.72 62.14 63.77 64.1
Table II: Performance of various methods in terms of Group Recall on Corel-1000 database.
Recall (%)
Jhanwar et al. RGB Hist. HSV Hist. PM
32.21 48.33 42.94 44.53
29.04 26.38 19.4 25.69
27.7 24.38 34.99 23.38
48.66 34.45 63.77 43.57
81.44 98.37 90.35 98.35
21.42 36.48 29.01 33.92
63.53 49.46 32.86 62.72
35.84 45.29 50.82 49.45
21.75 17.5 14.55 23.24
29.02 37.01 47.82 38.09
39.06 41.76 42.65 44.29

894 Vol. 7, Issue 3, pp. 887-895

Fig. 4: Two query results of the proposed method on Corel-1000 image database.
V. CONCLUSIONS AND FUTURE WORK
A new image indexing and retrieval algorithm using localized RGB histogram is proposed in this
paper. The performance of the proposed method is tested by conducting experimentation on Corel-
1000 natural image database. The results after being investigated show a significant improvement in
terms of precision, recall, average retrieval rate and average retrieval precision as compared to other
existing techniques on Corel-1000 image database.
In this paper, we propose a new color feature for image retrieval. Further, the performance of the
CBIR system can be improved by integrating the proposed color feature with the texture features like,
local binary patterns (LBP), local ternary patterns (LTP), local derivative patterns (LDP), local tetra
patterns (LTrP), etc,.
REFERENCE
[1] Y. Rui and T. S. Huang, Image retrieval: Current techniques, promising directions and open issues, J.. Vis.
Commun. Image Represent., 10 (1999) 3962.
[2] W.M. Smeulders, M. Worring, S. Santini, A. Gupta, and R. Jain, Content-based image retrieval at the end
of the early years, IEEE Trans. Pattern Anal. Mach. Intell., 22 (12) 13491380, 2000.
[3] M. Kokare, B. N. Chatterji, P. K. Biswas, A survey on current content based image retrieval methods,
IETE J. Res., 48 (3&4) 261271, 2002.
[4] Ying Liu, Dengsheng Zhang, Guojun Lu, Wei-Ying Ma, Asurvey of content-based image retrieval with
high-level semantics, Elsevier J. Pattern Recognition, 40, 262-282, 2007.
[5] M.J. Swain, D.H. Ballard, Color indexing, Int. J. Comput. Vis. 7, pp.1132, 1991.
[6] M. Flickner, H. Sawhney,W. Niblack, J. Ashley, Q. Hunag, B. Dom, M. Gorkani, J. Hafner, D. Lee, D.
Petkovic, D. Steele, P. Yanker, Query by image and video content: the QBIC system, IEEE Comput. 28
(9), pp.2332, 1995.
[7] Carson, M. Thomas, S. Blongie, J.M. Hellerstine, J. Malik, Blobworld, A system for region-based image
indexing and retrieval, Proceedings of the SPIE, Visual Information Systems, Netherland, pp. 509516,
1999.
[8] M. Stricker, A. Dimai, Spectral covariance and fuzzy regions for image indexing, Mach. Vis. Appl. 10,
pp. 6673, 1997.
895 Vol. 7, Issue 3, pp. 887-895

[9] J. Huang, S.R. Kumar, M. Mitra, W.J. Zhu, R. Zabih, Image indexing using color Correlogram,
Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, San Juan, pp. 762768,
1997.
[10] J. Huang, S.R. Kumar, M. Mitra, W.J. Zhu, R. Zabih, Spatial color indexing and applications, Int. J.
Comput. Vis. 35 (3), pp. 245268, 1999.
[11] F. Mahmoudi, J. Shanbehzadeh, A.M. Eftekhari-Moghadam, H. Soltanian-Zadeh, Image retrieval based
on shape similarity by edge orientation auto-correlogram, Pattern Recognition 36 (2), pp. 17251736,
2003.
[12] H.H.S. Ip, D. Shen, An affine-invariant active contour model (AI-snake) for model-based segmentation,
Image Vis. Comput. 16 (2), pp. 135146, 1998.
[13] Sajjanhar, G. Lu, A grid based shape indexing and retrieval method, Special issue of Austral. Comput. J.
Multimedia Storage Archiving Systems 29(4), pp.131140, 1997.
[14] Liu, F., Picard, R.W., 1996. Periodicity, directionality, and randomness: Wold features for image modeling
and retrieval. IEEE Trans. Pattern Anal. Machine Intell. 18, 722733.
[15] J. R. Smith and S. F. Chang, Automated binary texture feature sets for image retrieval, Proc. IEEE Int.
Conf. Acoustics, Speech and Signal Processing, Columbia Univ., New York, (1996) 22392242.
[16] Ahmadian, A. Mostafa, An Efficient Texture Classification Algorithm using Gabor wavelet, 25th Annual
international conf. of the IEEE EMBS, Cancun, Mexico, (2003) 930-933.M.
[17] M. N. Do and M. Vetterli, The contourlet transform: An efficient directional multi-resolution image
representation, IEEE Trans. Image Process., vol. 14, no. 12, pp. 20912106, 2005.
[18] M. Unser, Texture classification by wavelet packet signatures, IEEE Trans. Pattern Anal. Mach. Intell., 15
(11): 1186-1191, 1993.
[19] S. Manjunath and W. Y. Ma, Texture Features for Browsing and Retrieval of Image Data, IEEE Trans.
Pattern Anal. Mach. Intell., 18 (8): 837-842, 1996.
[20] M. Kokare, P. K. Biswas, B. N. Chatterji, Texture image retrieval using rotated Wavelet Filters, Elsevier J.
Pattern recognition letters, 28:. 1240-1249, 2007.
[21] M. Kokare, P. K. Biswas, B. N. Chatterji, Texture Image Retrieval Using New Rotated Complex Wavelet
Filters, IEEE Trans. Systems, Man, and Cybernetics, 33 (6): 1168-1178, 2005.
[22] M. Kokare, P. K. Biswas, B. N. Chatterji, Rotation-Invariant Texture Image Retrieval Using Rotated
Complex Wavelet Filters, IEEE Trans. Systems, Man, and Cybernetics, 36 (6): 1273-1282, 2006.
[23] L. Birgale, M. Kokare, D. Doye, Color and Texture Features for Content Based Image Retrieval,
International Conf. Computer Grafics, Image and Visualisation, Washington, DC, USA, (2006) 146 149.
[24] Subrahmanyam, A. B. Gonde and R. P. Maheshwari, Color and Texture Features for Image Indexing and
Retrieval, IEEE Int. Advance Computing Conf., Patial, India, (2009) 1411-1416.
[25] N. Jhanwara, S. Chaudhuri, G. Seetharamanc, and B. Zavidovique, Content based image retrieval using
motif co-occurrence matrix, Image and Vision Computing 22, (2004) 12111220.
[26] Corel1K image database. [Online]. Available: http://wang.ist.psu.edu/docs/rela-ted.shtml.
AUTHORS
K. prasanthi Jasmine, received her B.Tech in Electronics & Communication Engineering
and M.Tech in Digital Systems from Regional Engineering College(Now NIT), Warangal,
Andhra Pradesh, and Osmania University College of Engineering, Osmania University,
Andhra Pradesh, India in the years 2000 and 2003 respectively. Currently, she is pursuing
Ph.D from Andhra University, A.P, India. Her major fields of interest is Image Retrieval,
Digital image Processing and Pattern Recognition.

P. Rajesh Kumar received his M.Tech, Ph.D degrees from Andhra University,
Vishakhapatnam, India. He is currently working as professor, Department of Electronics &
Communication Engineering, Andhra University College of Engineering, Visakhapatnam,
Andhra Pradesh. He is also Assistant Principal of Andhra University college of
Engineering, Visakhapatnam, Andhra Pradesh. He has twenty years experience of teaching
undergraduate and postgraduate students and guided a number of post graduate thesis. He
has published twenty research papers in national and international journals and conferences.
Presently he is guiding twelve Research scholars. His research interests are digital signal and image processing,
computational intelligence, human computer interaction, radar signal processing.

896 Vol. 7, Issue 3, pp. 896-904

WITRICITY FOR WIRELESS SENSOR NODES
M. Karthika
1
and C. Venkatesh
2

1
PG Scholar, Erode Builder Educational Trusts Group of Institutions,
Kangayam, Tirupur, Tamilnadu, India
2
Professor- ECE, Erode Builder Educational Trusts Group of Institutions,
Kangayam, Tirupur, Tamilnadu, India

ABSTRACT
A major challenge in the wireless sensor networks is to develop a way of supplying power to sensor nodes in an
efficient and reliable manner. This paper explored a potential solution to this challenge by transferring energy
using Radio Frequency (RF) signals. The RF signal was generated using oscillator, amplified using power
amplifier, and transmitted through antenna. The wirelessly transmitted RF energy can be captured by the
receiving antennas, transformed into DC power by a rectifying circuit, and stored in a battery to provide the
required energy to the sensor node. Such a system has the advantage of eliminating battery replacement of the
sensor nodes.
KEYWORDS: Wireless Power Transmission, Wireless Electricity, Wireless Sensor Networks
I. INTRODUCTION
Energy constraints are widely regarded as a fundamental limitation of wireless sensor networks. For
sensor networks, a limited lifetime due to battery constraint poses a performance bottleneck and
barrier for large scale deployment.
Recently, wireless power transfer has emerged as a promising technology to address energy and
lifetime bottlenecks in a sensor network. As wireless sensor devices become pervasive, charging
batteries for these devices has become a critical problem. Existing battery charging technologies are
dominated by wired technology, which requires a wired power plug to be connected to an electrical
wall outlet. Wireless power transfer (WPT) achieves the same goal but without the hassle of wires.
A small, low-cost, wireless sensor node is important for sensing the environment. However, the need
for frequently replacing its battery has always been a problem, which has limited its use of WSNs.
WSNs based on energy harvesting are partly in practical use. The use of solar energy in energy
harvesting WSNs has increased for practical applications; this is because of the fact that solar panels
are easily available and they have a higher energy density as compared to other energy harvesting
techniques. This high energy density allows the development of smaller sensor nodes. However, solar
power strongly depends on sunlight and can therefore hardly harvest energy during the nighttime ant
the amount of harvested energy depends on the weather.
RF based Wireless Power Transfer is one of the key techniques used to solve this problem. This
system focuses on using an ambient RF field as an energy source to power wireless sensor nodes. The
use of this unutilized energy as a power source will not only reduce the battery replacement cost, but
also enable a long period operation in WSNs.
1.1 Related Works
In paper [1], an Improved Energy Efficient Ant Based routing Algorithm energy management
technique, which improves the lifetime of sensor network, was proposed. Using the harvested energy,
the time of charging the battery powering the sensor nodes drastically reduced, while requiring time
intervals of 91.9 hrs to recharge the battery. Paper [2] studied a power and data transmission system
for strain gauge sensors. Two kinds of receiver antennas (dipole and patch) are evaluated in terms of
897 Vol. 7, Issue 3, pp. 896-904

return loss and peak received power. Results show that dipole antenna have a better impedance
matching with the conversion circuit than that of the patch antenna. The rectenna optimization,
channel modeling and design of a transmit antenna are discussed in paper [3]. Paper [5] reviewed the
history of wireless power transfer and described its recent developments. It showed how such
technologies can be applied to sensor networks and addresses its energy constraints. Paper [7]
presented the concept of transmitting power without using wires i.e., transmitting power as
microwaves from one place to another is in order to reduce the cost, transmission and distribution
losses. Paper [8] aimed at designing a prototype of wireless power transmission system. Detailed
analysis of all the system components has been discussed.
This research paper is organized as follows. A brief description of existing and proposed systems is
presented in Section 2. Section 3 discusses the transmitter and receiver circuits and the output of the
each circuit. The brief discussion of result is described in Section 4. Finally, some concluding remarks
have been highlighted in Section 5.
II. SYSTEM OVERVIEW
2.1 Existing System
Comparing with sensor node or battery replacement approaches, the wireless charging technology
allows a mobile charger to transfer energy to sensor nodes wirelessly without requiring accurate
localization of sensor nodes or strict alignment between the charger and nodes [10].
The system consists of

Figure 1. Energy Transfer through Moving vehicle charger
A moving vehicle charger (MVC): a mobile robot carrying a wireless power charger, a network of
sensor nodes equipped with wireless power receivers, and an energy station that monitors the energy
status of the network and directs the MVC to charge sensor nodes.
A network of sensor nodes equipped with wireless power receivers: Sensor nodes perform application
tasks such as environment monitoring, generate sensory data, and periodically report the data to the
sink. In addition, that also monitor the voltage readings of their own batteries, estimate energy
consumption rates, based on which derive their own lifetime, and then report the information to the
sink periodically.
An energy station: it is responsible for monitoring the energy status of sensor nodes, deciding the
power charging sequences to be executed by the mobile charger (as shown in the Figure 1).
Problem in Existing System:
The existing system has the overhead of controlling and maintaining the MVC. The MVC takes more
than 30% of energy for travelling between the sensor nodes and charging. The distance between MVC
and sensor node is too small (in 5-40cm).
2.2 Proposed System
898 Vol. 7, Issue 3, pp. 896-904

To overcome the energy constraint problem in WSN, WiTricity (Wireless Electricity) is proposed.
Energy is transmitted from energy station to sensor node without wire connections. WiTricity is
directly transmitted from energy station to sensor nodes.WiTricity is effectively achieved by RF based
Wireless Power Transfer System. Wireless Power Transmission system using Radio frequency will be
designed for sensor nodes which require low power to operate. RF-based wireless power transmission
focus on passively powered sensor networks by improving the conversion efficiency. Attempting to
maximize the output power by designing efficient antennas. Antennas can be made more directional,
allowing longer distance power beaming.
2.2.1 Wireless Power Transfer System

Figure 2. Transmitter Block Diagram
The Wireless power transfer system consists of two parts, transmitter side and receiver side. The
block diagram of a typical transmitter unit of WPT system is shown in Figure 2. There is always a
large amount of signal power loss in the free space while the RF signal propagates through it. To
compensate this loss at the receiver side, the transmitter of the wireless power transfer system should
be capable of transmitting a high power. For this reason the transmitting antenna should have high
performance [6].
The receiver section contains the receiving antenna, rectifier, DC-DC converter and battery as shown
in Figure 3. The receivers main function is to receive the RF signal from the transmitter, convert it to
DC signal which is used to charge the connected devices battery. A simple battery charging theory is
to run current through the battery, and apply a voltage difference between the terminals of the battery
to reverse the chemical process. In this paper, the current is obtained from the radio wave signal
coming from the antenna [8].

Figure 3. Receiver Block Diagram
III. SYSTEM IMPLEMENTATION
System implementation is based on designing the transmitter and the receiver. The transmitter part
consists of oscillator, power amplifier and antenna. The receiver part consists of receiving antenna,
rectifier, DC-DC converter and battery.
3.1 Oscillator
899 Vol. 7, Issue 3, pp. 896-904

Figure 4. Oscillator Circuit Diagram
The oscillator circuit is based on the Hartley oscillato. The Hartley Oscillator is a particularly useful
circuit for producing good quality sine wave signals in the RF range. The Hartley design can be
recognized by its use of a tapped inductor. The frequency of oscillation can be calculated in the same
way as any parallel resonant circuit, using:

Where L = L1 + L2.
12V power supply is given to oscillator circuit. 82pF capacitor and the 120uH inductor which have 25
turns with 6mm diameter are acting as a tank circuit. Tank circuit is the responsible for the oscillator
frequency. The center point of the inductor is given to the amplifier circuit. So that, with two
inductors and one capacitor, the oscillator circuit is working with the principle of Hartley oscillator.
The oscillator circuit frequency from this circuit is design for 56MHz with 2V amplitude. 1K resister
and 0.5nF capacitor is used to give the basing voltage to the transistor.
3.2 Pre-Amplifier
Output from the oscillator circuit that is, 56MHz with 2V amplitude is given as an input signal to the
power amplifier circuit. 12V power supply is given to the power amplifier circuit. BFW16A is a RF
transistor that used to operate on the center frequency of 100MHz. This amplifier circuit is considered
as a voltage divider amplifier because of 47K and 5.6K resister. The gain of 2 is achieved from this
amplifier circuit. BFW16A is a multi emitter silicon planner epitaxial NPN transistor with extremely
inter modulation properties and high power gain. For this reason, BFW16A was chosen.

Figure 5. Power Amplifier Circuit Diagram
900 Vol. 7, Issue 3, pp. 896-904

Oscillator, Pre-Amplifier Output
In Figure 6, Channel 1 shows the output of the oscillator and the scale for the channel 1 is 5V. The
channel 2 shows the output of the amplifier and the scale for the channel 2 is 5V. The oscillator output
was measured as 5V and the amplifier output was measured as 13V.

Figure 6. Oscillator and Amplifier Output
3.3 Power Amplifier
A push pull amplifier is an amplifier which has an output stage that can drive a current in either
direction through the load. Advantages of push pull amplifier are low distortion, absence of magnetic
saturation in the coupling transformer core, and cancellation of power supply ripples which results in
the absence of hum. Push-Pull power amplifier is shown in Figure 7. MJE15032 and MJE15033 are
8.0 amperes, complementary silicon power transistors of 250 volts, 50 watts with high DC current
gain, high current gain bandwidth product. For this reason, MJE15032 and MJE15033 were chosen.
Output from the pre-amplifier is given to the power amplifier with frequency of 62 MHz and voltage
of 6V.

Figure 7. Push-Pull Power Amplifier Circuit Diagram
Power Amplifier Output
901 Vol. 7, Issue 3, pp. 896-904

The push-pull power amplifier circuit was shown in Figure 8. 10V was given to this circuit as the
input power supply. The output current and the voltage were measured as 300mA and 20V
respectively. The total power achieved from the circuit was 6W.

Figure 8. Power Amplifier Output
3.4 Antenna
The telescopic (whip) antenna is used as transmitting and receiving antenna as shown in Fig.9. The
whip antenna can be considered half of a dipole antenna, and like a vertical dipole has an
omnidirectional radiation pattern, radiating equal radio power in all azimuthal directions
(perpendicular to the antenna's axis), with the radiated power falling off with elevation angle to zero
on the antenna's axis.

Figure 9. Telescopic Antenna

3.5 Rectifier
Bridge rectifier is used to convert the RF energy to the dc voltage as shown in Fig.8. Its elements are
usually arranged in a multi element phased array with a mesh pattern reflector element to make it
directional. A simple rectifier can be constructed from a switching diode placed between antenna
dipoles. The diode rectifies the current induced in the antenna by the RF signal. Switching diodes are
used because they have the lowest voltage drop and highest speed and therefore waste the least
amount of power due to conduction and switching. Rectifier is highly efficient at converting RF
energy to electricity.

902 Vol. 7, Issue 3, pp. 896-904

Figure 10. Rectifier Circuit Diagram
The telescopic antenna is used to receive the RF signal. This AC signal is given to the bridge rectifier
circuit. The bridge rectifier has two ceramic capacitors of value 0.1pF, two electrolytic capacitors of
value 50uF and four PN diodes of 1N4148. This has converted the signal from the antenna to the DC
voltage.
3.6 DC-DC Converter
The output voltage of the rectifying circuit is too low for charging a battery. A DC-to-DC voltage
converter will adapt the rectified voltage to a stable output voltage high enough for charging a
rechargeable battery [3].

Figure 11. Dc to DC Converter Circuit Diagram
DC to DC converter is used to boost up the DC signal to the fixed DC voltage. 1V DC voltage is
given to the DC to DC converter circuit and the output DC voltage is measured as 9V because of the
DC to DC module LT1073. The LT1073 is gated oscillator switcher. This type architecture has very
low supply current because the switch is cycled only when the feedback pin voltage drops below the
reference voltage.
DC to DC Converter Output
903 Vol. 7, Issue 3, pp. 896-904

Figure 12. DC to DC Converter Output
Figure 12 shows the simulation output of the DC to DC converter module. 1V input was given to the
DC to DC converter and the output was measured as 9V. In practical experiment, 8V is measured
from the DC to DC converter circuit.
IV. RESULT
In this paper, transmitter produced the radio signal with the voltage of 20V and current 300mA. This
power was transmitter through the telescopic antenna. In receiver side, another antenna was place at
the distance of 2m from the transmitter to receive the radio signal and fed to the rectifier circuit. The
output of the rectifier was measured as 1.15V and that was given to the DC to DC converter. The
output of the DC to DC converter was measured as 8V.
V. CONCLUSION AND FUTURE WORK
WiTricity for wireless sensor nodes project propose a wireless charging system for sensor networks.
This project presents the design and implementation of oscillator, pre-amplifier, push-pull power
amplifier, rectifier and DC to DC convert. The transmitter produced the radio signal of 6W and
transmitted through the distance of 2m. The output of the rectifier circuit was measured as 1.15V.
The design of the antenna at the transmitter and the receiver plays an important role in transferring the
voltage level at 1V to the DC to DC converter. Also the design of either a Tor network at both the
transmitter and the receiver end plays a major role in power transfer. Integration of all the stages leads
to impedance mismatch which can be overcome using EDA tools like Advance Wave Research,
HFSS etc. Very low power DC to DC convertor can be designed to generate a reasonable propagation
of voltage and current.
REFERENCES
[1]. A.M. Zungeru, L.-M. Ang, SRS. Prabaharan, K.P. Seng A.M. Zungeru, Radio Frequency Energy
Harvesting and Management for Wireless Sensor Networks, In: Energy Scavenging and Optimization
Techniques for Mobile Devices. V. Hrishikesh, and G -M. Mountean (Eds.) USA: Chapter 13, pp. 341-
367, CRC Press, Taylor and Francis Group, 2011. (ISBN-13: 978-1439859896).
[2]. Guocheng Liu1, Nezih Mrad, George Xiao, Zhenzhong Li, and Dayan Ban, RF-based Power
Transmission for Wireless Sensors Nodes, SMART MATERIALS, STRUCTURES & NDT in
AEROSPACE Conference, NDT in Canada 2011, 2 - 4 November 2011, Montreal, Quebec, Canada.
[3]. Hubregt J. Visser, Indoor Wireless RF Energy Transfer for Powering Wireless Sensors,
Radioengineering, Dec2012, Vol. 21 Issue 4, p963-973. 11p.
[4]. Ke Wu, Debabani Choudhury, Hiroshi Matsumoto, Wireless Power Transmission, Technology, and
Applications, Proceedings of the IEEE, Vol. 101, No. 6, June 2013.
904 Vol. 7, Issue 3, pp. 896-904

[5]. Liguang Xie, Yi Shi, Y. Thomas Hou, and Wenjing Lou, Wireless power transfer and applications to
sensor networks, to appear in IEEE Wireless Communications Magazine (Accepted Dec. 2012, impact
factor: 2.575).
[6]. Md M. Biswas, Member, IACSIT, Umama Zobayer, Md J. Hossain, Md Ashiquzzaman, and Md Saleh,
Design a Prototype of Wireless Power Transmission System Using RF/Microwave and Performance
Analysis of Implementation, IACSIT International Journal of Engineering and Technology, Vol. 4, No.
1, February 2012.
[7]. M.Venkateswara Reddy, K.Sai Hemanth, CH.Venkat Mohan, Microwave Power Transmission A Next
Generation Power Transmission System IOSR Journal of Electrical and Electronics Engineering (IOSR-
JEEE), e-ISSN: 2278-1676 Volume 4, Issue 5 (Jan. - Feb. 2013), PP 24-28.
[8]. Olakanmi O. Oladayo ,Departement of Electrical and Electronic Engineering, University of Ibadan,
Ibadan Nigeria, A Prototype System for Transmitting Power through Radio Frequency Signal for
Powering Handheld Devices, International Journal of Electrical and Computer Engineering (IJECE)
Vol.2, No.4, August 2012.
[9]. P. Powledge, J.R. Smith, A. Sample, A. Mamishev, S. Roy, "A wirelessly powered platform for sensing
and computation, Proceedings of Ubicomp 2006: 8
th
International Conference on Ubiquitous
Computing. Orange Country, CA, USA, September 17-21 2006, pp. 495-506.
[10]. Yang Peng, Zi Li, Wensheng Zhang, Daji Qiao, "Prolonging Sensor Network Lifetime Through Wireless
Charging", In Proc. IEEE RTSS'10, San Diego, CA, November 30 - December 3, 2010.
BIBLIOGRAPHY OF AUTHOR
M KARTHIKA received her B.Tech in Information Technology from Bannari Amman
Institute of Technology, Sathyamanagalam in 2010. She started his career as a Test
Engineer in Wipro Technologies, Chennai (2010 to 2012). She is presently a student of M.E
Computer and Communication Engineering in Erode Builder Educational Trusts Group of
Institutions, Kangayam.

C. VENKATESH, graduated in ECE from Kongu Engineering College in the year 1988,
obtained his master degree in Applied Electronics from Coimbatore Institute of
Technology, Coimbatore in the year 1990. He was awarded PhD in ECE from Jawaharlal
Nehru Technological University, Hyderabad in 2007. He has a credit of two decade of
experience which includes around 3 years in industry. He has 18 years of teaching
experience. He is supervising 12 Ph.D., research scholars. He is a Member of IEEE, CSI,
ISTE, FIE and Fellow IETE. He is presently a Professor, Dean- Faculty of Engineering in
Erode Builder Educational Trusts Group of Institutions.
905 Vol. 7, Issue 3, pp. 905-910

STUDY OF SWELLING BEHAVIOUR OF BLACK COTTON SOIL
IMPROVED WITH SAND COLUMN
Aparna
1
, P.K. Jain
2
, and Rakesh Kumar
3

1
M. Tech. Student, Department of Civil Engineering M.A.N.I.T. Bhopal, (M.P) India.
2
Professor, Department of Civil Engineering, M.A.N.I.T. Bhopal, (M.P) India.
3
Assistant Professor, Department of Civil Engineering, M.A.N.I.T. Bhopal, (M.P) India.

ABSTRACT
This paper presents the result of an experimental study conducted for evaluating the effect of size of the sand
column on swelling of expansive soil. The sand columns of diameters25mm, 37.5mm and 50mm were made in
black cotton soil test beds in a cylindrical mould of diameter 100mm. The test beds were prepared at different
water contents (14, 18, 22, 26,30,36,40 and 44% by weight of dry soil) keeping the dry density of the soil as
constant. The soil with sand column was submerged and the swelling of the composite material was observed.
The test results show that the presence of sand column in the expansive black cotton soil reduces the swelling.
The reduction in swelling depends on the size of the sand column and the initial moisture content in the soil. A
column of diameter 50mm reduces swelling more than the smaller ones. For 14% initial moisture content in the
black cotton soil, the stone columns of diameters 25mm, 37.5mm and 50mm have shown reduction in swelling
by 11.5%, 23% and 42% respectively in comparison to that exhibited by the raw soil. The soil with high initial
moisture content shows less swelling than those with low moisture content. Thus, by manipulating the initial
moisture content and the diameter of the sand column, the expansive soil reinforced with sand columns can be
made volumetrically stable.
KEYWORDS: Sand column, swelling, expansive soil, composite ground, ground improvement.
I. INTRODUCTION
Expansive soils are encountered in arid and semi-arid regions of the world, where annual evaporation
exceeds annual precipitation. In India, expansive soils cover about 20% of the total land area (Ranjan
and Rao 2005, Shelke and Murthy 2010). These soils increase in volume on absorbing water during
rainy seasons and decrease in volume when the water evaporates from them (Chen, 1988). The
volume increase (swell) if resisted by any structure resting on it; then vertical swelling pressure is
exerted by the soil on the structure. This pressure if not controlled, may cause uplifting and distress in
the structure (Shelke and Murthy 2010). The strength loss on wetting is another severe problem with
such soils. Due to this peculiar behaviour many civil engineering structures constructed on expansive
soils get severely distressed. Pavements are in particular, susceptible to damage by expansive soils
because they are lightweight and extend over large areas. Dwelling houses transferring light loads to
such soils are also subjected to severe distress. Similarly, earth structures such as embankments,
canals built with these soils suffer slips and damages (Mishra et al., 2008).
Soil stabilization techniques are widely used for stabilizing expansive soils. Physico-chemical
stabilization using lime, cement, fly ash, enzymes, and other chemicals control the swelling in
expansive soil (Lopez-Lara et al., 1999). In these techniques, uniform mixing of the stabilizers in soil
must be ensured else erratic results may come. Mechanical stabilization of soil (without altering
chemical properties) includes controlling compaction (Sridharan and Gurtug, 2004), pre-wetting
(Chen, 1988), mixing with sand (Sridharan and Gurtug, 2004; Mishra et al., 2008), using cohesive
non-swelling soil (Katti et al., 1983), EPS geofoam .(Shelke and Murthy 2010), reinforcing the soil
906 Vol. 7, Issue 3, pp. 905-910

using geosynthetics (Sharma and Phanikumar, 2005; Ikizler et al., 2009) and by using polypropylene
fiber ( Muthukumar 2012) are used. Special foundation techniques such as lime under-reamed piles,
belled piers and granular pile anchors are also suggested (Phanikumar 1997).
Recently Kumar and Jain 2013 and Kumar 2014 have demonstrated that the concept of granular pile,
that is popular in improving the weak marine clays, could be utilized to improve the load carrying
capacity of soft expansive black cotton soils too. In the granular pile, also known as stone column,
technique about 10 to 35% weak soil is removed and replaced with granular material in the form of
piles. Kumar (2014) performed model test in the laboratory on granular piles of sand constructed in
soft expansive soil of different UCS values. End bearing granular Piles were casted in the soil and the
load test were performed. The test results showed that load carrying capacity of a footing resting on
granular pile is significantly more than the corresponding value for the footing placed directly on the
soft soil bed. The increase in load carrying capacity is observed for all soil consistencies of the
expansive soil. It is concluded from his study that loss in strength and excessive settlement of the
expansive soil due to wetting could be minimized to a large extent by installation of granular piles in
the soil.
As reported above, besides the strength loss, the swelling and volume instability are the other severe
problems with these soils. The present work is an attempt to fill this gap. The size of the sand column
and the consistency of the soil play an important role in changing the behaviour of composite ground.
Hence these two aspects are varied in the present work and the influence of the sand column diameter
and the initial moisture content on the swelling of an expansive black cotton soil has been studied.
The details of the experimental program, results of the tests and the conclusions drawn from the study
are described below.
II. EXPERIMENTAL PROGRAM
The experiments were carried out in a cylindrical mould. The soil beds of black cotton soil were
prepared in the mould at a dry density of 15kN/m
3
. The initial mixing water content in the soil and the
diameter of the sand columns were the variable of the study. The swelling of composite soil (black
cotton soil reinforced with a sand column) on wetting was recorded. Soil beds were prepared with
14%, 18%, 22%, 26%, 30%, 36%, 40%, 44% of water content and for each test bed three diameters of
sand columns (25 mm, 37.5 mm, 50 mm) were installed. One series of swell measurements were
taken for black cotton soil beds prepared at the above water contents i.e. without the sand columns.
III. MATERIAL PROPERTIES
Two basic materials used for this study are: the black cotton soil representing the soft soil to be
improved and, the fine river sand as sand column forming material. The properties of these materials
are as follows:
(i) Black cotton soil: The black cotton soil was taken from NIT Bhopal campus. Its properties are
given in Table -1.
Table -1: Properties of Black Cotton Soil
Properties Values
Liquid limit (L.L.), % 54
Plastic limit (P.L.)% 29
Plasticity index (P.I.), % 25
Maximum dry density (MDD), kN/m
3
15
Optimum water content (OMC), % 23.5
Differential free swell (DFS), % 40
Clay and silt content, % 95.0
Soil Classification (IS:1498-1970) CH, Clay of high
plasticity
(ii) Sand: Properties of the river sand used in the sand column are listed in Table -2.

907 Vol. 7, Issue 3, pp. 905-910

Table -2: Properties of Sand
Properties Values
Particle size corresponding to 10% finer, D10, mm 0.32
Coefficient of curvature, CC 1.12
Coefficient of uniformity CU 1.69
Minimum dry density, min, kN/m
3
16.00
Maximum dry density, max, kN/m
3
17.15
Specific gravity, G 2.76
Browns suitability Number (Brown 1976) 8.87
Soil Classification (IS:1498-1970) SP (Poorly
graded sand)
IV. TEST SETUP AND PROCEDURE
A typical test arrangement with a single column of sand is shown in Fig. 1. A swelling mould of 100
mm diameter and 128 mm height with one collar and two porous plates was used. The soil was oven
dried and a predetermined amount of water was mixed and compacted in three layers to attain a dry
density of 15kN/m
3
. A porous plate with filter paper was placed below the soil sample. Then a hole of
required diameter was formed by using an auger and a casing pipe. The granular material (fine river
sand) was filled in the hole and compacted in layers to get the required density. One porous plate with
a filter paper is placed above the soil sample and the collar is fitted to the mould. Heave stake was
placed on the soil sample inside the swelling mould and a dial gauge was fixed on the top of the heave
stake to measure the swelling. This entire arrangement is placed inside a tank filled with water. The
swelling was monitored continuously by taking the dial gauge readings from time to time till dial
reading ceases to change. Tests were conducted for all the specimens prepared with different water
contents and with different sizes of the sand columns.

V. RESULTS AND DISCUSSIONS
The variation of swelling with respect to the time, for 14% water content in the soil, is plotted in Fig.
2. It was observed that swelling continue to occur nearly for about 96 hours, beyond which there is no
change in swelling. Similar trend is obtained for test beds of other initial water content in the soil.
Fig. 1 Experimental setup
908 Vol. 7, Issue 3, pp. 905-910

The maximum swelling in each case was noted and its variation with the diameter of the sand column
is plotted (Fig. 3). It can be observed from this figure, that there is a decrement in swelling with the
increase in the diameter of the sand column. The sharp decrement in swelling is observed when the
test bed is prepared at lowest water content of 14%. In this case the 50mm diameter stone column
reduces swelling by 42% in comparison to the raw soil. The corresponding values for 37.5mm and
25mm diameter stone columns are 23% and 11.5% respectively.
The reason for observing reduction in swelling with installation of the sand columns is mainly due to
the replacement of expansive soil by non expansive sand. A large diameter column replaces more soil,
hence reduction in swelling is observed more than that by the smaller one for a particular value of the
initial water content in the soil.

Further, there is significant reduction in swelling with increase in the initial water content in the black
cotton soil. To show it, Fig. 4 is plotted. It can be noted from this figure that as moulding water
content (i.e. the initial water content) in the black cotton soil increases, the swelling decreases in all
Fig. 3 Effect of sand column diameter on swelling of expansive soil

Fig 2 Variation of swelling of soil with time at 14% water content
909 Vol. 7, Issue 3, pp. 905-910

the cases. This observation is obvious because an expansive soil containing high initial water content
has little scope for imbibing more water on submergence as all its mineral particles are already
saturated and therefore very small, practically nil, swelling is observed at high initial moisture content
of 44%.

Fig 4 Variation of swelling with water content for raw soil and sand column reinforced soil
VI. CONCLUSIONS
The results of the testing program show that installation of sand column in expansive soil controls the
swelling effectively. The size of sand column and initial water content in the black cotton soil affect
the swelling behaviour. A large size sand column reduces swelling more than by a smaller one.
Swelling is also reduced with increase in water content. At 44% water content there is no swelling
found in the soil. The reduction in swelling is mainly due to replacement of expansive soil by non-
expansive sand and also because of presence of water in the soil. Thus if sand columns are installed in
expansive soils in wet condition maximum benefit in terms of volume stability can be achieved.
VII. SCOPE FOR FUTURE WORK
Installations of granular pile in soil are easy in comparison to other methods of soil improvements
such as lime or cement stabilization, where proper mixing of stabilizer with the soil, the depth of soil
to be treated poses practical difficulty. The use of sand column/ granular pile technique in swelling
soil is a relatively new area. The density, mineral characteristics of the expansive soil, and the
properties of the granular pile forming material are expected to influence performance of sand
columns in expansive soils. Future research in this area will pave the way to develop a design
methodology for mitigating the problems of expansive soil by sand columns with confidence.
REFERENCES
[1]. Brown, R.E. (1976). Vibration compaction of granular hydraulic fills. ASCE, National Water Resources
And Ocean Engineering Convention, pp. 1-30.
[2]. Chen, F.H. (1988). Foundations on Expansive Soils. Elsevier Scientific Publishing Co., Amsterdam.
[3]. Ikizler, S. B., Aytekin, M. and Vekli, M. (2009). Reductions in swelling pressure of expansive soil
stabilized using EPS geofoam and sand. Geosynthetics International, 16(3), 216221.
910 Vol. 7, Issue 3, pp. 905-910

[4]. I.S: 1498 1970 Classification and Identification of soils for General Engineering Purposes.
[5]. Harishkumar. K and Muthukkumaran, K (2011). Study on swelling soil behaviour and its improvements.
International Journal of Earth Sciences and Engineering, ISSN 0974-5904, Volume 04, No 06 SPL, October
2011, pp. 19-25
[6]. Katti, R. K., Bhangle, E. S. and Moza, K. K. (1983). Lateral pressure of expansive soil with and without
cohesive non-swelling soil layer applications to earth pressures of cross drainage structures of canals and key
walls of dams (studies of K0 condition). Central Board of Irrigation and Power. Technical Report 32, New
Delhi, India.
[7]. Kumar, R. and Jain P. K. (2013). Expansive Soft Soil Improvement by Geogrid Encased Granular Pile.
Int. J. on Emerging Technologies, 4(1): 55-61(2013).
[8]. Kumar, R. (2014). A Study on soft ground improvement using fiber-reinforced granular piles, Ph. D.
thesis submitted to MANIT, Bhopal (India)
[9]. Lopez-Lara, T., Zepeta- Garrido, J. A. and Castario, V. M. (1999). A comparative study of the
effectiveness of different additives on the expansion behavior of clays. Electronic Journal of Geotechnical
Engineering, 4(5), paper 9904.
[10]. Mishra, A. K., Dhawan, S. and Rao, M. S. (2008). Analysis of swelling and shrinkage behavior of
compacted clays. Geotechnical and Geological Engineering, 26(3), 289298
[11]. Muthukumar. M (2012). Swelling pattern of polypropylene fiber reinforced expansive soils.
International Journal of Engineering Research and Applications, Vol. 2, Issue 3, May-Jun 2012, pp.1385-1387
[12]. Phanikumar, B.R.(1997), A study of swelling characteristics of and granular pile anchor foundation
system in expansive soils. Ph.D. thesis, Jawaharlal Nehru Technological Univ., Hyderabad, India.
[13]. Ranjan, G and Rao, A.S.R. (2005), Basic and applied soil mechanics, New Age International (P) Ltd,
New Delhi pp 753.
[14]. Sharma, R. S. and Phanikumar, B. R. (2005). Laboratory study of heave behaviour of expansive clay
reinforced with geopiles. Journal of Geotechnical and Geoenvironmental Engineering, 131 (4), 512520.
[15]. Shelke, A.P. and Murthy, D.S. (2010). Reduction of Swelling Pressure of Expansive Soils Using EPS
Geofoam. Indian Geotechnical Conference (2010).
[16]. Sridharan, A. and Gurtug, Y. (2004). Swelling behavior of compacted fine-grained soils. Engineering
Geology, 72(1-2), 918.
AUTHORS
Aparna was born in Lucknow (U.P.), India, in 1992. She received the Bachelor degree in
civil engineering from Babu Banarasi Das National Institute of Technology & Management,
in 2012 and she is currently pursuing the Master degree in geotechnical engineering from
Maulana Azad National Institute of Technology, Bhopal, which will complete in July, 2014.
Her research interests include geotechnical engineering.

Pradeep Kumar Jain was born in Jhansi (U.P), India, in 1964. He received the Bachelor
in civil engineering degree from the MITS, Gwalior, in 1986 and Master degree in
construction technology and management from MITS, Gwalior, in 1988. He has completed
his PhD from IIT Roorkee, Roorkee, in 1996. His research interests include geotechnical
engineering.

Rakesh Kumar was born in Gajrola (U.P), India, in 1977. He received the Bachelor degree
in civil engineering from the MMMEC, Gorakhpur, in 1999 and Master degree in
geotechnical engineering from IIT Roorkee, Haridwar, in 2001. He has completed his PhD
from Maulana Azad National Institute of Technology, Bhopal, in 2014. His research
interests include geotechnical engineering.

911 Vol. 7, Issue 3, pp. 911-916

EFFECTIVE FAULT HANDLING ALGORITHM FOR LOAD
BALANCING USING ANT COLONY OPTIMIZATION IN CLOUD
COMPUTING
Divya Rastogi and Farhat Ullah Khan
ASET, Amity University, Noida, India

ABSTRACT
Cloud computing is an emerging technology in distributed computing. It is a collection of interconnected
virtual machines as to facilitate pay per use model as per the demand and requirement of the user. The primary
aim of cloud computing is to provide efficient access to remote and geographically distributed resources without
losing the property of reliability. In order to make these virtual machines hassle free, a distribution of work load
(Load balancing) is done properly. Since in cloud computing the processing is done on remote computers hence
there are more chances of errors, due to the undetermined latency and lose control over computing node. So it
is very necessary to maintain the reliability of remote computers. That is why it is mandatory to create a fault
tolerant infrastructure for cloud computing. This paper focuses on to balance the load of entire system while
trying to maintain the reliability of the system by creating it as a fault tolerant system. Our objective is to study
the existing ACOs and to develop an effective fault tolerant system using ant colony optimization. Paper
describes an algorithm in order to make the system more reliable and fault tolerant.
KEY WORDS: Ant colony optimization (ACO), Cloud computing, fault tolerant, virtual machines.
I. INTRODUCTION
Internet and its technologies are fast growing and very popular now a days. With the grown popularity
of Internet cloud Computing became a hot topic of industry and academia as an emerging new
computing mechanism. It is supposed to provide computing as the utility to meet the everyday needs
of the general community [2]. Cloud computing delivers services such as infrastructure as a service
(IaaS), platform as a service (PaaS), and software as services (SaaS). In SaaS, software application is
provided to the user by the cloud provider. In PaaS an application development platform is provided
as a service to the developer to create a web based application. In IaaS computing infrastructure is
provided as a service to the requester in the form of Virtual Machine (VM). Above mentioned
services are made available on a subscription basis using pay-per-you-use model to customers,
regardless of their location.
Cloud Computing still under in its development stage and has many issues and challenges. One of the
major issues of cloud computing is fault management.
Fault is a situation where in the system deviates from its expected behavior. Fault management is to
manage the fault and to take the corrective action in order to make the system fault tolerant and
reliable. Fault tolerance or graceful degradation is the property that enables a system to continue
operating properly in the event of the failure of (or one or more faults within) some of its components
[6]. If its operating quality decreases at all, the decrease is proportional to the severity of the failure,
as compared to a navely designed system in which even a small failure can cause total breakdown.
Fault management is to manage the fault tolerance system without degrading the performance of the
system.
A fault managing mechanism enables the system to continue its intended operation, possibly at a
reduced level, rather than failing completely, when some part of the system fails.
912 Vol. 7, Issue 3, pp. 911-916

The main aim of fault management is to provide a high reliable computing and the best system
throughput [3]. A fault manager handles the situations of failure and can take the corrective actions in
order to maintain the flow.
This paper presents an algorithm for fault management at the time of load balancing in ACO (Ant
colony optimization) system in cloud computing. Rest of the paper is organized is as follows: section
II discusses the literature review, section III contains information about the basics of ant colony
optimization, section IV discusses the proposed work, Section V contains the expected outcome and
last section concludes the paper with the future prospects.
In this section, we have described the related work of fault management in cloud computing
environment. The author of paper [3] modeled that the system tolerates the fault and makes the
decisions on the basis of reliability of the processing nodes named virtual machines. After every
computing cycle the reliability of virtual machines changes thats why it is adaptive. But if a virtual
machine manages to produce a correct result within the time limit than the reliability increases but if it
fails then the reliability of virtual machine decreases. The authors of paper [14] proposed a low
latency fault tolerance (LLFT) system that provides fault tolerance for distributed applications within
a local-area network, by using a leaderfollower replication strategy. It provides application-
transparent replication. Its novel system model enables LLFT to maintain a single consistent infinite
computation, despite faults and asynchronous communication. The authors of paper [15] gave a model
named as Vega-warden, uniform user management system which helps to supply a global user
space for different virtual infrastructure and application services in cloud computing environment.
This model is made for virtual cluster base cloud computing environment to resolve the usability and
security arise at the time of infrastructure sharing.
Fumia Machida [12] proposed a component base availability modeling frame work known as Candy,
which creates a comprehensive availability model in a semi automatic fashion and described by
system modeling language. The base of this model is the high availability assurance of cloud service.
Authors of paper [11] proposed a model that would help to overcome the limitation of existing on
demand service methodologies. To achieve the reliability and resilience they proposed an innovative
perspective on creating and managing fault tolerance. By this particular methodology user can specify
and apply the desire level of fault tolerance without requiring any knowledge about its
implementation. FTM architecture this can primarily be viewed as an assemblage of several web
services components, each with a specific functionality. Authors in paper [8] has given a model
named as Magi-Cube to provide high reliability and low redundancy for the storage architecture for
cloud computing. The system is built on the top of HDFS and uses HDFS as a storage system for file
read /write and metadata management. A file scripting and repair component is also built to work in
the back ground independently. This model Magi-Cube is constructed with the view that high
reliability and performance and low cost (space) are the 3 conflicting component of storage system.
Authors of paper [16] has modeled FT-Cloud that is a component ranking based frame work and its
architecture for building cloud application. FT-Cloud employs the component invocation structure
and frequency for identify the component. There is an algorithm to automatically determine fault
tolerance stately.
In [8], suggested Ant Colony Optimization (ACO) algorithm for load balancing. In this approach,
incoming ants update the entries in the pheromone table of a node. For instance, an ant traveling from
source to destination will update the corresponding entry in the pheromone table. If an ant is at a
choice point when there is no pheromone, it makes a random decision. Here tasks are assigned to
server with non-preemption scheduling so waiting time of high priority tasks are increased.
III. ANT COLONY OPTIMIZATION
Dorigo M. has introduced the ant algorithm based on the behavior of real ants in 1996[14][15], a new
heuristic algorithm for the solving the combinatorial optimization problems. ACO is inspired from the
ant colonies that work together in foraging behavior Investigations show that: Ant has the intelligence
system to find an optimal path from nest to source of food. On the way of searching, ants act as the
913 Vol. 7, Issue 3, pp. 911-916

expert agents and in the direction of movement they lay some pheromone on the ground; while an
isolated ant runs into a previously set trail. The trailing ant can detect it and decide to follow it with
high probability. The probability of ant chooses a way is proportion to the concentration of a ways
pheromone. To a way, the more ants choose, the way which has denser pheromone, and vice versa.
By this feedback technique, ant can find out an optimal way finally [14][15]. The ants work together
in search of new sources of food and simultaneously use the existing food sources to shift the food
back to the nest [14][15].
In Paper [8], author has suggested that each node in ant based control system which was designed to
solve the load balancing in cloud environment was configured with
1) Capacity that accommodates a certain.
2) Probability of being a destination.
3) Pheromone (or probabilistic routing) table.

Figure 1. Updation through incoming ants
As shown in figure 1[8], incoming ants update the entries the pheromone table of a node. For
instance, an ant traveling from (source) to (destination) will update the corresponding entry in the
pheromone table in. Consequently, the updated routing information can only influences the routing
ants and calls that have as their destination. However, for asymmetric networks, the costs may be
different. Hence, in this approach for updating pheromone is only appropriate for routing in
symmetric networks.
IV. PROPOSED WORK
As cloud computing is a distributed system. For a distributed system to work properly, it is necessary
that the system should be in safe state. In case of any fault, there should be technique to handle this
fault so that working of the system should not get affect.
In paper [9], authors did not explain any method or mechanism for fault handling. It does not provide
any solution in case of any failure or error in system. Here in this paper, we propose a model for fault
management.
As discussed before, ants follow the route according to the updated entries in pheromone table so it is
necessary to store the status of this pheromone trail. To make the ACO more reliable with the capacity
of fault tolerance, we construct the system with the following features:
(i) Each ant will be having a small memory element.
(ii) Every node in the system will have the knowledge about its neighbors.
(iii) Each node can flood the message in the network.
Memory element is installed in each ant at the time of the construction of ACO system and this will
work as knowledge base. Memory element will store the information about each movement of the ant
in the system (distance covered by each ant), its neighboring ants information and routing
information (Based on pheromone table). Knowledge base is used when a fault is generated and helps
to take the corrective actions.
A) Working model
914 Vol. 7, Issue 3, pp. 911-916

Suppose there are N nodes in the system. Some of the nodes act as the source nodes (Surplus with the
workload) while other nodes work as destination node (Lightly loaded nodes). Our task is to provide a
fault management technique from source nodes to destination nodes. We are providing a model that
will improve the reliability of ACO system. Fault management is a two step process:
(i) Fault detection
(ii) Fault handling
(i) Fault Detection: for fault detection, a fault detector is installed into the system. This fault
detector will work on the concept of scholastic petri nets algorithm given by et al
Agrayasan. The partially stochastic Petri nets developed in this paper provide independent
behaviors to regions of the net that are not directly interacting. The diagnosis problem
addressed assumes a sequence of alarms. By generating the connecting tiles that match
the observed sequence of alarms and their causal dependence relations, we can find out
the fault generated in the system [16].
(ii) Fault Handling: To control the fault and to improve the reliability of the system, we
construct a modified algorithm of ACO with the implementation of checkpoints. Check
Points (CP) are those points from where backward error recovery is done in case of
complete failure of a system. Adding checkpoints helps to provide an automatic forward
recovery. If an ant fails to route the data properly at the time of load balancing system
will not fail. It will continue to operate with remaining nodes. This mechanism will work
properly until all the nodes fail.
B) Modified Algorithm
This algorithm will help to make the cloud environment more reliable. Here, in this algorithm,
whenever a request comes then the state of the system is checked. If the resource can be allocated
without any deadlock then resource is allotted, otherwise request is transferred to waiting state.
Start
N: = number of nodes
dis_cov:= 0 (Distance covered by the ant)
Set Checkpoint= 0 (covered distance is zero by far).
Do (Formation of load balancing using ACO)
Input processing node status
If input processing node status = pass then
Dis_cov= dis_cov+1
Set checkpoint
While (all nodes get balanced)
If (failure = yes) then
If (dis_cov<1) then
Roll back the process
Else if (dis_cov>1) then
Check for the next hop and nearest checkpoint.
Update the value according to the nearest checkpoint.
Else
Give a pass.
End
Here, in the above given algorithm we can see that every time an ant moves, the value is stored in
ants memory element. This covered distance shows the path value covered by individual ants in the
system from source to destination. If any fault occurs then the ant will look into its pheromone table
for the next hop and its distance to the net hope. The fault management system will check for the
nearest covered checkpoint and roll back the system to the covered checkpoint.
V. EXPECTED OUTCOME
The given algorithm will continuously allocate the resources as per the requests generated by the
client. Before the allocation this algorithm checks that the system is in safe state or not before any
allocation.
915 Vol. 7, Issue 3, pp. 911-916

Suppose that there are N nodes acting as active (current nodes). At each movement of the ant, distance
is stored in the memory element of the ant. As the ant reaches to a checkpoint, the value of the
checkpoint is also stored.
Suppose there are N1, N2, N3, N4, N5 nodes in the system. On receiving the request server allocates
the request to any one of nodes as per the matrix provided in paper [9] [10]. The node with minimum
cost is selected. A fault may generate at time when the server is busy in balancing the load among the
nodes. Suppose server selects node N2 at any time t1 for handling request r1 (by watching the matrix
of load at all nodes), at the time of transferring the load N2 stops responding to the server. Now at this
stage the checkpoint and nearest neighbour is checked. As the nearest neighbour is N1 and N3 so the
load is transferred to any of them by considering their load matrices.
Till now, there were many algorithms which can help to make the system fault tolerant but no
technique was provided to resolve the problem when fault occurs at the time of load balancing. In this
paper, we have given an algorithm that can deal with the situations of faults or failures. We tried to
make the system reliable with the implementation of this algorithm. In future we will try to calculate
the complexity of this algorithm so that a more optimized way can be find out.
REFERENCES
[1]. Ginu Chako, S. Induja, A. Abuthahir, S. Vidya, Improving Performance of Fault Tolerant Mechanism by
Optimal Strategy Selection in Cloud, IJST, March 2014.
[2].G. Shobha, M. Geetha, R.C. Suganthe, Load Balancing by Preemptive Task Scheduling Instigated Through
Honey Bee Behaviour in Cloud Datacenter, ICSCN, March 2014.
[3]. Shivam Nagpal, Praveen Kumar , A Study on Adaptive Fault Tolerance in Real Time Computing,
International Journal of Advanced Research in Computer Science and Software Engineering(IJARCSSE),
Volume 3, Issue 3, March 2013.
[4].Prasinjit Kumar Patra, Haspreet Singh, Gurpreet Singh, Fault Tolerance Techniques and comparative
Implementation in Cloud Computing, IJCA, February 2013.
[5]. Anjali D. Meshram, A.S. Sanbare, S.D. Zade, Fault Tolerance Model for Reliable Cloud Computing, Vol
1, issue 7, July 2013.
[6]. Ravi Jhawar, Vincenzo Piuri, Marco Santambrogio, Fault Tolerance Management in Cloud Computing: A
System Level Perspective, IEEE Systems Journal, Vol 7, No 2, June 2013.
[7] N.Chandrakala, Dr. P.Sivaprakasam, Analysis of Fault Tolerance Approaches in Dynamic Cloud
Computing, IJARCSSE, Volume 3, Issue 2, March 2013.
[8]. Qingqing Feng, Jizhong Han, Yun Gao, Dan Meng Magicube: High Reliability and Low Redundancy
Storage Architecture for Cloud Computing 2012 IEEE Seventh International Conference on Networking,
Architecture, and Storage.
[9]. Ratan Mishra, Anant Jaiswal, Ant colony Optimization: A Solution of Load Balancing in cloud, IJWest,
Vol 3, No 2, April 2012.
[10]. Nidhi Jain Kansal, Inderveer Chana, Cloud Load Balancing Techniques: A Step Towards Green
Computing, IJCSI, Vol 9, Issue 1, No1, January 2012.
[11]. Ravi Jhawar, Vincenzo Piuri and Marco Santambrogio A Comprehensive Conceptual System level
Approach to Fault Tolerance in Cloud Computing IEEE.
[12]. Fumio Machida, Ermeson Andrade, Dong SeongKim and Kishor S. Trivedi Candy: Component-based
Availability Modeling Framework for Cloud Service Management Using Sys-ML 2011 30th IEEE
International Symposium on Reliable Distributed Systems.
[13]. Arvind Kumar et.al, Rama Shankar Yadav, Ranvijay, Anjali Jain, Fault Tolerance in Real Time
Distributed System, IJCSE, Vol 3, No 2 February 2011.
[14]. W. Zhao, P. M. Melliar Smith, L.E. Moser, Fault Tolerance Middleware for Cloud Computing, in
proceedings of the 2010 IEEE 3
rd
International Conference on Cloud Computing, Ser. Cloud 10, Washington,
DC, USA:IEEE Computer Society , 2010, pp 67-74.
[15]. Jianlin, Xiaoyi Lu, Lin Yu, YongqiangZou and Li Zha Vega Warden: A Uniform User Management
System for Cloud Applications 2010 Fifth IEEE International Conference on Networking, Architecture, and
Storage.
916 Vol. 7, Issue 3, pp. 911-916

[16]. ZibinZheng, Tom Chao Zhou, Michel R. Lyu, and Irwin king FT-Cloud: A Component Ranking
Framework for Fault-Tolerant Cloud Applications 2010 IEEE 21
st
International Symposium on Software
Reliability Engineering.
[17]. Armen Aghasarayan et.al, Eric Fabre, Albert Benvenister, Fault Detection and Diagnosis in Distributed
Systems: An Approach by Partially Stochastic Petri Nets, Discrete Event Dynamic Systems: Theory and
Application, 8, 203-231, 1998.
AUTHORS
Divya Rastogi is a student of M.Tech in Amity University. Divya also works as Assistant Professor for G.L.
Bajaj Institute of Technology & Management, Greater Noida. Her expertise includes Networking, Network
security and cloud computing.

917 Vol. 7, Issue 3, pp. 917-922

HANDLING SELFISHNESS OVER MOBILE AD HOC NETWORK
Madhuri D. Mane and B. M. Patil
P.G. Dept., MBES College of Engineering Ambajogai, Maharashtra, India

ABSTRACT
Mobile Ad hoc Network (MANET) also called as mobile mesh network. It is a self-configuring network of mobile
devices connected by wireless links. Here we assume all mobile nodes cooperate fully with the functionalities of
the network. In reality, however, some nodes may selfishly decide only to cooperate partially, or not at all, with
other nodes [1]. In this paper we have to find selfish nodes and recover them. These selfish nodes could then
reduce the overall data accessibility and increase delay in the network. In this paper, we examine the impact of
these selfish nodes and reduce the delay.
KEYWORDS: MANET, Proactive, Reactive, selfish, SCF, eSCF, SCE-DS, SCF-CN.
I. INTRODUCTION
Wireless cellular system has been in use since 1980s. Wireless system operates with the aid of a
centralized supporting structure such as an access point. These access points assist the wireless users
to keep connected with the wireless system, when they roam from one place to other. In wireless
system the device communicate via radio channel to share resource and information between devices.
Due to presence of a fixed supporting structure, limits the adaptability wireless system is required
easy and quick deployment of wireless network. Recent advancement of wireless technologies like
Bluetooth, IEEE 802.11 introduced a new type of wireless system known as Mobile ad-hoc network
(MANETs) [13], which operate in the absence of central access point. It provides high mobility and
device portabilitys that enable to node connect network and communicate to each other. It allows the
devices to maintain connections to the network as well as easily adding and removing devices in the
network. User has great flexibility to design such a network at cheapest cost and minimum time.
MANETs has shows distinct characteristics, such as:
Weaker in Security
Device size limitation
Battery life
Dynamic topology
Bandwidth and slower data transfer rate
Ad hoc routing protocols can be broadly classified as:
Proactive (table-driven)
Reactive (on-demand)
Proactive protocols maintained nodes in a MANET and keep track of routes to all possible destinations
so that the route is already known and can be immediately used, when a packet needs to be forwarded.
On the other hand, reactive protocols employ a lazy approach whereby nodes only discover routes to
destinations on demand, i.e., a node does not need a route to a destination until that destination is to be
the sink of data packets sent by the node[14].In MANET, mostly we assume that all mobile nodes
cooperate fully in the network functionalities. But some nodes decide not to cooperate at all. From Past
there is a tendency in the nodes in an ad hoc network to become selfish. The selfish nodes are reluctant
to spend their resources such as memory, battery power and CPU time for others but they are not
malicious nodes [14]. The problem may become complicated, when with the passage of time the nodes
have small amount of residual power and want to conserve it for their own purpose.
918 Vol. 7, Issue 3, pp. 917-922

Figure 1. Selfishness in MANET
In general while handling Selfishness in MANET we have to improve data accessibility and reduce
query delay, i.e., query Response time, if the mobile nodes in a MANET together have sufficient
memory space to hold both all the replicas and the original data. For example, the response time of a
query can be substantially reduced, if the query accesses a data item that has a locally stored replica.
However, there is often a trade-off between data accessibility and query delay, since most nodes in a
MANET have only limited memory space [15]. For example, to reduce its own query delay a node
may hold a part of the frequently accessed data items locally. However, if there is only limited
memory space and many of the nodes hold the same replica locally, then some data items would be
replaced and missing. Thus, the overall data accessibility would be decreased. Hence, to maximize
data accessibility, a node should not hold the same replica that is also held by many other nodes [15].
In [1] it defines three types of behavioral states for nodes from the viewpoint of selfish replica
allocation:
Type-1 node: are non-selfish nodes i.e. the nodes hold replicas allocated by other nodes within
the limits of their memory space.
Type-2 node: called as fully selfish nodes. These nodes do not hold replicas allocated by other
nodes, but allocate replicas to other nodes for their accessibility.
Type-3 node: are partially selfish nodes. The nodes use their some part of memory space for
allocated replicas by other nodes. Their memory space may be divided logically into two parts
as selfish and public area. These nodes allocate replicas to other nodes for their data
accessibility.
The detection of the type-3 nodes is complex, because they are not always selfish. In some sense, a
type-3 node might be considered as non-selfish, since the node shares part of its memory space here
this is considered as (partial) selfish, because the node also leads to the selfish replica allocation
problem. Selfish and non-selfish nodes perform the same procedure when they receive a data access
request, although they behave differently in using their memory space [1].
MANETs rely on the cooperation of all the participating nodes. The more nodes cooperate to transfer
traffic, the more powerful a MANET gets. But supporting a MANET is a cost-intensive activity for a
mobile node. Detecting routes and forwarding packets consumes local CPU time, memory, network-
bandwidth, and last but not least energy [16]. Therefore there is a strong motivation for a node to deny
packet forwarding to others, while at the same time using their services to deliver own data. Some
resources, namely battery power (energy), are scarce in a mobile environment and can be depleted at
fast pace with the device utilization. This can lead to a selfish behavior of the device owner that may
attempt to take the benefit from the resources provided by the other nodes without, in return, making
available the resources of his own devices. In this scenario, open MANETs will likely resemble social
environments. A group of persons can provide benefits to each of its members as long as everyone
provides his contribution. For our particular case, each member of a MANET will be called to
forward messages and to participate on routing protocols. A selfish behavior threatens the entire
919 Vol. 7, Issue 3, pp. 917-922

community. Optimal paths may not be available. As a response, other nodes may also start to behave
in the same way.
In the [1] existing strategy consists of three parts: 1) detecting selfish nodes, 2) building the SCF-tree,
and 3) replica allocation.
The reason is that without forming any group or engaging in lengthy negotiations each node can
detect selfish nodes and makes replica allocation at its own discretion [15].
1) Detecting Selfish Node
The notion of credit risk can be described by the following equation:
(1)
i.e., the expected risk is calculated by number of requests served by the node. And the expected value
is calculated by number of memory spaces shared by nodes. In the existing strategy, each node
calculates a CR score for each of the nodes to which it is connected [1]. The calculated CR value is
called as degree of selfishness. Degree of selfishness tells that the node is seems to be Selfish node.
Each node shall estimate the selfishness degree for all of its connected nodes based on the CR score.
They first describe selfish features that may lead to the selfish replica allocation problem to determine
both expected value and expected risk [15].
2) Building SCF-Tree
The SCF Tree build based on human friendship management in the real world, where each person
makes their own friends forming a web and manages friendship by their self. They do not have to
discuss these with others to maintain the friendship [1]. The decision is only at their discretion. The
main goal of the replica allocation techniques discussed is to reduce traffic overhead, achieving data
accessibility to maximum level. If this replica allocation technique can allocate replica without
considering with other nodes, it will decrease the traffic overhead.
3) Allocating Replica
A node allocates replica at every relocation period, after building the SCF-tree. Within its SCF-tree
[1] each node asks non selfish nodes to hold replica when it cannot hold replica in its local memory
space. Each node determines replica allocation individually without any communication with other
nodes, since the SCF-tree based replica allocation is performed in a fully distributed manner. At first,
a node determines the priority for allocating replicas. The priority is based on Breadth First Search
(BFS) order of the SCF-tree. The dotted arrow in represents the priority for allocating replica.
III. PROPOSED SYSTEM
Although network issues are important in a MANET, replica allocation is also crucial, since the
ultimate goal of using a MANET is to provide data services to users. A selfish node may not share its
own memory space to store replica for the benefit of other nodes. We can easily find such cases in a
typical peer-to-peer application. A selfish node may not share its own memory space to store replica
for the benefit of other nodes. We can easily find such cases in a typical peer-to-peer application.
They are based on the concept of a self-centered friendship tree (SCF-tree) and its variation to achieve
high data accessibility with low delay in the presence of selfish nodes. We first describe selfish
features that may lead to the selfish replica allocation problem to determine both expected value and
expected risk. Here we can consider threshold value as main part for detecting selfishness. If node
transfers the packet within the threshold value, then it is a non selfish. If exceeds the threshold value,
then it is a selfish node.
Credit risk value can be calculated by means of threshold value. For every node ,we can store the data
and sent to another node, If there is a selfish node present, then the data will be lost, At that time,
Neighboring node will have copy of that data, so we can come one step backward route the data to
the another path. As stated in [1] the detection of the type-3 nodes is complex, because they are not
always selfish. In some sense, a type-3 node might be considered as non-selfish, since the node shares
part of its memory space, due to this here we can consider only two types of nodes as selfish and non-
selfish. A selfish node can silently drop some or all of the data packets sent to it for further forwarding
even when no congestion occurs. Selfish node attack presents a new threat to wireless ad hoc
networks since they lack physical protection and strong access control mechanism. An adversary can
easily join the network or capture a mobile node and then starts to disrupt network communication by
920 Vol. 7, Issue 3, pp. 917-922

silently dropping packets. Selfish node attack is a serious threat to the routing infrastructure of both
MANET and the Internet since it is easy to launch and difficult to detect. For this Session Key is used
for security purpose.
IV. SIMULATION
Simulations are made using NS-2 [17] simulation program that consists of the collection of all
network protocols to simulate many of the existing network topologies. The number of mobile nodes
is set to 50. Each node has its local memory space and moves with a velocity from 0 ~ 1 (m/s) over
1600X1000 meter flatland area. The movement pattern of nodes follows the random waypoint model,
where each node remains stationary for a pause time and then it selects a random destination and
moves to the destination [1]. After reaching the destination, it again stops for a pause time and repeats
this movement behavior. The radio communication range of each node is a circle with a radius of 1 _
19 (m). We suppose that there are 50 individual pieces of data, each of the same size. In the network,
node Ni (1 < i < 50) holds data Di as the original. The data access frequency is assumed to follow Zipf
distribution. And threshold value is considered as 4ns. The default relocation period is set to 256 units
of simulation time which we vary from 64 to 8,192 units of simulation time. Nodes were set to use
802.11 radios with 2 Mbps bandwidth and 250 meters nominal range. We considered only static
scenario so link breakage due to mobility is zero. The simulated time was 100 seconds. TABLE I
describes the simulation parameters.
Table 1. Simulation Parameters
Parameter (unit) Value (default)
Number of nodes 50
Number of data items 50
Radius of communication range (m) 1 ~ 19 (7)
Size of network (m) 1600 *1000
Percentage of selfish node (%) 0 ~ 100 (70)
Relocation period (ms) 64 ~ 8,192
Threshold (ns) 0 ~ 5 (4)
V. RESULT

Figure 2. Forwarding Data to handle selfishness
921 Vol. 7, Issue 3, pp. 917-922

In Figure 2 we have to maintain the data flow, where node 10 becomes selfish. Due to this selfishness
behavior packet send over link of node 2-10 gets lost. To handle this situation link cut between these
nodes, node 2 send request to node 3 for further data delivery. Hence Selfishness can be handled.
Effect of handling Selfishness over MANET under NS2 simulator is given below.

Figure 3. Query delay with respect to relocation period
Figure 3 shows average query delay with respect to relocation period. As expected, Our Proposed
technique shows the best performance in terms of query delay, since most successful queries are
served by local memory space. While in [1] SAF shows best result which gives minimum delay. Our
graph shows better than this as it passes below to SAF technique. In above DCG technique shows the
worst performance. This can be explained as follows: the distance in average query delay counts
among group members in the DCG technique is longer than that in the SCF technique. Since most
successful queries are served by group members in these techniques, the long distance among group
members affects query delay negatively [1].
In this paper, we have assumed an environment in which nodes simultaneously issue access requests
to correlated data items in MANETs. We have proposed method to handle selfishness condition that
is actually extensions of our previously proposed methods to adapt such an environment. The
simulation results show that the methods proposed in this paper which give lower query delay than the
corresponding methods used in [1]. The results also show that the proposed method is better than
DCG, SCF and SAF methods. Here delay is minimum than all this methods.
As part of our future work, we plan to address data replication in an environment where access
requests for correlated data items are issued with some intervals. And increase data accessesability.
We also plan to extend our proposed methods to adapt data updating.
REFERENCES
[1] Jae-Ho Choi, Kyu-Sun Shim, Handling Selfishness in Replica Allocation over a Mobile Ad Hoc
Network, IEEE Transactions on Mobile Computing, vol. 11, no. 2, pp 278-291, Feb 2012.
[2] Takahiro Hara And Sanjay Kumar Madria Consistency Management Strategies For Data Replication In
Mobile Ad Hoc Networks, IEEE Transactions On Mobile Computing, Vol. 8, No. 7, pp. 950-967, July
2009.
[3] Nikolaos Laoutaris, Orestis Telelis, Vassilios Zissimopoulos, And Ioannis Stavrakakis Distributed Selfish
Replication IEEE Transactions On Parallel And Distributed Systems, vol. 17, no. 12, pp.1401-1413, Dec
2006.
[4] Takahiro Hara and Sanjay Kumar Madria, Data Replication for Improving Data Accessibility in Ad Hoc
Networks, IEEE Transaction On Mobile Computing, vol. 5, no. 11, pp. 1515-1532, Nov. 2006
922 Vol. 7, Issue 3, pp. 917-922

[5] Alessandro Mei, Luigi V. Mancini, and Sushil Jajodiasecure Dynamic Fragment And Replica Allocation
In Large-Scale Distributed File Systems, IEEE Transactions On Parallel And Distributed Systems, Vol.
14, No.9, September 2003.
[6] G. Tamilarasi, Devi Selvam Allocation of Replicas by Solving Selfishness in MANET, International
Journal of Engineering Research & Technology (IJERT) Vol. 2 Issue 1, January- 2013 ISSN: 2278-0181.
[7] Jim Solomon , Immanuel John A Survey on Selfishness Handling In Mobile Ad Hoc Network,
International Journal of Emerging Technology and Advanced Engineering, Volume 2, Issue 11, November
2012, ISSN 2250-2459.
[8] Shiow-yang Wu, Member, IEEE, and Yu-Tse Chang A User-Centered Approach to Active Replica
Management in Mobile Environments, IEEE Transactions On Mobile Computing, Vol. 5, No. 11,
November 2006.
[9] Martin Schutte Detecting Selfish and Malicious Nodes in MANETs, Seminar: Sicherheit in
Selbstorganisierenden Netzen, Hpi/University Potsdam, sommersemester 2006.
[10] Kashyap Balakrishnan, Jing Deng and Pramod K. Varshney TWOACK: Preventing Selfishness in Mobile
Ad Hoc Networks, IEEE Conference, 0-7803-8966-2/05, 2005.
[11] Prasanna Padmanabhan, Le Gruenwald, Anita Vallur and Mohammed Atiquzzaman A survey of data
replication techniques for mobile ad hoc network databases, The VLDB Journal, ISSN 11431164 May
2008.
[12] Takahiro Hara Effective Replica Allocation in Ad Hoc Networks for Improving Data Accessibility,
IEEE INFOCOM, 2001
[13] Y. Hu, A. Perrig and D. Johnson, Ariadne: A Secure On-demand Routing Protocol for Ad Hoc Networks,
in Proceedings of ACM MOBICOM02, 2010.
[14] Devi Selvam and Tamilarasi. G, Selfish Less Replica Allocation in MANET, International Journal of
Computer Applications, Vol. 63 No.19, pp.33-37 February 2013.
[15] K.P.Shanmuga Priya and V.Seethalakshmi, Replica Allocation In Mobile Adhoc Network For Improving
Data Accessibility Using SCF-Tree, International Journal of Modern Engineering Research (IJMER),
Vol.3, Issue.2, pp-915-919 March-April. 2013.
[16] T.V.P.Sundararajan and Dr.A.Shanmugam, Performance Analysis of Selfish Node Aware Routing
Protocol for Mobile Ad Hoc Networks, ICGST-CNIR Journal, Volume 9, Issue 1, July 2009.
[17] The network simulator - ns2, http://www.isi.edu/nsnam/ns/.

AUTHORS
Madhuri D. Mane presently working P.G. student at the Department of Computer Science
& Engineering in MBES College of Engineering, Ambajogai, India. She completed her
Bachelors degree in Computer Science & Engineering Department from M.B.E. societys
College of Engineering, Ambajogai under Dr. B.A.M.University, Aurangabad, India. She is
pursuing her Masters Degree from the College of Engineering, Ambajogai. Her areas of
research interest include Computer Networks &Wireless Network.

B.M. Patil is currently working as a Professor in P.G. Computer Science & Engineering
Department in M.B.E. Societys College of Engineering, Ambajogai, India. He received his
Bachelors degree in Computer Engineering from Gulbarga University in 1993, MTech
Software Engineering from Mysore University in 1999, and PhD Degree from Indian
Institute of Technology, Roorkee, 2011. He has authored several papers in various
international journals and conferences of repute. His current research interests include data
mining, medical decision support systems, intrusion detection, cloud computing, artificial
intelligence, artificial neural network, wireless network and network security.
923 Vol. 7, Issue 3, pp. 923-929

A NEW APPROACH TO DESIGN LOW POWER CMOS FLASH
A/D CONVERTER
C Mohan and T Ravisekhar
2

M. Tech (VLSI) Student, Sree Vidyanikethan Engineering College (Autonomous), Tirupati,
India
2
Assistant Professor, ECE Department, Sree Vidyanikethan Engineering College
(Autonomous), Tirupati, India

ABSTRACT
The present investigation proposes an efficient low power encoding scheme intended for a flash analog to
digital converter. The designing of a thermometer code to binary code is one of the challenging issues in the
design of a high speed low power flash ADC. An encoder circuit translates the thermometer code into the
intermediate gray code to reduce the effects of bubble errors. The implementation of the encoder through
pseudo NMOS logic is presented. To maintain the high speed with low power dissipation, CMOS inverter has
been used as a comparator and by adjusting the ratio of channel width and length, the switching threshold of
the CMOS inverter is varied to detect the input analog signal. To maintain the high speed with low power
dissipation, the implementation of the ADC through pseudo NMOS logic. The proposed ADC is designed using
micro technology in 5 V power supply using PSPICE tool.
KEYWORDS: Analog to digital converter, Flash ADC, Pseudo NMOS logic, Pseudo Dynamic
CMOS logic, multi threshold CMOS inverters.
I. INTRODUCTION
1.1 Concept of ADC
The flash ADC is a fastest speed compared to other ADC architectures. Therefore, it is used for
high-speed and very large bandwidth applications such as radar processing, digital oscilloscopes,
and so on. The flash ADC is also known as the parallel ADC because of its parallel architecture.
Figure 1 illustrates a typical flash ADC block diagram. As shown in Fig. 1, this architecture needs 2
n
-1 comparators for a n-bit ADC; for example, a set of 7 comparators is used for 3-bit flash ADC.
Each comparator has a reference voltage that is provided by an external reference source. These
reference voltages are equally spaced by V
LSB
from the largest reference voltage to the smallest
reference voltage V1. An analog input is connected to all comparators so that each comparator output
is produced in one cycle. The digital output of the set of comparators is called the thermometer code
and is being converted to gray code initially (for minimizing the bubble errors) and further changed
into a binary code through the encoder [1]. However, the flash ADC needs a large number of
comparators as the resolution increases. For instance, a 6-bit flash ADC needs 63 comparators, but
1023 comparators are needed for a 10-bit flash ADC. This exponentially increasing number of
comparators requires a large die size and a large amount of power consumption [3].
The encoder is designed using pseudo NMOS logic style for achieving highest sampling frequency
of 5GS/s and low power dissipation.
924 Vol. 7, Issue 3, pp. 923-929

Figure 1: Flash ADC Block Diagram.
1.2 Design of the Encoder
Conversion of the thermometer code output to binary code is one of the bottlenecks in high speed
flash ADC design [2]. The bubble error usually results from timing differences between clock and
signal lines and it is a situation where a 1 is found above zero in thermometer code. For very fast
input signals, small timing difference can cause bubbles in the output code. Depending on the
number of successive zeroes, the bubbles are characterized as of first, second and higher orders. To
reduce the effect of bubbles in thermometer code, one of the widely used methods is to convert the
thermometer code to gray code [5, 6]. The truth table corresponding to 2 bit gray code is
presented in Table1.The relationship between thermometer code, gray code and binary code is
given below
G2=T1
G1 = T1 XOR T0
B0 = G2 XOR G1
B1 = G2
The equations for this encoder are derived from the truth table provided in Table 1.
Table1. Gray Code Encoder Truth Table

T2 T1 T0 G1 G0 B1 B0
0 0 0 0 0 0 0
0 0 1 0 1 0 1
0 1 0 1 1 1 0
0 1 1 1 0 1 1
1 0 0 0 0 0 0
1 0 1 0 1 0 1
1 1 0 1 1 1 0
1 1 1 1 0 1 1
925 Vol. 7, Issue 3, pp. 923-929

1.3 Implementation of Encoder
There are different logic styles to design the encoder. Generally the implementation will be do
using static CMOS logic style. The advantage of static CMOS logic style is have a lowest power
consumption with a lower speed. So for achieving a low power with high speed, other logic styles are
preferred. Here the design is implemented using logic style called pseudo NMOS logic [8].
The pseudo NMOS logic circuit consists of a PMOS transistor with gate connected to ground, a
bunch of NMOS transistors for the implementation of the logic style in the pull down network and
an inverter. For the implementation of a specific logic circuit with N inputs, pseudo NMOS logic re-
quires N+1transistors instead of 2N transistors in comparison with static CMOS logic. Pseudo
NMOS logic is an attempt to reduce the number of transistors with extra power dissipation and
reduced robustness.

Figure. 2 Schematic of two input AND Gate Using Pseudo NMOS Logic
The basic structure of two inputs and gate using pseudo NMOS logic style is shown in Fig. 2. The
PMOS transistor in the pull up network is connected to ground that will make the pull up network
to be pulled on all the time. The output will be evaluated conditionally depending upon the value
of the inputs in the pull down network. The inverter on the output transforms the inverted gate to
non inverted gate. Since the voltage swing on the output and the overall functionality of the gate
depend on the ratio of the NMOS and PMOS sizes, the transistor sizing is crucial in the
implementation design. The disadvantage with pseudo NMOS logic is that it has static power
consumption. (The power dissipation occurs when a direct current flows between VDD and ground.
That is when both pull up and pull down networks are switched on simultaneously). The nominal
high output voltage of (V
OH
) of pseudo NMOS logic is VDD inverter is added at the output side.
This will improve the noise margin of the circuit. In spite of static power dissipation, the pseudo
NMOS logic consumes less amount of power because of the reduced number of transistors and the
absence of other components (resistors) used for the implementation in comparison with current mode
logic.
The transistor sizes are given in Table 2.
Table 2. Transistor Sizes

(W/L) PMOS 300um/100um
(W/L) NMOS 120um/100um
926 Vol. 7, Issue 3, pp. 923-929

1.4 Simulation Results

Figure3: simulation results for General ADC.
II. MODIFIED FLASH ADC
A traditional n-bit flash ADC architecture uses 2n resistors and 2
n
-1 comparators to convert an analog
signal to digital. This architecture has drawbacks like large input signal driving, high reference
accuracy, high driving reference voltage and circuit complexity [10], [11]. CMOS inverters have been
reported to be used in ADC designs [9]-[10]. In this work, this novel idea of employing CMOS
inverters instead of analog comparators is considered for a flash ADC. CMOS is a combination of an
n-MOSFET (NMOS) and a p-MOSFET (PMOS). CMOS inverter switching threshold (Vth) is a point
at which input is equal to output voltage (Vin= Vout) and in this region both PMOS and NMOS
always operate in saturation region. If the input arrives at a particular threshold voltage, then the
output state changes. Vth can be obtained as [9]-[10]

927 Vol. 7, Issue 3, pp. 923-929

where kn and kp are constant trans conductance parameters. Vtn and Vtp are the threshold voltage
values of NMOS and PMOS respectively. As the voltages are constant, Vth depends on kn and kp
values which decide the transition point of CMOS inverter [11]. If the ratio of kn and kp is decreased,
the transited threshold voltage becomes high, otherwise, the transited inverter threshold voltage
becomes low. kn and kp can be controlled by adjusting the width (W) and length (L) of NMOS and
PMOS respectively. Based on this concept, various width/length ratio of CMOS inverters are
designed to change their threshold voltages.
Each CMOS inverter thus has a specified threshold depending upon this ratio. The W/L ratios are
define as Zn=Wn/Ln and Zp=Wp/Lp. By changing the ratio of Zn and Zp, we can obtain various
transit threshold voltages of CMOS inverters to quantize the input level. All inputs of CMOS inverter
are tied together to detect the analog input level. If the input arrives at a particular threshold voltage,
then the output state changes. The basic architecture of the proposed flash ADC is shown in Figure 3.
In a 4-bit ADC, 15 CMOS inverters are tied in parallel to detect the input signal level. The inverter
output is array from MSB to LSB. For LSB bit, the Zn/ Zp value should become small to increase the
threshold voltage and for MSB bit, the Zn/ Zp value should become large to decrease the threshold
voltage. The encoder is design in the same way of above using same technology.

Figure 4.The architecture of proposed ADC.
Table3. The transition point of 3 inverters in a flash ADC.

928 Vol. 7, Issue 3, pp. 923-929

2.1 Simulation Results

Figure 5: simulation results for modified flash ADC.
III. SIMULATION RESULTS AND DISCUSSION
The results show that the new design consumes less power than the traditional flash adc[2]. The
power dissipation is reduced because of the usage of the reduced op-amps and resistors for the
implementation of the logic. In comparison with the traditional flash adc, the proposed ADC is less
cost.
Table 4: Comparison with flash ADC

IV. CONCLUSION
Low power architecture for a 2-bit CMOS inverter based flash ADC is presented using 90nm
technology. The proposed ADC design can achieve very low power dissipation and compared with
the traditional flash ADC, this proposed method can reduce power consumption as well as uses
smaller silicon area. Hence the proposed ADC cost is reduced.
ACKNOWLEDGEMENTS
The authors would like to thank Mr.T.Ravisekhar, Department of ECE for her valuable guidance

Results Flash adc Cmos flash adc
Architecture Flash Flash
Resolution 2 bits 2bits
Technology Micro technology Micro technology
Sampling frequency 1Khz 1Khz
Vdd 5v 5v
Power consumption 1.88E+01 WATTS 1.59E+01 WATTS
929 Vol. 7, Issue 3, pp. 923-929

REFERENCES
[1]. D.Lee, J.Yoo, K.Choi and J. Ghaznavi, Fat-tree encoder design for ultrahigh speed flash analog to digital
converters I proc. IEEE Midwest Symp. Circuits Syst, pp 233-236, Aug 2002.
[2]. S. Sheikhaei, S. Mirabbasi, A. Ivanov, An Encoder for a 5GS/s 4bit flash A/D converter in 0.18um
CMOS, Canadian Conference on Electrical and Computer Engineering, pp 698-701, May 2005.
[3]. R.Baker, H.W.Li, and D.E. Boyce, CMOS Circuit Design, Layout and Simulation. Prentice Hall 2000.
[4]. Sunghyun Park, Yorgos Palaskas, Ashoke Ravi, Ralph.E.Bishop, and Michael P. Flynn, A 3.5 GS/s 5-b
Flash ADC in 90nm CMOS, IEEE Custom Integrated Circuits Conference 2006.
[5].Niket Agrawal, Roy Paily, An Improved ROM Architecture for Bubble error Suppression in High Speed
Flash ADCs, Annual IEEE Conference, pp 1-5,2005.
[6]. Mustafijur Rahman, K.L. Baishnab, F.A. Talukdar, A Novel ROM Architecture for Reducing Bubble and
Meta-stability Errors in High Speed Flash ADCs, 20th International Conference on Electronics,
Communications and Computer, pp 15-19, 2010.
[7].Vinayashree Hiremath, Design of High Speed ADC , M.S. Thesis, Wright State University, 2010.
[8]. Jan M Rabaey, Anantha Chandrakasan, Borivoje Nikolic, Digital Integrated Circuits, a design
perspective, second edition, Prentice Hall 2011.
[9] S. Chang Hsia, Wen- Ching Lee, A Very Low Power Flash A/D Converter Based on CMOS Inverter
Circuit, IDEAS05, pp. 107-110, July 2005
[10] A. Tangel and K. Choi, CMOS Inverter as a comparator in ADC design, in Proc. ICEEE, 2001,pp. 1-5
[11] S. M. Kang and Y. Leblebici, CMOS Digital Integrated Circuits Analysis and Design, Tata McGraw-Hill
Edition 2003.

Authors
C. Mohan received the B.Tech. degree in 2011 in Electronics and
Communication Engineering from VITT College of Engineering, Thanapalli. He is
presently pursuing M.Tech in VLSI in Sree Vidyanikethan Engineering College
(Autonomous), Tirupati and would graduate in the year 2014. His research interests
include Digital Logic Design, VLSI and FPGA.

T. Ravisekhar, M.Tech., is currently working as an Assistant Professor in ECE
Department of Sree Vidyanikethan Engineering College (Autonomous),Tirupati. She has
completed M.Tech in VLSI Design, in Satyabhama University and received B.Tech
from Sri Kalahastheeswara Institute of Technology, Srikalahasti. Her research areas are
RFIC Design, Digital Design and VLSI Signal Processing.

930 Vol. 7, Issue 3, pp. 930-937

OPTIMIZATION AND COMPARATIVE ANALYSIS OF NON-
RENEWABLE AND RENEWABLE SYSTEM
Swati Negi
1
and Lini Mathew
2

1
ME student, Department of E.E.E, Engineering, NITTTR, Chandigarh, India
2
Associate Professor, Department of E.E.E, NITTTR, Chandigarh, India

ABSTRACT
A Hybrid Renewable Energy System for rural electrification is a better solution to reduce environmental
pollution. Renewable energy systems offer power supply to those areas where grid electricity is not feasible or
cost of grid extension is relatively large. The main objective of this paper is to compare a non-renewable system
with a renewable energy system for remotely located site in Tirunelveli, Tamilnadu .The current system is
operated using a diesel generator and batteries and the proposed system integrates a hybrid combination of
wind and solar energy system into existing diesel generator and battery system to serve the same load. This
paper also performs cost optimization of the proposed system and the optimal cost analysis of HRES is done
using Hybrid Optimization Model for Electric Renewable (HOMER). The HOMER is a energy modelling
software for designing and analyzing hybrid power systems, which contain a combination of conventional
generators, wind turbines, solar photovoltaics, hydropower, batteries, fuel cells, hydropower, biomass and
other inputs. It is currently used all over the world by tens of thousands of people. The results show that the
proposed hybrid renewable energy system is more cost effective than the non-renewable system. The proposed
system significantly reduces the running time of diesel generator and this helps to reduce the emission level.
KEYWORDS: Hybrid Renewable Energy System, Non Renewable Energy System, Feasibility Study,
Photovoltaic, Wind Turbine Generator.
I. INTRODUCTION
Rapid depletion of fossil fuel resources on the world wide basis has necessitated an urgent search for
alternative energy sources to cater to present days demand. Another key reason to reduce our reliance
on fossil fuel is the growing evidence of the global warming phenomena. It is imperative to find
energy sources to cover the continuously increasing demand of energy while minimize the negative
environmental impacts [1].
Hybrid Renewable Energy System (HRES) is composed of one or more renewable sources combined
with conventional energy source, that works in stand alone or grid connected mode [2]. The most
popular alternative sources are wind and photovoltaic. Research indicates that hybrid
Wind/PV/battery system is a reliable source of electricity [3-5]. Since these sources are intermittent in
nature therefore, in most cases diesel generator and batteries are integrated for power storage
respectively [6].
The objective of this work is to optimize and compare a non-renewable energy system (existing
system) with a hybrid renewable energy system (proposed system) for a site in Tirunelveli. The
current system uses diesel generator and batteries for power generation and the proposed system adds
hybrid wind and solar system into the current system. HOMER is used to obtain the most feasible
configuration [7].
The paper is divided into four subsections. Section II describes the input variables like solar
irradiation, wind speed and load data. In section III the simulation of current system and proposed
system using HOMER as well as cost summary of the system components is explained. In section IV
931 Vol. 7, Issue 3, pp. 930-937

comparison between current system and proposed system is explained and discussed. Section V gives
some of the future trends of hybrid renewable energy system.
II. SYSTEM DESCRIPTION
The design of HRES is based upon the certain important sensitivity variables to optimize the cost and
size effectively. Hence, before designing the system, certain parameters like solar irradiation, wind
speed and load profile must be evaluated. It is presented in the following sections.
2.1. Solar Radiation
The latitude and longitude of Tirunelveli are 8.73N and 77.7E respectively. The hourly solar
radiation is collected for year from NASA website [8]. The average solar radiation is 4.91kWh/m
2
/d.
Clearness index and average solar irradiation for a year are shown in Table I, while Figure 1 shows
the solar irradiation in a year produced by HOMER.

Figure.1. Monthly solar radiation
Table I Clearness Index and Average Daily Irradiation for a Year
Month Clearness Index Daily Radiation
(kWh/m
2
/d)
January 0.537 4.840
February 0.578 5.580
March 0.598 6.140
April 0.526 5.520
May 0.495 5.130
June 0.438 4.460
July 0.443 4.530
August 0.469 4.870
September 0.498 5.130
October 0.451 4.420
November 0.435 3.970
December 0.479 4.200
Average 0.495 4.895
2.2. Wind Speed Data
The second renewable source implemented in the system is wind. Wind data for this site is collected
for year from NASA website. The average wind speed is 5.17m/s. The monthly average speed is
shown in table II and figure 2 shows the wind speed in a year produced by HOMER.
932 Vol. 7, Issue 3, pp. 930-937

Figure .2. Monthly wind speed
Table II Wind Speed for a Year

2.3. Load Profile
Load is an important consideration in any power generating system. In this case study, we have
considered a remote village in Tirunelveli, which lacks access to the utility grid. The measured
annual consumption is considered as 97kWh/d in the present study. Figure 3 shows monthly average
load profile. The peak load requirement decides the size of the system. Here peak load consumption is
9.7 KW.

Figure. 3. Load profile for a village in Tirunelveli
III. SYSTEM OPTIMIZATION
The non-renewable energy system (existing system) and the hybrid renewable energy system
(proposed system) are simulated in HOMER software.
3.1. Non-Renewable Energy System
Month Wind Speed (m/s)
January 5.010
February 3.810
March 3.690
April 4.030
May 5.820
June 7.330
July 6.460
August 6.460
September 5.800
October 4.530
November 4.000
December 5.090
Average 5.18
933 Vol. 7, Issue 3, pp. 930-937

The existing system model, which consists of a diesel generator and batteries to power the load, has
been modelled using micro grid optimization software HOMER as shown in Figure 4.

Figure .4. Existing power system at Tirunelveli
3.2. Renewable Energy System
The proposed hybrid renewable energy system, which consists of existing power system, wind turbine
and photovoltaic, is shown in figure 5. The proposed system is going to reduce diesel fuel
consumption and associated operation and maintenance cost. In this system PV and wind turbines will
be the primary power source and diesel generator will be using as a backup for long term storage
system and batteries for storage system.

Figure .5. Proposed hybrid Power System for Tirunelveli
3.3. Homer Input Summary
Table III and IV give the summary of the costs, other technical details of the components and other
required parameters which are given as inputs to the HOMER hybrid model.
TABLE III Cost summary of the system components
Component Size Capital
Cost $
Replacemen
t cost $
O & M
Cost
PV system 10 kW 26700 20000 10$/year
Wind
Turbine
10kW 20000 20000 500$/year
Battery 360 Ah 450 440 10$/year
Generator 10kW 5500 5475 0.5$/hr
15kW 6600 6600 0.6$/hr
21kW 7500 7500 0.7$/hr
25kW 8000 8000 0.8$/hr
30kW 8800 8800 0.9$/hr
Converter 30kW 23000 23000 10$/year

934 Vol. 7, Issue 3, pp. 930-937

Table IV specifications of the components used
PV System
Model Canadian SolarCS6P-240P
Peak Power 240W
Derating Factor 80%
Slope 6.81
Azimuth 0
Ground reflectance 20%
Temperature coefficient -0.43%/C
Nominal operating temperature 45C
Efficiency at standard test
condition
14.92%
Lifetime 20 years
Wind Turbine
Model BWC Excel-S
Rated Power 10kW
Hub height 15m
Lifetime 20 years
Battery
Nominal Voltage 6V
Nominal Capacity 360Ah
Lifetime throughput 1075kWh
Round trip efficiency 85%
Min. State of charge 30%
Float life 10 years
Maximum charge rate 1A/Ah
Maximum charge current 18A
Batteries per string 2(12 V DC bus)
Diesel Generator
Lifetime 25000
Minimum load ratio 50%
Fuel Diesel
Fuel cost $1.19
Converter
Lifetime 10 years
Efficiency 90%
Economics
Annual interest rate 5%
Project lifetime 20 years

Both the systems are simulated in HOMER software. The software finds the optimal results in each
case. Optimization result for non-renewable energy system is shown in Figure 6. As shown in the
figure the total Net Present Cost (NPC) is $339,909. Diesel generator burns 14,872L of fuel per year
and annual generator run time is 5291 hours. In twenty years, the diesel generator will burn 297,440L
of fuel. The probability of fuel prices increase is also high. The total cost is calculated with constant
price of fuel, which is $1.19 per litre. The total fuel cost during these 20 years will be 353,953.6$ and
the total cost for the whole system will be $693,862.6. Figure 7 shows the monthly average electric
production of the system which is totally produced by diesel generator.
935 Vol. 7, Issue 3, pp. 930-937

Figure .6. Optimized result for the non-renewable energy system

Figure .7. Monthly average electric production for non-renewable energy system
The hybrid renewable energy system was also simulated in HOMER software with four sensitivity
variables. These variables are wind speed, solar irradiation, load, and diesel price. Figure 8 shows the
optimized results for the proposed system. The total Net Present Cost (NPC) is $270,514. The system
will consume only 3010 liters of diesel fuel per year and annual generator run time is expected to be
1034 hours. The lifetime of this system is 25 years, but 20 years life is used to make the comparison
between two systems. In twenty years the diesel generator will burn 60200L of fuel and it will cost
$71,638. The total cost of the system will be around $342,152 Figure 9 shows the monthly average
electric production of the system. Photovoltaic production is 29% with 13,482kWh/yr. Diesel
generator production is 18% with 8,731kWh/yr. Finally, wind turbine is expected to supply the rest of
the load which is 53% with 25,068kWh/yr.

Figure 8. Optimized result for the renewable energy system

Figure 9. Monthly average electric production for renewable energy system
The difference cost between two systems is $351,710.6 which is a very significant number for a small
system. Diesel generator run times are reduced and diesel generator in the proposed system will
936 Vol. 7, Issue 3, pp. 930-937

produce only 18% of the total power production. Moreover, the reduction of yearly diesel fuel
consumption from 14,872L to 3010L has a large impact on the environment and it will reduce the net
cost of the system. Also, the diesel generator will require less maintenance and operation cost and
longer period of service before a replacement.
V. FUTURE TRENDS AND LIMITATIONS
The renewable technologies have come a long way in terms of research and development. However
there are still certain obstacles in terms of their efficiency and optimal use. Following are the
challenges faced by the designer.
The renewable energy sources, such as solar PV and FCs, need innovative technology to
harness more amount of useful power from them. The poor efficiency of solar is major
obstruction in encouraging its use.
The manufacturing cost of renewable energy sources needs a significant reduction because
the high capital cost leads to an increased payback time.
It should be ensured that there should be minimal amount of power loss in the power
electronic devices.
The storage technologies need to increase their life-cycle through inventive technologies.
These stand alone systems are less adaptable to load fluctuations. Large variation in load might
even lead to entire system collapse.
VI. CONCLUSIONS
The paper compares two different systems for providing uninterruptible power for a remote site. One
is the non-renewable energy system which consists of diesel generator and batteries and another is the
proposed system which is a combination of existing system and hybrid wind and PV system. HOMER
software is used for the comparison based on pre-feasibility study for each system. It is seen that the
proposed system will save extra cost associated with transporting diesel and maintenance. Analysis
indicates that renewable energy system will cost $351,710.6 less in its expected life than the existing
diesel generator system. Therefore a hybrid renewable energy based system is recommended for
Tirunelveli site.
REFERENCES
[1]. N.A.b.A. Razak,M.M.bin Othman, I. Musirin,"Optimal sizing and operational strategy of hybrid
renewable energy system using homer.Power Engineering and Optimization
Conference(PEOCO),2010 4th International, pp.495-501, 23-24 June 2010.
[2]. Ahmad Rohani, Kazem Mazlumi, Hossein Kord, Modeling of a hybrid power system for economic
analysis and environmental impact in HOMER.Electrical Engineering (ICEE), 2010.
[3]. R. Ramakumar, I. Abouzahr, K. Ashenayi, A knowledge-based approach to the design of integrated
renewable energy systems, Energy Conversion, IEEE Transaction on, vol&, no.4, pp.648-659,
Dec1992.
[4]. R.Chedid, H.Akiki, S. Rahman,A decision support technique for the design of hybrid solar-wind
power systems, Energy Coversion, IEEE Transaction on, vol.13, no.1.pp.76-83, Mar 1998.
[5]. W.D. Kellogg, M.H. Nehrir, G. Venkataramanan and V.Gerez,Generation unit sizing and cost analysis
for stand-alone wind, photovoltaic, and hybrid wind/PV systems, Energy conversion, IEEE
Transaction on, vol.13, no.1.pp.70-75, Mar1998.
[6]. M.S. Hossan, M.M.Hossan, A.R.N.M.R. Haque,Optimization and modelling of hybrid energy system
for off-grid electrification, Environment and Electrical Engineering (EEEIC), 2011 10th International
Conferenceon, pp.1-4, 8-11 May2011.
[7]. HOMER, http://www.homerenergy.com/ (accessed 5.2.11).
[8]. http://eosweb.larc.nasa.gov/

937 Vol. 7, Issue 3, pp. 930-937

AUTHORS BIOGRAPHY
Swati Negi is currently persuing M.E. Regular from National Instititute of Technical
Teachers Training & Research, Chandigarh Sec-26. She has completed her B-tech from
Graphic Era Institute of Technology, Dehradun. She has two year academic experience in
Graphic Era Institute of Technology, Dehradun. Her Interest areas are power electronics,
renewable technology.

Lini Mathew is presently working as Associate Professor in the Electrical Engineering
Department of National Institute of Technical Teachers Training and Research,
Chandigarh, India. She holds a Bachelor degree in Electrical Engineering from Kerala
University, and Masters and Ph.D from Panjab University. She has 28 years of experience
out of which 2 years are of industrial and the rest of teaching. She has guided more than 50
Masters degree theses and has more than 50 articles in her credit. Her areas of
specialization are Digital Signal Processing, Power Systems, ANN and Fuzzy Logic,
Virtual Instrumentation, MATLAB etc.

.

938 Vol. 7, Issue 3, pp. 938-947

A FEED FORWARD ARTIFICIAL NEURAL NETWORK BASED
SYSTEM TO MINIMIZE DOS ATTACK IN WIRELESS
NETWORK
Tapasya Pandit & Anil Dudy
Deptt. of Electronics and Communication Engineering,
Baba Mastnath College of Engineering and Technology, Rohtak, Haryana, India

ABSTRACT
Security has become more and more important in our life according to development. During the past years, the
concepts of security involved considering the process of assessing computer system, network and file, scanning,
analyzing system information from various areas, observing and analyzing both user and system activities to
identify possible security violations which include both intrusions (attacks from outsider) and misuse (attacks
from inside the organization ). A technology that is developed to assess the security of computer systems or
network is one of the most popular types of security management system for computers and networks which is
defined as intrusion detection system.
I NDEX TERMS Artificial Neural Network (ANN), Back Propagation Neural Network, Delay of Service
(DOS), Feed Forward Neural Network, Intrusion Detection System (IDS), Network Security
I. INTRODUCTION
The preservation of security has become more difficult by time because the possible technologies of
attack are becoming more superior. At the same time, less technical ability is required for the novice
snoopier because the verified past methods are easily accessed through the organization. The main
idea of protecting the information through the encrypted channel for data and also confirming the
identity of the connected device through the firewall, which will not accept any connection with a
stranger, firewalls do not provide full protection for the system (Rung-Ching , Kai-Fan and Chia-Fen
,2009). So, it is needed to extend the network security capabilities by complementing with other tools
or intrusion detection system (IDS is not a replacement for either a good antivirus program or
firewall). Since it is technically impossible to create computer systems (Hardware & Software)
without any defect or security failure, intrusion detection in computer systems researches is
specifically regarded as important. IDS is a protective system that can detect disorders occurring on
the network. The procedure goes as intrusion detection can report and control occurred disorders
through steps including collecting data, seeking ports, controlling computers, and finally hacking. So,
intrusion detection can report control intrusion sabotage that composed of phases collecting data,
probing port, gaining computers control and finally hacking. In this paper, we consider some
different agents, each of which can detect one or two DOS attacks. These agents interact in a way not
to interfere each other. Parallelization Technology is used to increase system speed. Since the
designed agents act separately and the result of each agent has no impact on the others, we can run
each system on discrete CPUs (depending on how many CPUs are used in IDS computers) to speed
up the performance. The purpose of an Intrusion Detection system is not to prevent an attack, but
only to discover and possibly detect the attacks and to recognize security problems in system or
computer networks and also to report it to the system administrator[5]. Intrusion Detection systems
are generally used with Firewalls as their security complements. Detection of anomaly outside-in
traffics of the network and reporting it to the administrator, or preventing suspected contacts is the
other feature of IDS. IDS is capable of detecting attacks by both internal and external users [9].
939 Vol. 7, Issue 3, pp. 938-947

Wang et al. have declared their ideas about intrusion detection and have explored different methods
of neural network. There are many different intelligent techniques for designing intrusion detection
systems, such as Machine learning, data mining, and fuzzy sets which are divided into two groups of
Fuzzy set and Fuzzy anomaly detection, to mention some. Neural network algorithms are also divided
into two groups of Supervised Learning and Unsupervised Learning [14].
II. INTRUSION DETECTION SYSTEM
Nowadays, Intrusion detection systems are most original and complete parts of a network monitoring
system. Intrusion detection system technologies relatively are new and promise us that we will do in
order to detect network intrusion that will help. Intrusion detection is the process in which events and
incidents on a system or network monitoring and monitoring of the network or system intrusion is
detected[1] .
The goal of intrusion detection is screening, evaluating and reporting of network activity. This system
acts on the data packets that have passed access control tool. Due to the reliability limitations,
internal threats, and the presence of required doubt and hesitation, the intrusion prevention system
should allow some cases of suspected attacks to pass in order to decrease the probability of false
detections (false positive). On the other hand, most of IDS methods are intellectual and use different
techniques to detect potential attacks, intrusions and abuses. Usually, an IDS uses the bandwidth in a
way that can keep on acting without making any effect on accounting and network architecture.
Intrusion detection systems (IDS), are responsible for identifying and detecting any unauthorized use
of the system, abuse or any damage caused by both internal and external users[4] .
Intrusion detection systems try to detect anomaly intrusions to the network by special algorithms
which can be divided into 3 categories of misuse-based, anomaly-based, and specification-based.
Analyzing the users behavior in the network, the anomaly-based system can find out the intrusions.
In anomaly-based method, an index of normal behavior is created. An abnormality may be an
indication of an intrusion. Indexes of normal behavior are created based on approaches like Neural
Networks, Machine Learning methods, and even life style safety systems. To detect anomalous
behaviors, normal behaviors should be identified and some specific patterns and rules should be
designed for them. The behaviors which follow these patterns are considered as normal and events
which show any deviation beyond the normal statistics of these patterns are detected as abnormal. Its
extremely difficult to detect abnormal intrusions, because there is no consistent pattern to monitor
them. Usually an event which shows more than two deviations from the normal behavior is assumed
to be normal. According to the rapid expansion of networks over the past century, system protection
has become one of the most important issues in Computer Systems due to the existence of gaps in
most of the components of protection systems such as FIREWALL systems. In the last past years,
several research were proposed, developed and designed to set ideas based on several techniques to
design systems intrusion detection to protect the system, analyze and expect the behaviors of users.
Misuse intrusion detection is the process that searches attack patterns in the source of data to identify
instances of network attacks by comparing current activity against the estimated actions of an
intruder. Thus intrusion detection systems (IDS) are used as secondary computer systems protector to
identify and avoid illegal activities or gaps. The intrusion detection problem is considered as a pattern
recognition, and the artificial neural network must be trained to distinguish between normal and
unusual patterns (DoS, Prob., R2L, U2R)[2].
Unfortunately, unusual anomaly-based intrusion detections and IDSs of this kind cause many false
alarms (false positive) due to the fact that the behavior patterns of the users and the system are very
previous attacks), abnormal behavior detection methods can detect any kind of new attacks. In
misuse-based technique, usually known as signature-based detection, pre-designed intrusion
templates (signatures) are stored as law, in a way that each template contains different types of a
specific intrusion and once a template of this kind appears in the system, the intrusion occurring is
alarmed. Usually, in these methods, the detector has data bases of attack signatures or templates and
tries to detect patterns that are similar to those stored in its own data base. This kind of methods are
able to detect known intrusions, and if new attacks appear anywhere in the network, they are not able
to detect them. The administrator should continuously add the templates (patterns) of new attacks to
940 Vol. 7, Issue 3, pp. 938-947

the intrusion detection system. One of the advantages of this method is the high accuracy applied in
detecting intrusions the templates of which have precisely given to the system .
III. DATASET
The Information System Technology Group at Massachusetts Institute of Technology Lincoln
Laboratory, sponsored by Defence Advanced Research Project Agency (DARPA) and Air Force
Research Laboratory, has collected and evaluated the first standard corpora for evaluation of
computer network Intrusion Detection Systems. This is called the DARPA Intrusion Detection
Evaluation[7].
The data sets used in this research are the data sets from the 1998 DARPA Intrusion Detection
Evaluation Program. In the dataset, the following attacks are present according to the actions and
goals of the attacker. Each attack type falls into one of the following four main categories:
A. Probing
Probing is a class of attacks where an attacker scans a network to gather information or find known
vulnerabilities. An attacker with a map of machines and services that are available on a network can
use the information to look for exploits. There are different types of probes: some of them abuse the
computers legitimate features, some of them use social engineering techniques. This class of attacks
is the most commonly heard and requires very little technical expertise. Attacks used was IP
sweep,Mscan, Nmap, Saint and Satan[2].
B. Denial of Service Attacks
Denial of Service (DoS) is a class of attacks where an attacker makes some computing or memory
resource too busy or too full to handle legitimate requests, thus denying legitimate users access to a
machine . There are different ways to launch DoS attacks:
Abusing the computers legitimate features.
Targeting the implementations bugs.
Exploiting the systems misconfigurations.
DoS attacks are classified based on the services that an attacker renders unavailable to legitimate
users. An attack used was Apache2, Back, Mail bomb, Neptune, Ping of death, Process table, Smurf,
Syslogd and UDP storm[2].
C. User to Root Attacks
User to root exploits are a class of attacks where an attacker starts out with access to a normal user
account on the system and is able to exploit vulnerability to gain root access to the system. Most
common exploits in this class of attacks are regular buffer overflows, which are caused by regular
programming mistakes and environment assumptions. Attacks used was Perl and Xterm[2].
Remote to User Attacks
A remote to user (R2L) attack is a class of attacks where an attacker sends packets to a machine over
a network, then exploits machines vulnerability to illegally gain local access as a user. There are
different types of R2L attacks; the most common attack in this class is done using social engineering.
Attacks used was Dictionary, FTP-write, Guest, Imap, Named, Phf, Sendmail, Xlock and Xnsnoop[2].
IV. ARTIFICIAL NEURAL NETWORK
Artificial neural networks born after McCulloc and Pitts introduced a set of simplified neurons in
1943. These neurons were represented as models of biological networks into conceptual components
for circuits that could perform computational tasks. The basic model of the artificial neuron is
founded upon the functionality of the biological neuron[10]. By definition, Neurons are basic
signaling units of the nervous system of a living being in which each neuron is a discrete cell whose
several processes are from its cell body. One can differentiate between two basic types of networks,
networks with feedback and those without it. In networks with feedback, the output values can be
traced back to the input values. However there are networks wherein for every input vector laid on
941 Vol. 7, Issue 3, pp. 938-947

the network, an output vector is calculated and this can be read from the output neurons. There is no
feedback. Hence only, a forward flow of information is present. Network having this structure are
called as feed forward networks. There are various nets that come under the feed forward type of
nets. A multilayer feed forward back propagation network with one layer of z-hidden units. The Y
output unit has Wok bias and Z hidden unit has Vok as bias. It is found that both the output units and
the hidden units have bias. The bias acts like weights on connection from units whose output is
always 1. This network has one input layer, one hidden layer and one output layer. There can be any
number of hidden layers. The input layer is connected to the hidden layer and the hidden layer is
connected to the output layer by means of interconnection weights. The bias is provided for both the
hidden and the output layer, to act upon the net input to be calculated[14].
V. TRAINING ALGORITHM
The training algorithm of back propagation involves four stages[14], viz.
1. Initialization of Weights
2. Feed Forward
3. Back Propagation of errors
4. Updation of the weights and the biases.
During first stage which is the initialization of weights, some small random values are assigned.
During feed forward stage each input unit (Xi) receives an input signal and transmits this signal to
each of the hidden units Z1Zp. Each hidden unit then calculates the activation function and
sends its signal Zj to each output unit. The output unit calculates the activation function to form the
response of the net for the given input pattern. During back propagation of errors, each output unit
compares its computed activation yk with its target value tk to determine the associated error for that
pattern with that unit. Based on the error, the factor k is computed and is used to distribute the error
at output unit yk back to all units in the previous layer. Similarly factor j is computed for each hidden
unit zj..During final stage, the weight and biases are updated using the factor and the activation .
x: input training vector[14]
x: (x1, .xi,., xn)
t: Output target vector
t: (t1, .ti,., tn)
k =error at output unit yk
j =error at hidden unit zj
= learning rate
Voj= bias on hidden unit j
zj= hidden unit j
wok=bias on output unit k
yk= output unit k.
The training algorithm used in the back propagation network is as follows. The algorithm is given
with the various phases:
D. Initialization of Weights
Step 1: Initialize weight to small random values.
Step 2: While stopping condition is false, do Steps 3-10.
Step 3: For each training pair do steps 4-9.
E. Feed Forward
Step 4: Each input unit receives the input signal xi and transmits this signals to all units in the layer
above i.e hidden units.
Step 5: Each hidden unit( zj, j=1,,p) sums its weighted input signals.
z-inj=voj+xivij (1)
applying activation function
Zj=f(zinj) (2)
942 Vol. 7, Issue 3, pp. 938-947

and sends this signal to all units in the layer above i.e. output units.
Step 6: Each output unit (yk) sums its weighted input signals.
y-ink=wok+zjwjk (3)
and applies its activation function to calculate the output signals.
Yk=f(y-ink) (4)
F. Back Propagation of Errors
Step 7: Each output unit receives a target pattern corresponding to an input pattern, error information
term is calculated as
k=(tk-yk)f(y-ink) (5)
Step 8: Each hidden unit (zj) sums its delta inputs from units in the layer above
-inj=jwjk (6)
The error information term is calculated as
j=-injf(z-inj) (7)

G. Updation of Weight and Biases
Step 9: Each output unit (yk) updates its bias and weights (j=0,..,p)
The weight correction term is given by
Wjk=kzj (8)
and the bias correction term is given by
Wok=k (9)
Wjk(new)= Wjk(old)+Wjk, Wok(new)=Wok(old)+Wok (10)
Each hidden unit (zj,j=1,.p) updates its bias and weights (i=0,..n)
The weight correction term
Vij=jxi (11)
The bias correction term
Voj=j (12) Vij(new)=
Vij (old) + Vij , Voj(new)= Voj (old) + Voj (13) Step 10: Test the stopping condition.
The stopping condition may be the minimization of the errors, number of epochs etc.

Fig. 1 Feed Forward Networks

Fig. 2 Back Propagation of Errors
VI. LEARNING
Learning is the process by which the neural network adapts itself to a stimulus and eventually it
produces desired output after making the proper parameter adjustments to itself. It is a continuous
943 Vol. 7, Issue 3, pp. 938-947

classification process of input stimuli; when a stimulus appears at the input nodes of a network it
either recognizes it or it develops a new classification. During the process of learning the synaptic
weights are adjusted in response to the input so that its actual output response converges to the
desired or target output response. When the actual output response is same as the desired response
then the network is said to be learned or acquired knowledge. There are different learning methods
and each learning method is described by a set of equations.
A. Supervised Learning
Training or learning of a neural network is accomplished by presenting a sequence of training vectors
or samples or patterns each with an associated target output vector. The weights are adjusted
according to the learning algorithm. This process is known as supervised learning. Fig. 3 represents
supervised learning process.

Fig. 3 Supervised Learning
B. Unsupervised Learning
In an unsupervised training a sequence of input vectors is provided but no target vectors are
specified. The net modifies the weights so that most similar input vectors are assigned to the same
output unit or cluster unit. It arbitrarily organizes the patterns into categories[14]. Even though
unsupervised learning does not require a teacher or target output it requires guidelines to determine
formation of clusters. Its representation is shown in Fig.4.

Fig. 4 Unsupervised Learning
VII. THE PROPOSED METHOD
In this section, the implementation of the proposed intrusion detection system, implementation steps
and evaluation criteria are described.
944 Vol. 7, Issue 3, pp. 938-947

H. Implementation
The network attacks can be divided in four groups of DoS, R2L, U2R, Probe. In the designed IDS, the
system can detect DoS -type attacks, in very high detection rate. In fact, this kind of IDS is
responsible for the detection attacks, which can be included in DoS category.
In order to design this type of IDS, we identified DoS attacks and designed a separate IDS for each
one to detect that specific attack. In general, considering the designed IDS, the system will detect
DoS attacks in the network (if there is any). The whole process of the system is shown in Figure 5.

Fig. 5.The Process of generating system of intrusion detection
As you see in figure 5, in order to train neural network and have a more qualified process, we made
some changes in database which as you see didnt affect on the totality of data base. It is just done for
improving the function of neural network.
I. Evaluation Criteria
To measure and detect the efficiency of the designed IDSs or the exact degree of their assurance and
correctness the following criteria can be used[2]:
True negative = correctly detect the normal data
True positive = correctly detect the attack
False Positive = distinguish normal events as attacks
False negative = distinguish the incidents of attack as normal
TNR = TN / (TN + FP) = the total number of normal incidents that are correctly detected / the total
number of normal incidents that are detected as normal.
TPR = TP / (TP + FN) = the number of incidents of attack that are correctly detected / the total
number of incidents that are detected as attacks.
FNR = FN / (FN + TP) = the number of attack incidents that are detected as normal / the total number
of incidents that are detected as normal.
FPR = FP / (FP + TN) = the number of normal incidents which are detected as attack / the total
number of incidents which are detected as attack.
In this implementation, we used the TPR criterion and as you see in the chart above, the efficiency of
this implementation is approximately more than 98%.

945 Vol. 7, Issue 3, pp. 938-947

VIII. RESULTS AND DISCUSSION
After executing the Program of Training of Neural Network values of y, W. W0, V, V0 are as under.
Y = 0.7593 0.9990 0.0949 0.9694 0.9234 0.1133 1.0000 1.0000 0.1660 0.8952
0.8883 1.0000 0.6756 1.0000 0.0000 0.0000 0.9569 0.0504
EPOCH = 990000
W =
[ 69.7724
-92.1417
75.1616 ]
WO =
[ 5.7396]
V =
[ -101.2418 -48.9699 238.5258
180.6523 215.8780 -0.1458
-245.0327 -305.1403 9.9378]
VO = [ 0.9382 4.7101 -1.8095
The Graphical representation of Epoch and Error is as shown in Fig. 6 which is plot between Epoch
Number and Error. As we know, when iterations increases error between target value and output
value decreases which is clear from Fig. 6.

Fig.6 Plot between Epoch Number and Error

946 Vol. 7, Issue 3, pp. 938-947

Now, as we know as the iterations increases output value reaches to target value it is clearly shown in
Fig. 7 which is a plot between epoch and output value.

Fig. 7 Plot between Epoch Number and Output Value

Now Using these values of W, W0, V, V0 in the program of Forecasting of Attack or Normal Traffic
by Neural Network value of y forecasted is 1 which is clearly a attack. So this is an excellent
technique of forecasting of attack or normal traffic
IX. FUTURE SCOPE
As mentioned above, there has been a lot of research on intrusion detection, and also on the use of
neural networks in intrusion detection. As showed in this thesis, back propagation neural networks
can be used successfully to detect attacks on a network. The same experiments should also be
conducted with other types of neural networks to see if these types can improve the detection rate we
got from the experiments with a back propagation neural network.
X. LIMITATIONS
As for many studies; there are some different challenges viewed in the intrusion detection systems. In
this study, some limitations were faced. They can be summarized as follows:
1) Intrusion detection systems need a periodic update to the training set and profiles.
2) Using a static training data might become outdated and deficient for prediction.
3) The accuracy of classification for the data do not 100%.
XI. CONCLUSION
There are various techniques of Artificial Neural Network, which can be applied to Intrusion
Detection System. Each technique is suitable for some specific situation. BPNN is easy to implement,
supervised learning artificial neural network. Number of the epochs required to train the network is
high as compare to the other ANN techniques. But, detection rate is very high. BPNN can be used
when one wants to not only detect the attack but also to classify the attack in to specific category so
947 Vol. 7, Issue 3, pp. 938-947

that preventive action can be taken. By combining the different ANN techniques, one can reduce the
number of the epochs required and hence can reduce the training time. As the DOS attack will be
resolved in this work, the throughput of the network will be improved and network delay will be
reduced. The work does not require any additional hardware and is software based. In the future this
system could be extended to an online system by little effort.
REFERENCES
[1] Bhavin Shah, Bhushan H Trivedi, Artificial Neural Network based Intrusion Detection System,
International Journal of Computer Applications Volume 39 No.6, February 2012.
[2] Manoranjan Pradhan, Sateesh Kumar Pradhan, Sudhir Kumar Sahu, Anomaly Detection using Artificial
Neural Network,International Journal of Engineering Sciences & Emerging Technologies, April 2012.
[3] Zahra Moradi1, Mohammad Teshnehlab , Intrusion Detection Model in MANETs using ANNs and
ANFIS, 2011 International Conference on Telecommunication Technology and Applications, Singapore
[4] Mehdi MORADI and Mohammad ZULKERNINE, A Neural Network Based Ssytem for Intrusion
Detection and Classification of Attacks.
[5] Przemysaw Kukieka, Zbigniew Kotulski, Adaptation of the neural network- based IDS to new attacks
detection.
[6] M. Dondo and J. Treurniet, Investigation of a Neural Network Implementation of a TCP packet Anomaly
Detection System, Defence Research and Development Canada, May 2004.
[7] V.Sivakumar1,T.Yoganandh,R.Mohan Das, Preventing Network From Intrusive Attack Using Artificial
Neural Networks, International Journal of Engineering Research and Applications (IJERA), Vol. 2, Issue
2,Mar-Apr 2012, pp.370-373.
[8] Samaneh Rastegari, M. Iqbal Saripan and Mohd Fadlee A. Rasid, Detection of Denial of Service Attacks
against Domain Name System Using Neural Networks, IJCSI International Journal of Computer Science
Issues, Vol. 6, No. 1, 2009.
[9] S. Devaraju, S. Ramakrishnan, Detection of Accuracy for Intrusion Detection System using Neural
Network Classifier, International Journal of Emerging Technology and Advanced Engineering (IJETAE).
[10] Afrah Nazir, A Comparative Study of different Artficial Neural Networks based Intrusion Detection
Systems International Journal of Scientific and Research Publications, Volume 3, Issue 7, July 2013.
[11] Sudhakar Parate, S. M Nirkhi, R.V Dharaskar, Application of Neural Forensics for detection of Web
Attack using Neural Network, National Conference on Innovative Paradigms in Engineering and
Technology(NCIPET-2013).
[12] Przemysaw Kukieka, Zbigniew Kotulski, Analysis of Neural Networks usage for detection of a new attack
in IDS, Annales UMCS Informatica AI X, 1 (2010) 51-59.
[13] Tariq Ahamad and Abdullah Aljumah, Hybrid Approach using intrusion Detection System, International
Journal of Computer Networks and Communications Security, VOL. 2, NO. 2, FEBRUARY 2014, 8792.
[14] Amit Garg and Ravindra Pratap Singh, Voltage Profile Analysis in Power Transmission System based on
STATCOM using Artificial Neural Network in MATLAB/SIMULINK, International Journal of Applied
Information Systems(IJAIS), Foundation of Computer Science, New York, USA, Volume 6- No. 1,
September 2013.

AUTHORS BIOGRAPHY
Tapasya Pandit has completed her B-Tech in Electronics and Communication Engineering from BPR college
of Engineering, Gohana, Haryana, India and now pursuing her M-Tech in Electronics and Communication
Engineering from Baba Mastnath College of Engineering and Technology, Rohtak, Haryana, India. Her interests
include Intrusion Detection System and Artificial Neural Network.

Anil Dudy is working as Assistant Professor in Electronics and Communication Engineering Department in
Baba Mastnath College of Engineering and Technology, Rohtak, Haryana, India and also pursuing Ph.D in
Electronics and Communication Engineering from Baba Mast Nath University, Rohtak, Haryana, India.

948 Vol. 7, Issue 3, pp. 948-956

IMPROVING PERFORMANCE OF DELAY AWARE DATA
COLLECTION USING SLEEP AND WAKE UP APPROACH IN
WIRELESS SENSOR NETWORK
1
Paralkar S. S. and
2
B. M. Patil
1
P.G. Dept., M. B. E. Societys College of Engineering. Ambajogai, Maharashtra, India
2
M. B. E. Societys College of Engineering, Ambajogai, Maharashtra, India

ABSTRACT
Even though there are many advancement and development in wireless sensor network, it experiences frequent
wireless sensor network are battery-powered devices. Hence for lifetime of wireless sensor network energy
saving is rather crucial one. The objective of the current work is sleep and wake up method in which nodes can save
energy when it is sleep. When node wake up it sending or receiving data to or from another node. Sub cluster head
method which save average energy and average delay of network. In this paper simulation result shows that sub
cluster head method using sleep and wake up method which shows that network require minimum average
energy as well as average delay from all other networks.
KEYWORDS: Sleep and Wake up Approach, Wireless sensor Network.
I. INTRODUCTION
Wireless sensor networks consist of independent number of nodes, in nature these nodes are low cost,
small in size, low weight and battery-powered devices. Nowadays WSNs are used in different
locations for data transmission and data retrieve purpose. These thousand nodes are working
simultaneously for specific operation. In the sense that improving technologies now days WSNs
experiences to reduce delay and energy to increase network lifetime.
A wireless sensor network (WSN) used in different conditions such as temperature, sound, vibration,
pressure, pollutants and to send the collected data through the network to a centred location of the
network.
A wireless sensor network classified into flat, hierarchical, location based network structure. These
further classified into multipath-based, query-based, negotiation-based, QoS-based, and coherent
based depending on the protocol operation [1].
The current work concentrates on improving energy and delay required for the network. It focuses
when the data send from node to node it require energy otherwise it is to be in sleep mode which
saves energy of network. The delay and energy is minimized by using the method. This technique
improves energy and delay require for network in the wireless sensor network.
1.1 Design issues of a wireless sensor network
In wireless sensor network there are designs issues consider for configuring network. In that some
design issues are as follows:-
Fault Tolerance: Node failure problem is common problem in WNS. In the sense that these nodes
can fail due to hardware problems or physical damage. The protocols which is designed for the
network should be able to detect these failures as soon as possible and to be robust enough to
handle a relatively number of node failures while it maintaining functionality of the network. The
949 Vol. 7, Issue 3, pp. 948-956

routing protocol designed in the sense that alternate paths are also available for rerouting of the
packets in network.
Scalability: Sensor networks are vary in scalable. The protocols are deployed in the networks to
maintain the performance.
Sensor Network Topology: WSNs can be evolved in many aspects, in that they are continue to be
networks with constrained resources in terms of energy, computing power, memory, and
communications capabilities. The network Topology Maintenance is to reduce energy
consumption in wireless sensor networks.
Power Consumption: In WNS size of the nodes limits can define the size of the battery. The
design of software and hardware it needs to carefully consider for the issues of energy use. The
required energy depends on the application; in some applications, nodes are to turn off when its
operation is over in network in order to conserve energy while in other applications it requires all
nodes operating continuously. In wireless sensor networks for wireless communications a lots of
the energy in a node is to be taken for network. For long transmission distance between number of
nodes and the base station is not to be considered.
In Heinzelmanet al. [2] have proposed LEACH a clustering algorithm. In networks using LEACH, In
LEACH it is a cluster-based protocol it select cluster head which collect data from nodes and send to
the base station. LEACH uses a TDMA/CDMA MAC to reduce collisions. In Lindsey and
Raghavendra [3] introduced algorithm PEGASIS, In PEGASIS, each node communicates only with a
nearest neighbor and takes turns transmitting to the base station, so that it reducing the energy
required for each round. In Tan and Krpeoglu [4] have proposed PEDAP which is define a
minimum spanning tree. In Fonseca et al. [5] have proposed the collection tree protocol . In A.
Manjeshwar and D. P. Agarwal [6,7] have proposed Threshold-Sensitive Energy Efficient Sensor
Network Protocol and APTEEN. TEEN protocol for reactive network. A CH sensor sends its
members a hard threshold and soft threshold value. APTEEN is a hybrid routing protocol. In Chi-
Tsun Cheng and Francis C. M. Lau [8] have proposed top down approach and bottom up approach. A
network consists of several clusters. In each cluster, one of the sensor nodes is work as a cluster
head(CH) and while other are cluster members (CM). In long distance transmission wireless sensor
nodes are involved in manner that energy required for network is reduced.
Top down approach: The top-down approach is a centralized control algorithm. The base station
is collect data from all sensor nodes in the network. After that the base station is then performing
further operation.
Bottom-Up Approach: The bottom-up approach is to join clusters of the same size together. It can
be implemented in either centralized or decentralized fashion.
In G. Lu, N. Sadagopan, B. Krishnamachari, and A. Goel [9] have proposed Sleep scheduling method
in which how to network life time increases and minimize end to end delay in network. In Marjan
Baghaie, and Bhaskar Krishnamachari [10] have proposed Delay Constrained Minimum Energy
Broadcast method. In this cooperative algorithm used for calculating delay and power efficiency and
after that compare the cooperative algorithm with smart non cooperative algorithm.
In P.M.Lokhande, A.P.Thakare [11] have proposed method anycast forwarding schemes to forward
the data packet to next hop node which minimizes the expected packet-delivery delays from the
sensor nodes to the sink node. In Abhay Raman et al. [12] have proposed the sleep-wake scheduling
protocol and the any cast packet-forwarding protocol not only to maximize the network lifetime but
also expected end to end packet-delivery delay. In Guofang Nan et al. [13] have proposed a coverage-
guaranteed distributed sleep/wake scheduling scheme. In this CDSWS network lifetime as well as
network coverage both considered and more than one node in cluster is active by using dynamic node
selection method. In Chih-Min Chao et al. [14] have proposed quorum-based MAC protocol that
enables sensor nodes to sleep longer under light loads which leads to save energy and transmission
latency minimum.
In Bo Jiang et al.[15] have proposed SSMTT method to support multiple target tracking sensor
networks which save the energy on proactive wake-up communication. In B. Chen, et al. [17] have
proposed Span to save the energy of the network increase system lifetime. Span is a distributed,
950 Vol. 7, Issue 3, pp. 948-956

randomized algorithm in which nodes make decisions to sleep or to join a forwarding backbone as a
coordinator. In Abtin Keshavarzian et al. [18] have proposed an algorithm which consist of multi-
parent schemes tries to maintain the longer network lifetime while satisfying the latency constraints.
In Yuanyuan Zhou, Muralidhar Medidi [19] have proposed method in which is not only reduce the
end-to-end delay but also extend the network lifetime. In Arjan Durresi, Vamsi Paruchuri, Leonard
Barolli[20] have proposed DEAP method, DEAP is enables a flexible range of trade offs between the
packet delay and the energy use. DEAP improves the packet delay and system lifetime.
In B. Liang, J. Frolik, and X. S. Wang [21] have proposed the method in which predictive QoS
control strategy for wireless sensor networks. In Giuseppe Anastasi et al.[22] have proposed an
Adaptive Staggered sleep Protocol (ASLEEP) for efficient power management in WNS targeted to
periodic data acquisition in which dynamically adjusting the sleep scheduling of node without
affecting on the network. This protocol reduces the energy required for nodes so that network lifetime
increases. In VamsiParuchuri et al. [23] have proposed Random Asynchronous Wakeup(RAW), a
power saving technique for sensor networks that reduces energy consumption without significantly
affecting the latency or connectivity of the network. In which node local decides to sleep or to be
active.
In Shunfu Jin et al. [24] have proposed a sleep/wakeup protocol is introduced in IEEE 802.15.4,
handover ratio, cost function for minimizing the energy consumption of the sensor node. In S.Kavitha,
S.Lalitha[25] have introduced a protocol to make local monitoring parsimonious in its energy
consumption and to integrate it with any extant sleep-wake protocol in the network.InP. kaur ,A.
Nayyar [26] have proposed an energy efficient dynamic power management technique which shuts
down the sensor node when there is no work and wake them up when necessary which yields better
savings of energy and enhance lifetime. In Zhihui Chen, AshfaqKhokhar [27] have proposed protocol
is based on Time Division Multiple Access (TDMA) principle and are referred to as TDMA in which
a sensor utilizes its assigned slot only when it is sending or receiving information, otherwise its
receiver and transmitter are turned off .
In N. Shrestha et al. [28] have proposed SWAP to reduce the packet latency of delay sensitive
packets. To evaluate the energy efficiency and performance of the network. An active periods
overlap at least once within a cycle of the sleep and wake-up slots between two neighbour node swap
scheduling scheme is used to communicate between those two nodes. In A. Sharma et al. [29] have
proposed an energy measurement system based on a node current consumption usage. In Y. S. Bae
[30] have proposed a new RF wakeup sensor, which is a dedicated small RF module to check
potential communications by sensing the presence of a RF signal. With RF wakeup sensor each node
no longer requires duty cycling, eliminating both sleep delay and idle listening.
The rest of the paper is organized in the following manner; section 2 describes proposed method,
section 3 presents simulation and analysis, section 4 shows experimental results and finally the
conclusion for new proposed scheme.
III. SLEEP/WAKE UP METHOD
The source initiates a route discovery by broadcasting the RREQ packet then route get RREP from
nodes if source get RREP then source send data to the node. if source not get RREP then send RREQ
all other nodes. When nodes are ready to send/receive the data then it wake up otherwise node are in
sleep mode then it takes minimum energy required for the network. Cluster members can send data to
the cluster head which are near to them using sleep and wake up method. The cluster head forward
data to next level.
IV. SIMULATION AND ANALYSIS
All the work done in this paper has been implemented and validated in NS-2.34[16] and experimented
on Linux Red hat as Operating System, Network Simulator and NAM, AWK, XGRAPH.
4.1 Method and Implementation
951 Vol. 7, Issue 3, pp. 948-956

As stated in Simulation and Analysis section, NS-2.34 and installed as extension class in NS. The
Topdown approach and bottom-up approach, SCHP and SCHP(SW) is further included in this
simulation for calculating average energy and average delay.
4.2 Work and Analysis
All the work done in this paper mainly focuses on sleep and wake up method using SCHP proposed
method in NS2, for which some value have set for set of parameters as shown in below table

Table 1. Simulation Parameters
Simulation Parameter Value
Simulation NS-2.34
Area 500*500 m
Number of Nodes 30
MAC 802.11
Queue type Queue/DropTail/PriQueue
Initial energy 90 joules
Queue length 200
4.3 Algorithm
Step 1:
Source sends RREQ to all the nodes.
Step 2:
if RREP send from node to start node then start node send data
Else RREQ update and retransmit to all other nodes.
For data transmission
Step 3:
If source equal to data received
Step 4:
If data received is not equal to RREQ then
Set flag 0
Step 5:
If flag 0 equal to 0 then Drop RREQ else
Transmit RREQ else if
Destination is equal to data received
Step 6:
If node_id RREQ is equal to node _id RREP
Set initialize RREP
For sleep RREP
Step 7:
If RREP is equal to received
Step 8:
If node_id received not equal to RREP
Set flag 1 else
Transmit RREP acknowledgement and data
Drop RREP else
Set flag 1
Step 9:
If node received not equal to source then
Transmit new and transmit RREP
Set flag 1 else
Set count if count is equal to 0
Transmit else
Stop RREP
Algorithm 1. Sleep and wake up method.
952 Vol. 7, Issue 3, pp. 948-956

4.4 Procedure for subcluster head
1. Set cluster head and normal head
2. For set I equal to 0 then compare the value of i with value (nn) and increment the value of i &
for j equal to 0 then compare the value of j with value (nn) and increment the value of j and
set flag 0.
3. For l equal to 0 then l compare with count nodes then increment the value of l.
4. If j equal to normal nodes(l) then Set flag value equal to 0 and If flag not equal to 1 then
set expressions for temp1,temp2,temp3,temp4 as follows:
temp1 [expr c1 * r1 *[expr $clusterhead($j) - X1(j) ] (1)
for temp 2 the expression is defined as
temp2 [expr c2 * r2 *[expr $cluster_head3 - X1($j) ]] (2)
temp3 [expr 1 * r1 *[expr cluster_head2(j) - X2(j) ] (3)
temp4 [expr$c2 * r2 *[expr cluster_head4 - X2(j) ] (4)
5. if the value of xid1(j) is less than 0 or xid2(j) is less than 0 then set the value of xid1(j) as
follows:
set the value xid1(j) is equal to 1 (5)
6. if the value of xid1(j)is greater than 90 or the value of xid2(j) greater than 90) then set the
value of xid1(j) as follows:
set the value xid1(j)is equal to 90 (7)
7. if the value of xid1(j) is equal to X1(j) and the value of xid2(j) is equal to X2(j) then
set count [expr count + 1] and set the value X1(j) xid1(j) and set the value X2(j) xid2(j)
8. for the value of set k is equal to 0 then k is less than anchor and increament the value of k.
set the value
disx [expr CLUSTER_X1(k) - X1(j)] (9)
disy [expr CLUSTER_X2(k) - X2(j)] (10)
9. set disxsq, disysq, distsq, dist;
disxsq [expr disx * disx] (11)
disysq [expr disy * disy] (12)
distsq [expr disxsq + disysq] (13)
dist [expr sqrt(distsq)] (14)
10. if dist is less than equal to transrange then set disxsq, disysq, distsq, dist;
disx [expr initposX1(i) - X1(i)] (15)
disy [expr initposX2(i) - X2(i)] (16)
disxsq [expr disx * disx] (17)
disysq [expr disy * disy] (18)
distsq [expr disxsq + disysq] (19)
dist [expr sqrt(distsq)] (20)
Est [expr Est + dist ] (21)
normalnodes(countnodes) j and print the cluster head: normalnodes(countnodes) then
countnodes [expr countnodes + 1]
V. SIMULATION AND RESULTS
Following figure shows that the simulation results for energy spent and an average delay for nodes. In
simulation total 30 nodes were considered.
953 Vol. 7, Issue 3, pp. 948-956

Figure 1. Calculating average energy spent between top down, bottom-up and schp
The energy spent for schp compare to topdown and bottom-up as shown in Figure 1. The X-axis
represents average energy and Y-axis represents nodes. Average energy spent by 30 nodes in schp is
requiring less than top down and bottom-up approach. As nodes increases average energy require for
nodes also increases. Schp takes minimum average energy as shown in figure 1. Low average energy
is better.
Figure 2. Calculating average delay between top down,bottom-up,schp
The average delay required for schp compare to topdown and bottom-up as shown in Figure 2. The X-
axis represents average delay and Y-axis represents nodes. As above figure shows that 30 nodes takes
average delay in schp is minimum as compare to topdown and bottom-up methods. Schp takes
minimum average delay as shown in figure 2. Low average delay is better.

Figure 3: Calculating average energy between topdown, bottom-up, schp, schp(sw)
954 Vol. 7, Issue 3, pp. 948-956

The energy spent for schp(sw) compare to topdown and bottom-up, schp as shown in Figure 3. The X-
axis represents average energy and Y-axis represents nodes. In figure 30 nodes require average energy
calculating using sleep and wake-up method takes minimum as compare to topdown, bottom-up and
schp. In figure 3 shows that sleep and wake up method require average energy is minimum. Nodes
increases average energy also increases as shown in above figure. Low average energy is better.
The average delay for schp(sw) compare to topdown and bottom-up, schp as shown in Figure 4. The
X-axis represents average delay and Y-axis represents nodes. In figure 30 nodes require average delay
nodes increases average delay also increases. Sleep and wake-up method shows that network takes
minimum average delay than topdown, bottom-up and schp method as shown in figure 4. Low
average delay is better.

Figure 4: Calculating average delay between topdown, bottom-up, schp, schp(sw)
Figure 5. Calculating average energy spent between schp, schp(sw)
The energy spent for schp(sw) compare to schp as shown in Figure 5. The X-axis represents average
energy and Y-axis represents nodes. In figure 30 nodes require average energy sleep and wake-up
method require minimum as compare to schp method. Number of nodes increases average energy also
increases as shown in figure 5. Low average energy is better.
The average delay for schp(sw) compare to schp as shown in Figure 6. The X-axis represents average
delay and Y-axis represents nodes. In figure 6 shows that 30 nodes takes average delay sleep and
wake-up method takes minimum average delay than schp method. Low average delay is better as
shown in figure.

955 Vol. 7, Issue 3, pp. 948-956

Figure 6. Calculating average delay between schp, schp(sw)
Subclusterhead method which saves average energy and minimize average delay. In this paper helps
sleep and wake up method in WSN to minimize the average delay for increasing the performance of
the network and save the average energy which is minimum require for the network as compare to
subclusterhead method, topdown and bottom-up methods. In this paper sleep and wake up method
shows that minimum average energy and average delay required for network for increase the network
lifetime. The values derived by objective method can be tuned according to different methods
implementation and application.
REFERENCES
[1] J. N. Al-Karaki, the Hashemite universityA. E. Kamal, Iowa State UniversityRouting Techniques In
Wireless sensor network: a survay IEEE Wireless Communications December 2004.
[2] W. B. Heinzelman, A. P. Chandrakasan, and H. Balakrishnan, An application- specific protocol
architecture for wireless microsensor networks, IEEE Trans. Wireless Commun., vol. 1, no. 4, pp. 660
670, Oct. 2002.
[3] S. Lindsey and C. S. Raghavendra, PEGASIS: Power-efficient gathering in sensor information systems,
in Proc. IEEE Conf. Aerosp., Big Sky, MT, USA, vol. 3, pp. 11251130, ,Mar. 2002.
[4] H. . Tan and . Krpeoglu, Power efficient data gathering and aggregation in wireless sensor networks,
ACM SIGMOD Record, vol. 32, no. 4, pp. 6671, Dec. 2003.
[5] R. Fonseca, O. Gnawali, K. Jamieson, S. Kim, P. Levis, and A. Woo, The collection tree protocol, TinyOS
Enhancement Proposals (TEP), vol. 123, Dec. 2007.
[6] A. Manjeshwar and D. P. Agarwal, TEEN: a Routing Protocol for Enhanced Efficiency in Wireless Sensor
Networks, 1st Intl. Wksp. on Parallel and Distrib. Comp. Issues in Wireless Networks and Mobile Comp.,
April 2001.
[7] A. Manjeshwar and D. P. Agarwal, APTEEN: A HybridProtocol for Efficient Routing and Comprehensive
Information Retrieval in Wireless Sensor Networks, Proc. Intl. Parallel and Distrib. Proc. Symp., pp.
195202 Proceedings of the International Parallel and Distributed Processing Symposium (IPDPS.02)2002.
[8] C.T. Cheng, Member, IEEE, C. K. Tse, Fellow, IEEE, and F. C. M. Lau, Senior Member, IEEE A Delay-
Aware Data Collection Network Structure for Wireless Sensor Networks IEEE Sensor Journal, Vol. 11,
No. 3, March 2011.
[9] G. Lu, N. Sadagopan, B. Krishnamachari, and A. Goel, Delay Efficient Sleep Scheduling in Wireless
Sensor Networks, Proc. 24
th
IEEE Intl Conf. Computer Comm., pp. 2470-2481, Mar. 2005.
[10] M. Baghaie, and B. Krishnamachari, Delay Constrained Minimum Energy Broadcast in Cooperative
Wireless Networks, IEEE International Conference on Computer Communication. 2008.
[11] P.M.Lokhande, A.P.Thakare An Efficient Scheduling Technique for the Improvement of WSN with
Network Lifetime & Delay Constraint International Journal of Recent Technology and Engineering
(IJRTE) ISSN: 2277-3878, Volume-2, Issue-1, March 2013.
[12]A.Raman, A. Kr. Singh, A. Rai Mininize delay and maximize lifetime for wireless sensor network with
anycastInternational Journal of Communication and Computer Technologies Volume 01 No.26, Issue: 04
April 2013
956 Vol. 7, Issue 3, pp. 948-956

[13] G. Nan, G. Shi, Z. Mao and M. Li CDSWS: coverage-guaranteed distributed sleep/ wake scheduling for
wireless sensor networks, EURASIP Journal on Wireless Communications and Networking 2012.
[14] C.M. Chao1 and Y.W. Lee, A Quorum-Based Energy Saving MAC Protocol Design for Wireless Sensor
Networks, IEEE, 2009.
[15] B. Jiang, B. Ravindran and H. Cho Energy Efficient Sleep Scheduling in Sensor Networks for Multiple
Target Tracking,2009.
[16]http://nsnam.isi.edu/nsnam/index.php/Downloading_and_installing_ns-2.
[17] B. Chen, K. Jamieson, H. Balakrishnan, and R. Morris, \Span: An energye _cient coordination algorithm
for topology maintenance in ad hoc wireless networks," in Mobile Computing and Networking, pp. 85-96,
2001.
[18] A. Keshavarzian, H. Lee, L. Venkatraman Wakeup Scheduling in Wireless Sensor Networks Proc.
Seventh ACM Intl Conf. Mobile Ad Hoc Networking and Computing, pp. 322-333, May 2006.
[19] Y. Zhou, M. Medidi Sleep-based Topology Control for Wakeup Scheduling in Wireless Sensor Networks
Washington State University, Pullman, WA 99163 USA
[20] A. Durresi, V. Paruchuri, L. Barolli Delay-Energy Aware Routing Protocol for Sensor and Actor
Networks Proceedings of the 2005 11th International Conference on Parallel and Distributed Systems
(ICPADS'05)2005.
[21] B. Liang, J. Frolik, and X. S. Wang, A predictive QoS control strategy for wireless sensor networks," in
Second IEEE International Conference on Mobile Ad Hoc and Sensor Systems (MASS), November 2005.
[22] G. Anastasi, M. Conti, M. D. Francesco Extending the Lifetime of Wireless Sensor Networks through
Adaptive Sleep IEEE Transactions on IndustrustrialInformatic.
[23] V. Paruchuri, S. Basavaraju, A. Durresi, R. Kannan and S.S. Iyengar Random Asynchronous Wakeup
Protocol for Sensor Networks Louisiana State University Department of Computer Science Baton Rouge,
LA 70803, USA Proceedings of the First International Conference on Broadband Networks
(BROADNETS04)2004.
[24] S. Jin, W. Yue and Q. Sun Performance analysis of the sleep/wakeup protocol in a wireless sensor
network ICIC International c 2012 ISSN 1349-4198 Volume 8, Number 5(B), pp. 3833-3844, May 2012.
[25]S.Kavitha, S.Lalitha Sleep scheduling for critical event monitoring in wireless sensor International
Journal of Advanced Research in Computer and Communication Engineering Vol. 3, Issue 1, January 2014.
[26]P. kaur ,A. Nayyar Conceptual representation and Survey of Dynamic Power Management (DPM) in
Wireless Sensor Network International Journal of Advanced Research in Computer Science and Software
Engineering Volume 3, Issue 3, March 2013.
[27] Z. Chen, A. Khokhar Self Organization and Energy Efficient TDMA MAC Protocol by Wake Up For
Wireless Sensor Networks IEEE University of Illinois at Chicago 851 S Morgan St., Chicago, IL, USA
2004.
[28] N. Shrestha, J. H. Youn, and N. Sharma A Code-Based Sleep and Wakeup Scheduling Protocol for Low
Duty Cycle Sensor Networks Journal of Advances in Computer Networks, Vol. 2, No. 3, September 2014.
[29] A. Sharma, K. Shinghal, N. Srivastava, R. Singh Energy Management for Wireless Sensor Network
Nodes International Journal of Advances in Engineering & Technology, Vol. 1, Mar 2011.
[30] Y. S. Bae, S. H. Lee, B. J. Park and L. Choi RF Wakeup Sensor for Wireless Sensor Networks
International Journal of Multimedia and Ubiquitous Engineering Vol. 7, No. 2, April, 2012.
AUTHORS BIOGRAPHY
Paralkar S.S. he is completed his Bachelors degree in Computer Science & Engineering
Department from BMIT, Solapur under solapur university, Solapur, India. he is pursuing his
Masters Degree from the College of Engineering, Ambajogai. His areas of research interest
include Computer Networks &Wireless Sensor Network.

B.M. Patil is currently working as a Professor in P.G. Computer Science & Engineering
Department in M.B.E.Societys College of Engineering, Ambajogai, India. He received his
Bachelors degree in Computer Engineering from Gulbarga University in 1993,MTech
Software Engineering from Mysore University in 1999, and PhD Degree from Indian
Institute of Technology, Roorkee, 2011. He has authored several papers in various
international journals and conferences of repute. His current research interests include data
mining, medical decision support systems, intrusion detection, cloud computing, artificial
intelligence, artificial neural network, wireless network and network security.

957 Vol. 7, Issue 3, pp. 957-966

IMPROVED NEW VISUAL CRYPTOGRAPHIC SCHEME USING
ONE SHARED IMAGE
Gowramma B.H
1
, Shyla M.G
1
, Vivekananda
2

1
PG Student,
2
Assistant Professor
Department of Computer Science and Engineering,

ABSTRACT
Visual cryptography scheme is a cryptographic technique which allows visual information (e.g. printed text,
handwritten notes, and picture) to be encrypted in such a way that the decryption can be performed by the
human visual system, without the aid of computers. The Improved Visual Cryptography Scheme (VCS) is differs
from any existing visual cryptography schemes; this scheme encodes one binary secret image into one shared
image with innocent-looking. The secret image is visually revealed by superimposing the portion of the share
and its other portion together. The Improved VCS has the advantages of resisting geometry distortion, easy
recognition and management, the size of the revealed image is same size as that of original secret image.
KEYWORDS: Visual cryptography scheme, geometric distortion.
I. INTRODUCTION
Digital information and data are transmitted more often over the Internet now than ever before. The
availability and efficiency of global computer networks for the communication of digital information
and data have accelerated the popularity of digital media. Digital images, video, and audio have been
revolutionized in the way they can be captured, stored, transmitted, and manipulated, and this gives
rise to a wide range of applications in education, entertainment, media and military, as well as other
fields. Computers and networking facilities are becoming less expensive and more widespread.
Creative approaches to storing, accessing and distributing data have generated many benefits for the
digital multimedia field, mainly due to properties such as distortion-free transmission, compact
storage and easy editing.
Visual Cryptography Scheme (VCS) was introduced by Naor and Shamir in 1995. The basic concept
of the conventional (k, n)-threshold VCS[1] is that one binary secret image is encoded into n random-
looking images called shares or shadows which are then distributed to n corresponding participants.
Any k or more participants print their shares on transparencies and superimpose the transparencies
together; the secret image is visually reconstructed. However, any k-1 or fewer participants cannot get
any information about the secret image.
Many visual cryptography schemes (VCS)[2][3][4] have been proposed to meet different
requirements. Various k-out- of-n VCS were also proposed by different researchers to support
variable threshold numbers, in which the secret image is split into n share images and the secret can
be retrieved by the cooperation of at least k of them. If the number of shares stacked is less than k, the
original image is not revealed. The other schemes are 2-outof- n and n-out-of-n VCS. In the 2-out-of-
n scheme n shares will be produced to encrypt an image, and any two shares must be stacked to
decrypt the image. In the n-out-of-n scheme, n shares will be produced to encrypt an image, and n
shares must be stacked to decrypt the image. If the number of shares stacked is less than n, the
original image is not revealed. Increasing the number of shares or participants will automatically
958 Vol. 7, Issue 3, pp. 957-966

increase the level of security of the encrypted message. The researchers also extended visual
cryptography schemes to support grayscale images and color images.
A novel EVCS [5] using one shared image encodes one secret image into one innocent-looking share.
The secret image is visually revealed by copying the share, shifting the copy and superimposing the
shifted copy and the original share together.
In this paper, an improved VCS using one shared image is proposed. The proposed scheme encodes
one secret image into one innocent-looking share. The secret image is visually revealed by
superimposing the portion of the share and its other portion together. The self-decrypt characteristic
of the proposed scheme make it more robust against geometry distortion, while conventional VCSs
and EVCSs are not.
In section 2, we discuss the related literature with respect to visual cryptography schemes. The section
3 discusses about general access structure. The section 4 discusses about the proposed scheme. The
section 5 gives results and discussions of the proposed scheme. The section 6 gives conclusion and
future enhancement of the work.
Moni Naor and Adi Shamir [1], introduces a new type of cryptographic scheme, which can decode
concealed images without any cryptographic computations. The scheme is perfectly secure and very
easy to implement. They extend it into a visual variant of the k out of n secret sharing problem, in
which a dealer provides a transparency to each one of the n users; any k of them can see the image by
stacking their transparencies, but any k - 1 of them gain no information about it.
The main results of [1] includes practical implementations of a k out of n visual secret sharing scheme
for small values of k and n, as well as efficient asymptotic constructions which can be proven optimal
within certain classes of schemes.
P. Tsai and M. Wang [3], proposes an improved (3, 3)-visual secret sharing scheme, which can be
used to embed three secret messages into three shares and improve security. First of all, the fist main
share image is resulted randomly and other two share images are based on the first share image and
the two coding tables designed by them. In the conventional (3, 3)-visual secret sharing scheme, it is
usually to embed one confidential messages, the proposed conventional (3, 3)-visual secret sharing
scheme has been extended to encrypt three secret images. It is also provides increased security.
Rijmen and Preneel [11] have proposed a visual cryptography approach for color images. In their
approach, each pixel of the color secret image is expanded into a 22 block to form two sharing
images. Each 22 block on the sharing image is filled with red, green, blue and white (transparent),
respectively, and hence no clue about the secret image can be identified from any one of these two
shares alone. Rijman and Preneel claimed that there would be 24 possible combinations according to
the permutation of the four colors. Because human eyes cannot detect the color of a very tiny
subpixel, the four-pixel colors will be treated as an average color. When stacking the corresponding
blocks of the two shares, there would be 242 variations of the resultant color for forming a color
image.
L. Duo and D. Yi-Qi [4], they proposed a new scheme on hiding a secret image into single shadow
image. It can also be used as a technique of digital watermarking. Differ from any existed methods of
visual hiding schemes using one secret image, the new scheme based on the rotation of the shadow
image. In this scheme, the single shadow image acted both encoder and decoder function. Therefore
the secret image can be recovered by stacking the shadow image and the shadow image after rotated
90 degree anticlockwise. The new method has the property of anti-compression, anti- distortion and
anti-shrink. Furthermore, the new scheme can use the shadow image sufficiently.
Xiaotian Wu and Wei Sun [5], they proposed a novel extended visual cryptography scheme using one
shared image. Differ from any existing visual cryptography scheme, the proposed scheme encodes
one secret image into one shared image with innocent-looking. The secret image is visually revealed
by copying the share, shifting the copy and superimposing the shifted copy and the original share
together. Extended visual cryptography scheme using one shared image has the advantage of resisting
geometry distortion. Comparing to the two existing one share based VCSs [4], the share generated by
the proposed scheme is easy to recognize and manage. The innocent-looking of the share make it
draw less attention from attackers.
959 Vol. 7, Issue 3, pp. 957-966

III. GENERAL ACCESS STRUCTURE
An access structure is a rule [6], which defines how to share a secret. The most familiar examples are
(n, n) and (t, n) threshold access structures. A (t, n) threshold access structure rules that any t or more
out of n participants can cooperate to reveal the secret image and any less than t participants together
get nothing about the secret image. Obviously, a (n, n) threshold access structure is one instance of the
(t, n) threshold access structure. It demands all participants to cooperate for a secret recovery and
hence nothing can be seen even if only one attendee is absent. It is easy to know that a (t, n) threshold
access structure is tolerant because the final secret still can be restored from the other t shares even
though up to (n t) shares are corrupted.
However, threshold access structure is only one special case of the so-called general access structure.
Usually, a general access structure is denoted as = {A0, A1}, where A0 and A1 are sets of subsets
of all participants and A0A1=. Furthermore, A0 denotes a collection of forbidden sets and A1
denotes a collection of qualified sets. It is easily known that stacking all the shares held by the
participants of any qualified set can recover the secret image; but stacking all the shares held by the
participants of any forbidden set cannot reveal any information about the secret image. For example,
in a system with four participants, we let A1 = {{1, 2}, {2, 3}, {3, 4}, {1, 2, 3}, {1, 2, 4}, {1, 3, 4},
{2, 3, 4}, {1, 2, 3, 4}}, which implies that A0 = {{1}, {2}, {3}, {4}, {1, 3}, {1, 4}, {2, 4}}.
Therefore, we can learn that stacking share 1 and share 2 can recover the secret image; however,
stacking share 1 and share 4 can reveal nothing about the secret image.
It is easy to know that a general access structure should follow the monotone property: if A1 and
, then A1; if A0 and , then A0. So we can learn that the fact {1, 2} A1 implies
it is truly sure that {1, 2, 3} A1, {1, 2, 4} A1 and {1, 2, 3, 4} A1; the fact {1, 4} A0 implies
that {1} A0 and {4} A0. Furthermore, we also can let A1 = {{1, 2}, {2, 3}, {3, 4}} and A0+ =
{{1, 3}, {1, 4}, {2, 4}} to represent above mentioned A1 and A0 respectively in terms of the
monotone property. In fact, A1 is usually named the family of minimal qualified sets and A0+ is the
family of maximal forbidden sets. In many situations, it is more convenient to refer to them instead of
A1and A0.
IV. THE PROPOSED SCHEME
In this section, an improved VCS using one shared image is proposed. The binary secret image is
encoded into one share with innocent-looking [5]. The secret image is visually revealed by
superimposing the portion of the share and its other portion together. The decryption of the proposed
scheme is very simple. By selecting upper most portion of share and lower most portion of the share
and superimposing them together gives the revealed secret image and exor-ing the each pixel in
blocks of size 2x2 from a revealed secret image to improve the reconstructed image.
4.1. Encryption Process
In encryption process binary image with m x n pixels is considered as the secret image. Then, a cover
image with innocent-looking is selected. The cover image has to be larger than the secret image. Here,
the size of the cover image is assumed to be M x N, where M >m, N>=n or M >=m, N>n. The
encryption of the proposed scheme is divided into two parts. Part 1 of the encryption algorithm
encodes Area 1 of the share, Part 2 encodes Area 2. Area 1 and 2 of the share are demonstrated in
Figure 1.The share is generated based on the secret image and cover image. Each pixel of secret
image is expanded into a 2x2 subpixel block. Therefore, the size of the output share is 2M x2N.
The basis matrices of the proposed scheme used in the encryption are illustrated in Table 1. Matrices
used in Area 2 are generated based on the three conditions of EVCS [7]. In the basis matrices, 1
represents black pixel and 0 represents white pixel.

960 Vol. 7, Issue 3, pp. 957-966

Figure 1. Area1 and Area2 of the share
Table 1. Basis matrices of the evc
Area Of The
Share
Basis Matrices
Area1 T
w
=[0 0 1 1]

T
b
= [0 1 1 1]

Area 2

T
ww
w = 0 0 1 1
1 0 1 0
T
ww
b = 0 0 1 1
1 1 0 0
T
wb
w = 0 0 1 1
1 0 1 1
T
wb
b = 0 0 1 1
1 1 1 0
T
bw
w = 1 0 1 1
0 0 1 1
T
bw
b = 1 1 0 0
0 0 1 1
T
bb
w = 1 0 1 1
1 0 1 1
T
bb
b = 0 1 1 1
1 1 1 0

In encryption algorithm, step 2-7 encode Area 1 of the share. Step 8-13 encodes Area 2 of the share.
Area 1 is firstly encoded, then Area 2. In Area 1, the subpixel block is determined by the
corresponding pixel in cover image. But in Area 2, the subpixel block is determined by cover image
pixel, secret image pixel and the corresponding subpixel block in Area 1.

2n
2N
2m

The Revealed Secret
Image
2M
Area 1
Area 2
961 Vol. 7, Issue 3, pp. 957-966

Algorithm for encryption process
Input: Secret image S (m x n pixels), cover image C (M x N pixels)
Output: Innocent-looking share SH (2M x 2N pixels)
Step 1: Compute p=M-m and q=N-n.
Step 2: For i=1 to p and for j=1 to N, repeat step 3-4.
Step 3: Obtain the corresponding basis matrix TC(i, j) from Table1 Area 1 according to
C(i,j).
Step 4: Randomly permute the columns of the determined basis matrix and assign the
elements of the permuted basis matrix to the pixels SH (2i-1, 2j-1), SH (2i, 2j-1),
SH(2i-1,2j), SH(2i,2j) of the share.
Step 5: For i = p+1 to M and for j=1 to q , repeat step 6- 7.
Step 6: Obtain the corresponding basis matrix TC(i, j) from Table 1 Area 1 according to
C(i, j).
Step 7: Randomly permute the columns of the determined basis matrix and assign the
elements of the permuted basis matrix to the pixels SH (2i-1, 2j-1), SH (2i, 2j-1),
SH(2i-1,2j), SH (2i, 2j) of the share.
Step 8:For i =p+1 to M and for j=q+1 to N, repeat step 9-13.
Step 9:Find out the corresponding basis matrices TS(i,j) C(i,j),D TS(i,j) D,C(i,j) fro
Table.1 Area 2 according to C(i, j) and S(i, j) where D can be black or white
Step10: Permute the columns of the corresponding basis matrices TS(i,j)C(i,j),D and TS(i,j)
D,C(i,j) for all possible. All these basis matrices and their permuted matrices form a
matrix set MS.
Step 11: Fetch one matrix from MS. Check the first row of this matrix is equal to
SH(2(i-p)-1,2(j-q)-1),SH(2(i-p),2(j-q)-1),SH(2(i -p)-1,2(j-q)),SH(2(i-p),2(j-q)) or not.
If yes, put the second row of this matrix into the candidate vector set CVS . If no,
move to the second row and conduct the above processing.
Step 12: Repeat step 11 until all the matrices in MS are checked and the candidate vector set
CVS is formed.
Step 13: Randomly choose one vector in CVS and assign this vector to the pixels
SH(2i-1,2j-1) , SH(2i,2j-1) , SH(2i-1,2j), SH(2i,2 j) of the share.
Step 14: Output the innocent-looking share SH.
The vital parts of encryption algorithm are finding the corresponding basis matrices and forming the
candidate vector set when Area 2 is encoding. For instance, assume that pixels in C(i, j) and S(i, j) are
1 and 0, respectively. The obtained basis matrices in Table.1 Area 2 are

T
wb
w = 0 0 1 1
1 0 1 1
T
bw
w = 1 0 1 1
0 0 1 1
T
bb
w = 1 0 1 1
1 0 1 1
If the corresponding four pixels SH(2(i p) 1,2(j q) 1) , SH(2(i p),2( j q) 1) , SH(2(i p)
1,2( j q)) , SH(2(i p),2( j q)) in Area 1 of the share are 1, 0, 0, 1, respectively. The formed
candidate vector set is {1011,1101}.
Then, one vector is randomly selected from the candidate vector set and assigned to the pixels
SH(2i 1,2 j 1) , SH(2i,2 j 1) , SH(2i 1,2 j) , SH(2i,2 j) of the share.
Let us consider an example
S be secret image of size 2x2 and C be cover image of size 3x3

S=

1 1
1 0
962 Vol. 7, Issue 3, pp. 957-966

C =

The generated share is of size 6x6.

Share=

4.2. Decryption Process
The decryption process of the proposed scheme is very simple. By selecting upper most portion of
share and lower most portion of the share and superimposing them together gives the revealed secret
image and exor-ing the each pixel in blocks of size 2x2 from a revealed secret image to improve the
reconstructed image.
The decryption process of the proposed scheme is illustrated in Figure 2. Superimposing the
uppermost portion of the share of size 2mx2n is selected i.e plain area in the below figure and
lowermost portion of the share of size 2mx2n is selected i.e shaded area in the below figure reveals
the secret image.

Figure 2: Decryption Process
Considering the above mentioned example, the share generated in the encryption process is

1 1 1
0 1 0
1 1 1
1 1 1 1 1 1
1 0 1 0 0 1
1 1 0 1 0 1
0 0 1 1 0 1
1 1 0 1 0 1
1 0 1 1 1 0
2M
2m
2N
2n
963 Vol. 7, Issue 3, pp. 957-966

Share=

The uppermost portion of the share is =

The lowermost portion of the share is=

Superimposing the 2 portions of the share reveals the secret image of size 2mx2n and exor-ing the
each pixel in blocks of size 2x2 from a revealed secret image and then negating it improves the
reconstructed image quality and also reduces its size to original secret image.

Revealed secret image=

Final revealed image=

V. RESULTS AND DISCUSSION
To demonstrate the effectiveness of the proposed scheme, examples are illustrated in this section. The
secret image with 200x200 pixels is shown in figure 3 and cover image with 249x260 pixels is shown
in Figure.4. The innocent-looking share of size 498x520 shown in Figure.5 Figure.6 shows the
superimposed image of size 400x400. The final reconstructed image after xor-ing the each pixel in
1 1 1 1 1 1
1 0 1 0 0 1
1 1 0 1 0 1
0 0 1 1 0 1
1 1 0 1 0 1
1 0 1 1 1 0
1 1 1 1
1 0 1 0
1 1 0 1
0 0 1 1
0 1 0 1
1 1 0 1
0 1 0 1
1 1 1 0
1 1 1 1
1 1 1 1
1 1 0 1
1 1 1 1
1 1
1 0
964 Vol. 7, Issue 3, pp. 957-966

blocks of size 2x2 is shown in Figure 7. The PSNR value between secret image and reconstructed
image is 11.6052 dB and histogram error between secret image and final reconstructed image is
0.00078408. The histogram error of superimposed image and secret image is 0.1185.The histogram
error of proposed VCS is smaller than the histogram error of the existing VCS it shows that the
proposed scheme gives more similar image of secret image.

Figure 3. Secret image of size 200x200

Figure 4. Cover image of size 249x260.

Figure 5. Share image of size 498x520.
965 Vol. 7, Issue 3, pp. 957-966

Figure 6. Superimposed image of size 400x400.

Figure 7. Final secret image of size 200x200.
VI. CONCLUSION
In this paper, an Improved VCS using single shared image is presented. The proposed scheme hides
one secret image into one innocent-looking share. The secret image is visually revealed by
superimposing the portion of the share and its other portion together. The proposed scheme has the
advantage of resisting geometry distortion and the revealed secret image size is same as that of
original secret image. Comparing to the two existing one share based VCSs, the share generated by
the proposed scheme is easy to recognize and manage. The innocent-looking of the share make it
draw less attention from attackers and image quality is also improved.
VI. FUTURE ENHANCEMENT
Several future enhancements can be made to this VCS. This proposed scheme can be enhanced by
using the morphological operations and using filters to remove the noise in the image. The proposed
system is designed to run on single host machine, the work can be extended to run on multiple host
machines.
REFERENCES
[1]Moni Naor and Adi Shamir, Visual Cryptography, advances in cryptology Eurocrypt, pp 1-12, 1995.
966 Vol. 7, Issue 3, pp. 957-966

[2] H. Wu and C. Chang, Sharing visual multi-secrets using circle shares, Computer Standards & Interfaces,
vol. 28, no. 1, pp. 123 135, 2005.
[3]P. Tsai and M. Wang, An (3, 3)-visual secret sharing scheme for hiding three secret data, in Proceedings of
the 2006 Joint Conference on Information Sciences, 2006.
[4]L. Duo and D. Yi-Qi, A New Visual Hiding Scheme Using One Secret Image, Chinese Journal Of
Computers, vol. 32, no. 11, pp. 22472251, 2009.
[5]Xiaotian Wu and Wei Sun A Novel Extended Visual Cryptography Scheme Using One Shared Image In
ICITIS, pp.216-220, 2010.
[6]G. Ateniese, C. Blundo, A. Santis, and D. Stinson, Extended capabilities for visual cryptography,
Theoretical Computer Science, vol. 250, no. 1-2, pp. 143161, 2001.
[7]Carlo Blundo, Alfredo De Santis, Douglas R. Stinson, On the contrast in visual cryptography schemes,
Journal of Cryptology, 12: pp. 261-289, 1999.
[8]C.C. Wu, L.H. Chen, A Study On Visual Cryptography, Master Thesis, Institute of Computer and
Information Science, National Chiao Tung University, Taiwan, R.O.C., 1998.
[9]E. Verheul and H. V. Tilborg, Constructions And Properties Of K Out Of N Visual Secret Sharing
Schemes. Designs, Codes and Cryptography, 11(2) , pp.179196, 1997.
[10]Tzung-Her Chen, Kai-Hsiang Tsao, and Kuo-Chen Wei, Multi-Secrets Visual Secret Sharing,
Proceedings of APCC2008, IEICE, 2008.
[11] V. Rijmen, B. Preneel, Efficient color visual encryption for shared colors of Benetton, Eurocrypto96,
Rump Session, Berlin, 1996.
[12] Rafel C. Gonzalez, Digital image processing. Englewood cliffs: Pearson education, 2002.
AUTHORS

Gowramma B H, is currently Pursuing 4
th
Semester, Master of Technology in Computer
Science and Engineering at AIT, Chickmagalur. She has completed her Bachelor of
Engineering from PES Institute of Technology and management, Shivamogga. She had
published a paper. Her areas of interests include visual cryptography and Information
Security.

Shyla M G, is currently Pursuing 4
th
Semester, Master of Technology in Computer Science
and Engineering at AIT, Chickmagalur. She has completed her Bachelor of Engineering
from Adichunchanagiri Institute of Technology, Chikmagalur. She had published a paper.
Her areas of interests include image processing and Information Security.

Vivekananda is presently working as assistant professor in the Department of Computer
Science and Engineering, Adichunchanagiri Institute of Technology, Chikmagalur,
Karnataka, India. He had 9 years of teaching experience. He obtained his Master of
Technology degree from SJCE Mysoor under Visveswaraya Technological University,
Belgaum. His research interests include Image Stegonography, Visual Cryptography, and
Language Processing.


967 Vol. 7, Issue 3, pp. 967-973

SEGMENTATION OF BRAIN TUMOUR FROM MRI IMAGES BY
IMPROVED FUZZY SYSTEM
Sumitharaj.R,

Shanthi.K
Assistant Professor, Department of ICE,
Sri Krishna College of Technology, Coimbatore, India

ABSTRACT
The objective of this work is to detect the brain tumour by using symmetric analysis. In this paper the brain
tumour is first detected and then it is segmented. The brain tumour is quantified by using MRI of an image. The
problem of segmenting tumour by using MRI is considered by multiple step. Segmentation of images is a
significant position in the region of image processing. It becomes more significant while using the medical
images; Magnetic Resonance (MRI) Imaging gives more and perfect information during medical examinations
than that of the other medical images such as ultrasonic, CT images and X-rays. Here we propose a novel
method of segmenting brain tumour. An image with the brain tumour is first used and then from that image the
brain tumour is detected and segmented. A segmentation method which is used to segment the tumour from a
particular area. In the first step, we propose a tumour image to detect .then it is to be segment the tumour. In
some process the morphological analysis is used as an image processing tools for sharpening the regions and
filing the gaps for binarised image.
KEY WORDS: MRI image, Segmentation, Tumour detection, Fuzzy system.
I. INTRODUCTION
A technique in which the image is digitized in mathematical operation, it is to enhance the image with
recognition task. An image which is used in several parts of the region such as pattern recognition, TV
screens, 2D, 3D images. An image can be processed optically or an image can be processed optically
or digitally with a computer. Image processing is a well diversified field using in such of medical,
defense, quality control and in entertainment. Brain tumour is defined as an abnormal growth of cells
within the brain or the central spinal canal. Brain tumours include all tumours inside the central spinal
canal. The uncontrolled growth of cell division in brain of neueron,glial cell, myelin ,blood vessels, in
the cranial nerves, in the brain envelopes. Any brain tumour is inherently serious and life threatening
because of its invasive and infiltrative character in the limited space of the intracranial cavity.
However, brain tumours (even malignant ones) are not invariably fatal. Brain tumours or intracranial
neoplasm can be cancerous (malignant) or non-cancerous (benign); however, the definitions of
malignant or benign neoplasm differs from those commonly used in other types of cancerous or non -
cancerous neoplasm in the body. Level depends on the combination of factors like the type of tumour,
its location, its size and its state.
While in earlier period the brain tumour is detected by using the CT scans. In CT the not only the
brain tumour part detects the tumour portion whole area will be detected. But in MRI the tumour
region only will be detects. The primary brain tumours that commence within the brain. An inferior
or metastatic brain tumour takes place when cancer cells extend to the brain from a primary cancer in
a different component of the body. Brain tumours are classified as primary and second tumour.

968 Vol. 7, Issue 3, pp. 967-973

Primary brain tumours are the tumours that originated in the brain from where they occur. They are
benign that is they are non-cancerous cells it do not attack the neighbor cells. But the secondary type
that which affects the whole region of the brain. Most frequently cancers that increase to the brain to
reason secondary brain tumours begin in the lumy, breast, and kidney or from melanomas in the skin.
A systematic brain tumour using MRI image of brain is used by type II fuzzy system. The proposed
type II fuzzy it has two modules. On this preprocessing and the other is segmentation method .In
preprocessing method a rule base is developed. In segmentation method the PCM is analyzed.
II. DESCRIPTION EXISTING METHODOLOGY
In the existing methodology, the brain tumour can be diagnosed by CT scan, ultrasonic, MRI scan.
While dealing with CT scans the whole area of brain part is detected. But in MRI the brain tumor part
only that get detects. In olden technique the brain tumour is diagnose with the whole region. So we
cannot segment the tumor with accurate .For these reasons, the brain tumor is detected by MRI .while
using MRI the brain tumor is accurately detected and can be segmented with accurate measurements.
The Segmentation of an image entails the division or separation of the image into regions of similar
attribute. The ultimate aim in a large number of image processing applications is to extract important
features from the image data, from which a description, interpretation, or understanding of the scene
can be provided by the machine. Segment the brain to remove non-brain data. However, in
pathological cases, standard segmentation methods fail, in particular when the tumour is located very
close to the brain surface.
Therefore we propose an improved segmentation method, relying on the approximate plane. In this
fuzzy system there are two types: Type I and type II fuzzy system. The type I fuzzy system the
framework of fuzzy sets, systems, and relations is very useful to deal with the absence of sharp
boundaries of the sets of symptoms, diagnoses, and phenomena of diseases. However, there are many
uncertainties and vaguenesss in images, which are very difficult to handle with Type-I fuzzy sets.
These fuzzy sets are not able to model such uncertainties directly because their memberships
functions are crisp. But in Type-II fuzzy sets are able to model such uncertainties as their membership
functions are themselves fuzzy. Therefore, Type-II fuzzy logic systems have the potential to provide
better performance. For these reasons, Type-II fuzzy modeling is used. The proposed Type-II expert
system has been tested and validated to show its accuracy in the real world. The results show that the
proposed system is superior in recognizing the brain tumour and its grade than Type-I fuzzy expert
systems.
In this paper first the image is preprocessed and then it is segmented. In segmentation method the
image is first calculated and then it is segmented through symmetric analysis methods or it varies
upon the rule base methods. In pre-processing some fundamental image enhancement and noise
lessening procedure are applied. The noise is reducing by the conversion of gray scale image. Then
this gray scale image pass in to the filter. We use here a high pass filter imfilter in Matlab to filter an
image, replaces each pixel of the image with a weighted average of the surrounding pixels. The
weights are determined by the values of the filter, and the number of surrounding pixels is determined
by the size of the filter used. Then the gray image and filtered image are merged together to enhanced
the image quality. Here we use Median filtering which is a nonlinear operation often used in image
processing to reduce "salt and pepper" noise. A median filter is more effective than convolution when
the goal is to simultaneously reduce noise and preserve edges. Then we convert the filtered image into
binary image by the thresholding method which computes a global threshold that can be used to
convert intensity image to a binary image with normalized intensity value between 0 and 1.
III. MATERIALS AND METHODS
The basic concept to detect tumour, is the component of the image hold the tumour generally has
extra concentration then the other segment and we can guess the area, shape and radius of the tumour
in the image. We calculate the area in pixel. Noise existing in the image can decrease the capability of
region growing filter to grow large regions or may result as a fault edges.

969 Vol. 7, Issue 3, pp. 967-973

Figure 1aTumourwith noise Figure 1bTumour without noise
The figure 1a is tumour with noise and figure 1b is tumour without noise. The basic type names which
are Anaplasia, Neoplasia, and Necrosis. More generally a neoplasm may cause release of metabolic
end products.
Table 1Tumour types
Anaplasia Loss of differentiation of
cells
Neoplasia Uncontrolled division of
cells
Necrosis Premature death of cells
In the type II fuzzy system, the figure 1b astrocytoma tumour is took for an example. The
astrocytoma tumour that is graded into four types. GradeI, GradeII, GradeIII and Grade IV.These
grades that are tabulated with the diagnosis list.
Table 2 Grade types
Grades % of brain tumour Survival of years
Grade I 1.8 91
Grade II 1.3 67
Grade III 4.3 46
Grade IV 22.6 9

Astrocytoma can be subdivided into four grades: Grade I (Pilocytic Astrocytoma), Grade II (Diffuse
Astrocytoma), Grade III (Anaplastic Astrocytoma), and Grade IV (Glioblastoma Multiforme).These
grading systems have been shown with survival. So the survival ranges are more than 5 years for
grade II, between 2 and 5 years for grade III, and less than 1 year for Grade IV.
3.1 Proposed methodology- type II fuzzy system
In this paper, the type II fuzzy system is proposed. The fuzzy is divided into two types. One is type I
and type II fuzzy system. The type I fuzzy system that has the membership functions of crisp data sets
.so it cannot be used for the sets and algorithms. But the type II fuzzy system, the data sets and
membership functions are fuzzy. In any uncertain vagueness or images the type II fuzzy is used.

Figure 2 Fuzzy System Classifications
Step 1-- defining the initial membership functions: For tuning the parameters of initial interval Type-
II membership functions, the Cluster the output data.
In this process, the CSF is cerebrospinal fluid is the black area, so the membership function is black
intensity. On the other hand, the Abnormality is the brightest part of MRI, so its initial membership
Fuzzy system
Type I fuzzy Type II fuzzy
Membership
functions are crisp
Membership
functions are fuzzy

970 Vol. 7, Issue 3, pp. 967-973

function must include the white intensities. The White Matter (WM) is bright and the Gray Matter
(GM) is gray, but their intensity levels are very similar.
Table 3: Normal tissue in brain image
CSF dark
GM gray
WM bright
CSF: cerebrospinal fluid; GM: gray matter; WM: white matter
Step2--Tuning the parameters of Type-II membership functions: Tuning the parameters of a Type I
membership function is possible formula. However, the output of a Type-II membership function
cannot be represented. Given an input and output training pair (xi, ij), an interval Type-II fuzzy is
designed so that the error function (ij).
=
1
2
(
)2
Where j (xi) is the real membership value of ith data in jth class and ij is the calculated membership
value of ith data in jth class.
3.2approaches for segmenting tumour image
In processing the tumour method, it consists of certain blocks of systematic images. Here the salt and
pepper noise is added.

Figure 3Flow Chart of Proposed Segmenting Tumour
In this system first the image is stated as a tumour image, then the image gets binarised .The binarised
image that gets filtered, and the added image is involved to get the segmented image.
3.3Algorithm for detecting brain tumour
Input: MRI of brain image.
Output: Tumour portion of the image.
Step1:- Read the input color or grayscale image.
Step2:- Converts input colour image in to grayscale image this is done by forming a weighted sum of
each three (RGB) component, eliminating the saturation and hue information while retaining the
luminance and the image returns a gray scale color map.
Step3:- Resize this image in to 200 200 image matrix.
Step4:- Filters the multidimensional array with the Multidimensional filter. Each element of the
output an integer or in array, then output elements that exceed the certain range of the integer type is
shortened, and fractional values are rounded.
Step5:- Add step2, step4 image and a integer value 45 and pass it in to a median filter to get the
resultant enhanced image.
Step6:- Computes a global threshold that can be used to convert an intensity image (Step5) to a binary
image with a normalized intensity value which lies in between range 0 and 1.
Step7:- Compute watershed segmentation by Matlab Command watershed (step6 image).
Input image
Tumour image
Binarised image
Edge image
Segmented image

971 Vol. 7, Issue 3, pp. 967-973

Step8:-Compute the morphological operation by two Matlab Command imerode and imdilate and
strel with arbitrary shape.
Step9:- Store the size of the step 8 image into var1 and var2 i.e. no. Of rows and column in pixels by
[var1 var2] =size (step8 image)
Step10:- Convert in to binary image and traces the exterior boundaries of objects, as well as
boundaries of holes inside these objects, in the binary image and into an RGB color image for the
purpose of visualizing labeled regions.
Step11:- Show only tumour portion of the image by remove the small object area.
Step12:- Compute edge detection using sobel edge detection Technique
IV. RESULTS
In this paper an interactive segmentation method that enables users too quickly and efficiently
segment tumours in MRI of brain. The tumour is a general problem in medical field. In the tumour
process there are primary and secondary tumours .while in the secondary tumour type many variety of
tissue is mostly depends on the primary one. So based on these types of sets we can detect or segment
the tumour.

Figure 4 Tumour image
The tumour image that gets binarised. The salt and pepper noise is added When the median filter is
used the image of noise that is reduced.

Figure 5 binary image
By adding the median filter, theedge image gets extracted to optimise the featured data .

Figure 6 Edge image
Finally the tumour that is segmented

972 Vol. 7, Issue 3, pp. 967-973

Figure 7 Segmented image
By using the type II fuzzy system ,the tumour image is segmented and the denoised threshold image is
obtained.then the tumour part is clustered .

Figure 8 Denoised threshold image
The threshold image that gets denoised. Fuzzy clustering are more advantageous over crisp clustering
.the total vector to a given class is not required in each iteration. The thresholding method is used to
recognize the characteristics of the pixels values.

Figure 9 clustered tumour image
V. CONCLUSION
In this project an iterative method is to enable the tumours by detecting and segmenting the tumour
from MRI image. Segmentation is used to segment the particular area or part of the image. In this
paper the goal is to detect and segment the tumour from MRI of brain. The type II fuzzy system is
used in this paper .Rather comparing with Type I fuzzy system, the type II is used because if there is
any uncertainties or vagueness in images the type I fuzzy has a membership function of crisp. But in
type II fuzzy the membership functions are fuzzy.
ACKNOWLEDGEMENTS
I would like to thank my Head of the Department,my institution,my parents and my collegues.
REFERENCES
[1] Sudipta Roy, Samir K. Bandyopadhya, Detection and Quantification of Brain Tumor from MRI of Brain
and its Symmetric Analysis,in International Journal of Information and Communication Technology
Research,Volume 2 No. 6, June 2012 .
[2]T.Logeswari and M. Karnan, An improved implementation of brain tumour detection using segmentation
based on soft computing Journal of Cancer Research and Experimental Oncology Vol. 2(1) pp. 006-014,
March, 2010.
0 50 100 150 200 250
0
50
100
150
200
250

973 Vol. 7, Issue 3, pp. 967-973

[3]M.H Fazel, M. Zarandi, M.Izadi, Systematic image processing for diagnosing brain tumor: A type II fuzzy
expert system approach,Applied Soft Computing 11 (2011) 285294.
[4]J. J. Corso, E. Sharon, and A. Yuille, Multilevel Segmentation and Integrated Bayesian Model Classification
with an Application to Brain Tumour Segmentation, in Medical Image Computing and Computer Assisted
Intervention, vol. 2, 2006, pp. 790798.
[5]A.W. Chung Liew, H. Yan, Current methods in the automatic tissue segmentation of 3D magnetic resonance
brain images, Current Medical Imaging Reviews2 (2006).
[6] Dou, W., Ruan, S., Chen, Y., Bloyet, D., and Constans, J. M. (2007), A framework of fuzzy information
fusion for segmentation of brain tumour tissues on MR images, Image and Vision Computing, 25:164171.
[7] Hassan Khotanlou, Olivier Colliot and Isabelle Bloch, Automatic brain tumour segmentation using
symmetry analysis and deformable models, GET-EcoleNationale Superior des Telecommunications, France.
[8]T.Logeswari and M.Karnan, An Improved Implementation of Brain Tumour Detection Using Segmentation
Based on Hierarchical Self Organizing Map, International Journal of Computer Theory and Engineering, Vol.
2, No. 4, August, 2010,pp.1793-8201.
[9] R. Rajeswari and P. Anandhakumar, Segmentation and Identification of Brain Tumour MRI Image with
Radix4 FFT Techniques, European Journal of Scientific Research, Vol.52 No.1 (2011), pp.100-109.
[10] P.Narendran, V.K. Narendira Kumar, K. Somasundaram, 3D Brain Tumours and Internal Brain Structures
Segmentation in MR Images, I.J. Image, Graphics and Signal Processing, 2012, 1, 35-43.
[11] T. Logeswari and M. Karnan, An improved implementation of brain tumour detection using segmentation
based on soft computing Journal of Cancer Research and Experimental Oncology Vol. 2(1) pp. 006-014,
March, 2010.
[12]E. Nasibov, G. Ulutagay, A new unsupervised approach for fuzzy clustering, Fuzzy Sets and Systems 158
(2007) 21182133.
[13]M.H. Fazel Zarandi, M. Zarinbal, I.B. Turksen, Type-II possibilistic C-mean clustering, IFSA-USEFLAT
(2009) 3035.
[14] R.N. Strickland, Image Processing Techniques for Tumour Detection, Marcel-Dekker, 2002.
[15] R. Krishnapuram, J.M. Keller, A possibilistic approach to clustering, IEEE Transactions on Fuzzy Systems
1 (1993).
[16] J.M. Mendel, R.I. John, F. Liu, Interval Type-2 fuzzy logic systems made simple,IEEE Transactions on
Fuzzy Systems 14 (2006) 808821.

ABOUT AUTHOR

Sumitharaj.R, Assistant professor of Dept. Instrumentation and control engineering, Sri
Krishna college of technology, Coimbatore. She received M.E in the Dept. of electrical and
electronics engineering in the specialization of control and instrumentation engineering,
Anna University, regional center, Coimbatore. She received B.E in the Dept of Electronics
and instrumentation engineering, Bannari Amman institute of technology, sathyamangalam.

Shanthi.K, Assistant professor of Dept. Instrumentation and control engineering, Sri
Krishna college of technology, Coimbatore. She received M.E in the Dept. of electronics
and communication engineering in the specialization of communication systems, Anna
university regional center, Coimbatore. She received B.E in the Dept of Electronics and
instrumentation engineering, Maharaja Engineering College, Coimbatore

974 Vol. 7, Issue 3, pp. 974-979

IMPLEMENTATION OF CLASSROOM ATTENDANCE SYSTEM
BASED ON FACE RECOGNITION IN CLASS
Ajinkya Patil
1
, Mrudang Shukla
2

1
Mtech (E&TC),
2
Assisstant Professor
Symbiosis institute of Technology, Pune, Maharashtra, India

ABSTRACT
The face is the identity of a person. The methods to exploit this physical feature have seen a great change
since the advent of image processing techniques. The attendance is taken in every schools, colleges and
library. Traditional approach for attendance is professor calls student name & record attendance. It takes some
time to record attendance. Suppose duration of class of one subject is about 50 minutes & to record attendance
takes 5 to 10 minutes. For each lecture this is wastage of time. To avoid these losses, we are about use
automatic process which is based on image processing. In this novel approach, we are using face detection &
face recognition system. This face detection differentiates faces from non-faces and is therefore essential for
accurate attendance. The other strategy involves face recognition for marking the students attendance. The
Raspberry pi module is used for face detection & recognition. The camera will be connected to the Raspberry pi
module. The student database is collected. The database includes name of the students, there images & roll
number. This raspberry pi module will be installed at the front side of class in such a way that we can capture
entire class. Thus with the help of this system, time will be saved. With the help of this system, it is so convenient
to record attendance. We can take attendance on any time.
KEYWORDS: Viola Jones algorithm, PCA, LDA, Image processing, Raspberry pi
I. INTRODUCTION
Organizations of all sizes use attendance systems to record when student or employees start and stop
work, and the department where the work is performed. Some organizations also keep detailed
records of attendance issues such as who calls in sick and who comes in late. An attendance system
provides many benefits to organizations. There was a time when the attendance of the students and
employees was marked on registers.
However, those who have been a part of the classes when attendance registers were used know how
easy it was to abuse such a method of attendance and mark bogus attendances for each other. Of
course, technology had to play its role in this field just as well as it has done in other fields. The
attendance monitoring system was created and it changed the way attendances were marked. The
attendance monitoring system has made the lives of teachers and employers easier by making
attendance marking procedure a piece of cake
When it comes to schools and universities, the attendance monitoring system is a great help for
parents and teachers both. Parents are never uninformed of the dependability of their children in the
class if the university is using an attendance monitoring system. The registers could easily be
exploited by students and if information was mailed to the parents, there were high chances that mails
could be made to disappear before parents even saw them. With the monitoring system in place, the
information can easily be printed or a soft copy can be sent directly to parents in their personal email
accounts.
The system started with two basic processes - Manual processes and Automatic processes. Manual
processes are eliminated as the staff needed to maintain them. It is often difficult to comply with
regulation, but an automated attendance system is valuable for ensuring compliance with regulations
regarding proof of attendance.
975 Vol. 7, Issue 3, pp. 974-979

II. HISTORY
Naveed Khan Balcoh proposes that students attendance in the classroom is very important task and if
taken manually wastes a lot of time. There are many automatic methods available for this purpose i.e.
biometric attendance. All these methods also waste time because students have to make a
queue to touch their thumb on the scanning device. This work describes the efficient
algorithm that automatically marks the attendance without human intervention. This attendance
is recorded by using a camera attached in front of classroom that is continuously capturing images
of students, detect the faces in images and compare the detected faces with the database and mark
the attendance. The paper review the related work in the field of attendance system then describes the
system architecture, software algorithm and results. ErikHjelmas face detection is a necessary first-
step in face recognition systems, with the purpose of localizing and extracting the face region from the
background. It also has several applications in areas such as content-based image retrieval, video
coding, video conferencing, crowd surveillance, and intelligent humancomputer interfaces. However,
it was not until recently that the face detection problem received considerable attention among
researchers. The human face is a dynamic object and has a high degree of variability in its appearance,
which makes face detection a difficult problem in computer vision. A wide variety of techniques have
been proposed, ranging from simple edge-based algorithms to composite high-level approaches
utilizing advanced pattern recognition methods.
III. METHODOLOGY
For this system we are using a two-step mechanism. First comes to be face detection then followed by
face recognition. For face detection we are using Viola Jones face detection algorithm while for face
recognition we are using hybrid algorithm from PCA and LDA.
1) Viola-Jones algorithm
There are three major blocks in Viola-Jones algorithm; Integral Images, Ada-Boost Algorithm and
Attentional cascade. The integral image computes a value at each pixel for example (x,y) that is the
sum of the pixel values above to the left of (x,y). This is quickly computed in one pass through the
image. Viola jones algorithm uses Haar like features. This is nothing but scalar product between the
image & some haar like structures. Feature is selected through adaboost. Ada-Boost provides an
effective learning algorithm and strong bounds on generalization performance. The overall form of the
detection process is that of a degenerate decision tree, what we call a cascade. A positive result
from the first classifier triggers the evaluation of a second classifier which has also been adjusted to
achieve very high detection rates. A positive result from the second classifier triggers a third
classifier, and so on. A negative outcome at any point leads to the immediate rejection of the sub-
window. The cascade training process involves two types of tradeoffs. In most cases classifiers with
more features will achieve higher detection rates and lower false positive rates. At the same time
classifiers with more features require more time to compute. In principle one can use following stages.

Figure 1: Cascade classifier
i) the number of classifier stages, ii) the number of features in each stage, and iii) the threshold of
each stage, are traded off in order to minimize the expected number of evaluated features.
Unfortunately finding this optimum is a tremendously difficult problem. In practice a very simple
framework is used to produce an effective classifier which is highly efficient. Each stage in the
976 Vol. 7, Issue 3, pp. 974-979

cascade reduces the false positive rate and decreases the detection rate. A target is selected for the
minimum reduction in false positives and the maximum decrease in detection. Each stage is trained by
adding features until the target detection and false positives rates are met these rates are determined by
testing the detector on a validation set. Stages are added until the overall target for false positive and
detection rate is met.
2) Flow chart

Figure 2: face detection & recognition
The flowchart is shown in fig.no.2
2.1) Histogram Normalization
Captured image sometimes have brightness or darkness in it which should be removed for good
results. First the RGB image is converted to the gray scale image for enhancement. Histogram
normalization is good technique for contrast enhancement in the spatial domain.
2.2) Noise Filtering
Many sources of noise may exist in the input image when captured from the camera. There are many
techniques for noise removal. Low pass filtering in the frequency domain may be a good choice but
this also removes some important information in the image. In our system median filtering in is used
for the purpose of noise removal in the histogram normalized image.
2.3) Skin classification
This is used to increase the efficiency of the face detection algorithm. Voila and Jones algorithm is
used for detection..
the images of faces and then applied on the class room image for detection of multiple faces
in the image.
2.4) Face Detection:
Haar classifiers have been used for detection. Initially face detection algorithm was tested
on variety of images with different face positions and lighting conditions and then
algorithm was applied to detect faces in real time video. Algorithm is trained for the
images of faces and then applied on the class room image for detection of multiple
faces in the image.

977 Vol. 7, Issue 3, pp. 974-979

Figure 3: Face Detection
2.5) Face Recognition and Attendance
After the face detection step the next is face recognition. This can be achieved by cropping the first
detected face from the image and compare it with the database. This is called the selection of region
of interest. In this way faces of students are verified one by one with the face database using the Eigen
Face method and attendance is marked on the server.

The system consists of a camera that captures the images of the classroom and sends it to the image
enhancement module. After enhancement the image comes in the Face Detection and
Recognition modules and then the attendance is marked on the database server. At the time of
enrollment templates of face images of individual students are stored in the Face database. Here all
the faces are detected from the input image and the algorithm compares them one by one with the face
database. If any face is recognized the attendance is marked on the server from where anyone can
access and use it for different purposes. This system uses a protocol for attendance. A time table
module is also attached with the system which automatically gets the subject, class, date and time.
Teachers come in the class and just press a button to start the attendance process and the
system automatically gets the attendance without even the intensions of students and teacher. In this
way a lot of time is saved and this is highly securing process no one can mark the attendance of other.
Attendance is maintained on the server so anyone can access it for it purposes like administration,
parents and students themselves. Camera takes the images continuously to detect and recognize all
the students in the classroom. In order to avoid the false detection we are using the skin
classification technique. Using this technique enhance the efficiency and accuracy of the detection
process. In this process first the skin is classified and then only skin pixels remains and all
other pixels in the image are set to black, this greatly enhance the accuracy of face detection
978 Vol. 7, Issue 3, pp. 974-979

process. Two databases are displayed in the experimental setup Figure. Face Database is the
collection of face images and extracted features at the time of enrollment process and the second
attendance database contains the information about the teachers and students and also use to mark
attendance.

Figure 4: Face recognition System in Class
IV. FUTURE SCOPE
For security reasons, we can use detection & recognition system. To identify culprits on bus stations,
railway stations 7 other public places, we can use this system. This will be helping hand to the police.
In this system, we will use GSM module. Suppose if culprit is detected, then detected signal can be
transmitted using GSM module to the central control room of police station. With the help of ISDN
number of GSM, culprit surviving area will be recognized
V. CONCLUSION
We come to know that there are wide range of methods such as biometric, RFID based etc. which are
time consuming and non-efficient. So to overcome this above system is the better and reliable solution
from every perceptive of time and security. Thus we have achieved to develop a reliable and efficient
attendance system to implement an image processing algorithm to detect faces in classroom and to
recognize the faces accurately to mark the attendance.
REFERENCES
[1]. Arulogun O. T., Olatunbosun, A., Fakolujo O. A., and Olaniyi, O. M. RFID-Based Students
Attendance Management System. International Journal of Scientific & Engineering Research Volume
4, Issue 2, February-2013.
[2]. Chitresh Saraswat and Amit Kumar. An Efficient Automatic Attendance System using Fingerprint
Verification Technique. International Journal on Computer Science and Engineering Vol. 02, No. 02,
2010, 264-269.
979 Vol. 7, Issue 3, pp. 974-979

[3]. Jobin J., Jiji Joseph, Sandhya Y.A, Soni P. Saji, Deepa P.L.. Palm Biometrics Recognition and
Verification System. International Journal of Advanced Research in Electrical, Electronics and
Instrumentation Engineering Vol. 1, Issue 2, August 2012.
[4]. Nirmalya Kar, Mrinal Kanti Debbarma, Ashim Saha, and Dwijen Rudra Pal. Study of Implementing
Automated Attendance System Using Face Recognition Technique. International Journal of Computer
and Communication Engineering, Vol. 1, No. 2, July 2012.
[5]. Yogesh Tayal, Ruchika Lamba, Subhransu Padhee. Automatic Face Detection Using Color Based
Segmentation. International Journal of Scientific and Research Publications, Volume 2, Issue 6, June
2012.
[6]. Omaima N. A. AL-Allaf. Review Of Face Detection Systems Based Artificial Neural Networks
Algorithms. The International Journal of Multimedia & Its Applications (Ijma) Vol.6, No.1, February
2014.
[7]. Deepak Ghimire and Joonwhoan Lee. A Robust Face Detection Method Based on Skin Color and
Edges. J Inf Process Syst, Vol.9, No.1, March 2013.
[8]. Sanjay Kr. Singh, D. S. Chauhan, Mayank Vatsa, Richa Singh. A Robust Skin Color Based Face
Detection Algorithm. Tamkang Journal of Science and Engineering, Vol. 6, No. 4, pp. 227-234 (2003).
[9]. Viola P. & Jones J...Robust Real-Time Face Detection. International Journal of Computer Vision 57(2),
137154, 2004.
[10]. Mayank Agarwal, Nikunj Jain, Mr. Manish Kumar and Himanshu Agrawal. Face Recognition Using
Eigen Faces and Artificial Neural Network. International Journal of Computer Theory and Engineering,
Vol. 2, No. 4, August, 2010.
[11]. Bayan Ali Saad Al-Ghamdi, Sumayyah Redhwan Allaam. Recognition of Human Face by Face
Recognition System using 3D. Journal of Information & Communication Technology Vol. 4, No. 2,
(Fall 2010) 27-34.
[12]. Hafiz Imtiaz and Shaikh Anowarul Fattah. A Face Recognition Scheme Using Wavelet - Based
Dominant Features. Signal & Image Processing: An International Journal (SIPIJ) Vol.2, No.3,
September 2011.
[13]. Jagadeesh H S, Suresh Babu K, and Raja K B. DBC based Face Recognition using DWT. Signal &
Image Processing: An International Journal (SIPIJ) Vol.3, No.2, April 2012
[14]. Aruna Bhat. Medoid Based Model For Face Recognition Using Eigen And Fisher Faces. International
Journal of Soft Computing, Mathematics and Control (IJSCMC), Vol. 2, No. 3, August 2013.
[15]. Kandla Arora. Real Time Application of Face Recognition Concept. International Journal of Soft
Computing and Engineering (IJSCE) ISSN: 2231-2307, Volume-2, Issue-5, November 2012.

AUTHORS BIOGRAPHY
Ajinkya Patil pursuing the M.Tech. degree in Electronics & Telecommunication
Engineering from Symbiosis International university, Pune. His research interests include
Image processing, Robotics, Embedded systems.

and Telecommunication department. His research interest is in image processing and
defining vision and path for automated vehicle in agricultural field.

980 Vol. 7, Issue 3, pp. 980-990

A TUNE-IN OPTIMIZATION PROCESS OF AISI 4140 IN RAW
TURNING OPERATION USING CVD COATED INSERT
C. Rajesh
Department of Mechanical Engineering, Srikalahasteeswara Institute of Technology,
Srikalahasthi -517640, Andhra Pradesh, India

ABSTRACT
In today's rapidly changing scenario in every manufacturing industry, aims at producing a large number of
quality products within relatively lesser time. This invites optimization methods in metal cutting processes,
considered to be a vital tool for continual improvement of output quality in products. In order to maximize the
gains from raw turning operation of AISI 4140 alloy steel, an accurate model of process must be constructed. In
this paper, an approach which incorporates Surface roughness, Material removal rate & Power consumption is
presented for optimizing the cutting parameters in finish turning. And a tutorial is presented on constructed
optimization process describing with Taguchi method, Artificial Neural networks and genetic algorithm (GA)
developed specifically for problems with multiple objectives by using specialized fitness function and GONNs is
adapted for better acceptance. ANOVA are used to analyze the effect of cutting parameters on the quality
characteristic of machined work piece. The approach is suitable for fast determination of optimum cutting
parameters during machining, where there is not enough time for deep analysis.
KEYWORDS: CVD coated insert, AISI 4140 alloy steel, Taguchi, ANOVA, ANN, GA, GONNs
I. INTRODUCTION AND LITERATURE SURVEY
The recent developments in science and technology have put tremendous pressure on manufacturing
industries. The manufacturing industries are trying to decrease the cutting costs, increase the quality
of the machined parts and machine more difficult materials. Machining efficiency is improved by
reducing the machining time with high speed machining. When cutting ferrous and hard to machine
materials such as steels, cast iron and super alloys, softening temperature and the chemical stability of
the tool material limits the cutting speed. High speed machining has been the main objective of the
Mechanical Engineering through ages. The trend to increase productivity has been the instrumental in
invention of newer and newer cutting tools with respect to material and designs [1].
The development of machining technologies and practices over recent years has meant that designs
that were difficult to manufacture can now be produced relatively easily. Also, tolerances and the
resultant component alterations that were only a short time ago only achievable by the most highly
perfected facilities can now be attained by much more ubiquitous equipment. Generally the industries
having metal cutting operations have been suffering from various big problems since the optimum
operating conditions for the machine tools cannot be easily achieved. The Industrial practitioners and
researchers have been dealing with this area to overcome such problems [2].
In machining, determining optimal cutting conditions or parameters under the given machining
situation is difficult in practice. Conventional way for selecting these conditions such as speed and
feed rate has been based upon data from machining handbooks and/or on the experience and
knowledge on the part of operator. Turning is the first most common method for cutting and
especially for the finishing machined parts. Turning is a cutting operation in which the part is rotated
as the tool is held against it on a machine called a lathe [4]. The raw stock that is used on a lathe is
usually cylindrical, and the parts that are machined on it are rotational parts. In a turning operation, it
is important task to select cutting parameters for achieving high cutting performance. Cutting
parameters are reflected on surface roughness, surface texture and dimensional deviations of the
981 Vol. 7, Issue 3, pp. 980-990

product (Nalbant, 2006). Surface finish obtained in manufacturing processes mainly depends on the
combination of two aspects: the ideal surface finish provided by marks that manufacturing process
produces on the surface and the actual surface finish which is generated taking into account
irregularities and deficiencies that may appear in the process, changing manufacturing initial
conditions (Arbizu and Perez, 2002).
The surface texture of an engineering component was very important. The surface of every part
includes some type of texture created by any combination of the following factors; the microstructure
of the material, the action of the cutting tool, cutting tool instability, errors in tool guide ways and
deformations caused by stress patterns in the component. It was affected by the machining processes
by changes in the conditions of either the component or machine (Abouelatta and Madl, 2000).
A machined surface carries a lot of valuable information about the process including tool wear, built-
up edge, vibrations, damaged machine elements etc. Consequently, a machined surface is a replica of
the cutting edge which carries valuable information related to the tool condition i.e., sharpness or
bluntness (Kasim et. al. 2006). (Chien & Chou, 2001) applied neural networks for modeling
surface roughness, cutting forces and cutting-tool life and applied a GA to find optimum cutting
conditions for maximizing the material removal rate under the constraints of the expected surface
roughness and tool life [7]. (Cus & Balic, 2003) also applied GA for optimizing a multi-objective
function based on minimum time necessary for manufacturing, minimum unit cost and minimum
surface roughness. All the process models applied in his research were empirical formulas from
machining handbooks which were fitted through regressions [11]. More complex models have also
been applied for surface roughness and tool wear modeling to optimize off-line cutting parameters.
(Zuperl & Cus, 2003) also applied and compared feed- forward neural networks for learning a multi-
objective function similar to the one presented in (Cus & Balic, 2003). It is nearly impossible to
discuss all the works related to Taguchi methods. We have tried to mention the main articles that
discuss the pros and cons of Taguchis contributions. There are several other papers that are listed in
the Bibliography but specifically not discussed here.
Organization of the work:
Step1: Design of Experiments
Step2: Taguchi Method Optimization
Step3: Test Analysis of variance
Step4: Multi-objective optimization - GA
Step5: Artificial Intelligence (AI): Simulation of ANN
Step6: Hybridization: GA + ANN = GONNs
Step7: Is the objective satisfies?
Step8: No, Tune-in optimum process
Step9: yes, Store the optimum solution
Step10: Compare the optimum solutions and report the best solution
II. MATERIALS AND EXPERIMENTATION
Machine: The experiment was carried out on the precision centre lathe (PSG A141) which enables
high precision machining and production of jobs. The main spindle runs on high precision roller
taper bearings and is made from hardened and precision drawn nickel chromium steel.
Technical Specifications are: centre height: 177.5mm, main motor power: 3hp, 30 longitudinal and
transverse feeds.
Work piece: Work piece of standard dimensions was used for machining: work piece diameter:
50mm, work piece length: 300mm (approx.).
The Cutting tool insert: The cutting selected for machining of AISI 4140 alloy steel was Chemical
Vapor Deposition coated cermet insert of 0.4 and 0.8 mm nose radii, 2 - 15 micro-m thick. Coated tips
typically have lives 10 times greater than uncoated tips. Common coating materials include titanium
nitride, titanium carbide and aluminum oxide, because it had been found that CVD tool is best choice
for machining of CRMOs due to its high wear resistance.
Lathe Tool Dynamometer: The instrument used for the measurement of cutting force was multi-
component force indicator. This instrument comprises independent DC excitation supply for feeding
982 Vol. 7, Issue 3, pp. 980-990

strain gauge bridges, signal processing systems to process and compute respective force values for
direct independent display. Instrument operates on 230V, 50Hz AC mains.
Surface Roughness measurement: The instrument used to measure surface roughness was
SURFTEST (Mitutoyo Make). For a probe movement of 5mm, surface roughness readings were
recorded at three locations on the work piece and the average value was used for analysis.

Table 1: Process variables and their limits

Fig 1: Experimental Setup

Table 2: Chemical composition of AISI 4140 [3]

Chemical Composition of AISI 4140
C Si Mn S P Cr Ni Mo Fe
0.20(max) 0.35(max) 0.50-1.00 0.040 0.040 0.75-1.25 1.00-1.50 0.08-0.15 Balance

In this experiment, in order to investigate the surface roughness of the machined work piece and
material removal rate, power consumption during cutting, a CVD coated tool was used. A view of the
cutting zone and Experimental setup is shown in Fig. 1. The surface roughness of the finished work
surface was measured with the help of a surface roughness tester and material removal rate, Power
consumption are calculated as below type.
The working ranges of the parameters for subsequent design of experiment, based on Taguchis L27
Orthogonal Array (OA) design have been selected. In the present experimental study, spindle speed,
feed rate and depth of cut have been considered as Process variables. The process variables with their
units (and notations) are listed in Table 1.

2.1 Experimental procedure
Turning is a popularly used machining process. The lathe machines play a major role in modern
machining industry to enhance product quality as well as productivity. In the present work, three
levels, three factors and twenty seven experiments are identified. Appropriate selection of orthogonal
array is the first step of Taguchi approach. According to Taguchi approach L27 orthogonal array has
been selected. Cutting tests were carried out on lathe machine under dry conditions. A pre-cut with a 1
mm depth of cut was performed on work piece of actual turning. This was done in order to remove the
rust layer or hardened top layer from the outside surface and to minimize any effect of in homogeneity
on the experimental results. After that, the weight of each samples have been measured accurately
with the help of a high precision digital balance meter. Then, using different levels of the process
parameters have been turned in lathe accordingly. Machining time, cutting forces for each sample has
been calculated accordingly. After machining, weight of each machined parts have been again
measured precisely with the help of the digital balance meter. Then surface roughness has been
measured precisely with the help of a portable Mitutoyo SJ-201P surftest. The results of the
experiments have been shown in table 3.a, 3.b & 3.c.

2.2 Calculation of the Material removal rate
Material removal rate (MRR) has been calculated from the difference of weight of workpiece before
and after experiment by using the following formula.
Factors Units Low Medium High
Speed rpm 740 580 450
Feed mm/rev 0.09 0.07 0.05
Depth of
cut
mm 0.25 0.20 0.10
983 Vol. 7, Issue 3, pp. 980-990

MRR = W i W f mm3 / min
* t
Where, Wi is the initial weight of work piece in g; Wf is the final weight of work piece in g; t is the
machining time in minutes; is the density of alloy steel (7.8 x 10-3 g/mm3).
2.3 Calculation of the power consumed
The cutting force system in 3D-turning consists of three forces: Fc is the largest force that accounts for
99% the power required, Fa requires very small power because feed rates are very small, and Fr the
radial force contributes very small also because velocity in the radial direction is negligible. Tool
Dynamometer is used for determining forces
Ignoring the thrust and radial forces, the total input power to cutting is given by:
PC = Fc * V
V= Volume of MR / t * Fc * Fa

III. DATA COLLECTION
The results of the experiments have been shown in Table 3 (a) to (c). Analysis has been made based
on those experimental data in the following session. Optimization of surface roughness, material
removal rate and power consumption of the cutting tool has been made by Taguchi method and
Genetic Algorithm coupled with Regression analysis, Analysis of Variance as well as Entropy
concept. Confirmatory tests have also been conducted finally to validate optimal results.

Table 3.a: Experimental Results L27 Orthogonal array

Table 3.b: Measurement of Material removal rate
Speed Feed DOC Surface Roughness, Ra(m) Speed Feed DOC
1 1 1 3.7624 2.8209 2 2 3
1 1 2 2.5408 4.2722 2 3 1
1 1 3 2.6463 1.9945 2 3 2
1 2 1 4.0304 1.9765 2 3 3
1 2 2 2.2705 3.1275 3 1 1
1 2 3 2.4472 1.9613 3 1 2
1 3 1 4.3392 2.2436 3 1 3
1 3 2 2.3151 3.5635 3 2 1
1 3 3 2.5792 2.0062 3 2 2
2 1 1 4.1644 2.8849 3 2 3
2 1 2 2.3815 3.0145 3 3 1
2 1 3 2.936 2.258 3 3 2
2 2 1 4.3271 2.883 3 3 3
2 2 2 2.6254 --
-- -- --
Sampl
e
Numb
er
Material
removal
rate
mm
3
/min
Sampl
e
Numb
er
Material
removal
rate
mm
3
/min
1 0.6234
15
0.1417
2 0.2564
16
0.6594
3 0.2685
17
0.3846
4 0.7792
18
0.1948
984 Vol. 7, Issue 3, pp. 980-990

Table 3.c: Measurement of Power consumed

IV. DATA ANALYSES
Experiment was conducted to assess the effect of Spindle speed, feed rate and depth of cut on the
surface finish and machine power consumption. Table 3 illustrates the experimental results of Ra,
MRR and PC.

A. Taguchi method
Taguchi Method is developed by Dr.Genichi Taguchi, a Japanese quality management Consultant, He
has developed both the philosophy and methodology for the application of factorial design
experiments that has taken the design of experiments from the exclusive world of the statistician and
brought it more fully into the world of manufacturing. His contributions have also made the
practitioners work simpler by advocating the use of fewer experimental designs, and providing a
clearer understanding of the nature of variation and the economic consequences of quality engineering
in the world of manufacturing and uses a statistical measure of performance called Signal-to-Noise
(S/N) ratio. Taguchi methods seek to remove the effect of noises, he pointed out that the key element
for achieving high quality and low cost is parameter design. The S/N ratio takes both the mean and
5 0.3550
19
0.1699
6 0.3568
20
0.1687
7 0.8727
21
0.0886
8 0.3550
22
0.3108
9 0.4918
23
0.3422
10 0.5647
24
0.2331
11 0.5273
25
0.1626
12 0.5367
26
0.1553
13 0.5648
27
0.1608
14 0.1417
-- --
Sample
Number
Power
consumed
KW
Sample
Number
Power
consumed
KW
1 9.8829 15 3.3306
2 3.1600 16 11.2389
3 10.5607 17 3.7078
4 20.8746 18 5.0348
5 4.6130 19 3.6868
6 10.1059 20 3.3031
7 15.3542 21 2.1369
8 3.6462 22 6.5768
9 11.7180 23 4.2897
10 17.9277 24 5.5792
11 9.6427 25 2.9818
12 5.0271 26 1.9107
13 13.8077 27 3.1278
14 1.4118 -- --
985 Vol. 7, Issue 3, pp. 980-990

the variability into account. The ratio depends on the quality Characteristics of the product/process to
be optimized.
The optimal setting is the parameter combination, which has the highest S/N ratio. The standard S/N
ratios generally used are as follows: - Nominal is Best (NB), Lower the Better (LB) and Higher the
Better (HB). Taguchi approach has potential for savings in experimental time and cost on product or
process development and quality improvement. Quality is measured by the deviation of a functional
characteristic from its target value. Through parameter design, levels of product and process factors
are determined, such that the products functional characteristics are optimized and the effect of noise
factors is minimized.
Taguchis ideas can be distilled into two fundamental concepts:
(a) Quality losses must be defined as deviations from targets, not conformance to specifications
(b) Quality is designed, not manufactured, into the product.
Main effect plot:
The analysis is made with the help of a software package MINITAB 16. The main effect plots are
shown inFig.2, 3 and 4.These show the variation of individual response with the three parameters i.e.
speed, feed, and depth of cut separately. In the plots, the x-axis indicates the value of each process
parameter at three level and y-axis the response value. Horizontal line indicates the mean value of the
response. The main effects plots are used to determine the optimal design conditions to obtain the
optimum surface roughness. Fig.2 shows the main effect plot for surface roughness.

Fig.2 Main effect plot for Ra Fig.3 Main effect plot for MRR Fig.4 Main effect plot for PC
According to main effect plots for S/N ratio Fig. 3, the optimal conditions for minimum Ra, maximum
MRR and minimum PC are:
Table 4: Optimal turning conditions

Responses Best-Levels
Ra 3-1-2
MRR 1-1-1
Power Consumed
3-1-2
B. Mathematical correlation
The linear polynomial models are developed using commercially available Minitab 16 software for
various turning parameters. The predictors are speed, feed and depth of cut. Linear regression
equations are used to develop a statistical model with an objective to establish a correlation between
the selected turning parameters with the quality characteristics of the machined work piece. The
regression equation for Surface roughness (Eq. 1) (Ra) = 1.07 + 0.218 *(S) - 0.023 *(F) + 0.277 *(D);
(Eq. 2) Material removal rate (MRR) = 1.67 x 10-5 + 1.76 *(S) + 0.110 *(F) + 0.586 *(D) and (Eq. 3)
Power consumption (PC) = 3.04 x 10-5 + 1.79 *(S) - 0.409 *(F) + 0.175 *(D)
Where Ra denotes the arithmetic average roughness-height (m), D the depth of cut (mm), S the
spindle speed (rpm), F the feed (mm/rev), R the coefficient of regression.
986 Vol. 7, Issue 3, pp. 980-990

C. Analysis of variance (ANOVA)

Analysis of Variance (ANOVA) is a powerful analyzing tool which is used to identify significant of
each parameter on output response. This is achieved by separating total variability of the S/N ratio,
which is measured by the sum of squared deviation (SST) from total mean S/N ratio into contribution
by each design parameters and the error. The F-value of each design parameter is a ratio of mean
squared deviations to the mean squared error. The Minitab 16 Software is used to identify various
terms in ANOVA. The table5, 6, and 7 shows the ANOVA for surface roughness, MRR and Power
consumption. Study of ANOVA table for a given analysis helps to determine which of the factors
need control and which do not. The ANOVA table shows effect individual effect of each parameter
and interaction effect of each parameter on output response.
ANOVA Terms & Notations: D.F =Degree of Freedom, S.S = Sum of Squares, M.S = Mean Square,
F =Variance Ratio & C = Percentage Contribution.
Table 5: ANOVA result for surface roughness Ra [95% confidence level]

SOURCE D.F S.S M.S F C%
(S) 2 0.8110 0.4055 1.89 5.69
(F) 2 0.1219 0.0609 0.28 0.86
(D) 2 12.499 6.2493 29.24 87.67
SXF 4 0.0379 0.0095 0.05 0.27
SXD 4 0.6894 0.1724 0.80 4.84
FXD 4 0.0953 0.0239 0.11 0.67
ERROR 8 1.6495 0.2062
TOTAL 26 14.254

100

Table 6: ANOVA result for Material removal rate (MRR) [95% confidence level]
SOURCE DOF S.S M.S F C%
(S) 2 0.3963 0.1982 0.93 48.03
(F) 2 0.0037 0.0018 0.08 0.45
(D) 2 0.3384 0.1691 0.81 41.02
SXF 4 0.0203 0.0101 0.04 2.46
SXD 4 0.0561 0.0280 0.14 6.80
FXD 4 0.0101 0.0050 0.03 1.24
ERROR 8 0.3611 0.0452

TOTAL 26 0.824997

100
Table 7: ANOVA result for Power consumption (PC) [95% confidence level]

SOURCE DOF S.S M.S F C%
(S) 2 204.0423 102.0212 477.38 36.46
(F) 2 16.2641 8.1320 38.05 2.90
(D) 2 247.1854 123.5927 578.33 44.18
SXF 4 54.142 13.5355 63.34 9.67
SXD 4 20.7271 5.1818 24.25 3.70
FXD 4 17.2881 4.3220 20.23 3.09
ERROR 8 207.5954 25.9494

TOTAL 26 559.6492

100
From ANOVA table for surface roughness it is clears that depth of cut (87.67%) is the major factor to
be selected effectively to get the good surface finish. The interaction SXD (4.84%) has more influence
than other two interactions. For material removal rate, speed (48.03%) is the major factor to be
selected to get high removal of material. The interaction SXD (6.80%) has more influence than other
two interactions. For power consumption (44.18%) is the major factor to be selected to get low
machining power. The interaction SXD (9.67%) has more influence than other two interactions.

987 Vol. 7, Issue 3, pp. 980-990

D. Artificial Neural networks (ANN)
Artificial Neural Networks (ANN) are a branch of the field known as "Artificial Intelligence" (AI)
which may also consists of Fuzzy logic (FL) and Genetic Algorithms (GA). ANN is based on the
basic model of the human brain with capability of generalization and learning. The purpose of this
simulation to the simple model of human neural cell is to acquire the intelligent features of these cells.
The term "artificial" means that neural nets are implemented in computer programs that are able to
handle the large number of necessary calculations during the learning process [12].
After conducting the DOE, an experimental database composed of 27 runs of 3 different cutting
conditions was generated. Each sample was defined by: surface roughness (Ra), material removal
rate (MRR) and power consumption (PC).The experimental database was used to learn three ANN
process models: (1) Ra, (2) MRR, (3) PC state model. Before training the ANN, a statistical study
was conducted for each model in order to discard those inputs variables that were not significant. The
final inputs variables applied for each model and the main characteristics of the ANN models are
shown in Table 8.
Table 8: Characteristics of ANN models
E. Genetic algorithm (GA)
MATLAB is another software package that has many capabilities to solve engineering problems
especially for constrained optimization problems. Genetic algorithms (GA) belong to the class of
stochastic search optimization methods. The concept of GA was developed by Holland in the 1960s
and 1970s [11]. These are the algorithms based on mechanics of natural selection and natural genetics,
which are more robust and more likely to locate global optimum. Chromosomes are made of discrete
units called genes, they assumed to be binary digits. GA operates with a collection of chromosomes
called a population. The population is normally randomly initialized. Three genetic operators are used
to accomplish this task: reproduction, crossover, and mutation. Reproduction involves selection of
chromosomes for the next generation. This is also called the selection process. The crossover operator
is the most important operator of GA. In crossover, generally two chromosomes, called parents, are
combined together to form new chromosomes, called offspring. The parents are selected among
existing chromosomes in the population with preference towards fitness so that offspring is expected
to inherit good genes which make the parents fitter. The mutation operator introduces random changes
into characteristics of chromosomes. Mutation is generally applied at the gene level. In typical GA
implementations, the mutation rate is very small and depends on the length of the chromosome.
Mutation plays a critical role in GA. Mutation reintroduces genetic diversity back into the population
and assists the search escape from local optima. The foregoing three steps are repeated for successive
Ra model MRR model PC model
Type Back propagation Type Back propagation Type Back propagation
Inputs S, F, doc Inputs S, F, doc Inputs S, F, doc
Output Ra Output MRR Output PC
Hidden
Layers

1
Hidden
Layers

1
Hidden
Layers

1
Neurons 3 Neurons 3 Neurons 3
Mapping
functions
Tran-sig
Mapping
functions
Tran-sig
Mapping
functions
Tran-sig
Training
Method
Lev-Marq.
Training
Method
Lev-Marq.
Training
Method
Lev-Marq.
Epochs 300 Epochs 300 Epochs 300
Learning
Rate
0.1
Learning
Rate
0.1
Learning
Rate
0.1
988 Vol. 7, Issue 3, pp. 980-990

generations of the population until no further improvement in fitness is attainable. Besides weighting
factors and constraints, suitable parameters of GA are required to operate efficiently.
Genetically optimized neural networks (GONNs) are the new inviting methods for complex
optimization of cutting parameters, describes the multiobjective technique of optimization of cutting
conditions by means of the neural networks taking into consideration the technological, economic and
organizational limitations. The GA parameters along with relevant objective functions and set of
machining performance constraints are imposed on GA optimization methodology to provide
optimum cutting conditions. The multi-Objective fitness function is as following type.
function y=simple_fitness(x)
y=(w1*(k*x(1)â*x(2)^b*x(3)^c)^-
1)+(w2*(k*x(1)â*x(2)^b*x(3)^c))+(w3*(k*x(1)â*x(2)^b*x(3)^c)^-1);
FitnessFunction=@simple_fitness;
E.1 Multi-objective GA
GA optimization methodology is based on machining performance predictions models developed
from a comprehensive system of theoretical analysis, experimental database and numerical methods.
Being a population-based approach, GA are well suited to solve multi-objective optimization
problems. A generic single-objective GA can be modified to find a set of multiple non-dominated
solutions in a single run. There are several multi-objective evolutionary algorithms; here Weight-
based Genetic Algorithm (WBGA) is implemented and its characteristics are shown in table 9.
E.2 Weighted sum approaches to Fitness function
The classical approach to solve a multi-objective optimization problem is to assign a weight Wi to
each normalized objective function f(x) so that the problem is converted to a single objective
problem with a scalar objective function as follows:
y= w1 f1 (x) + w2 f2 (x) + .. +wk f k (x) . (1)
Where f1 (x) is the normalized objective function and SUM (Wi) =1. This approach is called the priori
approach since the user is expected to provide the weights, here the estimation of weights are done by
Entropy Method [13]. Solving a problem with the objective function (1) for a given weight vector Wi
= {w1, w2,...wk} yields a single solution, and if multiple solutions are desired, the problem must be
solved multiple times with different weight combinations. The main difficulty with this approach is
selecting a weight vector for each run. To automate this process; weight based genetic algorithm for
multi-objective optimization, each solution in the population uses a different weight vector Wi = {w1,
w2,.. wk} in the calculation of the summed objective function (1). The weight vector Wi is embedded
within the chromosome of solution x i. Therefore, multiple solutions can be simultaneously searched
in a single run. In addition, weight vectors can be adjusted to promote diversity of the population.
Table 9: Characteristics of Genetic search algorithms
Genetic Algorithm
Variables to optimize Speed, Feed, Depth of cut
Population Size 10
Generations 15
Crossover Frac. 0.8
Elite Count 2
Mutation function Gaussian
Selection function Stochastic
Boundary ranges S=[450,580,740], F=[0.05,0.07,0.09], Doc=[0.10,0.20,0.25]

Stop criterium
Stall Time: 6s
Stall Generations: 7
Table 10: Optimal Turning Conditions & Pin Point Optimal Values

Responses S F D EXP MODEL
(Eq. 1,2,3)
GA GONNs
Ra(m) 450 0.09 0.20 1.96 2.18 2.24 2.10
MRR(mm3/min) 740 0.09 0.25 0.62 0.59 0.66 0.62
PC(KW) 450 0.09 0.20 3.30 3.41 3.53 3.35
989 Vol. 7, Issue 3, pp. 980-990

F. Confirmation Experiment
Table 10 shows the turning conditions, the pin point optimal values of controlling factors, results
obtained from the confirmation test, calculated from the developed model [Eq. 1,2,3], and identify in
between the results. Therefore, Eq. (1,2,3) correlates the arithmetic mean surface roughness, material
removal rate and power consumed with the turning conditions (depth of cut, speed, and feed) with a
realistic degree of approximation.
V. FUTURE WORK
AI models are called black-box models, and they cannot be optimized with conventional optimization
methods. Due to this limitation, a cutting parameter optimization methodology based on AI models
requires an advanced search methods for global optimization, such as Genetic Algorithms (GA) and
Mesh Adaptive Direct Search (MADS) algorithms. For better and better acceptance hybrid of GA-
ANN and MADS-ANN ate can be used to find the global optimal solution in complex
multidimensional search spaces.
VI. CONCLUSION
There is general agreement that off-line experiments during product or process design stage are of
great value. Reducing quality loss by designing the products and processes to be insensitive to
variation in noise variables is a novel concept to statisticians and quality engineers.
The fluctuation of average surface roughness (Ra), Material removal rate (MRR) and Power
consumed (PC) values resulted from machining is explained by the fact that they depends on the
turning parameters.
Multiple regressions produced a satisfactory correlation for the prediction of output responses with
reasonable degree of approximation. The ANOVA revealed that the depth of cut is the dominant
parameter followed by speed for surface Roughness. In case of MRR response, the speed is the
dominant one followed by the depth of cut and incase of PC response, depth of cut is the dominant
one followed by feed
Therefore, the above results (table 10) conclude that with-in the domain genetically optimized Neural
Networks having better capability. GONNs has been the most popular heuristic approach to multi-
objective design and optimization problems. The aforesaid methods Taguchi, Regression, Artificial
neural networks and Genetic-Algorithm can be applied for continuous quality improvement of the
product/process and off-line quality control.
REFERENCES
[1] Meyers A.R. and Slattery T.J., Basic Machining Reference Handbook, Industrial Press Inc, (2001)
[2] Shaw M., Metal Cutting Principles, Oxford University Press, 1984.
[3] Kalpakjian S. and Schmid S.R., Manufacturing Processes for Engineering Materials, 5th ed., Pearson
Education, Inc., SBN 9-78013227-271-1 (2008)
[4] Boothroyd G. and Knight W.A., Fundamentals of Machining and Machine Tools, 3rd Ed., CRC Publication.
ISBN 1-57444-659-2, (2006)
[5] Groover M.P., Fundamentals of Modern Manufacturing, Prentice-Hall, Upper Saddle River, NJ (now
published by John Wiley, New York), 634, 1996.
[6] Kalpakjian S. and Schmid S.R., Manufacturing Processes for Engineering Materials, 5th ed., Pearson
Education, Inc., SBN 9-78013227-271-1, 2008.
[7] Chien, W. T. and Chou, C. Y. (2001). The predictive model for machinability of 304 stainless steel.
Journal of Materials Processing Tech., 118, 1-3, (2001), 442447, ISSN 0924-0136.
[8] Rodrigues L.L.R., Kantharaj A.N., Effect of Cutting Parameters on Surface Roughness and Cutting Force in
Turning Mild Steel., Research Journal of Recent Sciences- ISSN 2277-2502, Vol. 1(10), 19-26, 2012.
[9] Hardeep Singh1*, Rajesh Khanna2, M.P. Garg2., Effect of Cutting Parameters on MRR and Surface
Roughness in Turning EN-8., Current Trends in Engineering Research Vol.1, No.1, 2011.
[10] Anirban Bhattacharya, Santanu Das, P. Majumder (2009), Ajay Batish, Estimating the effect of cutting
parameters on surface finish and power consumption during high speed machining of AISI 1045 steel using
Taguchi design and ANOVA, Prod. Eng. Res. Devel, vol. 3, pp 3140
990 Vol. 7, Issue 3, pp. 980-990

[11] Cus, F. and Balic, J. (2003). Optimization of cutting process by GA approach and Computer-
Integrated Manufacturing, 19, 1-2, 113121, (2003), ISSN 0736-5845.
[12] Asiltrk I. and unkas M., Modeling and prediction of surface roughness in turning operations using
artificial neural network and multiple regression method, Expert Systems with Applications, 38(5), 5826-5832
(2010)
[13] K.srinivasa Raju & D.Nagesh Kumar: Multi criterion analysis in engineering and management, text book.
ABOUT THE AUTHOR
Chittam Rajesh was born in 25th Sep 1988. He is currently working as a Lecturer in
Mechanical Engineering Department of Sri kalahasteeswara Institute of Technology,
Srikalahasthi (A.P). He has 2 years experience in teaching, one year industry. He received
his B.Tech in Mechanical Engineering from Jawaharlal Nehru Technological University,
Anantapur (A.P.) in 2009. He has done M.Tech from SRI VENKATESWARA
UNIVERSITY, Tirupathi (A.P) in 2012 and planning for Research programme. His research
interests are Advanced machining and Manufacturing technologies.

991 Vol. 7, Issue 3, pp. 991-997

A MODIFIED SINGLE-FRAME LEARNING BASED SUPER-
RESOLUTION AND ITS OBSERVATIONS
Vaishali R. Bagul
1
and Varsha M. Jain
2

1
Student,
2
Professor,
PG department, MBES College of Engineering, Ambajogai, Maharashtra, India

ABSTRACT
In this paper, we would focus on various super-resolution image reconstruction techniques. Super-resolution is
nothing but conversion of a low resolution image into a high resolution one using a set of training images. In this
paper, we have discussed the learning based and total variation regularization methods for super resolution. Also
we have discussed the methods with their advantages and disadvantages and compared the performance of every
method including their PSNR values and the computational time. Higher number of training images leads to
increase in the computational time of that method, which in turn makes the method difficult to implement in real
time. We will also combine single-image learning based method along with total variation regularization and
obtain its results. Super-resolution technology has applications in many areas of image processing such as
medical imaging, image sensing systems, satellite imaging, space technology and surveillance systems etc. By
comparison, we have put forward the areas where we need to make improvements along with the future scope.
KEYWORDS: Super resolution, learning based, total variation, Regularization, TV up sampling.
I. INTRODUCTION
Now-a-days image enhancement is applied in various image processing areas. Be it in surveillance,
imaging systems, TV signal conversions etc. Low resolution cameras give images with low-resolution
i.e. images from which one cannot extract exact or minute details. In order to extract those details there
needs an increase in the resolution of the image. There are various up-sampling methods and
interpolation techniques to increase the resolution of an image [1]-[12]. Super-resolution is a
technique to increase the resolution of the image for which certain pairs of low and high resolution
training images are utilized. Increasing the resolution means to increase the size of the image and the
pixel number [2].
The main aim of this paper is to check the performance of single-image learning based super-resolution
method [8]-[12] using total variation method [13]-[18] and compare it with the existing methods. We
have to check the performance of the super resolution technique by reducing the computational time
and complexity of the system and getting a good quality image. In order to do so we are extending
previous method which was described in [2]. Also it has to be checked that the artifacts are reduced
while a super resolved image is reconstructed. We call image super-resolution as image up scaling,
image zooming, image magnification, image up sampling etc.
Li et al in [1] proposed a non-iterative adaptive interpolation scheme for natural image sources. In this
method, a switching between two methods i.e. bilinear interpolation and covariance based adaptive
interpolation is done in order to reduce the computational complexity of the system. Sun et al in [4]
proposed a Bayesian approach to image hallucination. Primal sketch priors are constructed and used to
enhance the quality of the hallucinated high resolution image. Chang et al [5] have generated a high
resolution image from a single low resolution image, with the help of a set of one or more training
images from scenes of the same or different types.
992 Vol. 7, Issue 3, pp. 991-997

Various methods are used to recover or obtain a high resolution (HR) image from one or more low
resolution (LR) images [2]-[12]. Pixel replication and bilinear interpolation are the standard
interpolation techniques that tend to increase the pixel count without actually adding the image details.
These techniques blur edges and other sharp details of the images but perform well in smoother regions
[3]. In conventional multi- frame super- resolution methods, [2] - [4] many low resolution images of
the same scene with varying pixel shifts are taken as inputs and correspondence between high and low
resolution patches is learned from a database which consists of low and high resolution image pairs and
then by applying this knowledge to a new low resolution (LR) image, a high resolution (HR) image is
reconstructed.
Super-resolution image reconstruction has got applications in areas related to image processing. It has
got applications in medical imaging, satellite imaging, TV signal conversion (NTSC to HDTV),
surveillance systems, various computer tools like printers, media players, image editing softwares, etc.
[11]. Super resolution (SR) can be achieved using a single image or multiple images (image sets). On
that basis, SR image reconstruction methods may be classified into 2 classes: (i) Classical multi-image
and (ii) Example- based super- resolution. Glasner et. al [7] combined the above two methods so as to
obtain super - resolution from a single image. These single-frame super-resolution methods are also
called Example based super-resolution described by Freeman et al. in [6] which states that to generate
an upscaled image with desired number of pixels it takes two steps: first one is interpolation to double
the number of pixels and the other being prediction of missing details.
Purkait et al in [10] have proposed image zooming technique using fuzzy rule based prediction. In [18],
single image super-resolution algorithm based on spatial and wavelet domain is presented. In [19], a
survey on techniques and challenges in image super resolution reconstruction are discussed.
Regularization literally means to remove the noise from an image. Total variation regularization leads
to reduction in the artifacts while maintaining the sharp edges [20]. In this paper, Learning based method
using single image is combined with total variation regularization in order to get an improved image
quality. It is different from [2] in the fact that in this case a single image is used for learning. Thus, a
huge database of images is not needed for checking correspondence between the image patches so there
is no problem of data redundancy as only high resolution patches constitute its database.
The rest of the paper is organized as follows. In section II, we have described the Learning based, Total
variation Regularization method and Single image learning based using total variation regularization.
In section III, we have described experimental results of combined total variation regularization method
and learning based method and our approach towards it. In section IV and V, we have concluded the
paper along with its future scope.
II. COMPARISON OF SUPER RESOLUTION METHODS
This section describes the Learning based and the Total variation regularization methods for Super
resolution of an image.

2.1. Learning based methods
Learning based super resolution can be achieved by use of either a multiple pairs of LR-HR images or
by using single image. Single image super- resolution can be categorized into three classes: functional
interpolation, reconstruction based and learning based. Learning based method is the most used method
for super resolution of an image. There are several learning based methods proposed for super-
resolution of images [2] - [11]. As described in [2],[3],[4], etc., multiple image super resolution which
needs a huge database of low and high resolution images. Due to this, the computational complexity
and time increases. So this paper describe single image learning based method for super resolution of
an image. A general flow of learning based method is given below -
1. Take a low resolution input image
2. Pass it through a High pass filter and get HF and LF components of the image
3. Upsample the low frequency (LF) using some interpolation method
4. Upsample the high frequency (HF) LR components using training image sets from database
5. Perform a correlative search between LR and HR components of training image
993 Vol. 7, Issue 3, pp. 991-997

6. Select the corresponding patches and add to upsampled low frequency HR output to get final
HR output image.
In this method, low resolution patch size used is 3 3 pixels. Using this method, one can obtain images
with better quality. The drawback of this method is patch redundancy or database redundancy and large
processing time required for searching similar patches in the database [2]. Using this method, a better
quality super resolution image is obtained.
2.2. Total variation regularization methods
Total variation regularization also known as total variation decomposition is a method used to separate
the low and high frequency components of an image. Regularization, in mathematics and statistics and
particularly in the fields of machine learning and inverse problems, refers to a process of introducing
additional information in order to solve an ill-posed problem or to prevent over fitting [17] . As per
Rudin et al [12], regularization is based on the principle that signals with excessive and possibly
spurious detail have high total variation, that is, the integral of the absolute gradient of the signal is high
[20].Various regularization methods are described in [11] - [16], etc.
Two terms, well - posed and ill- posed were described by Hadamard in [15]. He stated that a problem
is said to be well-posed if a unique solution is present for it which is continuously dependent on the data
and when this is not the case, the problem is said to be ill-posed. The flow of this method is as follows
[2]:
1. Take a low resolution image as input
2. Apply total variation decomposition to obtain structure and texture components
3. The low frequency, Structure (cartoon) part is upsampled by using TVR upsampling.
4. The high frequency, Texture part is upsampled using linear interpolation
5. At last, add the high resolution structure and texture components to obtain a high resolution
output image.
The structure part consists of low frequency components along with edge components whereas the
texture component consists of high frequency components minus edge components. In total variation
regularization, ROF method [12] or Chambolle method [13] can be used. These two methods are
iterative methods amongst which the Chambolle method is generally used because its convergence is
faster. In [12], Non-linear methods for noise removal are used where the denoising of images is done
by minimizing the total variation norm of the estimated solution. The structure and texture can be
upsampled both by using either TV upsampling or Linear interpolation. When both components are TV
upsampled the image is quite sharp and looks quite clear and natural
2.3. Single image learning based method using TVR
In our method, there is a low resolution image Xi as an input. Figure 1 below shows a general schematic
of the process. The input image Xi is decomposed using total variation regularization method. The input
image is decomposed into structure (cartoon) and texture components [2]. Structure component (u0) is
the low frequency component and texture component (v0) contains high frequencies. Now the structure
component is upsampled using total variation upsampling method to get HR structure component (U0)
and the texture component is upsampled using bilinear interpolation to get HR texture component (V0).
The texture component is learning processed for which we take an image for training which is same as
the input image. A patch size of 3 3 pixels without any overlapping of pixels is taken. The
corresponding HR patches from learning image are added to the upsampled texture component. Finally
the HR structure and HR texture components are added in order to get an output HR image.
One can upsample the structure and texture components both either by using simply linear interpolation
or by using Total variation upsampling or by using a good interpolation method. Various linear or non-
linear filters that would help sharpen the image can be used at the output. In this paper, the output image
obtained after this processing is goes through 2D Haar Wavelet transform. The input image has 3
dimensions so the wavelet transformation is applied and to any 2 dimensions [18]. Instead of using haar
wavelets, we can use various filters such as those used in [11] in order to get sharper edges. Our method
is different form [2] in the fact that we are using a single image for the learning based processing and
that our output image gets smoother edges, which can be later on sharpened using various techniques.
994 Vol. 7, Issue 3, pp. 991-997

Figure 1. Proposed system
III. EXPERIMENTAL RESULTS
This method has been implemented on various images using MATLAB 2013a software. The system
has got a dual core processor with speed 2.20 GHz and a 4 GB RAM. For images with lower resolutions
it needs less time to process. In Table 1. Gives the comparison between PSNR and processing time
taken by images of two different sizes. Figure 2 (a),(b) and (c) respectively shows a flower image to
which learning based method using single image is applied, TVR method and LBM + TVR using Single
Image respectively. Figure 3 (a), (b), (c), (d) displays the Learning based image, TVR image, combined
learning and total variation method [2] image and finally image obtained from combination of single
image learning and total variation regularization method respectively.
The processing speed can be increased by using a system with high configuration for implementation.
We have used images of various sizes and converted them into images of size 128 128 pixels and 256
256 pixels. With 128 128 pixel size image, one can get the results as shown in Table 1. The number
of iterations for processing in this method is equal to 30. Patch size is taken as 3 3 pixels for LR
image. As we increase the resolution of the input image the processing time increases and so do the
artifacts get introduced. Using this method, one can reconstruct images which can be zoomed upto 2X.
(Magnification factor = 2). As seen in table above, high PSNR values are obtained and time utilised is
also low.
Table 2. The comparison between the PSNR values and processing time required for 2 methods of Total
variation Regularization for different input images is done. In the 1
st
TV method, we have Total
variation upsampled both the structure and texture components and obtained output images whereas in
the 2
nd
method one can upsample structure component using TV upsampling and texture component by
Bicubic interpolation [2]. Using TVR, one can get sharp edges and clear image. Super-resolved image
is very similar to the original image
Table. 1. Comparison of PSNR and Processing time of LBM + TVR using 2- D Haar wavelet transform

Image 128 128 pixels 256 256 pixels
PSNR(dB) Time (sec) PSNR (dB) Time (sec)
Sono 1 46.4692 22.621522 56.6856 90.053187
Red rose 46.1070 24.007834 57.1931 91.943678
CT image 51.4386 23.304213 62.9225 93.818613
Pink rose 47.7972 23.661270 57.1931 92.304604
995 Vol. 7, Issue 3, pp. 991-997

Table 2. Comparison of PSNR and Processing time for TVR method using 2 different methods

Figure 2. Flower image processed using (a) Learning based method (b) TVR method and (c) LBM + TVR
method using Haar wavelet transform

Figure 3. Lighthouse image processed using (a) Learning based method (b) TVR method (c) LBM + TVR [2]
(d) LBM+TVR using Single image
IV. CONCLUSIONS
When we combined single image learning based method and total variation regularization method we
got resulting images which had smoother edges. In order to sharpen the image we used 2D - Haar
wavelet transform which helped us to combine the structure and texture components so that the image
looks better. The computational time required for processing is reduced. In Table 1 and 2, we have
compared the PSNR values and Computational time for every method. Our method is simple to execute,
though it needs images sharpening and contrast enhancement. The system complexity is reduced and
the computational time too. We get high PSNR values for the output images. Fig 2 shows the
comparison of images super resolved and reconstructed using learning based method, total variation
regularization method and our method. This method is quite useful in medical imaging for processing
CT images. We get reduced artifacts using this approach
V. FUTURE WORK
Super resolution techniques can be applied in various image processing areas. We can also apply fuzzy
rule based prediction to individual components so as to get an improved image quality. The process is
simple in computation, no huge databases required for processing the images. A developed algorithm
can also improve the image quality. In order to get a good quality image, we can combine the Tikhonov
regularization with TV regularization so that the images are not too smooth. We can also change the
color model in case of color image processing. Convert RGB model into YIQ or YCbCr model and then
apply interpolation to individual components for better results.

Input Image Total variation method 1 Total variation method 2
Time (sec) PSNR (dB) Time (sec) PSNR (dB)
Red rose 18.800081 66.8998 30.225065 62.8053
Old man 16.046236 64.7839 30.346805 60.3586
Baby 23.692761 63.6995 32.417537 60.9683
Mango 29.777048 65.3266 55.420420 62.1022
Pink rose 19.397311 70.0629 30.902189 64.8487
996 Vol. 7, Issue 3, pp. 991-997

ACKNOWLEDGEMENTS
The authors would like to thank Dr. B. I. Khadakbhavi, Principal, College of Engineering, Ambajogai
and also Dr. V. G. Kasbegoudar, PG Dean, College of Engineering, Ambajogai for their time to time
and valuable guidance. Also we would express our sincere gratitude towards Dr. B. M. Patil for their
immense support and boost.
REFERENCES
[1]. X. Li and M. T. Orchard, New edge-directed interpolation (201) IEEE Transactions on Image
Processing, Vol. 10, No. 10, pp. 1521-1527.
[2]. T. Goto, Y. Kawamoto, Y. Sakuta, A. Tsutsui , M. Sakurai, Learning-based Super-resolution Image
Reconstruction on Multi-core Processor (2012) IEEE Transactions on Consumer Electronics, Vol. 58,
No. 3, pp. 941- 946.
[3]. P. P. Gajjar and M. V. Joshi, New learning based super-resolution: Use of DWT and IGMR-F prior
(2010) IEEE Trans. on Image Processing, Vol. 19, No. 5, pp. 1201-1213
[4]. J. Sun , N. N. Zheng, H. Tao and H. Y. Shum, Image Hallucination with primal sketch priors (2003)
IEEE Computer society Conference on Computer Vision and Pattern Recognition, Vol. 2, pp. 729-736.
[5]. H. Chang, D. Y. Yeung and Y. Xiong, Super-resolution through neighbor embedding (2004) IEEE
Computer society Conference on Computer Vision and Pattern Recognition, Vol.1, pp. 275-282.
[6]. S. Dai, M. Han, W. Xu, Y. Wu, and Y. Gong, Soft edge smoothness prior for alpha channel super
resolution (2007) IEEE Computer Vision and Pattern Recognition, pp. 1-8.
[7]. A. Lukin , A. S. Krylov and A. Nasanov, Image Interpolation by Super- resolution [2006] International
conference Graphicon, Novosibirsk Akademgorodok, Russia.
[8]. W. T. Freeman, T. R. Jones, and E. C. Paztor, Example based super- resolution (2002) IEEE Computer
Graphics and Applications, Vol. 22, No. 2, pp. 56-65.
[9]. D. Glasner, S. Bagon, M. Irani, Super-Resolution form a Single Image (2009). ICCV
[10]. P. Purkait and B. Chanda, Fuzzy- Rule based Approach for Single Frame Super Resolution (2013) In
Proceedings of the IEEE International Conference on Fuzzy Systems (Fuzz-IEEE 13).
[11]. M. Sakurai , Y. Sakuta, M. Watanabe, T. Goto, S. Hirano, Super-resolution through Non-linear
Enhancement filters (2013) IEEE International Conference on Image Processing, pp. 854-858
[12]. S. Naik and N. Patel, Single image Super- resolution in Spatial and Wavelet domain (2013) The
International Journal of Multimedia & its Applications, Vol. 5, No. 4, pp. 23-32.
[13]. H. A. Aly and Eric Dubois, Image Up-sampling using Total Variation Regularization with a New
Observation Model (2005) IEEE Transactions on Image Processing, Vol. 14, No. 10, pp. 1647-1659.
[14]. L. Rudin, S. Osher, E. Fatemi, Nonlinear total variation based noise removal algorithm (1992)
Physica D, Vol. 60, pp. 259 268.
[15]. A. Chambolle , An algorithm for total variation minimization and applications (2004) J. Mathematical
Imaging and Vision, Vol. 20, No. 1, pp. 89-97.
[16]. F. Guichard and F. Malgouyres, Total variation based interpolation (1998) Proceedings of Eusipco98,
pp. 1741-1744.
[17]. J. Hadamard (1952) Lectures on Cauchys Problem in Linear Partial Differential Equations, New York:
Dover.
[18]. Michael. K. Ng, H. Shen, E. Y. Lam, L. Zhang, A Total Variation Regularization based Super-
resolution Reconstruction Algorithm for Digital Video (2007) EURASIP Journal on Advances in Signal
Processing, Volume 2007, Article ID 74584,2007.
[19]. H. Pandey, P. Swadas, M. Joshi, A Survey of Techniques and Challenges in mage Super-resolution
Reconstruction (2013), International Journal of Computer Science and Mobile Computing, Vol. 2,
Issue. 4, pp. 317 -325.
[20]. Wikipedia contributors, 11 April 2014, Total variation denoising,
en.wikipedia.org/wiki/Total_variation_denoising
AUTHORS
Vaishali R. Bagul, received B. E. degree from Department of Electronics and
Telecommunications Engineering, GECA, Aurangabad, India in 2010. She is currently
pursuing her M.E. degree from COE, Ambajogai, India. Her interests are image and signal
processing. Her research interests are image segmentation, image enhancement, etc.

997 Vol. 7, Issue 3, pp. 991-997

Varsha M. Jain, received B. E. degree from Department of Electronics and
Telecommunications, WIT, Solapur, India in 1996. She has completed her M.Tech from
Department of Electronics Engineering, VJTI, Mumbai, India in 2007. She is currently
working as an Associate Professor with the Department of Electronics Engineering, COE,
Ambajogai. Her interest areas include Matlab Signal and Image processing. Her research
interests are VLSI and Embedded system.

998 Vol. 7, Issue 3, pp. 998-1002

DESIGN MODIFICATION AND ANALYSIS OF TWO WHEELER
COOLING FINS-A REVIEW
Mohsin A. Ali
1
and S.M Kherde
2

1
Mechanical Engineering Department, KGIET, Amravati, India
2
Professor, Mechanical Engineering Department, KGIET, Amravati, India

ABSTRACT
Engine life and effectiveness can be improved with effective cooling. The cooling mechanism of the air cooled
engine is mostly dependent on the fin design of the cylinder head and block. Insufficient removal of heat from
engine will lead to high thermal stresses and lower engine efficiency. The cooling fins allow the wind and air to
move the heat away from the engine. Low rate of heat transfer through cooling fins is the main problem in this
type of cooling. The main of aim of this work is to study various researches done in past to improve heat
transfer rate of cooling fins by changing cylinder block fin geometry and climate condition.
KEYWORDS: cooling fins, Heat transfer, Convection & Thermal Stresses.
I. INTRODUCTION
In Engine When fuel is burned heat is produced. Additional heat is also generated by friction between
the moving parts. Only approximately 30% of the energy released is converted into useful work. The
remaining (70%) must be removed from the engine to prevent the parts from melting. For this purpose
Engine have cooling mechanism in engine to remove this heat from the engine some heavy vehicles
uses water-cooling system and almost all two wheelers uses Air cooled engines, because Air-cooled
engines are only option due to some advantages like lighter weight and lesser space requirement. The
heat generated during combustion in IC engine should be maintained at higher level to increase
thermal efficiency, but to prevent the thermal damage some heat should remove from the engine. In
air-cooled engine, extended surfaces called fins are provided at the periphery of engine cylinder to
increase heat transfer rate. That is why the analysis of fin is important to increase the heat transfer
rate. Computational Fluid Dynamic (CFD) analysis have shown improvements in fin efficiency by
changing fin geometry, fin pitch, number of fins, fin material and climate condition.
In the research of J. Ajay Paul and Sagar Chavan Vijay [2] Parametric Study of Extended Fins in the
Optimization of Internal Combustion Engine they found that for high speed vehicles Engines thicker
fins provide better efficiency. When fin thickness increases, the gap between the fins reduces that
resulted in swirls being created which helped in increasing the heat transfer. Large number of fins
with less thickness can be preferred in high speed vehicles than thick fins with less numbers as it
helps inducing greater turbulence. [2]
Author plotted the experimental results in figure 1, it shows the variation of the heat Transfer with
respect to velocity. Ansys fluent software was used to predict the behavior or wind flow and analysis.
At zero velocity it is seen that the heat transfer from the 4mm and 6mm fins are the same. When the
velocity is increased it can be seen that the heat transfer is increased with due to forced convection
and also due to the swirl generated between two fins which induces turbulences and hence higher heat
transfer. For a larger fin thickness, the corresponding fin spacing is comparatively small. As a
consequence, the generated swirled flow may mingle with the main flow and result in a higher heat
transfer performance. [2]
999 Vol. 7, Issue 3, pp. 998-1002

The heat transfer from 6mm fins is found to be the higher at high velocities. For high speed vehicles
thicker fins provide better efficiency. When fin thickness was increased, the reduced gap between the
fins resulted in swirls being created which helped in increasing the heat transfer. Large number of fins
with less thickness can be preferred in high speed vehicles than thick fins with less numbers as it
helps inducing greater turbulence. [2]

Figure 1. Results of Analysis
In the research of J.C. Sanders et.al. [3] to find out overall heat transfer coefficient from the barrel he
conducted experiment to of cooling tests on two cylinders, one with original steel fins and one with 1-
inch spiral copper fins brazed on the barrel. The copper fins improved the overall heat transfer
coefficient from the barrel to the air 115 percent. They also concluded that in the range of practical
fins dimensions, copper fins having the same weight as the original steel fins will give at least 1.8
times the overall heat transfer of the original steel fins.
On the other hand Kumbhar D.G et.al. [4] Heat transfer augmentation from a horizontal rectangular
fin by triangular perforations whose bases parallel and towards the fin base under natural convection
has been studied using ANSYS. They have concluded that the heat transfer rate increases with
perforation as compared to fins of similar dimensions without perforation. The perforation of the fin
enhances the heat dissipation rates at the same time decreases the expenditure for fin materials also.
N.Nagarani and K. Mayilsamy, Experimental heat transfer analysis on annular circular and elliptical
fins.[6]This other had analyzed the heat transfer rate and efficiency for circular and elliptical annular
fins for different environmental conditions. Elliptical fin efficiency is more than circular fin. If space
restriction is there along one particular direction while the perpendicular direction is relatively
unrestricted elliptical fins could be a good choice. Normally heat transfer coefficient depends upon
the space, time, flow conditions and fluid properties. If there are changes in environmental conditions,
there is a change in heat transfer co-efficient and efficiency also. [5]
To compare the rate of heat transfer with solid and permeable fins Ashok Tukaram Pise and Umesh
Vandeorao Awasarmol [7] conducted the experiment. Permeable fins are formed by modifying the
solid rectangular fins with drilling three holes per fins incline at one half lengths of the fins of two
wheeler cylinder block. Solid and permeable fins block are kept in isolated chamber and effectiveness
of each fin of these blocks were calculated. Engine cylinder block having solid and permeable fins
were tested for different inputs (i.e. 75W, 60W, 45W, 30W, 15W). It was found that permeable fins
block average heat transfer rate improves by about 5.63% and average heat transfer coefficient 42.3%
as compared to solid fins with reduction of cost of the material 30%.
Optimal Design of an I C engine cylinder fin array using a binary coded genetic algorithm by
G.Raju, Dr. Bhramara Panitapu, S. C. V. Ramana Murty Naidu. [8]This study also includes the effect
of spacing between fins on various parameters like total surface area, heat transfer coefficient and
total heat transfer .The aspect rations of a single fin and their corresponding array of these two
profiles were also determined. Finally the heat transfer through both arrays was compared on their
weight basis. Results show the advantage of triangular profile fin array. Heat transfer through
1000 Vol. 7, Issue 3, pp. 998-1002

triangular fin array per unit mass is more than that of heat transfer through rectangular fin array.
Therefore the triangular fins are preferred than the rectangular fins for automobiles, central processing
units, aero-planes, space vehicles etc where weight is the main criteria. At wider spacing, shorter
fins are more preferred than longer fins. The aspect ratio for an optimized fin array is more than that
of a single fin for both rectangular and triangular profiles. [8]
R.P. Patil and H.M. Dange [9] conducted CFD and experimental analysis of elliptical fins for heat
transfer parameters, heat transfer coefficient and tube efficiency by forced convection. The
experiment is carried for different air flow rate with varying heat input. The CFD temperature
distribution for all cases verifies experimental results. At air flow rate of 3.7 m/s, the heat transfer rate
decreases as heat input increases. Also h is higher at above atmospheric temperature and lower at
below atm. Temperature. At air flow rate of 3.7 m/s the efficiency, increases as heat input increases.
Magarajan U. , Thundil karrupa Raj R. , Elango T. Numerical study on heat transfer I C Engine
cooling by extended fins using CFD [4]In this study, heat release of an IC engine cylinder cooling
fins with six numbers of fins having pitch of 10 mm and 20 mm are calculated numerically using
commercially available CFD tool Ansys Fluent. The IC engine is initially at 150 and the heat release
from the cylinder is analyzed at a wind velocity of 0 km/h. It is observed from the CFD result that it
takes 174.08 seconds (pitch=10mm) and 163.17 secs ( pitch =20mm) for ethylene glycol domain to
reach temperature of 423 K to 393 K for initially. The experiment results shows that the value of heat
release by the ethylene glycol through cylinder fins of pitch 10mm and 20mm are about 28.5W and
33.90 W.[10]
Pulkit Agarwal et.al [11] simulated the heat transfer in motor-cycle engine fins using CFD analysis. It
is observed that when the ambient temperature reduces to a very low value, it results in overcooling
and poor efficiency of the engine. It is observed that when the ambient temperature reduces to a very
low value, it results in overcooling and poor efficiency of the engine. They have concluded that
overcooling also affects the engine efficiency because of overcooling excess fuel consumption occurs.
This necessitates the need for reducing air velocity striking the engine surface to reduce the fuel
consumption. It can be done placing a diffuser in front of the
Mr. N. Phani Raja Rao, Mr. T. Vishnu Vardhan. Thermal Analysis of Engine Cylinder Fins By
Varying Its Geometry And Material.[12] The principle implemented in the project to increase the
heat dissipation rate by using the invisible working fluid nothing but air. The main aim of the project
is to varying geometry, material. In present study, Aluminium alloy 6061 and magnesium alloy are
used and compared with Aluminium Alloy A204. - The various parameters (i.e., shape and geometry
of the fin) are considered in the study, shape (Rectangular and Circular), thickness (3 mm and 2.5
mm). By reducing the thickness and also by changing the shape of the fin to circular shaped, the
weight of the fin body reduces thereby increasing the heat transfer rate and efficiency of the fin. The
weight of the fin body is also reduced when Magnesium alloy is used. The results shows, by using
circular fin with material Aluminium Alloy 6061 is better since heat transfer rate, Efficiency and
Effectiveness of the fin is more. By using circular fins the weight of the fin body reduces compare to
existing engine cylinder fins.
S.M. Wange and R.M. Metkar [13] have done experimental and computational analysis of fin array
and shown that the heat transfer coefficient is more in notch fin array than without notch fin array.
Geometric parameters of fin affects on the performance of fins, so proper selection of geometric
parameter such as length of fin, height of fin, spacing between fins, depth of notch is needed.
Heat Transfer Augmentation of Air Cooled 4 stroke SI Engine through Fins- A Review Paper. [14]
The author had study number of research paper and concludes that the phenomenon by which heat
transfer takes place through engine fins must frequently be improved for these reasons. Fins are
extended surface which are used to cool various structures via the process of convection. Generally
heat transfer by fins is basically limited by the design of the system. But still it may be enhanced by
modifying certain design parameters of the fins. Hence the aim of this paper is to study from different
literature surveys that how heat transfer through extended surfaces (fins) and the heat transfer co-
efficient affected by changing cross-section, climatic conditions, materials etc. It is to be noted that
heat transfer of the fin can be augmented by modifying fin pitches, geometry, shape, and material and
wind velocity. As per available literature surveyed there is a little work available on the wavy fins
geometry pertaining to current research area to till date. So there is a scope of research in the field of
heat transfer study on wavy fins on cylinder head block assembly of 4 stroke SI engine.
1001 Vol. 7, Issue 3, pp. 998-1002

N.Nagarani and K. Mayilsamy, Experimental heat transfer analysis on annular circular and elliptical
fins.[15]This other had analyzed the heat transfer rate and efficiency for circular and elliptical annular
fins for different environmental conditions. Elliptical fin efficiency is more than circular fin. If space
restriction is there along one particular direction while the perpendicular direction is relatively
unrestricted elliptical fins could be a good choice. Normally heat transfer coefficient depends upon
the space, time, flow conditions and fluid properties. If there are changes in environmental conditions,
there is a change in heat transfer co-efficient and efficiency also.[15]
2.1 Examples of Commercially Available Engines

Figure 1. Examples of Engines
Courtesy by KGN AUTOMOBILES, GACHIBOWLI, HYERABAD, AP, INDIA
III. CONCLUSION
The summary of the present literature review is as follows:
1. Design of fin plays an important role in heat transfer. The fin geometry and cross sectional area
affects the heat transfer co efficient there is a scope of improvement in heat transfer of air-cooled
engine cylinder fin if mounted fins shape varied from conventional one.
2. From the all the research and experiment that covered in this paper it can be conclude that Contact
time for the air flows over the fin is also important factor in heat transfer rate. If we can increase the
turbulence of air by changing the design and geometry of the fins it will increase the rate of heat
transfer and it is found that Curve and Zig-zag fin shaped cylinder block can be used for increasing
the heat transfer from the fins by creating turbulence of upcoming air. Improvements in heat transfer
with new curve and Zig-Zag design fin can be compare with conventional one by CFD Analysis
(Ansys Fluent).
REFERENCES
[1]. Shri. N.V. Hargude and Dr. S.M. Sawant, Experimental Investigation of Four Stroke S.I. Engine using
Fuel Energizer for Improved Performance and Reduced Emissions, International Journal of Mechanical
Engineering & Technology (IJMET), Volume 3, Issue 1, 2012, pp. 244 - 257, ISSN Print: 0976 6340, ISSN
Online: 0976 6359
[2].J.Ajay Paul, SagarChavan Vijay, Magarajan&R.ThundilKaruppaRaj, Experimental and Parametric Study of
Extended Fins In The Optimization of Internal Combustion Engine Cooling Using CFD, International Journal
of Applied Research in Mechanical Engineering
[3]. J.C.Sanders, et al. Cooling test of an air-cooled engine cylinder with copper fins on the barrel, NACA
Report E-103
[4]. D.G.Kumbhar, et al. (2009). Finite Element Analysis and Experimental Study of Convective Heat Transfer
Augmentation from Horizontal Rectangular Fin by Triangular Perforations. Proc. Of the International
Conference on Advances in Mechanical Engineering.
[5].N.Nagarani and K. Mayilsamy (2010). "EXPERIMENTAL HEAT TRANSFER ANALYSIS ON
ANNULAR CIRCULAR AND ELLIPTICAL FINS." International Journal of Engineering Science and
Technology 2(7): 2839-2845.
1002 Vol. 7, Issue 3, pp. 998-1002

[6]. S.S.Chandrakant, et al. (2013). "Numerical and Experimental Analysis of Heat Transfer through Various
Types of Fin Profiles by Forced Convection." International Journal of Engineering Research & Technology
(IJERT)
[7]. A. T. Pise and U. V. Awasarmol (2010). "Investigation of Enhancement of Natural Convection Heat
Transfer from Engine Cylinder with Permeable Fins." International Journal of Mechanical Engineering &
Technology (IJMET) 1(1): 238-247.
[8]. G.Raju, Dr. BhramaraPanitapu, S. C. V. RamanaMurty Naidu. Optimal Design of an I C engine cylinder
fin array using a binary coded genetic algorithm. International journal of Modern Engineering Research. ISSN
2249-6645,Vol. 2 Issue.6, Nov-Dec.(2012),pp.4516-4520
[9]. R.P. Patil and P. H. M. Dange (2013). "Experimental and Computational Fluid Dynamics
Heat Transfer Analysis on Elliptical Fin by Forced Convection." International Journal of Engineering
Research & Technology (IJERT) 2(8).
[10]. Magarajan U., Thundilkaruppa Raj R. and Elango T., Numerical Study on Heat Transfer of
Internal Combustion Engine Cooling by Extended Fins Using CFD, Research Journal of Recent Sciences ISSN
2277-2502 Vol. 1(6), pp.32-37, June (2012).
[11]. P. Agarwal, et al. (2011). Heat Transfer Simulation by CFD from Fins of an Air Cooled Motorcycle
Engine under Varying ClimaticConditions. Proceedings of the World Congress on Engineering.
[12]. Mr. N. Phani Raja Rao, Mr. T. Vishnu Vardhan. Thermal Analysis Of Engine Cylinder Fins By Varying
Its Geometry And Material. International journal of Engineering Research & Technology.ISSN:2278-
0181,Vol. 2 Issue 8, August(2013)
[13]. S. Wange and R. Metkar (2013). "Computational Analysis of Inverted Notched Fin Arrays Dissipating
Heat by Natural Convection." International Journal of Engineering and Innovative Technology (IJEIT) 2(11)
[14]. Heat Transfer Augmentation of Air Cooled 4 stroke SI Engine through Fins- A Review Pape. International
Journal of Recent Development in Engineering and Technology. ISSN 2347 6435, Volume 2, Issue 1, January
(2014)
[15]. N.Nagarani and K. Mayilsamy, Experimental heat transfer analysis on annular circular and elliptical fins.
International Journal of Engineering Science and Technology 2(7): 2839-2845
[16]. Islam Md. Didarul, Oyakawa Kenyu, Yaga Minoru and Senaha Izuru, Study on heat transfer and fluid
flow characteristics with short rectangular plate fin of different pattern Experimental Thermal and Fluid
Science, Volume 31, Issue 4, February 2007
[17]. P. R. Kulkarni, Natural Convection heat transfer from horizontal rectangular fin arrays with triangular
notch at the center. Paper presented at NCSRT-2005. (Nov 18-19, 2005), Pg. No.: 241-244.
ABOUT AUTHOR
Mohsin A. Ali BTech(MECH) From GCOE Amravati. Pursuing ME(CAD/CAM) from
SGBA University, Amravati
1003 Vol. 7, Issue 3, pp. 1003-1008

VIRTUAL WIRELESS KEYBOARD SYSTEM WITH CO-
ORDINATE MAPPING
Souvik Roy, Ajay Kumar Singh, Aman Mittal, Kunal Thakral
Department of Electronics and Communication Engineering,
Guru Gobind Singh Indraprastha University, New Delhi, India

ABSTRACT
This paper presents more efficient algorithm used to detect the finger stroke of a projected keyboard layout on
any flat non-reflecting surface. The virtual keyboard consist of variable intensity projector for projecting
keyboard layout, a Camera with infra-red filter for capturing only infra-red wavelength light reflected objects,
an infra-red Diode for object detection, photo-diode with simplified circuitry to on-off the key stroke detection
and key board layout projection. The camera used is connected to PC or laptop with a wireless connection
operating at 2.4GHz frequency of IEEE standard. An image processing algorithm is designed in open source to
extract the keystroke on the surface and display the exact key on the screen. The integration of components with
the software that is designed to run on any Operating system with even lower level of processor gives expected
keystroke. Comparison algorithm calculates and checks the nearest value suitable for key injection to the system
API. Height of device made has been kept as low as possible so that it does not interrupt the view of display
screen of laptop or PC. Present work can be more upgraded with more gesture based feature that controls the
virtual keyboard device and making the extra surface for virtual mouse option.
KEYWORDS: Keyboard, Object Detection, Finger Detection, Camera, Infra-Red Filter, Frame Capturing
I. INTRODUCTION
With an increasing demand for switching into new and virtual environment from the old physical
environment where hardware ware out has been a problem due to continuous use of it. Virtualization
[1] of keyboard to display has been carried out since decades but the precision of functionality has
become a problem. Virtual keyboard is the most important part of virtualization as it cannot be
obsoleted from the computer part. Hence there is a need for upgrading the method of keystroke
detection that will make it more precise and adapt to human tendency of pressing.
II. DESIGN
The design of the Virtual keyboard system consists of minimum use of hardware devices and
maximum dependence on software algorithm.
2.1. System design
Virtual keyboard system is designed keeping in mind the maximum efficiency of algorithm [2] and
lower error rate that will not make user feeling different response rate then physical keyboard. The
design consists of following modules:
CMOS wireless camera
Infra-red band pass filter
Keyboard Projector with intensity control
Infra-red laser diode
Double Concave lens
Diffraction grating
Holographic film
1004 Vol. 7, Issue 3, pp. 1003-1008

Current controller circuit
Software for Key-stroke detection
LM-1117 voltage convertor
All these modules are used to design a proper and highly efficient system.
2.2. Hardware design
The Hardware of the system is most important part of the system as exact configuration of
components is required to simplify the algorithm and error rate. The first sensor part of the system
CMOS [3] wireless camera is integrated with Infra-red band pass filter which limit the camera sensing
capacity to only Infra-red light. Limiting the sensing capacity of camera is useful in detection of
finger touching the virtual projected keyboard layout as it is the least interfering light in any
environment that could generate error and algorithm more complex. Second part of the design is the
Keyboard projector [4] that is used to project a fixed designed keyboard with two different language
characters printed on them. This projector projects a visible laser beam passing through biconcave
lens [5] and a holographic [6] keyboard layout plate thus projecting the shadow of keyboard. The
intensity of the laser diode is controlled by limiting the current flowing to the laser diode. Current
limiting is controlled through current controlling circuit which consist of variable resistor to change
the intensity of laser diode. The third part of the design is Infra-red laser diode that is used to generate
a layer of linear Infra-red light parallel to the surface on which keyboard layout is projected. Since the
infra-red laser cannot emit a linear diffracted [7] beam to cover the whole keyboard layout without
diverging. Therefore we need to pass the laser light through Biconcave lens adjusted in such a
distance that emerging of parallel laser beam to pass through Diffraction plate which diverge the beam
linearly hovering over the whole surface of keyboard layout.
2.3. Software/Algorithm Design
The design of software starts from the image processing algorithm which is the most important part of
key-stroke detection. Now the camera is only able to detect objects blocking infra-red light from the
Infra-red laser [16] beam plane or objects emitting infra-red light in the camera field of view. Starting
with capturing image frames from a CMOS camera and transmitting frames through Wi-Fi [8]
network to the laptop or PC using frame capturing code written. This capturing of image should be
such that the generated image is mirror image of the original image. Image frames generating from
CMOS camera consist of infra-red images as the camera lens is integrated with infra-red band pass
filter. The image contains lot of noise and high frequency regions where object edges to be extracted,
therefore a two steps are performed before moving into key-stroke detection. Firstly Gaussian [9]
filter is used to remove the white noise that is present everywhere in the whole frequency range of
image spectrum [10]. It also has linear graph according to which there is no sharp/abrupt change in
frequency band, thereby increasing the accuracy of detecting sharp edges of object which are
generally in high frequency region of image histogram. Secondly threshold function is used to omit
the region of non-object and showing the region of objects touching the surface. Key-stroke detection
is started with point of interests extraction where centroid of all the touching points is calculated using
cvFindContour [11] function. This function provides a co-ordinate that is average area of the finger
tip. The co-ordinate value extracted is need to find out for actual key pressed at that location, which is
done through a method of predefined [12] keyboard with fixed key co-ordinate location are stored in a
one-dimensional array. The mapping of generated co-ordinate with beforehand preserved co-ordinate
of keyboard pattern is done through comparison of co-ordinated to find out minimum difference of
displacement between co-ordinates. This comparison does a very efficient computation as uses tree
[13] structure for comparison. After the comparison of co-ordinates the nearest key value is found out
which is injected into Operating system through available API for windows system allowing to inject
the nearest key found out. Finally the key pressed is found out.
III. IMPLEMENTATION
Virtual keyboard use image processing algorithm in OpenCV to decease the processing time. It gives
user a idea to connect with their laptop or PC using a virtual environment.
1005 Vol. 7, Issue 3, pp. 1003-1008

3.1. Hardware
The described components for hardware are used to make a proper prototype that could match the
requirement of software design. Positions of the three main components are kept at a fixed height and
angle such that the camera can capture the whole keyboard image. Height between projector and
camera is kept to be 3cm which makes linear relationship between camera view and projector
constant. Camera used wide angle lens for complete view of the keyboard layout at minimum height.
As per the pre-defined keyboard key location stored in one dimensional array the height of both
projector and camera is fixed. The only one laser plane generating component is kept at the lowest
section touching the surface so to minimum distance between surface and laser plane. Laser keyboard
generating diode is connected with current controlling circuit using potentiometer used to control the
intensity of keyboard layout. Intensity of keyboard layout is independent of image processing
algorithm. The whole hardware is powered with rechargeable 9v battery.

Figure 1. Working prototype
3.2. Algorithm
Design of algorithm is needed to be integrated with the hardware for proper working of the hardware.
The Wi-Fi enabled camera is connected with laptop or PC for transmission of video frames. Rate of
sending video frames is able to match with the transfer speed bandwidth allowing 30fps to be
transferred easily. Software designed is completed after writing the code in openCV and compiling
using g++ compiler generating object file, configuration file and application file. Application file
generated is used to do the real time image processing using data send from Wi-Fi camera. With the
start of application file an option pops up for selection of camera and then checking establishment of
proper connection with camera. The processing of real time video frames [14] coming from wireless
camera is continuously going through. If any key event is found the API informs the OS to print that
corresponding key. The applications runs in the background to make user feel like actual keyboard.
1006 Vol. 7, Issue 3, pp. 1003-1008

Figure 2. Key Stroke Identification process
3.3. Wireless Communication
Connection of Wi-Fi CMOS camera to the laptop or PC is done with simply using the Wi-Fi
transmitter at 2.4GHz to establish a connection with the laptop or PC. The CMOS camera is used as
transmitter and laptop or PC Wi-Fi connects to that network
3.4. Efficiency
The efficiency of algorithm is found out to be 90%. It is calculated using the test cases where 3
different paragraphs are written with their accuracy being calculated. At the same time key-stroke
location mapping efficiency found out to be 80% giving correct result.

Figure 3. Probability of correct key stroke event v/s Area of Key
The probability of correct key-stroke event within a key area of projected keyboard layout is a slope
down from value 1 to 0 as the finger stroke value shifts from centre to the boundary of key.
3.5. New Feature

1007 Vol. 7, Issue 3, pp. 1003-1008

The virtual keyboard has a feature of auto switch off mode. This feature allows the projected key-
board layout to switch off the keyboard projector laser diode until the user finger comes near the
keyboard region. An photodiode [15] is put just above the plane laser which continuously checks if
the no signal is received for 300seconds then it switch off of the key-board layout projector to save
power It switch on automatically by just hovering hand over the key-board layout.
3.6. ADVANTAGE
The most important advantage of this system is to make the system independent of on-board
processing which uses most of the battery power as well as the independency of each module makes it
easier for upgrade to new feature. The wireless feature makes it useful to work from a distant location
with virtually controlling the laptop or PC.
IV. CONCLUSIONS
The virtual keyboard system with keyboard co-ordinate mapping has been implemented with
minimum use of complex hardware structure and algorithm that provides a better result without much
complexity. More gesture based feature can be added that will make it more close to the virtual
device.
REFERENCES
[1]. Celluon keyboard (http://www.gizmag.com/celluon-epic-laser-keyboard-review/28342/) An
Introduction to Virtualization by Amit http://www.kernelthread.com/publications/virtualization/
[2]. www.wikipedia.org (http://en.wikipedia.org/wiki/Laser_projection_keyboard) Yael Ben-Haim & Elad
Tom-Tov A Streaming Parallel Decision Tree AlgorithmIBM Haifa Research Lab, Journal of
Machine Learning Research 11 (2010) 849-872
[3]. www.robopeak.com/blog/ Igor Brouk, Kamal Alameh Design and Characterization of CMOS/SOI
Image Sensors, IEEE TRANSACTIONS ON ELECTRON DEVICES, VOL. 54, NO. 3, MARCH
2007
[4]. www.dfrobot.com
http://www.dfrobot.com/index.php?route=product/product&product_id=931#.U0rB4vTIuuI)
[5]. BI-CONCAVE LENS (BCC), http://www.lambda.cc/bi-concave-lens-bcc/
[6]. http://www.slideshare.net/priyalbhagat/laser-protection-virtual-keyboard
[7]. Kolsch, M. and Turk, M. Keyboards without Keyboards: A survey of Virtual Keyboards, n. pag.
University of California, Nov. 05, 2003 (http://www.create.ucsb.edu/sims/PDFs/Koelsch_an
d_Turk_SIMS.pdf).
[8]. Lawrence Erlbaum Associates Inc, Human Computer Interaction, pp 89-
129,2002(http://www.almaden.ibm.com/u/zhai/papers/Zhai HunterSmithHCIGalley.pdf).
[9]. Alpern, M., Projection Keyboards, Oct. 28, 2003 (http://www.alpern.org/weblog/stories/2003/01/09/
projectionKeyboards.html).
[10]. Baxes, G. A. Digital Image Processing Principles and Applications, America, 1994.
[11]. Gonzalez, R. and Woods, R.,Digital Image Processing,, India, 2002.
[12]. Wijewantha N. S., VistaKey: A Keyboard Without A Keyboard A Type Of Virtual Keyboard,
Final year project thesis, Informatics Institute of Technology, Wellawatta, Sri Lanka, April 2004.
[13]. Hirsh, L., Virtual Keyboard Replicates Real Typing, Oct. 23, 2003
(http://www.wirelessnewsfactor.com/perl/story/147 62.html).
[14]. Wrolstad, J., Laser Device Projects Full-Size Keyboard for Handhelds, Oct. 28, 2003
(http://www.wirelessnewsfactor.com/perl/story/177 56.html).
[15]. Khar, A., Virtual Typing, Oct. 28, 2003 (http://www.pcquest.com/content/technology/10303
0402.asp).
[16]. Aid Inc, LightKey Visualize A Virtual Keyboard. One With No Moving Parts, Nov. 12,
2003(http://www.advancedinput.com/AIDpdfDownloads/AIDLightKey.pdf)

1008 Vol. 7, Issue 3, pp. 1003-1008

AUTHORS
Aman Mittal was born in 1992.He is pursuing bachelors in Electronics and Communication
Engineering from Guru Gobind Singh Indraprastha University, New Delhi. His main
research interests include Robotic systems, computer vision and System Integration. He has
participated in several projects and currently working on a generic device.

Kunal Thakral was born in 1992. He is pursuing bachelors in Electronics and
Communication Engineering from Guru Gobind Singh Indraprastha University,New Delhi.
His main research interests include Robotic systems, micro-controller and Gesture control.
He has participated in several projects and currently working on a generic device and
UART.

Souvik Roy was born in 1992.He is pursuing bachelors in Electronics and Communication
Engineering from Guru Gobind Singh Indraprastha University, New Delhi. His main
research interests include Robotic systems, computer vision and machine learning. He has
filled patent on new technique to save water during irrigation. He has implemented in
several ideas into working prototype and currently working on freescale semiconductors
smart car coding competition using ARM microcontroller.

Ajay Kumar Singh was born in 1992.He is pursuing bachelors in Electronics and
Communication Engineering from Guru Gobind Singh Indraprastha University, New Delhi.
His main research interests include Robotic systems, computer vision and System
Integration. His main research interests include Robotic systems, UAVs, computer vision,
Embedded and System Integration. He has participated in several projects and currently
working on a generic device.

1009 Vol. 7, Issue 3, pp. 1009-1017

SECURE KEY MANAGEMENT IN AD-HOC NETWORK:
A REVIEW
Anju Chahal
1
and Anuj Kumar
2
, Auradha
2

1
Department of Computer Science Engineering, AMITY University, Haryana, India
2
Assistant Professor, AMITY University, Haryana, India

ABSTRACT
An ad hoc network is a decentralized type of network. The network is ad hoc because it does not have a pre-
existing infrastructure. An ad hoc network is a common wireless network that can communicate with each other
without any centralized administration or pre-existing infrastructure. Due to nature of Inconstant Wireless
medium Data Transfer is a major problem in ad hoc it lacks Security and Reliability of Data. Cryptographic
techniques are often used for secure Data transmission wireless networks. Most cryptographic technique can be
symmetric and asymmetric, depending on the way they use keys. However, all cryptographic techniques is good
for nothing if key management is weak. There are various type of key management schemes that have been
proposed for ad hoc. In this survey, we present a complete study of various key management techniques to find
an efficient key management for Secure and Reliable Data transmission in ad hoc.
KEYWORDS: Ad-Hoc network, Security issues, Key Management.
I. INTRODUCTION
An Ad Hoc network is a collection of wireless nodes that are communicated with each other without
any centralized administration (node). Ad Hoc network is kind to peer-to-peer networks, where there
is no fixed infrastructure (i.e. network is formed on demand, and have a fully dynamic network
topology. There is no central authority and ad hoc network is self organizing and adaptive .Node
forming in ad hoc network is often low energy, portable small devices. Ad Hoc network may be ideal
different from the other network in computer science classrooms an Ad Hoc network could form
between students PDA and the workstation of the teacher
[1]

Ad hoc network is dynamic in nature e.g. Consider in its the 8 nodes show in fig 1.They are
connected to other nodes within their individual range. Now consider a node 3 move from its present
position and comes near to the node 7.Thens in previous link of the node s 4 is broken and its forms
a new link through its new neighbor node 7.This scenario shows an example that an ad hoc network in
dynamic in nature.
II. SECURITY ISSUES
2.1 Security Goal for Ad-Hoc Network
[2]

2.1.1 Confidentiality: Confidentiality ensures that only authorized person can have access
to certain information. Classified information application that uses ad hoc network
like in military operations, certain information can be appropriate. So disclosure of
such information can be high in price.
1010 Vol. 7, Issue 3, pp. 1009-1017

Figure1. Ad-Hoc Network
2.1.2 Availability: Availability ensures that the requested service should be available as when
requested. So availability opposes Denial of Service (DOS). With denial of service attack
competitor can also break down important services like key management. So, availability is
an important security goal that should be achieved with any kind of ad hoc network
application.
2.1.3 Integrity: Integrity implies that message should be unaltered during its transmission from
source to destination. Message can be modified Un-intentionally during transmission
considering of radio propagation. A malicious attacker can also modify a message
intentionally during its transmission.
2.1.4 Authentication: Authentication is the process of identification, that a receiving entity is
assured that message he receives come from an authorized source. In an ad hoc network,
mobile node is vulnerable to compromise without proper authentication an attacker can
authenticate user and thus can have the full control of the entire network.
2.1.5 Non Repudiation: Non Repudiation implies that once a message has been sent, the sender
cannot deny that they ever sent or received such a message. It is important security services
by which compromised node can detect and isolate.
2.2 security attack
[3]

2.2.1 Passive Attack: In passive attacks an originator captures the data without altering it. The
attacker does not modify the data and does not edit any additional data. The main goal of
attacker is to obtain information that is being transmitted.
2.2.2 Active Attack: In Active attacks an attacker actively participates in distort the normal
operation of the network services. An attacker can create an active attack by modifying
packets or by modifying the information or giving the false information.
Active attack can be divided into two major groups:
a) Internal attack: are forms compromised nodes that were once an authorized part of the
network. Since the already part of the network as authorized nodes, they are much more
secure and difficult to detect as compared to external attack.
b) External attack: are carried by node that is not an authorized part of the network.

2.3 Key Issues and Challenges [4]
1011 Vol. 7, Issue 3, pp. 1009-1017

2.3.1 Link Level Security: Nodes in ad hoc network communicate by wireless link, which is much
vulnerable to various active and passive attacks. Absence of security mechanism like firewall,
access control leads a node in an ad hoc network attack to be attacked from any direction.
Attacks like impersonating a node, traffic re-direction, denial of service etc. Make it hard to
achieve the prime security goals.
2.3.2 Secure Routing: Most of the researches in the area an ad hoc networking lead to obtain a
secure routing protocol for mobile ad hoc network, which is hard to achieve. In most of the
routing protocol in mobile ad hoc networks, intermediate nodes are acting as disseminate. So
if a node is compromised, then it can generate false routing protocol information, spread
previous routing information, insert new information with existing information, which will
finally break down the whole network. Sometime node can act meanly to save own battery
power. A compromised node can also send malicious information to other nodes, which in
turn attacks other nodes in the network.
2.3.3 Key Management: Key management is one of the prime requirements in any secure network.
In ad hoc networks have no fixed infrastructure, no central authority and connectivity is not
always guaranteed, Key management becomes a key issue for securing ad hoc networks. Key
management issues are discussed in detail.
2.3.4 Dynamic Mobility: One of the major characteristic of an ad hoc network is the node is
dynamic in nature. Nodes can join or leave in ad hoc network at any time and thus there is no
guaranteed connectivity between nodes. Static security mechanisms are not always suitable
for ad hoc network. So this property of ad hoc network makes it difficult for the researchers
come up with secure key management and routing protocol.
III. KEY MANAGEMENT IN AD HOC NETWORK
Cryptography is a powerful tool in achieving security. Mostly cryptography is used for secure, Robust
and efficient key management subsystem. Key management is a basic main part of security of ad hoc
network. Some of symmetric and asymmetric key management scheme has been purposed in the ad
hoc network. Key management used with Key Generation Key distribution, Key storage, updating
keys, Revocation, deleting, arching and using the key acing to secure.
Key management schemes in Ad-Hoc Network
Symmetric Key
management
scheme
Asymmetric key
management
scheme
Group key
management
schemes
Hybrid composite
key management
service
1. DKPS
2. PKIE
3. INF
1. SRP
2. URSA
3. INF
4. SOKM
5. SEKM
6. Z&H
7. SOKS
8. ID-C
9. Identity based
10. Three level key
1. SGEK
2. PGSK
1. Cluster based
composite key.
2. Hybrid schema
zone based

Figure2: Key Management Scheme in Ad hoc Network
1012 Vol. 7, Issue 3, pp. 1009-1017

3.1 Symmetric Key Management in Ad hoc Network
In symmetric key management same keys are used by sender and receiver. This key is used for
encrypting the data as well as for decrypting the data. If n nodes want to communicate in Ad -hoc
network number of keys are required, where k = n (n-1)/2. In public key cryptography, two keys are
used, one private key and other public key. Different keys are used for encryption and decryption. The
private key is used for decryption. The public key is used for encryption and it available to the public.
In each communication new pair of public and private key is created. It requires less no of keys as
compared to symmetric key cryptography. Symmetric key is used for long massage. Now discuss
about some of the symmetric key management schemes in Ad-Hoc
3.1.1. Distribute key Pre-Distribution Scheme (DKPS): In its consist of three important
phase
a) Distributed Key Selection (DKS) In these phase every node takes the
Random key from the universal set by using exclusive property.
b) Secure Shared-key Discovery (SSD) In the second phase of DKPS in
which every node having a shared key with another node. Node cant find
that which key on the ring are in common with which node. This method
is not providing security, but easy to evaluate due to eavesdropping can
occur in DKS phase.
c) Key Exclusion Property Testing (KEPT) - Last phase of DKPS
symmetric key management scheme is KEPT. In its matrix is used for
present the relationship between mobile nods key and shared keys it uses
binary values for constructing the matrix. A KEPT phase test that is all
keys of mobile nodes fulfilling the exclusive property of CFF. Features of
DKPS are no need of TTP. DKPS needs less storage as compared to pair-
wise key agreement approach. This scheme is more efficient as compared
to group key agreement
[5]

3.1.2. Peer Intermediaries for Key Establishment (PIKE): In this uses the sensor nodes
to establish the shared key. PKIE is a symmetric key agreement scheme, it uses
unique secret key in a set of nodes .This model uses the concept of random key pre-
distribution, and in 2-D case with each of the O (n) nodes every mobile node shares a
unique secret key in horizontal and vertical dimension. This scheme can be extended
to 3D or any other dimension. Features of this model are good security services, and
fair scalability
[6]

3.1.3. Key Infection (INF): This model is simple and every mobile node participates
equally to making the key establishment process. INF model has no need of
collaborative effort due to node acts as a trust component; this component broadcasts
their symmetric key. This model having weak security services, but INF having low
storage cost, low encryption, and low operation. It has fair scalability with the
problem of late entry of mobile node
[7].

3.2 Asymmetric key management in Ad hoc Network
Asymmetric keys, use two-part public and private key. Each recipient has a private key that is kept
secret and a public key that is used for everyone. The sender sent the recipients public key and uses it
to encrypt the message. The recipient uses the private key to decrypt the message and never publishes
or transmits the private key to anyone. Thus, the private key is never passing over and remains
invulnerable. This system is sometimes using public keys. This reduces the risk of data loss and
increases compliance management when the private keys are properly managed.
3.2.1 Secure Routing Protocol (SRP): This scheme is composed with three nodes and an
administrative authority which work as dealer in this model. Dealer is the entity
which provides the initial certificate to the mobile nodes. Three nodes are defined as:
1. Client Node. 2. Server Node. 3. Combiner Node. SRP node plays the important
task in SRP model.
1013 Vol. 7, Issue 3, pp. 1009-1017

3.2.2 Ubiquitous and Robust Access Control (URSA): URSA is efficient and provides
reliable Availability with having the feature of encrypted local communication. This
model uses efficient threshold scheme to broadcast the certificate (RSA Certificate)
signing keys to all nodes. Each mobile node of Ad-Hoc updates their certificates
periodically. This scheme provides communication delay, search failure, and degrades
the system security. To protect the network from DOS attack and the compromise the
signing key URSA using verifiable and proactive secret sharing mechanisms
[10]
.
3.2.3 Mobile Certificate Authority (MOCA): The mobile nodes having great
computational power, physically more secure. When the nodes are equally equipped
than, MOCA nodes are selected randomly. This scheme is decentralized and the
services of a CA are distributed to MOCA nodes [11]. In their scheme , a node could
locate k+ MOCA node either randomly through the shortest path in its route cache
.But the critical question is how nodes can discover those paths securely since most
routing protocols are based on the establishment of a key services
[11]
.
3.2.4 Partially Distributed Threshold CA Scheme: Partially Distributed Threshold CA
Scheme was discovered by Zhou, L. and Hass, Z. in 1999. When the mobile ad-hoc
network is constructed, this scheme is using the concept of CA distribution in
threshold fashion. Security services like off line authentication, great intrusion
tolerance, and trust management by CA (certification authority) are provided by
asymmetric key management scheme. The key is generated by this model are
accepted by self-organized network and partial distributed threshold CA. This scheme
having the scalability of CRL (certificate revocation list), and certification
[12]
.
3.2.5 Self-Organized Key Scheme (SOKS): In the self-organized network each mobile
node acts as a distinct CA.SOKS was disclosed by Capkun, S., Buttya, L., and
Hubaux, P. in 2003. It has poor scalability and poor resource efficiency, but having
the off line authentication and limited intrusion detection security services. SOKS
having high intermediate encryption operations and high storage cost
[13]
.

3.2.6 Key Distribution Technique (ID-C): In this scheme node create or initialize the Ad-
Hoc network with using the threshold private key generator identity based scheme.
The generated key is accepted by self-organized network. Off net authentication, trust
management and intrusion tolerances type security services are provided by ID-C
asymmetric key management scheme. Scalability is provided through an Id
Revocation list with greater resources efficiency. This scheme has medium storage
coast, operation and encryption
[14]
.
3.2.7 Identity-Based Key Asymmetric Management Scheme: In Secured ID-based key
management scheme for Ad-Hoc network which allows nodes to use their public keys
directly from their known network identities and with some other common
information. This scheme provides inline Certification Authority (PKI) to share a
secret key. It also provides end-to-end authentication and enables mobile user to
ensure the authenticity of user of the peer node. The significant advantage of solution
is to avoid users to generate their own public keys and to then distribute these keys
throughout the network. This scheme solved the security problem in the ad hoc
network and is also suitable for application to other wired and wireless network. In
this major problem of security
[15].

3.2.8 Three Level Key Management Scheme: Secure and Highly Efficient Three Level
Key Management scheme for Ad-Hoc network is proposed by Wan AnXiong, Yao
Huan Gong in 2011. To achieve three level security in ad hoc this model uses ID-
Based Cryptography with threshold secret sharing, Elliptic Curve Cryptography
(ECC) and Bilinear Pairing Computation. ECC provides short keys to mobile nodes
and high security level. Key generation and key distribution security services in the
prevention from adversaries attack are done by (t, n) threshold secret sharing
algorithm. ECC provides an enhanced security level with using 160 bits key and 1024
bits equivalent strength of RSA. Pairing technology provides confidentiality and
authentication with less computational cost and reduced communication overhead
[16].

1014 Vol. 7, Issue 3, pp. 1009-1017

3.3. Group Key Management Scheme in Ad hoc Network
Group key in cryptography is a single key which is assigned only for one group of nodes in Ad-Hoc
network. For create a group key, group key is creating and distributing a secret for group members.
There are specifically three categories of group key protocol

a) Centralized, in which the controlling of group is being done by one entity.
b) Distributed, group members or a mobile node which comes in the group are equally responsible
for making the group key, distribute the group key.
c) Decentralized, more than one entity is responsible for making, distributing group key. Let us
discuss about some important Group key Management schemes in Ad-Hoc network.

3.3.1 Simple and Efficient group key Management (SEGK): This scheme presents the reliable
double multicast tree formation and maintenance protocol, which ensures that it
covers all group members. The initialization process is starting by the group
coordinator with sending the join message into the ad-hoc network. No of nodes are
directly propositional to compute cost. In SEGK model, any mobile node or group
member can join and leave the network. To ensure the backward and forward security
updating of group key is done very frequently. Two detection methods are described
in SEGK model
[17]
.
a) Tree Links, when the node mobility is not a significant detection is done through tree
links.
b) Periodic Flooding of Control Messages, for the high mobility environment this method is
used.
3.3.2 Private Group Signature Key (PGSK): Group signatures are proposed in
[18]
,
provide anonymity for signers. Any member of the group can sign messages, but the
resulting signature keeps the identity of the signers secret. In some systems there is a
third party that can trace the signature, or undo its anonymity, using a special
trapdoor. Some systems support revocation where group membership can be
selectively disabled without affecting the signing ability of unprovoked members.
Currently, the most efficient constructions are based on the Strong-RSA assumption.
A Private Group Signature key is generated by a Key Server for each node in the
Network, which ensures full anonymity which means a signature does not reveal the
signers identity but everyone can verify its validity.

3.4 Hybrid key management scheme in Ad hoc network

Hybrid or composite keys are those keys which are made from the combination of two or more
keys and it may be combination of symmetric & asymmetric key. Let us discuss about some of
the important Hybrid key management schemes in Ad-Hoc network.

3.4.1 Cluster Based Composite Key Management: This scheme takes the concept of off-line CA,
mobile agent, hierarchical clustering and partial distributes key management. Public
key of the members are maintained by cluster head that reduces the problem of
storage in PKI. On the basis of current, trust value and the old public key, cluster
heads public key is computed. Using the timestamp in key number key renewal
process can be done easily. It supports network extendibility through hierarchical
clustering. This model saves network bandwidth and storage space
[19]
.

3.4.2 Zone-Based Key Management Scheme: This scheme uses ZRP (Zone Routing Protocol)
proposed in [20], in this model for each mobile node zone is defined. Some pre-
defined number is allocated to each node which depends on the distance in hops.
Symmetric key management is used by node only for intra or inside zone (zone
radius). Without depends on clustering node uses asymmetric key management for
inter-zone security. It provides an efficient way to making the public key without
losing the capability of making the certificates.
1015 Vol. 7, Issue 3, pp. 1009-1017

IV. COMPARATIVE SURVEY
In the Previous Section we have discussed about some of the most important Key Management
Techniques in Mobile ad hoc networks. In Comparative survey, we are going to compare these Key
Management techniques based upon some of the Features like Reliability, Security, Scalability and
Robustness. The Comparative Survey is made depending upon the results that are analyzed from
various research works and journals. Table I shows the Comparative Survey of Key Management
schemes in ad hoc Networks. Let us discuss about the features of Key Management schemes that we
are going to compare.
4.1 Security: Central security issues are trust management and vulnerability. Trust relations may
change during the network lifetime. The system should enable the exclusion of compromised
nodes. In order to judge the security of a key-management scheme, possible vulnerabilities
should be important. Security services enabled one or a combination of confidentiality, integrity,
authentication and non-repudiation.
4.2 Scalability: Key management operations should finish in a timely. The fraction of the available
bandwidth occupied by network management traffic should be kept as low as possible. Any
increase in management traffic reduces the available bandwidth for payload data accordingly.
Hence, scalability of key-management protocols is essential.
4.3 Reliability: The Reliability of a Key Management scheme depends upon the Key Distribution,
Storage and Maintenance. It is necessary to make sure that the Keys are Properly Distributed
among the nodes, safely stored where hacker arent able to hack the keys and should be Properly
Maintained.
4.4 Robustness: The key-management system should survive despite denial-off service attacks and
unavailable nodes. The key-management operations should be able to be completed despite
faulty nodes and nodes exhibiting behavior, that is, nodes that deliberately deviate from the
protocol. Necessary key management operations caused by dynamic group changes should
execute in a timely manner. Key management operations should not require network wide and
strict synchronization. It is resistance to security attacks (e.g. man-in-the-middle).

TABLE1. Comparative survey of key management
Security Scalability Robustness Reliability
DKPS Medium Medium Medium High
PKIE Medium Low Medium Medium
INF Low High High Low
URSA Medium High Low High
MOCA High High Low Medium
SOKM Medium Medium High Medium
SEKM High Medium High High
Identity Based High High Medium High
SEGK Low High High Low
PGSK High Medium High High
Cluster based key Low Low Low Medium
Zone based key Low Low Medium Low
V. CONCLUSION
Different types of key management schemes are covered in this survey paper. In summary, symmetric
key management schemes are described in three categories DKPS, PIKE and INF. DKPS symmetric
1016 Vol. 7, Issue 3, pp. 1009-1017

key management scheme is much efficient as compared to group key schemes and pair wise key
agreement. PIKE scheme has good security services with fair scalability. INF model has no need of a
collaborative effort with having low storage cost. In this paper observe that DKPS is highly secure
and efficient schemes as compared to other symmetric key management schemes. The identity-based
key management is reliable. This scheme solved the security problem in the ad hoc network and is
also suitable for application to other wired and wireless network. This is a major problem of security
in the identity-based key management. SEGK is group key scheme in Ad-Hoc network double
multicast tree is constructed in this model. . Two detection methods are introduced into SEGK
scheme. Cluster based &Zone based key schemes come in hybrid or composite key management
scheme. In future work, we will focus in Identity based secure key management using Elliptic Curve
Cryptography. Advantages of using elliptic curve they seems to offer a level of security comparable to
classical that use much larger key size .Minimum key sizes for ECC should be 132 bits Vs. 952 bits
for RSA .
ACKNOWLEDGEMENTS
Author is thankful to his guides Mr. Anuj singh and Ms. Anuradha who give him an opportunity to
carry out this work. Author is also thankful to NPTEL which provides E-learning through online Web
and Video courses in Engineering, Science and humanities streams
REFERENCES
[1] Panagiotis Papadimitraos and Zygmunt J. Hass, Securing Mobile Ad Hoc Networks, in Book The
Handbook of Ad Hoc Wireless Networks (Chapter 31), CRC Press LLC, 2003.
[2] J.Kong P.Zerfos H.Luo, S.Lu and L. Zhang, Providing robust and ubiquitous security support for mobile
ad hoc networks, in Proceedings of the 9th International Conference on Network Protocols(ICNP),
November 2001, pp. 251-260.
[3] Preetida Vinayakray-Jani, Security within Ad hoc Networks, Nokia Research Center, Helsinki, Finland.
Position Paper, PAMPAS Workshop, Sept. 16/17 2002, London.
[4] Wu, B., Chen, J., Wu, J., and Cardei, M. (2006). A Survey on Attacks and Countermeasures in Mobile Ad
Hoc Networks. Wireless/Mobile Network Security, Springer. Chapter 12.
[5] Aldar C-F. Chan, Distributed Symmetric Key Management for Mobile Ad hoc Networks, IEEE, 2004.
[6] Aziz, B., Nourdine, E. and Mohamed, E., A Recent Survey on Key management Schemes in
MANETICTTA08, pp. 1-6, 2008.
[7] R. Anderson, Haowen and Perring, Adrian, Key Infection: Smart trust for smart dust, 12th IEEE
International Conference on Network Protocol ICNP, 2004.
[8] Valle, G. and Cerdenas, R., Overview the key Management in Ad Hoc Networks, ISSADS pp. 397 406,
2005.
[9] Wu, B., Wu, J., Fernandez, E., Ilyas, M. and Magliveras, S., Secure and Efficient key Management in
mobile ad hoc networks, Network and Computer Applications, Vol. 30, pp. 937-954, 2007.
[10] Luo, H. and Lu, S., URSA: Ubiquitous and Robust Access Control for Mobile Ad Hoc Networks, IEEE /
ACM Transactions on Networking Vol. 12, pp. 1049-1063, 2004.
[11] Yi, S., Naldurg, P. and Kravets , R, Security-aware ad hoc routing for wireless networks ,MobiHoc, pp.
299-302, 2001.
[12] Zhou, L. and Hass, Z.,Secure Ad Hoc Networks, IEEE Network Magazine vol. 13, no. 6, pp.24-30, 1999.
[13] Capkun, S., Buttya, L., and Hubaux, P.,Self-Organized Public Key Management for Mobile AdHoc
Networks, IEEE Trans. Mobile Computing, vol. 2, no. 1, pp. 52-64, 2003.
[14] A. Khalili, Katz, Jonathan and Arbaugh, William A., Towards secure key distribution in truly ad hoc
networks, IEEE Workshop on Security and Assurance in ad hoc Networks i conjunction with the 2003
International Symposium on Application and the Internet, 2003.
[15] AnilKapil and SanjeevRana, Identity-Based Key Management in MANETs using Public Key
Cryptography, International journal of Security, vol. (3): Issue (1).
[16] Wan AnXoing, Yao Huan Gong, Secure and Highly Efficient Three Level Key Management Scheme for
MANET, WSEAS TRANSACTIONS on COMPUTERS, Vol. 10, Issue 10, 2011.
[17] Bing Wu, Jie Wu and YuhongDong,An efficient group key management scheme for mobile ad hoc
network, International Journal and Networks, Vol. 2008.
[18] D. Boneh, X. Boyen, and H. Shacham, Short group signatures, in Advances in CryptologyCrypto04,
Lecture Notes in Computer Science, vol. 3152, 2004, pp. 4155
1017 Vol. 7, Issue 3, pp. 1009-1017

[19] R. PushpaLakshmi, A. Vincent Antony Kumar,Cluster Based Composite Key Management in Mobile Ad
Hoc Networks, International Journal of Computer Applications, vol. 4- No. 7, 2010.
[20] ThairKhdour, Abdullah Aref, A HYBRID SCHEMA ZONE-BASED KEY MANAGEMENT FOR
MANETS, Journal of Theoretical and Applied Information Technology, vol. 35 No. 2, 2012.
AUTHORS
Anju Chahal M.Tech (CSE) AMITY University, GURGAON, HARYANA, Anju Chahal
was born in Bhiwani, Haryana, India, in 1992. She received the Degree in Bachelors of
Technology in Computer Science Engineering from Shri Baba Mastnath Engineering
College (SBMN), MDU Rohtak, in Year 2008-2012 and currently pursuing Masters in
Computer Science engineering, degree from AMITY University, Haryana. Her research
interests include Network Security, Cloud Computing, and Data Security.

Anuj Kumar Assistant Professor at the Department of Computer science in AMITY University, GURGAON,
HARYANA, India. His research interests include Data storage , Network security , and Data Security.

Anuradha Rani

Assistant Professor, AMITY University, HARYANA ,INDIA Anuradha rani is born in hansi a
professor at the Department of Computer science in AMITY University , GURGAON, HARYANA, India. She
received degree in Masters of science in Information Technology from G.J.U, Hisar ,2005, Masters of
Technology, in Computer Science, from Banasthali university 2007. Beachelors of science from K.U.K
University 2003. Her research interests include Network Security, data mining, and Networking.
1018 Vol. 7, Issue 3, pp. 1018-1026

PREDICTION OF STUDY TRACK BY APTITUDE TEST USING
JAVA
Deepali Joshi and Priyanka Desai
Department of Computer Engineering,
Thakur College of Engineering & Technology, Mumbai, Maharashtra, India

ABSTRACT
In todays competitive world everyone wants to be successful to achieve this it is essential to be successful in
academics. The basic education is from 1st to 10th standard and once 10th standard is complete there are various
courses that can be selected by the student. The students get confused for selecting the appropriate field. The
proposed system can be used to solve this problem in order to achieve this aptitude test is implemented. Aptitude
Test is utilized by the proposed system which predicts the suitable stream depending upon the intellectual
capability of the student. To find the suitable stream aptitude test method can be used as compared to the
traditional process. The proposed system is beneficial as compared to traditional system as the accuracy of results
is better.
KEYWORDS: Aptitude, ssc, ssc marks, Accuracy, Streams
I. INTRODUCTION
Each and every person wants to be successful in all phases of life. To be successful it depends on
whether or not the correct field is selected. If correct field is selected by students then they will be
successful in their careers [1]. If appropriate field is not selected then students face lot of problems.
The education is categorized in various phases. Once 10
th
standard is complete number of options are
available to students and depending upon the field the career will be decided. After 10
th
various courses
are available like Science, Commerce and Arts [6]. As many options are available it becomes difficult
to choose the suitable field. A method or technique is required through which the students can find
suitable stream. Some solutions are available in order to solve the problem but they do not provide
appropriate results. One method which is used to specify the stream is the Aptitude Test Method. The
aptitude test consists of Questions and along with questions answers are also displayed. The student
have to find out the correct answer from the given options. The questions specify various streams like
Science, Commerce, Arts and Diploma. Through the aptitude test intellectual capability of the student
can be assessed and the suitable stream can be predicted to the student. The paper contains detailed
information about Building the Model, DATA COLLECTION, Tools, Implementation, Solution,
Results and Conclusion and followed by references.
II. BUILDING THE MODEL
Aptitude Test is an efficient method through which prediction of the field can be done. The main
advantage of aptitude test is that it completely depends upon the intellectual capability of the student.
The test will be carried out on every individual student and the result will be generated. The result of
students will be different because it will be specific to that student only. In order to build the model
Java Net beans is used and along with this a database is connected so that the questions as well as
answers are stored. The database used is SQLite.
1019 Vol. 7, Issue 3, pp. 1018-1026

III. DATA COLLECTION
To implement the proposed system data is required and the data consists of questions and their correct
answer along with three options. The data is gathered from the 10
th
standard text books and the internet
is also utilised for this purpose. The questions and answers belong to the streams like Science,
Commerce, Arts, Diploma and Engineering.
IV. TOOLS
The tools which are used to implement the proposed system are Java Net beans and the database is
SQLite.
4.1 Java
Java is a set of several computer software products and specifications from Sun Microsystems (which
has since merged with Oracle Corporation), that together provide a system for developing application
software and deploying it in a cross-platform computing environment. Java is used in a wide variety
of computing platforms from embedded devices and mobile phones on the low end, to enterprise
servers and supercomputers on the high end. While less common, Java applets are sometimes used to
provide improved and secure functions while browsing the World Wide Web on desktop computers.
4.2 SQLite
SQLite is an in-process library that implements a self-contained, server less, zero-configuration,
transactional SQL database engine. The code for SQLite is in the public domain and is thus free for use
for any purpose, commercial or private. SQLite is currently found in more applications than we can
count, including several high-profile projects. SQLite is an embedded SQL database engine. Unlike
most other SQL databases, SQLite does not have a separate server process. SQLite reads and writes
directly to ordinary disk files. A complete SQL database with multiple tables, indices, triggers, and
views, is contained in a single disk file. The database file format is cross-platform - you can freely copy
a database between 32-bit and 64-bit systems or between big-endian and little-endian architectures.
These features make SQLite a popular choice.
V. IMPLEMENTATION
In order to predict the suitable stream, the student have to give input such as name, gender, phone
number, email id, hobbies, railway line, nearby station. Once this is complete a option is provided to
the student that is take aptitude test. As soon as the student clicks on this option the aptitude test screen
will be displayed. The aptitude test is implemented so that the student can find the suitable stream
depending upon the intellectual capability. The aptitude test consists of questions from various streams.
For each stream there are three levels of questions and if the student gives correct answer then the level
will be incremented. The questions of first level are of one mark, the second level questions are of two
marks and the third level questions are of three marks. The first level consists of three questions, if the
student gives correct answer of first question then the level will be incremented and the questions of
level two will be displayed to the student. If the answer to the question one of level one is incorrect then
the second question of level one will be displayed and there is no increment of the level. Similarly if
the answer of first question of level two is correct then the level will be incremented and the question
of level three will be displayed to the student. Thus if the student gives correct answer of all three levels
then the student has faced only three questions and the suitable stream will be displayed to the student.
The questions and their answers are stored in database and will be displayed to the student on the onset
of the test. There are various attributes such as id, Stream, Number, Questions, ans1, ans2, ans3, ans4,
correctans. The attribute id is used for identification. Point is basically the marks allotted for each and
every question. Stream specifies to which stream the question belongs. The streams are MED which
specifies the question belongs to Science field, COM specifies Commerce, ARTS as the name suggests
Arts, ENG specifies diploma. Number indicates the question number. Question consists of questions to
be asked or displayed to the student. Every question will have four answer options one of which is
correct. The correctans consists of answer number which is correct. For each stream there are three
1020 Vol. 7, Issue 3, pp. 1018-1026

levels of questions and if the student gives correct answer then the level will be incremented. The
questions of first level are of one mark, the second level questions are of two marks and the third level
questions are of three marks. The first level consists of three questions, if the student gives correct
answer of first question then the level will be incremented and the questions of level two will be
displayed to the student. If the answer to the question one of level one is incorrect then the second
question of level one will be displayed and there is no increment of the level. Similarly if the answer of
first question of level two is correct then the level will be incremented and the question of level three
will be displayed to the student. Thus if the student gives correct answer of all three levels then the
student has faced only three questions and the suitable stream will be displayed to the student.
Table 1. Database Table.

The above table contains various attributes such as id, point, stream, number, questions, ans1, ans2, ans3,
ans4 and correctans. Id is used for identification, point specifies the marks allotted to each and every
question. The stream specifies to which stream the question belongs such as Commerce, Arts, Diploma
and Science. Number is nothing but the question number. Questions contain the detailed questions along
with answer option ans1, ans2, ans3, ans4 and the correctans contains the numeric which specifies correct
answer.
1021 Vol. 7, Issue 3, pp. 1018-1026

Figure 1. Student Details
In the above figure the student will enter their basic detail.
The first step in proposed system is to gather the student details such as the Name, Gender, Phone No,
Email id, Hobbies, Railway Line, Near by Station. Once the basic details are filled by the student the first
screen appears where the question, answers, number of questions attempted will be displayed to the student.
It also consists of buttons such as Start, Next, End which perform the operation as the name suggests.

1022 Vol. 7, Issue 3, pp. 1018-1026

Figure 2. Main Page
The above screen is the first screen that appears when the student clicks on aptitude.
As soon as the student clicks on Start button the first question will be displayed to the student along
with the options. Question no will also be displayed to the student. On the screen it displays Question
No. 1/12, which indicates the first question out of twelve. In the next screen the student has entered the
option 1 and then clicks on Next the next question will be displayed. Similarly the student has to answer
twelve questions and then the result will be displayed. The result that is generated is engineering and
along with this the Aptitude score is also displayed.

1023 Vol. 7, Issue 3, pp. 1018-1026

Figure 3. Aptitude Question with options.
In the above figure the aptitude question along with the answer options is displayed.

Figure 4. First question with answer.
1024 Vol. 7, Issue 3, pp. 1018-1026

In the above figure once the question has been read and understood the student enters the suitable answer
option.

Figure 5. Next question with options.
The second question along with the options are displayed.

Figure 6. Result
In the above figure the suitable stream is predicted.
1025 Vol. 7, Issue 3, pp. 1018-1026

VI. SOLUTION
The proposed system uses Aptitude Test method in order to specify the suitable stream to the student.
It takes input such as name, contact number, email-id, hobbies, railway line, nearby station this is done
for registration purpose. Once the registration is complete the aptitude screen will be displayed and as
soon as the student clicks on start the question along with the answers will be displayed. There are four
options for the answers and the student has to select one. The questions belong to different streams.
Depending upon the correct answer given by the student maximum number of times for the specific
stream indicates that the student has prerequisite knowledge and that stream is more suitable for that
student. The aptitude test method provides efficient result because it depends upon the intellectual
capability of the student.
VII. RESULTS
The proposed system is more efficient as compared to the existing system and generates appropriate
results. Through the aptitude test, the student can answer the questions and find their intellectual
capability for the specific stream and according to the correct answers the stream will be predicted. The
Aptitude Test method was conducted for twenty-five students of Abhinandan Classes. Each and every
student personally gave the aptitude test on separate machines and the questions displayed to them were
different to make sure that their intellectual capability is tested in a efficient way. The aptitude test
method worked successfully and predicted the suitable stream for individual student.
The pie chart specifies the result of the Aptitude Test. There are four streams Science, Commerce, Arts,
Diploma. The test was conducted on twenty-five students, out of which science was the suitable stream
for eight students and in percentage it can be specified as 32%, commerce for five students and
percentage is 20%, arts for three students and percentage is 12%, diploma for nine students and can be
specified as 20%.

Figure 7. Pie Chart
The above pie chart shows the percentage of the specific field being predicted.
VIII. CONCLUSION
The proposed system was tested on students from Abhinandan Classes. Each and every student gave
the test individually on separate machines and distinct questions displayed to them. The number of
students who appeared for the aptitude test was 25. The suitable stream was predicted to them through
32%
20% 12%
36%
Stream
Science Commerce Arts Diploma
1026 Vol. 7, Issue 3, pp. 1018-1026

the aptitude test method. Science was predicted for eight students and in percentage it can be specified
as 32%, commerce for five students, arts for three students, diploma for nine students. Once the aptitude
test was completed an interactive session was carried out and the students commented that the streams
predicted to them were suitable from their point of view. The aptitude test worked successfully and
predicted the suitable stream for individual student. The students as well as their parents have clarity
about the stream and their wards future.
IX. FUTURE WORK
In this paper, we proposed and built a model to predict a suitable stream to the students depending on
their intellectual capability. The proposed system aims to improve the quality of education by helping
the student to select the stream which is appropriate for them. The proposed system is implemented for
ssc students and predicts the stream. The research can be enhanced by implementing it for the hsc
students so that once the stream can be predicted to them.
REFERENCES
[1] Ahmad I., Manarvi, I., Ashraf, N. Predicting university performance in a subject based on
high school majors, 978-1-4244-4136-5/09/ 2009 IEEE
[2] Zhiwu Liu, Xiuzhi Zhang. Prediction and Analysis for Students' Marks Based on Decision Tree Algorithm,
Intelligent Networks and Intelligent Systems (ICINIS), 2010 3rd International Conference on Digital Object
Identifier:10.1109/ICINIS.2010.59 Publication Year: 2010 , Page(s): 338 341
[3] Anupama Kumar S, Vijayalakshmi M.N. Mining of student academic evaluation records in higher education,
Recent Advances in Computing and Software Systems (RACSS), 2012 International Conference on Digital Object
Identifier: 10.1109/RACSS.2012.6212699
Publication Year: 2012 IEEE , Page(s): 67 70
[4] Bunkar, K, Singh U.K., Pandya B, Bunkar R, Data mining: Prediction for performance
improvement of graduate students using classification,
Wireless and Optical Communications Networks (WOCN), 2012 Ninth International Conference on Digital
Object Identifier:10.1109/WOCN.2012.6335530 Publication Year: 2012 IEEE, Page(s): 1 5
[5] Garcia, E.P.I. ; Mora, P.M., Model Prediction of Academic Performance for First Year Students, Artificial
Intelligence (MICAI), 2011 10th Mexican International Conference on Digital Object
Identifier: 10.1109/MICAI.2011.28 Publication Year: 2011 IEEE , Page(s): 169 174
[6] Qasem A. Al-Radaideh, Ahmad Al Ananbeh, and Emad M. Al-Shawakfa A Classification Model For
Predicting The Suitable Study Track For School Students. IJRRAS 8 (2) August 2011
[7] Pumpuang, P., Srivihok, A., Praneetpolgrang, Comparisons of classifier algorithms: Bayesian network,
C4.5, decision forest and NBTree for Course Registration Planning model of undergraduate students, Systems,
Man and Cybernetics, 2008. SMC 2008. IEEE International Conference on Digital Object
Identifier: 10.1109/ICSMC.2008.4811865
Publication Year: 2008 IEEE, Page(s): 3647 3651

AUTHORS
Deepali Joshi, M.E (Pursuing), B.E (CMPN). Area of Specialization Data Mining,
Operating System

Priyanka Desai, Ph.D (Pursuing), M.Tech (CSE), B.E (CSE). Area of
Specialization Networks, Web/Text Mining, Software Engineering,
Database/Object Oriented Technology.

1027 Vol. 7, Issue 3, pp. 1027-1037

UNSTEADY MHD THREE DIMENSIONAL FLOW OF
MAXWELL FLUID THROUGH A POROUS MEDIUM IN A
PARALLEL PLATE CHANNEL UNDER THE INFLUENCE OF
INCLINED MAGNETIC FIELD
L.Sreekala
1
, M.VeeraKrishna
2
*, L.HariKrishna
3
and E.KesavaReddy
4

1
Assistant Professor, Department of Mathematics, CRIT, Anantapur, Andhra Pradesh, India
2
Department of Mathematics, Rayalaseema University, Kurnool, Andhra Pradesh, India
3
Assistant Professor, Department of Mathematics, AITS, Kadapa, Andhra Pradesh, India
4
Professor, Department of Mathematics, JNTU, Anantapur, Andhra Pradesh, India

ABSTRACT
In this paper, we discuss the unsteady hydro magnetic flow of an electrically conducting Maxwell fluid in a
parallel plate channel bounded by porous medium under the influence of a uniform magnetic field of strength
Ho inclined at an angle of inclination with the normal to the boundaries. The perturbations are created by a
constant pressure gradient along the plates. The time required for the transient state to decay and the ultimate
steady state solution are discussed in detail. The exact solutions for the velocity of the Maxwell fluid consists of
steady state are analytically derived, its behaviour computationally discussed with reference to the various
governing parameters with the help of graphs. The shear stresses on the boundaries are also obtained
analytically and their behaviour is computationally discussed in detail.
KEYWORDS: Maxwell fluids, unsteady flows, porous medium, parallel plate channels, MHD flows
I. INTRODUCTION
Several fluids including butter, cosmetics and toiletries, paints, lubricants, certain oils, blood, mud,
jams, jellies, shampoo, soaps, soups, and marmalades have rheological characteristics and are referred
to as the non-Newtonian fluids. The rheological properties of all these fluids cannot be explained by
using a single constitutive relationship between stress and shear rate which is quite different than the
viscous fluids [1, 2]. Such understanding of the non-Newtonian fluids forced researchers to propose
more models of non-Newtonian fluids. In general, the classification of the non-Newtonian fluid
models is given under three categories which are called the differential, the rate, and the integral types
[3]. Out of these, the differential and rate types have been studied in more detail. In the present
analysis we discuss the Maxwell fluid which is the subclass of rate-type fluids which take the
relaxation phenomenon into consideration. It was employed to study various problems due to its
relatively simple structure. Moreover, one can reasonably hope to obtain exact solutions from
Maxwell fluid. This motivates us to choose the Maxwell model in this study. The exact solutions are
important as these provide standard reference for checking the accuracy of many approximate
solutions which can be numerical or empirical in nature. They can also be used as tests for verifying
numerical schemes that are being developed for studying more complex flow problems [49]. On the
other hand, these equations in the non-Newtonian fluids offer exciting challenges to mathematical
physicists for their exact solutions. The equations become more problematic, when a non-Newtonian
fluid is discussed in the presence of MHD and porous medium. Despite this fact, various researchers
are still making their interesting contributions in the field (e.g., see some recent studies [115]). Few
investigations which provide the examination of non-Newtonian fluids in a rotating frame are also
1028 Vol. 7, Issue 3, pp. 1027-1037

presented [119]. Recently Faisal Salah [20] discussed two explicit examples of acceleration subject
to a rigid plate are taken into account. Constitutive equations of a Maxwell fluid are used and
modified Darcys law has been utilized. The exact solutions to the resulting problem are developed by
Fourier sine transform. With respect to physical applications, the graphs are plotted in order to
illustrate the variations of embedded flow parameters. The mathematical results of many existing
situations are shown as the special cases of that study. Such studies have special relevance in
meteorology, geophysics, and astrophysics. Hayat et.al [21] investigated to analyze the MHD rotating
flow of a Maxwell fluid through a porous medium in parallel plate channel. M.V. Krishna [22]
discussed analytical solution for the unsteady MHD flow is constructed in a rotating non-Newtonian
fluid through a porous medium taking hall current into account. In this paper, we examine the MHD
flow of Maxwell fluid through a porous medium in a parallel plate channel with inclined magnetic
field, the perturbations in the flow are created by a constant pressure gradient along the plates. The
time required for the transient effects to decay and the ultimate steady state solution are discussed in
detail. The exact solutions of the velocity in the Maxwell fluid consists of steady state are analytically
derived, its behaviour computationally discussed with reference to the various governing parameters
with the help of graphs. The shear stresses on the boundaries are also obtained analytically and their
behaviour is computationally discussed.
II. FORMULATION AND SOLUTION OF THE PROBLEM
We consider the unsteady flow of an electrically conducting Maxwell fluid through porous medium in
a parallel plate channel subjected to a uniform transverse magnetic field of strength Ho inclined at an
angle of inclinationo normal to the channel walls. The boundary plates are assumed to be parallel to
xy-plane and the magnetic field to the z-axis in the transverse xz-plane. The component along z-
direction induces a secondary flow in that direction while its x-components changes perturbation to
the axial flow. At 0 > t the fluid is driven by a prescribed pressure gradient parallel to the channel
walls. We choose a Cartesian system O(x, y, z) such that the boundary walls are at 0 = z and l z = ,
since the plates extends to infinity along x and y directions, all the physical quantities except the
pressure depend on z and t alone. The unsteady hydro magnetic equations governing the electrically
conducting Maxwell fluid under the influence of transverse magnetic field with reference to a frame
are
R B J S div p V V.
t
V
+ + + V =
(
V +
c
c
) ( (2.1)
0 V . = V (2.2)
0 B . = V (2.3)
J B
m
= V (2.4)
t
B
E
c
c
= V (2.5)
Where, J is the current density, B is the total magnetic field, E is the total electric field,
m
is the
magnetic permeability, V = (u, v, w) is the velocity field, T is the Cauchy stress tensor, B is the total
magnetic field so that B=B0 o Sin + b, where B0 is the applied magnetic field parallel to the z-axis and
b is the induced magnetic field. The induced magnetic field is negligible so that the total magnetic
field B = (0, 0, B0 o Sin ), the Lorentz force V Sin B B J o o
2 2
0
= , o is the electrical conductivity
of the fluid, is the density of the fluid, and
Dt
D
is the material derivative and R is the Darcy
resistance. The extra tensor S for a Maxwell fluid is
S I p T + = (2.6)
A SL LS
Dt
DS
S
T
=
|
.
|
\
|
+ (2.7)
1029 Vol. 7, Issue 3, pp. 1027-1037

where I p is the stress due to constraint of the impermeability, here p is the static fluid pressure, I
is the identity tensor, is the viscosity of the fluid, is the material time constants referred to as
relaxation time, it is assumed that 0 > . The first Rivlin-Ericksen tensor A1 is defined as
A1 = (grad V) + (grad V)
T
(2.8)

It should be noted that this model includes the viscous Navier-Stokes fluid as a special case for 0 =
. Let us indicate the stress tensor and the velocity component as
V(z, t) = (u, 0, w) (2.9)
According to Tan and Masuoka [4] Darcys resistance in an Oldroyd-B fluid satisfies the following
expression:
V
t
1
k
R
t
1
r
|
.
|
\
|
c
c
+ =
|
.
|
\
|
c
c
+
|
(2.10)
where is the retardation time, | is the porosity (0<| <1), and is the permeability of the porous
medium. For Maxwell fluid = 0, and hence,
V
k
R
t
1
|
=
|
.
|
\
|
c
c
+ (2.11)
Making use of the equations (2.6), (2.7) and (2.8), the equation (2.1) reduces to
x
2 2
0
xz
R u Sin B
z
S
x
p
t
u
+
c
c
+
c
c
=
c
c
o (2.12)
z
2 2
0
yz
R w Sin B
z
S
t
w
+
c
c
=
c
c
o (2.13)
Where and z are and z-components of Darcys resistance ;
z
u
S
t
1
xz
c
c
=
|
.
|
\
|
c
c
+ and
z
w
S
t
1
yz
c
c
=
|
.
|
\
|
c
c
+ (2.14)
The equations (2.12) and (2.13) reduces to
u
k

u Sin B
z
S
x
p
t
u
2 2
0
xz

c
c
+
c
c
=
c
c
o (2.15)
w
k

v Sin B
z
S
t
w
2 2
0
yz

c
c
=
c
c
o (2.16)
Let iw u q + = Combining equations (2.15) and (2.16), we obtain
q
k

q Sin B iS S
z x
p
t
q
2 2
0 yz xz
+
c
c
+
c
c
=
c
c
o ) ( (2.17)
Since
z
) (
t
1
c
c
= +
|
.
|
\
|
c
c
+
q
iS S
yz xz
(2.18)
Substituting the equation (2.18) in the equation (2.17), we obtain the equation for the governing the
flow through a porous medium with respect to the rotating frame is given by
2
2 2
0
z
q
x
p
t
1
1
q
t
1
k
Sin B
t
q
t
1
c
c
+
c
c
|
.
|
\
|
c
c
+ =
|
.
|
\
|
c
c
+ + +
c
c
|
.
|
\
|
c
c
+ ) (
2
| o
(2.19)
The boundary and initial conditions are
0 z 0, t q = > = 0 (2.20)
l z , 0 t , 0 q = = = (2.21)
z all for , 0 t , 0
dt
t) dq(z,
, 0 t) (z, q s = = (2.22)
We introduce the following non dimensional variables are

2
2
* *
2
*
2
* * *
Pl
P ,
l
l
,
l
t
t ,
l q
q ,
l
z
z = = = = = =
1030 Vol. 7, Issue 3, pp. 1027-1037

Using non dimensional variables the governing equations are (dropping asterisks in all forms)
2
2
1 1
1 2
1
z
q
P
t
1
1
q
t
1 D Sin M
t
q
t
1
c
c
+ |
.
|
\
|
c
c
+ = |
.
|
\
|
c
c
+ + +
c
c
|
.
|
\
|
c
c
+
) (
2
| o (2.23)
where,
v
l H
M
2 2
0
2
e 2
= is the Hartmann number,
k
l
D
2
1
=
is the inverse Darcy Parameter,

2 1
l
= is the material parameter related to relaxation time and

x
p
P
c
c
= is the pressure gradient.
The corresponding initial and boundary conditions are
0 z 0, t q = > = 0 (2.24)
1 z , 0 t , 0 q = = = (2.25)
z all for , 0 t , 0
dt
t) dq(z,
, 0 t) (z, q s = = (2.26)
supposing the pressure is given by
<
> +
=
z t
z t e P P
P
t i
0 , 0
, 0 ,
1
1 0
e
(2.27)
Taking Laplace transforms of equations (2.23) and (2.27) using initial conditions (2.26) the
governing equations in terms of the transformed variable reduces to
| | = + + + + +

q D Sin M s D Sin M 1 s
dz
q d
1 2 1 2
1
2
1 2
2
) ( )) ( (
2 2
| o | o | |

s
P
i - s
P
i 1
0
1
1
1 1
+ ) ( | (2.28)
Solving equation (2.28) subjected to the conditions (2.24) and (2.25), we obtain

) i (s
z) Cosh( ) i (1 P
q
1
2
1
1 1 1 1
+
=
s
z) ( Cosh P
2
1
1 0

+
+
+
) Sinh( s.
z) ( Sinh ). Cosh( P
) Sinh( ). i (s
z) ( )Sinh )Cosh( i (1 P
1
2
1
1 1 0
1 1
2
1
1 1 1 1 1

) Sinh( s.
z) ( .Sinh P
) Sinh( ). i (s
z) ( ).Sinh i (1 P
1
2
1
1 0
1 1
2
1
1 1 1 1
) s s(1
P
) i (s
) i (1 P
2
1
0
1
2
1
1 1 1
o +
+
+
+ (2.29)
Where ) D Sin (M ))s D Sin (M (1 s
1 2 1 2
1
2
1
2
1

+ + + + + = o o
2 2

Taking the inverse Laplace transforms to the equations (2.29) on both sides, We obtain
+ + + =
2
0
0
0
2
0
0 0
0
2
0
0 0 0
2
0
0 0
b
P
) Sinh(b b
z) Sinh(b P
) Sinh(b b
z) Sinh(b ). Cosh(b P
b
z) Cosh(b P
q

+

+
+
) Sinh(b
z) Sinh(b ). Cosh(b
z) Cosh(b
) s )(i s (i
) i (1 P
4
4 4
4
2 1 1 1
1 1 1

+

+
+
)
`
+
) i )(s s (s
z) Cosh(b ). i (1 P
e 1
) Sinh(b
z) Sinh(b
1 1 2 1
3 1 1 1
t i
4
4 1

) Sinh(b ) i )(s s (s
z) Sinh(b ). i (1 P
) Sinh(b ). i )(s s (s
z) Sinh(b ). Cosh(b ). i (1 P
3 1 1 2 1
3 1 1 1
3 1 1 2 1
3 3 1 1 1

+

+
+

+
+
) Sinh(b ). )(s s (s
s ) s (s
z) Cosh(b P
) i )(s s (s
) i (1 P
3 1 2 1
3 3 0
1 2 1
3 0
1 1 2 1
1 1 1

+
)
`
t s
1 2 1
0
3 1 2 1
3 0
1
e
) )(s s (s
P
) Sinh(b ). )(s s (s
z) Sinh(b . P

1031 Vol. 7, Issue 3, pp. 1027-1037

+

+
+
) i )(s s (s
z) Cosh(b ). i (1 P
1 2 1 2
7 1 1 1

) Sinh(b ) i )(s s (s
z) Sinh(b ). i (1 P
) Sinh(b ). i )(s s (s
7 1 2 1 2
7 1 1 1
7 1 2 1 2
7 7 1 1 1

+

+
+

+
+
) Sinh(b . s ) s (s
s ) s (s
z) Cosh(b P
) i )(s s (s
) i (1 P
7 2 1 2
7 7 0
2 1 2
7 0
1 2 1 2
1 1 1

+
)
`
t s
2 1 2
0
7 2 1 2
7 0
2
e
s ) s (s
P
) Sinh(b . s ) s (s
z) Sinh(b . P

{
=0 n
+

+

+
) i )(s s (s b
z) Sinh(b ). i (1 P
) i )(s s (s b
1 3 4 3
2
6
6 1 1 1
1 3 4 3
2
6
6 6 1 1 1

t s
3 4 3
2
6
6 0
3 4 3
2
6
6 6 0 3
e
) )(s s (s b
z) Sinh(b P
) )(s s (s b
)
`
+
{
=
+
0 n
+

+

+
) i )(s s (s b
z) Sinh(b ). i (1 P
) i )(s s (s b
1 4 3 4
2
5
5 1 1 1
1 4 3 4
2
5
5 5 1 1 1

t s
4 3 4
2
5
5 0
4 3 4
2
5
5 5 0 4
e
) )(s s (s b
z) Sinh(b P
) )(s s (s b
)
`
+ (2.30)
(Where the constants are mentioned in the appendix)
The shear stresses on the upper and lower plate are given by

1 z
U
dz
dq
=
|
.
|
\
|
=
and
0 z
L
dz
dq
=
|
.
|
\
|
= (2.31)
III. RESULTS AND DISCUSSION
We discuss the unsteady flow of an electrically conducting Maxwell fluid through a porous medium
in parallel plate channel subjected to uniform magnetic field. In unperturbed state the perturbation are
created by performing to imposition of constant pressure gradient along the axis (OX) of the channel
walls the velocity component along the imposed pressure gradient and normal to it. Under the
boundary layer assumptions these velocity components are to functions of z and t alone, where z
corresponds to the direction of axis of the channel. The transverse magnetic field once arising give
rise to Lorentz forces resisting the flow along normal to the channel wall.
The constitutive equations relating the stress and rate of strain are chosen to depict the Maxwell fluid.
The Brinkmans model has been chosen to analyses the flow through a porous medium. The equation
governing the velocity components and with reference to frame ultimately can be combined into a
single equation by defining the complex velocity iw u q + = The expression for the components of
the stresses are manipulated from the stress and strain relationships. Under these assumptions the
ultimate governing equations for the unsteady flow through a porous medium with reference to frame
is formulated the corresponding boundary and initial conditions. This boundary value problem has
been solved using non-dimensional variables making use of Laplace transform technique.
The solution for the combined velocity q consists of two kinds of terms 1. Steady state 2. The
transient terms involving exponentially varying time dependence. The analysis of transient terms
indicates that this transient velocity decay exponentially in dimensionless time to of order i.e.,
>
4 3
1
1
,
1
, max
s s
t | . This decay in the transient term depends on the non-dimensional parameters
1, M and D
-1
. When these transient terms decay the ultimate velocity consists of steady and
oscillatory components.
1032 Vol. 7, Issue 3, pp. 1027-1037

2
0
0
0
2
0
0 0
0
2
0
0 0 0
2
0
0 0
steady
b
P
) Sinh(b b
z) Sinh(b P
) Sinh(b b
b
z) Cosh(b P
q + + == ) ( (3.1)
The flow governed by the non-dimensional parameters namely viz. M the magnetic field parameter
(the Hartmann number), D
-1
the inverse Darcy parameter,
1
| is the material time parameter referred
as relaxation time. The computational analysis has been carried out to discuss the behaviour of
velocity components u and w on the flow in the rotating parallel plate channel and the lower plate
executes non-torsional oscillations in its own plane with reference to variations in the governing
parameters may be analyzed from figures (1-3) and (4-6) respectively (P0=P1=10, t =0.1,
3 / , 4 /
1
t o t e = = ).
We may note that the effect of the magnetic field on the flow from figures (1 and 4). The magnitude
of the velocity component u reduces and the velocity component w increases with increase in the
Hartmann number M. However, the resultant velocity reduces throughout the fluid region with
increase in the intensity of the magnetic field (the Hartmann number M). The figures (2 and 5)
represent the velocity profiles with different variation in the inverse Darcy parameter D
-1
. We find that
the magnitude of u reduces with decrease in the permeability of the porous medium, while the
magnitude of w experiences a slight enhancement with increase in the inverse Darcy parameter D
-1
. It
is interesting to note that lesser the permeability of the porous medium lower the magnitude of the
resultant velocity. i.e., the resultant velocity reduces throughout the fluid region with increase in the
inverse Darcy parameter D
-1
. Both the velocity components u and w enhances with increase in the
relaxation time entire fluid region. These displayed in the figures (3 and 6). The resultant velocity
enhances throughout the fluid region with increase in the relaxation time. The shear stresses on the
upper and lower plates have been calculated with reference to variations in the governing parameters
and are tabulated in the tables (I-IV). On the upper plate the magnitude of the stresses
x
t enhances
with increase in M and
1
| , while it reduces with increase in the inverse Darcy parameter D
-1
. The
magnitude of the stresses
y
t enhances with increase in for all governing parameters M, D
-1
and
1
|
(tables. I-II). On the lower plate the magnitude of the stresses
x
t and
y
t enhances with increase in M
and
1
| , while these reduces with increase in the inverse Darcy parameter D
-1
(tables. III-IV).
IV. CONCLUSIONS
1. The resultant velocity reduces throughout the fluid region with increase in the intensity of the
magnetic field (the Hartmann number M).
2. Lesser the permeability of the porous medium lower the magnitude of the resultant velocity. i.e.,
the resultant velocity reduces throughout the fluid region with increase in the inverse Darcy
parameter D
-1
.
3. Both the velocity components u and w and the resultant velocity enhances with increase in the
relaxation time in the entire fluid region.
4. On the upper plate the magnitude of the stresses
x
t enhances with increase in M and
1
| , while it
reduces with increase in the inverse Darcy parameter D
-1
.
5. The magnitude of the stresses
y
t enhances with increase in for all governing parameters M, D
-1

and
1
| . On the lower plate the magnitude of the stresses
x
t and
y
t enhances with increase in
M, and
1
| , while these reduces with increase in the inverse Darcy parameter D
-1
.

1033 Vol. 7, Issue 3, pp. 1027-1037

V. GRAPHS AND TABLES

Fig. 1: The velocity profile for u with M.
1
1
= | , D
1
=2000, E=0.01

Fig. 2: The velocity profile for u with D
-1
.
, 1
1
= | E=0.01, M=2

Fig. 3: The velocity profile for u with
1
| .
E=0.01, D
1
=2000, M=2

0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0 0.2 0.4 0.6 0.8 1
u
z
M=2
M=3
M=4
M=5
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0 0.2 0.4 0.6 0.8 1
u
z
D=2000
D=3000
D=4000
D=5000
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0 0.2 0.4 0.6 0.8 1
u
z
1=1
1=2
1=3
1=4
1034 Vol. 7, Issue 3, pp. 1027-1037

Fig. 4: The velocity profile for w with M.
, 1
1
= | D
1
=2000, E=0.01

Fig. 5: The velocity profile for w with D
-1
.
, 1
1
= | E=0.01, M=2

Fig. 6: The velocity profile for w with
1
| .
E=0.01, D
1
=2000, M=2
-0.14
-0.12
-0.1
-0.08
-0.06
-0.04
-0.02
0
0 0.2 0.4 0.6 0.8 1
v
z
M=2
M=5
M=8
M=10
-0.12
-0.1
-0.08
-0.06
-0.04
-0.02
0
0 0.2 0.4 0.6 0.8 1
v
z
D=2000
D=3000
D=4000
D=5000
-0.16
-0.14
-0.12
-0.1
-0.08
-0.06
-0.04
-0.02
0
0 0.2 0.4 0.6 0.8 1
v
z
1=1
1=2
1=3
1=4
1035 Vol. 7, Issue 3, pp. 1027-1037

Table I: The shear stresses (
x
t ) on the upper plate
P0 = P1 I II III IV V VI VII
2 0.084673 0.156783 0.246352 0.062501 0.046782 0.107466 0.145336
4 0.121453 0.186299 0.268751 0.116002 0.083146 0.144236 0.181673
6 0.146755 0.208888 0.278752 0.118208 0.121482 0.180083 0.256335
10 0.163752 0.408755 0.544799 0.127436 0.118442 0.207853 0.501652

I II III IV V VI VII
M 2 5 8 2 2 2 2
D
1

2000 2000 2000 3000 4000 2000 2000
1
|
5 5 5 5 5 6 8
Table II: The shear stresses (
y
t ) on the upper plate
2 -0.01467 -0.02561 -0.03216 -0.01565 -0.01682 -0.01512 -0.01811
4 -0.01814 -0.02848 -0.04821 -0.01255 -0.02845 -0.02147 -0.02533
6 -0.02107 -0.03245 -0.04552 -0.02856 -0.03215 -0.02658 -0.03275
10 -0.04251 -0.06837 -0.07550 -0.05478 -0.06253 -0.05865 -0.08314

M 2 5 8 2 2 2 2
D
1

2000 2000 2000 3000 4000 2000 2000
1
|
5 5 5 5 5 6 8
Table III: The shear stresses (
x
t ) on the lower plate
2 0.000048 0.000054 0.000064 0.000041 0.000032 0.000052 0.000084
4 0.000066 0.000072 0.000084 0.000042 0.000035 0.000062 0.000098
6 0.000072 0.000078 0.000089 0.000052 0.000042 0.000082 0.000099
10 0.000084 0.000094 0.000132 0.000062 0.000048 0.000092 0.000147

Table IV: The shear stresses (
y
t ) on the lower plate
2 -0.00467 -0.00599 -0.00653 -0.00321 -0.00301 -0.00546 -0.00675
4 -0.00521 -0.00684 -0.00744 -0.00427 -0.00357 -0.00584 -0.00748
6 -0.00633 -0.00744 -0.00831 -0.00524 -0.00427 -0.00752 -0.00846
10 -0.00801 -0.00856 -0.00946 -0.00622 -0.00582 -0.00942 -0.00999

M 2 5 8 2 2 2 2
D
1

2000 2000 2000 3000 4000 2000 2000
1
|
5 5 5 5 5 6 8

M 2 5 8 2 2 2 2
D
1

2000 2000 2000 3000 4000 2000 2000
1
|
5 5 5 5 5 6 8
1036 Vol. 7, Issue 3, pp. 1027-1037

ACKNOWLEDGEMENTS
The authors very much thankful to authorities of JNTU, Anantapur, Andhra Pradesh, India, providing
necessary facilities to have done this work and IJAET journal for the support to develop this
document.
REFERENCES
[1]. P. Puri, Rotary flowof an elastico-viscous fluid on an oscillating plate, Journal of Applied Mathematics
and Mechanics, Vol. 54, No. 11, pp. 743745, 1974.
[2]. M. Hussain, T. Hayat, S. Asghar, and C. Fetecau, Oscillatory flows of second grade fluid in a porous
space, Nonlinear Analysis: Real World Applications, Vol. 11, No. 4, pp. 24032414, 2010.
[3]. C. Fetecau, S. C. Prasad, and K. R. Rajagopal, A note on the flow induced by a constantly accelerating
plate in an Oldroyd-B fluid, Applied Mathematical Modelling, Vol. 31, No. 4, pp. 647 654, 2007.
[4]. W. Tan and T. Masuoka, Stokes first problem for a second grade fluid in a porous half-space with
heated boundary, International Journal of Non-Linear Mechanics, Vol. 40, No. 4, pp. 515522, 2005.
[5]. C. Fetecau, M. Athar, and C. Fetecau, Unsteady flow of a generalized Maxwell fluid with fractional
derivative due to a constantly accelerating plate, Computers andMathematics with Applications, Vol.
57, No. 4, pp. 596603, 2009.
[6]. M. Husain, T. Hayat, C. Fetecau, and S. Asghar, On accelerated flows of an OldroydB fluid in a
porous medium, Nonlinear Analysis: Real World Applications, Vol. 9, No. 4, pp. 13941408, 2008.
[7]. M. Khan, E. Naheed, C. Fetecau, and T. Hayat, Exact solutions of starting flows for second grade fluid
in a porous medium, International Journal of Non-LinearMechanics, Vol. 43, No. 9, pp. 868879, 2008.
[8]. F. Salah, Z. A. Aziz, and D. L. C. Ching, New exact solution for Rayleigh-Stokes problem ofMaxwell
fluid in a porous medium and rotating frame, Results in Physics, Vol. 1, No. 1,pp. 912, 2011.
[9]. M. Khan, M. Saleem, C. Fetecau, and T. Hayat, Transient oscillatory and constantly accelerated non-
Newtonian flow in a porous medium, International Journal of Non-Linear Mechanics, Vol. 42, No. 10,
pp. 12241239, 2007.
[10]. F. Salah, Z.Abdul Aziz, andD. L.C.Ching, New exact solutions for MHD transient rotating flow of a
second-grade fluid in a porous medium, Journal of Applied Mathematics, Vol. 2011, Article ID 823034,
8 pages, 2011.
[11]. C. Fetecau, T. Hayat, M. Khan, and C. Fetecau, Erratum: Unsteady flow of an Oldroyd-B fluid induced
by the impulsive motion of a plate between two side walls perpendicular to the plate, Acta Mechanica,
Vol. 216, No. 14, pp. 359361, 2011.
[12]. C. Fetecau, T. Hayat, J. Zierep, and M. Sajid, Energetic balance for the Rayleigh-Stokes problem of an
Oldroyd-B fluid, Nonlinear Analysis: Real World Applications, Vol. 12, No. 1, pp. 113, 2011.
[13]. K. R. Rajagopal and A. S.Gupta, On a class of exact solutions to the equations of motion of a second
grade fluid, International Journal of Engineering Science, Vol. 19, No. 7, pp. 10091014, 1981.
[14]. M. E. Erdogan and C. E. Imrak, On unsteady unidirectional flows of a second grade fluid,
International Journal of Non-Linear Mechanics, Vol. 40, No. 10, pp. 12381251, 2005.
[15]. F. Salah, Z. A. Aziz, and D. L. C. Ching, Accelerated flows of a magnetohydrodynamic (MHD) second
grade fluid over an oscillating plate in a porous medium and rotating frame, International Journal of
Physical Sciences, Vol. 6, No. 36, pp. 80278035, 2011.
[16]. C. Fetecau andC. Fetecau, Starting solutions for some unsteady unidirectional flows of a second grade
fluid, International Journal of Engineering Science, Vol. 43, No. 10, pp. 781789, 2005.
[17]. T. Hayat, K. Hutter, S. Asghar, and A.M. Siddiqui, MHD flows of an Oldroyd-B fluid, Mathematical
and Computer Modelling, Vol. 36, No. 9-10, pp. 987995, 2002.
[18]. T.Hayat, S.Nadeem, S.Asghar, andA.M. Siddiqui, Fluctuating flow of a third-grade fluid on a porous
plate in a rotating medium, International Journal of Non-Linear Mechanics, Vol. 36, No. 6, pp. 901
916, 2001.
[19]. S. Abelman, E. Momoniat, and T. Hayat, Steady MHD flow of a third grade fluid in a rotating frame
and porous space, Nonlinear Analysis: Real World Applications, Vol. 10, No. 6, pp. 33223328, 2009.
[20]. Faisal Salah, Zainal Abdul Aziz, Mahad Ayem, and Dennis Ling Chuan Ching, MHD Accelerated
Flow of Maxwell Fluid in a Porous Medium and Rotating Frame, ISRN Mathematical Physics, Vol.
2013, Article ID 485805, 10 pages, 2013, http://dx.doi.org/ 10.1155/2013/485805.
[21]. Hayat.T, C.Fetecau, M.Sajid., Physics Letters A, Vol. 372, pp. 1639-1644, 2008.
[22]. M.VeeraKrishna, S.V.Suneetha and R.SivaPrasad, Hall current effects on unsteady MHD flow of
rotating Maxwell fluid through a porous medium, Ultra Scientist of Physical Sciences, Vol. 21(1)M, pp.
133-144, 2010.
1037 Vol. 7, Issue 3, pp. 1027-1037

Appendix
,
i 1
z iz
b ,
z
b ), D (M 1 z , D M z
1
1 2
2
1
1
1
1
0
1 2
1 2
1 2
1
+
+ +
= = + + = + =

1
2 2
1 1
2
2 2
4
1
2 2
1 1
2
2 2
3
1
1 1
2
2 2
2
1
1 1
2
2 2
1 1 2 2
2
2 1 7
2 2
1 3 2
2
3 1 6
2 2
1 4 2
2
4 1 5
1 1
1 1 2
2
1 1
4
1 2
1 1 2
2
1 1
3
1
1 2
2
1
2
2
) n (z 4 z z
s
2
) n (z 4 z z
s
,
2
z 4 z z
s ,
2
z 4 z z
s z s z s b
) n (z s z s b ), n (z s z s b
,
i 1
z iz
b ,
s 1
z s z s
b ,
i 1
z iz
b
+
=
+ +
=

=
+
= + + =
+ + + = + + + =
+
+ +
=
+
+ +
=
+
=
,
,

AUTHORS BIOGRAPHY

L. Sreekala, presently working as a Asst Professor in the Department of Mathematics in
Chiranjeevi Reddy Institute of Technology, Anantapur, Andhra Pradesh India, I have seven
years of experience in teaching and three years in Research. I am doing my Ph.D in area of
fluid dynamics. I have published many papers in national and international well reputed
journals.

M. Veera Krishna received the B.Sc. degree in Mathematics, Physics and Chemistry from
the Sri Krishnadevaraya University, Anantapur, Andhra Pradesh, India in 1998, the M.Sc. in
Mathematics in 2001, the M.Phil and Ph.D. degree in Mathematics from same, in 2006 and
2008, respectively. Currently, He is an in-charge of Department of Mathematics at
Rayalaseema University, Kurnool, Andhra Pradesh, India. His teaching and research areas
include Fluid mechanics, Heat transfer, MHD flows and Data mining techniques. He
awarded 1 Ph.D from Monad University, Hapur (U.P), India and 28 M.Phils from DDE,
S.V. University, Tirupati, A.P., He has published 52 research papers in national and international well reputed
journals. He has presented 18 papers in National and International seminars and conferences. He attended four
national level workshops. He is a life member of Indian Society of Theoretical and Applied Mechanics
(ISTAM).

L. Hari Krishna acquired M.Sc. degree in mathematics from S.V. University, Tirupati in
the year 1998. He obtained his M.Phil from M.K. University, Madhurai in the year 2004. He
obtained his Ph.D in the area of Fluid Dynamics from JNTU, Anantapur in 2010. He has
got 14 years of experience in teaching to engineering students and seven years of research
experience. He has published 8 international publications. He has presented 8 papers in
National and International seminars. He has attend two national level workshops. He was a
Editorial and Article review board member for AES Journal in Engineering Technology and
Sciences (2010-2014). He has periyar university Guide ship approval and successfully guided 3 students to take
their M.Phils. He has memberships in professional bodies.

K. Keshava Reddy, presently working as professor of Mathematics in JNT University
college of Engineering Anantapur, He has 14 years of experience in teaching and 10 years
in research, He obtained his Ph.D degree in mathematics from prestigious University
Banaras Hindu University varanasi, His areas of interest include functional Anaysis.
Optimization Techniques, Data mining, Neural Networks and Fuzzy logic. He produced 2
Ph.D, 1 M.Phil and has published more than 35 Research papers in National and
International Journals and conference. He authored 06 books on engineering mathematics
and Mathematical Methods for various Mathematics for JNTUA both at UG level and PG level presently he is
the chairman, PG Board of studies for Mathematics of JNTUA. He is a member of Board of studies for
Mathematics of various universities in India.
1038 Vol. 7, Issue 3, pp. 1038-1043

VIDEO STREAMING ADAPTIVITY AND EFFICIENCY IN
SOCIAL NETWORKING SITES
G. Divya
1
and E R Aruna
2

1
Department of CSE, Vardhaman College of Engineering, India.
2
Associate Professor, Department of IT, Vardhaman College of Engineering, India.

ABSTRACT
While hassle on video traffic over cell phone networks have been unpleasant the wireless link capability cannot
keep up with the traffic exact. The space between the traffic demand and the link capability, all along with time-
varying link position, results in reduced service excellence of video streaming over cell phone like as lengthy
buffering time and blinking disturbance Leveraging the cloud compute technology, we suggest a new video
streaming structure of mobile, AMES-Cloud dubbed, which has couple of parts: ESoV (efficient social video
sharing) and AMoV (adaptive mobile video streaming). ESoV and AMoV create a private mediator to give video
streaming services capably for every mobile client. For a particular client, AMoV lets her secret mediator
adaptively alter her streaming pour with a coding of a scalable video technique depending upon the reaction of
link excellence. Similarly efficient social video sharing observes the social network acquaintances between
mobile clients and their private mediators try to pre hold video pleased in advance. We realize a trial product of
the AMES-Cloud framework to expose its presentation. It is shown that the confidential agents in the clouds can
efficiently provide the adaptive streaming, and achieve prefetching (i.e. video sharing,) depending upon the
social network research.
KEYWORDS: Adaptive video streaming, cloud computing, mobile networks, scalable video coding, social
video sharing.
I. INTRODUCTION
In excess of the past decade, more and more traffic is accounted by video streaming and downloading.
In exacting, video tributary services over mobile networks have turn out to be widespread over the
earlier period little years. Although the video streaming/ prefetching is not so demanding in wired
networks, mobile networks have been affliction from video traffic communication over inadequate
bandwidth of wireless links. Regardless of network operators anxious efforts to improve the wireless
connection bandwidth (e.g., 3G and LTE), soaring video traffic burden from mobile customers
quickly devastating the wireless link capability. While receiving video streaming traffic through
3G/4G mobile networks, mobile customer often put up with from long buffering time and irregular
disruptions due to the partial bandwidth and link circumstance fluctuation caused by multi-path loss
and user mobility [2][4]. Thus, it is vital to pick up the service quality of mobile video streaming
while via the networking and computing assets competently [5][8]. Recently there have been
numerous lessons on how to improve the service excellence of mobile video streaming on two
aspects: Scalability: Mobile video streaming services should sustain a broad spectrum of mobile
devices; they have unlike video resolutions, different computing authority various wireless links (like
LTE and 3G) and therefore. Also, the accessible link ability of a mobile device may differ over time
and break based upon signal potency, other clients traffic in the identical cell, and link circumstance
difference. Preserving various versions (with variant bit rates) of the similar video contented may earn
high transparency in terms of storage space and announcement. To tackle this concern, the Scalable
Video Coding (SVC) modus operandi of the H.264 AVC video solidity standard classify a BL (base
1039 Vol. 7, Issue 3, pp. 1038-1043

layer) with multiple boost layers (ELs). These sub streams can be programmed by developing 3
scalability qualities: (i) by utilizing layering image motion spatial scalability (screen pixels), (ii) with
the help of layering the casing rate temporal scalability, and (iii) With the help of layering the image
density quality scalability. By the SVC, a video can be played / decoded at the buck quality if only the
BL is delivering. Yet, the more ELs can be deliver, the enhanced excellence of the video stream is
attained.
Adaptability: conventional video streaming method intended by considering relatively stable traffic
links among servers and users achieve poorly in mobile environment [2]. Thus the unpredictable
wireless link status supposed to be accurately compact with to provide tolerable video streaming
services. To deal with this situation, we have to alter the bit rate settle of video at present time-varying
available link bandwidth of every mobile user. Such adaptive streaming technique can efficiently
diminish packet losses and bandwidth dissipates.
Scalable video coding and adaptive streaming techniques can be jointly combined to accomplish
effectively the best possible quality of video streaming services. That is, we can dynamically adjust
the number of SVC layers depending on the current link status [9], [12].

Figure 1: Cloud framework usage growth periodically
However most of the proposals seeking to jointly utilize the video scalability and adaptability rely on
the active control on the server side. That is, every mobile user needs to individually report the
transmission status (e.g., packet loss, delay and signal quality) periodically to the server, which
predicts the available bandwidth for each user. Thus the problem is that the server should take over
the substantial processing overhead, as the number of users increases.
II. PROBLEM STATEMENT
Existing system: Cloud computing assure lower expenses, fast scaling, easier safeguarding, and
service ease of use anywhere, anytime; a key confront is how to make certain and build pledge that
the cloud can knob user data firmly. A topical Microsoft review says that public percentage of fifty
eight and business percentage is eighty six best are excited about the cloud computing potential.
Although almost ninety percent of those people are worried about protection, ease of use, and
confidentiality of their data as it respite in the cloud computing.
Proposed system: We put forward an mobile video pour which is adaptive out and allocation
framework, identify AMES-Cloud, which capably provisions videos in the clouds (VC), and exploit
1040 Vol. 7, Issue 3, pp. 1038-1043

cloud figure to construct personal mediator (subVC) for every mobile client to attempt to present
video streaming adapt which is non-terminating to the changeability of link superiority based on the
Scalable Video Coding procedure. Also AMES-Cloud can auxiliary seek to provide nonbuffering
familiarity of video streaming by surroundings near enough occupation among the localVB, subVBs
and VB of mobile clients. We appraise the AMES-Cloud by prototype achievement and shows that
the cloud computing procedure transport noteworthy perfection on the mobile streaming adaptivity.
We unseen the encoding workload worth in the cloud computing while realize the sample.
III. SYSTEM DEVELOPMENT
1. Module of Admin
2. Module of User one
3. Module of User two
1. Module of Admin: In this unit, Admin contains sub modules count of three. They are,
Video Uploading: In this Admin is capable of adding a new video. Its used for user for
screening more sets
Details of User: Admin is capable of watching the clients details h which have record in
this website.
Videos Rating: This Videos Rating module for stay away from surprising videos from
clients. Later reject/accept videos then only user can/cannot view their individual videos.
2. Module of user one: In this Module of user one, it enclose the subsequent modules which are
sub ones and those are,
Feeding News: Here client of this social network can vision status from his friends like
communication or videos.
Search Friends: Here they can explore for a acquaintances and send a application to them
and also capable of watching their information.
Video Sharing: They can allocate videos with his associates by adding new videos also
they contribute to their standing by sending communication to friends.
Update Details: In this component, the user can modernize their own information.
3. Module of User two: In this Module of User two, client can catalog their information like
name, gender, password, age, and followed by. There the client can create associates by
accepting friend request or sending friend request.
They can share their standing by messages/chat also videos sharing with pals and acquire
comments/remarks from those.
IV. RELATED WORK
A. Adaptive Video Streaming Techniques
In the adaptive streaming, the video transfer rate is in tune on the fly so that a user can incident the
maximum possible video excellence based on his or her links time-varying bandwidth capability [2].
There are mostly two types of adaptive torrent techniques, depending on whether the adaptivity is
proscribed by the user or the head waiter. The Microsofts Smooth Streaming [27] is a live adaptive
streaming tune which can toggle among diverse bit rate section encoded with configurable bit rates
and video declaration at servers, while clients energetically request videos based on local monitoring
of link superiority. Adobe and Apple also industrial client-side HTTP adaptive live streaming
resolution operating in the similar manner.

1041 Vol. 7, Issue 3, pp. 1038-1043

Figure2: Client downloading a tagged video (direct recommendation video file)
.There are also some similar adaptive streaming services where servers controls the adaptive
transmission of video segments, for example, the Quavlive Adaptive Streaming. However, most of
these explanation maintain numerous copies of the video content with unusual bit rates, which brings
gigantic trouble of storage on the server. Regarding rate adaptation controlling techniques, TCP
friendly rate control methods for streaming services over mobile networks are proposed [28], [29],
where TCP throughput of a flow is predicted as a function of packet loss rate, round trip time, and
packet size. Considering the estimated throughput, the bit rate of the streaming traffic can be adjusted.
A rate adaptation algorithm for conversational 3G video streaming is introduced by [30]. Then, a few
cross-layer adaptation techniques are discussed [31], [32], which can acquire more accurate
information of link quality so that the rate adaptation can be more accurately made. However, the
servers have to always control and thus suffer from large workload. Recently the H.264 Scalable
Video Coding (SVC) technique has gained a momentum [10]. An adaptive video streaming system
based on SVC is deployed in [9], which studies the real-time SVC decoding and encoding at PC
servers. The work in [12] proposes a quality-oriented scalable video delivery. using SVC, but it is
only tested in a simulated LTE Network. Regarding the encoding performance of SVC, Cloud Stream
mainly proposes to deliver high-quality streaming videos through a cloud-based SVC proxy [20],
which discovered that the cloud computing can significantly improve the performance of SVC coding.
The above studies motivate us to use SVC for video streaming on top of cloud computing.
B. Mobile Cloud Computing Techniques
The cloud computing has been well located to make available video streaming services, especially in
the agitated Internet because of its scalability and ability [13]. For example, the quality-assured
bandwidth auto-scaling meant for VoD streaming foundation on the cloud computing is planned [14],
and the CALMS framework [33] is a cloud-assisted live media streaming service for internationally
dispersed users. yet, enlarge the cloud computing-based services to mobile environments necessitate
more factors to believe: wireless link dynamics, user mobility, the imperfect capability of mobile
devices [34], [35]. More recently, new designs for users on top of mobile cloud computing
environments are proposed, which virtualize private agents that are in charge of satisfy in the
requirements (e.g., QoS) of individual users such as Cloudlets [21] and Stratus [22]. Thus, we are
1042 Vol. 7, Issue 3, pp. 1038-1043

motivated to design the AMES-Cloud framework by using virtual a gents in the cloud to provide
adaptive video streaming services.
ames-cloud framework
In this section we explain the AMES-Cloud framework includes the Adaptive Mobile Video
streaming (AMoV) and the Efficient Social Video sharing (ESoV).
As shown in Fig. 1, the whole video stockpile and streaming system in the cloud is describe the Video
Cloud (VC). In the VC, there is a large-scale video bottom (VB), which stores the most of the trendy
video clips for the video service contributor (VSPs).A chronological video support (tempVB) is used
to accumulation new aspirant for the popular videos, while tempVB tot up the access regularity of
each video. The VC keeps running a satellite dish to seek videos which are previously popular in
VSPs, and will re-encode the peaceful videos into SVC format and store into tempVB first. By this 2-
tier cargo space, the AMES-Cloud can keep allocation most of popular videos perpetually. Note that
supervision employment will be handle by the regulator in the VC. specific for each transportable
user, a sub-video cloud (subVC) is fashioned vigorously if there is any video torrent require from the
user. The sub-VC has a sub video base (subVB), which provisions the in recent times obtain video
segments. Note that the video deliveries among the subVCs and the VC in most cases are actually not
copy, but just link operations on the same file ternally within the cloud data center [36]. There is
also encoding function in subVC (actually a smaller-scale encoder instance of the encoder in VC), and
if the mobile user demands a new video, which is not in the subVB or the VB in VC, the subVC will
fetch, encode and transfer the video. During video streaming, mobile users will always report link
conditions to their corresponding subVCs, and then the subVCs offer adaptive video streams. Note
that each mobile device also has a temporary caching storage, which is called local video base
(localVB), and is used for buffering and prefetching.
Note that as the cloud service may across different places, or even continents, so in the case of a video
delivery and prefetching between different data centers, an transmission will be carried out, which can
be then called copy. And because of the optimal deployment of data centers, as well as the capable
links among the data centers, the copy of a large video file takes tiny delay [36].
V. FUTURE WORK
As one important future work, we will carry out large-scale implementation and with serious
consideration on energy and price cost. In the future, we will also try to improve the SNS-based
prefetching, and security issues in the AMES-Cloud.
VI. CONCLUSION
In this paper, we discus about our app which indicates an adaptive mobile video streaming and
allocation framework, called AMES-Cloud, which proficiently supplies videos in the billows (VC),
and utilize cloud compute to erect classified agent (subVC) for each mobile customer to try to offer
non-terminating video streaming get used to the instability of link quality pedestal on the Scalable
Video Coding technique. Also AMES-Cloud can auxiliary seek to offer non-buffering practice of
video brook by background almost function among the VB, subVBs and localVB of mobile user. We
evaluate the AMES-Cloud by trial product execution and shows that the cloud compute technique
brings momentous improvement on the adaptivity of the mobile streaming. The focal point of this
document is to confirm how cloud computing can get better the program compliance and prefetching
for mobile client We ignored the cost of programming workload in the cloud while implement the
model. As one important possible work, we will carry out large-scale awareness and with somber
consideration on force and price cost. In the opportunity, we will also try to advance the SNS-based
pre-setting, and security issue in the AMES-Cloud.
REFERENCES
[1] Cisco Visual Networking Index: Global Mobile Data Traffic Forecast Update, 20112016, CISCO, 2012.
1043 Vol. 7, Issue 3, pp. 1038-1043

[2] Y. Li, Y. Zhang, and R. Yuan, Measurement and analysis of a large scale commercial mobile Internet TV
system, in Proc. ACM Internet Meas. conf., 2011, pp. 209224.
[3] T. Taleb and K. Hashimoto, MS2: A novel multi-source mobile- streaming architecture, IEEE Trans.
Broadcasting, vol. 57, no. 3, pp. 662673, Sep. 2011.
[4] X. Wang, S. Kim, T. Kwon, H. Kim, and Y. Choi, Unveiling the bittorrent performance in mobile WiMAX
networks, in Proc. Passive Active Meas. Conf., 2011, pp. 184193.
[5] A. Nafaa, T. Taleb, and L. Murphy, Forward error correction adaptation strategies for media streaming over
wireless networks, IEEE Commun. Mag., vol. 46, no. 1, pp. 7279, Jan. 2008.
[6] J. Fernandez, T. Taleb, M. Guizani, and N. Kato, Bandwidth aggregation- aware dynamic QoS negotiation
for real-time video applications in next-generation wireless networks, IEEE Trans. Multimedia, vol. 11, no. 6,
pp. 10821093, Oct. 2009.
[7] T. Taleb, K. Kashibuchi, A. Leonardi, S. Palazzo, K. Hashimoto, N. Kato, and Y. Nemoto, A cross-layer
approach for an efficient delivery of TCP/RTP-based multimedia applications in heterogeneous wireless
networks, IEEE Trans. Veh. Technol., vol. 57, no. 6, pp. 38013814, Jun. 2008.
[8] K. Zhang, J. Kong, M. Qiu, and G. L. Song, Multimedia layout adaptation through grammatical
specifications, ACM/SpringerMultimedia Syst., vol. 10, no. 3, pp. 245260.
[9] M. Wien, R. Cazoulat, A. Graffunder, A. Hutter, and P. Amon, Real-time system for adaptive video
streaming based on SVC, IEEE Trans. Circuits Syst. Video Technol., vol. 17, no. 9, pp. 12271237, Sep. 2007.
[10] H. Schwarz, D. Marpe, and T. Wiegand, Overview of the scalable video coding extension of the
H.264/AVC standard, IEEE Trans. Circuits Syst. Video Technol., vol. 17, no. 9, pp. 11031120, Sep. 2007.
[11] H. Schwarz and M. Wien, The scalable video coding extension of the H. 264/AVC standard, IEEE Signal
Process. Mag., vol. 25, no. 2, pp. 135141, Feb. 2008.
[12] P. McDonagh, C. Vallati, A. Pande, and P. Mohapatra, Quality-oriented scalable video delivery using H.
264 SVC on an LTE network, in Proc. WPMC, 2011.

AUTHORS BIOGRAPHY
G. Divya, pursuing her M.tech in computer science from Vardhaman college of
Engineering, Kacharam village, Shamshabad Mandal, Ranga Reddy District A.P, India.
Affiliated to Jawaharlal Nehru Technological University, Hyderabad. Approved by AICTE,
NEW DELHI.

E R Aruna Asst. Prof. of IT, in Vardhaman college of Engineering, Kacharam village,
Shamshabad Mandal, Ranga Reddy District, A.P, India. She graduated from Computer
Science Dept., at JNTUH, Hyderabad in 2004. She finished her master degree (M.tech) from
Sathyabama University, Chennai, TN in 2008. She Pursuing her Ph.D in Computer Science
and Engineering at JNTU, Hyderabad

1044 Vol. 7, Issue 3, pp. 1044-1052

INTRUSION DETECTION SYSTEM USING DYNAMIC AGENT
SELECTION AND CONFIGURATION

Manish Kumar
1
, M. Hanumanthappa
2

1
Assistant Professor, Dept. of Master of Computer Applications,
M. S. Ramaiah Institute of Technology, Bangalore
and Research Scholar, Department of Computers Science and Applications,
Bangalore University, Bangalore, India
2
Dept. of Computer Science and Applications, Jnana Bharathi Campus, Bangalore University,
Bangalore -560 056, India

ABSTRACT
Intrusion detection is the process of monitoring the events occurring in a computer system or network and
analysing them for signs of possible incidents, which are violations or imminent threats of violation of computer
security policies, acceptable use policies, or standard security practices. An intrusion detection system (IDS)
monitors network traffic and monitors for suspicious activity and alerts the system or network administrator. It
identifies unauthorized use, misuse, and abuse of computer systems by both system insiders and external
penetrators. Intrusion detection systems (IDS) are essential components in a secure network environment,
allowing for early detection of malicious activities and attacks. By employing information provided by IDS, it is
possible to apply appropriate countermeasures and mitigate attacks that would otherwise seriously undermine
network security. However, Increasing traffic and the necessity of stateful analysis impose strong computational
requirements on network intrusion detection systems (NIDS), and motivate the need of architectures with
multiple dynamic sensors. In a context of high traffic with heavy tailed characteristics, static rules for
dispatching traffic slices among sensors cause severe imbalance. The current high volumes of network traffic
overwhelm most IDS techniques requiring new approaches that are able to handle huge volume of log and
packet analysis while still maintaining high throughput. This paper shows that the use of dynamic agents has
practical advantages for intrusion detection. Our approach features unsupervised adjustment of its
configuration and dynamic adaptation to the changing environment, which improvises the performance of IDS
significantly.
KEYWORDSIntrusion Detection System, Agent Based IDS, Dynamic Sensor Selection.
I. INTRODUCTION
Intrusion Detection is the process of monitoring and analysing the information sources, in order to
detect malicious information. It has been an active field of research for over two decades. John
Andersons Computer Security Threat Monitoring and Surveillance was published in 1980 and has
embarked upon this field. It was one of the earliest and most famous papers in the field. After that in
1987, Dorothy Denning published An Intrusion Detection Model, provided a methodological
framework that inspired many researchers around the world and has laid the groundwork for the early
commercial products like Real Secure, Trip Wire, Snort, Shadow, and STAT etc.
Intrusion Detection technology has evolved and emerged as one of the most important security
solutions. It has several advantages and it is unique compared to other security tools. As information
systems have become more comprehensive and a higher value asset of organizations, intrusion
detection systems have been incorporated as elements of operating systems and network.
Intrusion detection systems (IDS) have a few basic objectives. Among these objectives are
Confidentiality, Integrity, Availability, and Accountability.
1045 Vol. 7, Issue 3, pp. 1044-1052

Intrusion Detection Systems (IDS) are important mechanisms which play a key role in network
security and self-defending networks. Such systems perform automatic detection of intrusion attempts
and malicious activities in a network through the analysis of traffic captures and collected data in
general. Such data is aggregated, analysed and compared to a set of rules in order to identify attack
signatures, which are traffic patterns present in captured traffic or security logs that are generated by
specific types of attacks. In the process of identifying attacks and malicious activities an IDS parses
large quantities of data searching for patterns which match the rules stored in its signature database.
Such procedure demands high processing power and data storage access velocities in order to be
executed efficiently in large networks. The next part of the paper discuss about classification of
Intrusion Detection Systems. The section II of the paper discuss about dynamic sensor agents for
improvising the performance of IDS. Section III discuss about the algorithm for using the dynamic
agent for improvising the performance of IDS. Section IV analyse and show the improvement in
performance of IDS implementation using agent followed by conclusion and future work in section V.
1.1 Classification of Intrusion Detection Systems
Intrusions can be divided into 6 main types:-
1. Attempted break-ins, which are detected by a typical behaviour profiles or violations of
security constraints.
2. Masquerade attacks, which are detected by atypical behaviour profiles or violations of
security constraints.
3. Penetration of the security control system, which are detected by monitoring for specific
patterns of activity.
4. Leakage, which is detected by atypical use of system resources.
5. Denial of service, which is detected by atypical use of system resources.
6. Malicious use, which is detected by a typical behaviour profiles, violations of security
constraints, or use of special privileges.
However, we can divide the techniques of intrusion detection into two main types. IDSs issue security
alerts when an intrusion or suspect activity is detected through the analysis of different aspects of
collected data (e.g. packet capture files and system logs). Classical intrusion detection systems are
based on a set of attack signatures and filtering rules which model the network activity generated by
known attacks and intrusion attempts [8]. Intrusion detection systems detect malicious activities
through basically two approaches: anomaly detection and signature detection [9][21][20].
i. Anomaly Detection
This technique is based on the detection of traffic anomalies. The deviation of the monitored traffic
from the normal profile is measured. Various different implementations of this technique have been
proposed, based on the metrics used for measuring traffic profile deviation.
Anomaly detection techniques assume that all intrusive activities are necessarily anomalous. This
means that if we could establish a "normal activity profile" for a system, we could, in theory, flag all
system states varying from the established profile by statistically significant amounts as intrusion
attempts. However, if we consider that the set of intrusive activities only intersects the set of
anomalous activities instead of being exactly the same, we find a couple of interesting possibilities:
(1) Anomalous activities that are not intrusive are flagged as intrusive. (2) Intrusive activities that are
not anomalous result in false negatives (events are not flagged intrusive, though they actually are).
This is a dangerous problem, and is far more serious than the problem of false positives.
The main issues in anomaly detection systems thus become the selection of threshold levels so that
neither of the above 2 problems is unreasonably magnified, and the selection of features to monitor.
Anomaly detection systems are also computationally expensive because of the overhead of keeping
track of, and possibly updating several system profile metrics. Some systems based on this technique
are discussed in Section 4 while a block diagram of a typical anomaly detection system is shown in
Fig 1.
1046 Vol. 7, Issue 3, pp. 1044-1052

Fig 1:- IDS Anomaly Detection System
ii. Misuse Detection
This technique looks for patterns and signatures of already known attacks in the network traffic. A
constantly updated database is usually used to store the signatures of known attacks. The way this
technique deals with intrusion detection resembles the way that anti-virus software operates.
The concept behind misuse detection schemes is that there are ways to represent attacks in the form of
a pattern or a signature so that even variations of the same attack can be detected. This means that
these systems are not unlike virus detection systems -- they can detect many or all known attack
patterns, but they are of little use for as yet unknown attack methods. An interesting point to note is
that anomaly detection systems try to detect the complement of "bad" behaviour. Misuse detection
systems try to recognize known "bad" behaviour. The main issues in misuse detection systems are
how to write a signature that encompasses all possible variations of the pertinent attack, and how to
write signatures that do not also match non-intrusive activity. Several methods of misuse detection,
including a new pattern matching model are discussed later. A block diagram of a typical misuse
detection system is shown in Fig 2 below.

Fig 2:-IDS Misuse Detection System
Intrusion detection systems are further can also be classified in two groups, Network Intrusion
Detection Systems (NIDS), which are based on data collected directly from the network, and Host
Intrusion Detection Systems (HIDS), which are based on data collected from individual hosts. HIDSs
are composed basically by software agents which analyse application and operating system logs, file
system activities, local databases and other local data sources, reliably identifying local intrusion
attempts. Such systems are not affected by switched network environments (which segment traffic
flows) and is effective in environments where network packets are encrypted (thwarting usual traffic
analysis techniques). However, they demand high processing power overloading the nodes resources
and may be affected by denial-of-service attacks. In face of the growing volume of network traffic and
high transmission rates, software based NIDSs present performance issues, not being able to analyses
all the captured packets rapidly enough. Some hardware based NIDSs offer the necessary analysis
throughput but the cost of such systems is too high in relation to software based alternatives.
From the above, it is clear that as IDS grow in function and evolve in power, they also evolve in
complexity. Agents of each new generation of IDS use agents of the previous generation as data
sources, applying ever more sophisticated detection algorithms to determine ever more targeted
responses. Often, one or more IDS and management system(s) may be deployed by an organization
within its own network, with little regard to their neighbours or the global Internet. Just as all
individual networks and intranets connect to form "The Internet", so can information from stand-alone
1047 Vol. 7, Issue 3, pp. 1044-1052

internal and perimeter host- and network-based intrusion detection systems be combined to create a
distributed Intrusion Detection System (dIDS).
Current IDS technology is increasingly unable to protect the global information infrastructure due to
several problems:
i. The existence of single intruder attacks that cannot be detected based on the observations
of only a single site.
ii. Coordinated attacks involving multiple attackers that require global scope for assessment.
iii. Normal variations in system behaviour and changes in attack behaviour that cause false
detection and identification.
iv. detection of attack intention and trending is needed for prevention
v. Advances in automated and autonomous attacks, i.e. rapidly spreading worms, require
rapid assessment and mitigation, and
vi. The sheer volume of attack notifications received by ISPs and host owners can become
overwhelming.
vii. If aggregated attack details are provided to the responsible party, the likelihood of a
positive response increases.
II. DYNAMIC SENSOR SELECTION
In our proposed architecture, IDS LOG can be collected from multiple sensors or agent. In this
section, we present a trust-based algorithm which dynamically determines the best aggregation agent
and also the optimal number of malicious or legitimate behaviour necessary for the reliable
identification of the best aggregation agent, while taking into account the: (i) past effectiveness of the
individual aggregation agents and (ii) number of aggregation agents and the perceived differences in
their effectiveness. We decided to use a trust-based approach for evaluating the aggregation agents,
because it not only eliminates the noise in the background traffic and randomness of the challenge
selection process, but accounts for the fact that attackers might try to manipulate the system by
inserting misleading traffic flows. An attacker could insert fabricated flows [15] hoping they would
cause the system to select an aggregation agent that is less sensitive to the threat the attacker actually
intends to realize. When using trust, one tries to avoid this manipulation by dynamically adapting to
more recent actions of an attacker [4][16].
The problem features a set of classifier agents } ,..., {
g
A that process a single, shared open-ended
sequence ... ,...,.
1 i
of incoming events and use their internal models to divide these events into
two categories: normal and anomalous. The events are inherently of two fundamental types:
legitimate and malicious, and the goal of the classifier agents is to ensure that the normal class as
provided by the agent is the best possible match to the legitimate traffic class, while the anomalous
class should match the malicious class. The classification thus has four possible outcomes [17] for
each event , two of them being correct classifications and two of them the errors (see also the
confusion matrix in Table 1).
Table 1: Confusion Matrix
actual class
legitimate malicious
classification normal true positive false positive
anomalous false negative true negative
The classifier agents actually provide more information, as they internally annotate the individual
events with a continuous normality value in the [0, 1] interval, with the value1 corresponding to
perfectly normal events and the value 0 to completely anomalous ones. This continuous anomaly
value describes an agents opinion regarding the anomaly of the event, and the agents apply adaptive
or predefined thresholds to split the [0, 1] interval into the normal and anomalous classes.
Given that the characteristics of the individual classifier agents k are unknown in the dynamically-
changing environment, the system needs to be able to identify the optimal classifier autonomously.
Furthermore, the system can have several users with different priorities regarding the detection of
specific types of malicious events. In the network monitoring use-case, some of the users concentrate
1048 Vol. 7, Issue 3, pp. 1044-1052

on major, infrastructure-type events only (such as denial of service attacks),while the other users seek
information about more subtle attack techniques targeting individual hosts. The users are represented
by the user agents and these agents are assumed to know their users preferences. Their primary goal is
to use their knowledge of user preferences to dynamically identify the optimal information source and
to change the source when the characteristics of the environment or user preferences change. To reach
this goal in an environment where they have no abstract model of classifier agents performance, they
rely on empirical analysis of classifier agents response to a pre-classified set of challenges [11][19].
In the following, we will analyse the problem from the perspective of a single user agent, which tries
to select the best available classification agent, while keeping the number of challenges as low as
possible. The challenges are events with known classification, which can be inserted into the flow of
background events as observed by the system, processed by the classifier agents together with the
background events and finally removed before the system reports the results to the users. The
processing of the challenges allows the user agents to identify the agent which achieves the best
separation of the challenges that represent known instances of legitimate behaviour from the
challenges that represent known malicious behaviour[13][7][18].
III. ALGORITHM
In this section we present a simple but adaptive algorithm for choosing the best classifier agent. For
each time step
i
, the algorithm proceeds as follows:
For each time step i N, the algorithm proceeds as follows:
i. Let each aggregation agent classify a set of known instances of malicious or legitimate
behaviour from different attack classes and selected legitimate known instances of
malicious or legitimate behaviour.
ii. Update the trust value of each aggregation agent, based on its performance on the known
instances of malicious or legitimate behaviour in time step i.
iii. Accept the output of the aggregation agent with the highest trust value as classification of
the remaining events of time step i.
Known instances of malicious or legitimate behaviour detection and aggregation agents in each time
step i with the sets of flows for which we already know the actual class, i.e. whether they are
malicious or legitimate. So, we challenge an aggregation agent with a set of malicious events,
belonging to K attack classes and a set of legitimate events drawn from a single class. With respect to
each class of attacks k, the performance of the agent is described by a mean and a standard deviation:
) , (
k
x
k
x for the set of malicious challenges and ) , (
x
y for the set of legitimate challenges. Both
means lie in the interval [0, 1], and
k
x close to 0 and y close to 1 signify accurate classifications of
the agent respectively.
The system used to perform the experiments described in this paper incorporates five different
anomaly detection [5] techniques presented in literature [10][1] [12][2]. Each of the methods works
with a different traffic model based on a specific combination of aggregate traffic features, such as:
Entropies of flow characteristics for individual source IP addresses.
Deviation of flow entropies from the PCA-based prediction model of individual sources.
Deviation of traffic volumes from the PCA-based prediction for individual major sources.
Rapid surges in the number of flows with given characteristics from the individual sources
and
Ratios between the number of destination addresses and port numbers for individual
sources.
These algorithms maintain a model of expected traffic on the network and compare it with real traffic
to identify the discrepancies that are identified as possible attacks. They are effective against zero-day
attacks and previously unknown threats, but suffer from a comparatively higher error rate
[17][10][11], frequently classifying legitimate traffic as anomalous(false positives), or failing to spot
malicious flows (false negatives). The classifier agents can be divided to two distinct classes:
Detection agents analyse raw network flows by their anomaly detection algorithms,
exchange the anomalies between them and use the aggregated anomalies to build and
1049 Vol. 7, Issue 3, pp. 1044-1052

update the long-term anomaly associated with the abstract traffic classes built by each
agent. Each detection agent uses one of the five anomaly detection techniques mentioned
above. All agents map the same events (flows), together with the same evaluation of these
events, the aggregated immediate anomaly of these events determined by their anomaly
detection algorithms, into the traffic clusters built using different features/metrics, thus
building the aggregate anomaly hypothesis based on different premises. The aggregated
anomalies associated with the individual traffic classes are built and maintained using the
classic trust modelling techniques (not to be confused with the way trust is used in this
work).
Aggregation agents represent the various aggregation operators used to build the joint
conclusion regarding the normality/anomaly of the flows from the individual opinions
provided by the detection agents. Each agent uses a distinct averaging operator (based on
order-weighted averaging or simple weighted averaging) to perform the Rgdet R
transformation from the gdet-dimensional space to a single real value, thus defining one
composite system output that integrates the results of several detection agents. The
aggregation agents also dynamically determine the threshold values used to transform the
continuous aggregated anomaly value in the [0, 1] interval into the crisp normal/
anomalous assessment for each flow.
The user agent functionality is implemented as a collection of the agents. The user agent creates
individual challenge agents, each of them representing a specific incident in the past, and these
temporary, single purpose agents interact with the data-provisioning layers of the system in order to
insert the flows relative to the incident into the background traffic and to retrieve and analyse the
detection results provided by the classifier agents.
IV. RESULTS AND PERFORMANCE ANALYSIS OF AGENT BASED IDS
We have simulated and tested the IDS using the KDD Cup 1999 dataset. The implementation gives us
the expected results. Our Agent based IDS prototype we are testing detects the simulated attacks. The
question is : why the realization of the system with agents is advantageous? We implement a
centralized system with local sensor that forward filtered data to a central analysis node and compare
it with Agent IDS.
Agent based IDS has proven itself to be capable of handling very high traffic. In such a design, the
incoming network traffic is disseminated to a pool of agents, which process a fraction of the whole
traffic, reducing the possibility of packet loss caused by overload. Agent IDS could support a load of
up to 56 Mbps (450 packets/second) with zero traffic loss. Moreover, we focus on a second important
criterion for IDS: detection delay which is defined as the duration from the time the attack starts to the
time epoch that the attack is detected. We generate a set of packets varied from 1000 to 8000. For
each set we simulate the attack and we calculate the detection delay. Figure3 plots the measurement
results. The detection delay is significantly reduced; Agent IDS is much faster than the centralized
IDS. For example, in the case of 8000 packets, we observe that detection delay is reduced by 56%
(7.91second vs 4.4 second). This can be explained by the fact that agents operate directly on the host,
where an action has to be taken, their response is faster than systems where the actions are taken by
central coordinator.
In fact, one of the most pressing problems facing current IDSs is the processing of the enormous
amounts of data generated by the network traffic monitoring tools and host-based audit logs. IDSs
typically process most of this data locally. Agents offer an opportunity to reduce the network load by
eliminating the need for this data transfer. Instead of transferring the data across the network, agents
can be dispatched to the machine on which the data resides, essentially moving the computation to the
data, instead of moving the data to the computation. It is obvious to see that the code-shipping versus
data-shipping argument is only valid if, the agents code and state that have to be transmitted are not
larger than the amount of data that can be saved by the use of an agent. Agent IDS does not only
perform better in terms of effectiveness but also in terms of detection delay.

1050 Vol. 7, Issue 3, pp. 1044-1052

Fig 3:- Performance of Centralize IDS vs Agent Based IDS
V. CONCLUSION AND FUTURE WORK
We take advantage of the multi-agent paradigm especially concerning reducing Network Load.
Indeed, agents offer the possibility to eliminate the need of transferring a huge amount of data to be
analysed. In this paper we have explained the architectural design and performance analysis of a
Centralize IDS vs. Agent based IDS. The experimental result was positive and we found that this
work can be continued with several other improvement and performance analysis As network attacks
are becoming more and more alarming, exploiting systems faults and performing malicious actions
the need to provide effective intrusion detection methods increases. Network-based, distributed
attacks are especially difficult to detect and require coordination among different intrusion detection
components or systems. The experiments emphasize the aim of applying agent to detect some kind of
intrusions and compete others IDS.
ACKNOWLEDGMENT
I would like to thank MSRIT Management, my colleagues and Dept. of Computer Science and
Applications, Bangalore University, for their valuable suggestion, constant support and
encouragement.
REFERENCES
[1] A. Lakhina, M. Crovella, and C. Diot. Mining Anomalies using Traffic Feature Distributions. In ACM
SIGCOMM, Philadelphia, PA, August 2005,pages 217228, New York, NY, USA, 2005. ACM Press.
[2] A. Sridharan, T. Ye, and S. Bhattacharyya. Connectionless port scan detection on the backbone.
Phoenix, AZ, USA, 2006.
[3] Axelsson, Stefan, Intrusion Detection Systems: A Taxonomy and Survey, Technical Report No 99-
15, Dept. of Computer Engineering, Chalmers University of Technology, Sweden, March 2000.
[4] Chang-Lung Tsai; Chang, A.Y.; Chun-Jung Chen; Wen-Jieh Yu; Ling-Hong Chen, "Dynamic intrusion
detection system based on feature extraction and multidimensional hidden Markov model
analysis," Security Technology, 2009. 43rd Annual 2009 International Carnahan Conference on , vol.,
no., pp.85,88, 5-8 Oct. 2009.
[5] D. E. Denning. An intrusion-detection model. IEEE Trans. Softw. Eng., 13(2):222232, 1987.
[6] F. A. Barika & N. El Kadhi & K. Ghedira, Agent IDS based on Misuse Approach, Journal of
Software, Vol. 4, No. 6, 495-507, August 2009.
[7] Guangcheng Huo; Xiaodong Wang, "DIDS: A dynamic model of intrusion detection system in wireless
sensor networks," Information and Automation, 2008. ICIA 2008. International Conference on , vol.,
no., pp.374,378, 20-23 June 2008.
1051 Vol. 7, Issue 3, pp. 1044-1052

[8] H.Debar, M. Dacier, and A. Wespi, Towards a taxonomy of Intrusion Detection Systems, The
International Journal of Computer and Telecommunications Networking - Special issue on computer
network security, Volume 31 Issue 9, Pages 805 822, April 23, 1999,
[9] Jun Wu; Chong-Jun Wang; Jun Wang; Shi-Fu Chen, "Dynamic Hierarchical Distributed Intrusion
Detection System Based on Multi-Agent System," Web Intelligence and Intelligent Agent Technology
Workshops, 2006. WI-IAT 2006 Workshops. 2006 IEEE/WIC/ACM International Conference on ,
vol., no., pp.89,93, Dec. 2006
[10] K. Xu, Z.-L. Zhang, and S. Bhattacharrya. Reducing Unwanted Traffic in a Backbone Network. In
USENIX Workshop on Steps to Reduce Unwanted Traffic in the Internet (SRUTI), Boston, MA, July
2005.
[11] Kumar, G.V.P.; Reddy, D.K., "An Agent Based Intrusion Detection System for Wireless Network with
Artificial Immune System (AIS) and Negative Clone Selection," Electronic Systems, Signal Processing
and Computing Technologies (ICESC), 2014 International Conference on , vol., no., pp.429,433, 9-11
Jan. 2014.
[12] L. Ertoz, E. Eilertson, A. Lazarevic, P.-N. Tan,V. Kumar, J. Srivastava, and P. Dokas. Minds -
minnesota intrusion detection system. In NextGeneration Data Mining. MIT Press, 2004.
[13] Lin Zhao-wen; Ren Xing-tian; Ma Yan, "Agent-based Distributed Cooperative Intrusion Detection
System," Communications and Networking in China, 2007. CHINACOM '07. Second International
Conference on , vol., no., pp.17,22, 22-24 Aug. 2007.
[14] Martin Rehak, Eugen Staab, Michal Pechoucek, Jan Stiborek, Martin Grill, and Karel Bartos. 2009.
Dynamic information source selection for intrusion detection systems. In Proceedings of The 8th
International Conference on Autonomous Agents and Multiagent Systems - Volume 2(AAMAS '09),
Vol. 2. International Foundation for Autonomous Agents and Multiagent Systems, Richland, SC, 1009-
1016.
[15] Martin Rehak, Eugen Staab, Volker Fusenig, Michal Pechoucek, Martin Grill, Jan Stiborek, Karel
Bartos, and Thomas Engel, Runtime Monitoring and Dynamic Reconfiguration for Intrusion
Detection Systems, Proceedings of 12th International Symposium, RAID 2009, Saint-Malo, France,
September 23-25, pp 61-80, 2009.
[16] Paez, R.; Torres, M., "Laocoonte: An agent based Intrusion Detection System," Collaborative
Technologies and Systems, 2009. CTS '09. International Symposium on , vol., no., pp.217,224, 18-22
May 2009.
[17] S. Northcutt and J. Novak. Network Intrusion Detection: An Analysts Handbook. New Riders
Publishing, Thousand Oaks, CA, USA, 2002.
[18] Sun-il Kim; Nwanze, N.; Kintner, J., "Towards dynamic self-tuning for intrusion detection
systems," Performance Computing and Communications Conference (IPCCC), 2010 IEEE 29th
International, vol., no., pp.17,24, 9-11 Dec. 2010.
[19] Weijian Huang; Yan An; Wei Du, "A Multi-Agent-Based Distributed Intrusion Detection
System," Advanced Computer Theory and Engineering (ICACTE), 2010 3rd International Conference
on , vol.3, no., pp.V3-141,V3-143, 20-22 Aug. 2010.
[20] Yinan Li; Zhihong Qian, "Mobile Agents-Based Intrusion Detection System for Mobile Ad Hoc
Networks," Innovative Computing & Communication, 2010 Intl Conf on and Information Technology
& Ocean Engineering, 2010 Asia-Pacific Conf on (CICC-ITOE) , vol., no., pp.145,148, 30-31 Jan.
2010.
[21] Yu Cai, Hetal Jasani, Autonomous Agents based Dynamic Distributed (A2D2) Intrusion Detection
System, Innovative Algorithms and Techniques in Automation, Industrial Electronics and
Telecommunications 2007, pp 527-533
AUTHORS

Manish Kumar is working as Asst. Professor in Department of Computer Applications, M.
S. Ramaiah Institute of Technology, Bangalore, India. His areas of interest are
Cryptography and Network Security, Computer Forensic, Mobile Computing and
eGovernance. His specialization is in Network and Information Security. He has also
worked on the R&D projects relates on theoretical and practical issues about a conceptual
framework for E-Mail, Web site and Cell Phone tracking, which could assist in curbing
misuse of Information Technology and Cyber Crime. He has published several papers in
International and National Conferences and Journals. He has delivered expert lecture in various academic
Institutions.

1052 Vol. 7, Issue 3, pp. 1044-1052

M Hanumanthappa is currently working as Associate Professor in the Department of
Computer Science and Applications, Bangalore University, Bangalore, India. He has over
17 years of teaching (Post Graduate) as well as Industry experience. He is member of Board
of Studies /Board of Examiners for various Universities in Karnataka, India. He is actively
involved in the funded research projects and guiding research scholars in the field of Data
Mining and Network Security.

1053 Vol. 7, Issue 3, pp. 1053-1059

EVALUATION OF CHARACTERISTIC PROPERTIES OF RED
MUD FOR POSSIBLE USE AS A GEOTECHNICAL MATERIAL
IN CIVIL CONSTRUCTION
Kusum Deelwal
1
, Kishan Dharavath
2
, Mukul Kulshreshtha
3

1
Ph.D. Scholar,
2
Assistant Professor,
3
Professor,
MANIT, Bhopal, M.P., India

ABSTRACT
Red mud is a byproduct produced in the process of extraction of alumina from bauxite. The process is called
Bayers Process. It insoluble product and is generated after bauxite digestion with sodium hydroxide at elevated
temperature and pressure. This paper describes the characteristic properties of Red Mud and possible use as a
geotechnical material. Basics properties like Specific gravity, Particle size distribution, Atter Bergs limit, OMC
and MDD are determined. Engineering properties like shear strength, permeability and CBR values are also
determined in conformity with the Indian Standard Code and test results are discussed in geotechnical point of
view. It revealed that the behavior of red mud is likely as clay soil with considerably high strength compared to
conventional clay soil.
KEY WORDS: Red mud, Bayers process, Bauxite residue.
I. INTRODUCTION
Industrialization and urbanization are the two world wide phenomena. Though these are the necessity
of the society and are mostly inevitable, one has to look into their negative impacts on the global
environment and social life. The major ill effect of these global processes is the production of large
quantities of industrial wastes and the problems related with their safe management and disposal.
Second problem is the scarcity of land, materials and resources for ongoing developmental activities,
including infrastructure.
Red Mud is produced during the process for alumina production. Depending on the raw material
processed, 12.5 tons of red mud is generated per ton of alumina produced [1]. In India, about 4.71
million tons/annum of red mud is produced which is 6.25% of worlds total digestion with sodium
hydroxide at elevated temperature and pressure [2]. It is a mixture of compounds originally present in
the parent mineral bauxite and of compounds formed or introduced during the Bayer cycle. It is
disposed as slurry having a solid concentration in the range of 10-30%, pH in the range of 10-13 and
high ionic strength.
Considerable research and development work for the storage, disposal and utilization of red mud is
being carried out all over the world [3]. This article provides an overview of the basic characteristics
of red mud. The main ways of comprehensive utilization are also summarized. It describes the
progress of experimental research and comprehensive utilization. The aim is to provide some valuable
information to further address the comprehensive utilization of red mud.
II. ORGANIZATION OF MANUSCRIPT
The present work is divided into two main two stages. i) Experimental work and ii) Utilization of red
mud as a geotechnical construction material. In experimental works, it is further divided into two
parts; first, tests determining index properties, second, tests determining the engineering properties. In
second stage of the work, it is described the possible utilization of red mud as a geotechnical material.
1054 Vol. 7, Issue 3, pp. 1053-1059

III. MATERIAL: RED MUD
The red mud is one of the major solid wastes coming from Bayer process of alumina production. For
the present work it was collected from HINDALCO, At Renukoot, Uttar Pradesh. The conventional
method of disposal of red mud in ponds has often adverse environmental impacts as during monsoons,
the waste may be carried by run-off to the surface water courses and as a result of leaching may cause
contamination of ground water: Further disposal of large quantities of Red mud dumped, poses
increasing problems of storage occupying a lot of space.
IV. CHARACTERISTIC PROPERTIES OF RED MUD
Index Properties:
Specific gravity of the red mud has been carried out as per the IS: 2720 (Part II) 1980. The
experiment was performed from both pycnometer method and density bottle. The specific gravity of
the red mud was found to be 3.04.
Particle Size distribution of the red mud was carried out as per the IS: 1498 1970 As the materials
consists near about 90 percent silt and clay, wet sieve analysis is carried out. The particles passing
through 75 micron was collected and allowed to Hydrometer analysis to determine the particle size
variation. About 87.32% percentage of the total mass was passed through the 75 micron sieve. Fig:1
shows the graph plot between the particle diameter and the percentage finer.

Fig-1: Graph showing particle size distribution with % finer
Standard Proctor test was carried out to determine the maximum dry density and optimum moisture
content of the red mud. The test is carried out as per the IS: 2720 (Part VII) Light compaction was
adopted. The variation of water content and the corresponding water content variation is shown in
thefig:2
.
Fig-2: Graph showing water content and dry density.
1055 Vol. 7, Issue 3, pp. 1053-1059

Atter Bergs limits of the red mud was determined as per the IS: 2720 (Part V). Corresponding liquid
limit, Plastic limit and plasticity index is shown in the table: 1
Table: 1. Strength and physical parameters of red mud
S.NO Tests Values
1 Maximum dry Density (g/cc) 1.53
2 Optimum moisture Content (%) 33.5
3 Specific Gravity 3.04
4 Liquid Limit (%) 45.5
5 Plastic Limit (%) 32.3
6 Classification ML
7 Cohesion (kg/cm
2
) 0.123
8 Angle of internal friction(in Degrees) 26.8
9 CBR (%)
Soaked 4.2
Unsoaked 7.8
Chemical and Mineral Compositions of Red Mud
Red mud is mainly composed of fine particles of mud. Its composition, property and phase vary with
the origin of the bauxite and the alumina production process, and will change over time when stocked.
Chemical analysis shows that red mud contains silicium, aluminium, iron, calcium, titanium, sodium
as well as an array of minor elements namely K, Cr, V, Ba, Cu, Mn, Pb, Zn, P, F, S, As, and etc.
Tables 2 and 3 list the chemical and mineral compositions of red mud that are produced by the Bayer
process [4].
Table 2. Typical composition of red mud
Composition Percentage
Fe2O3 30-60%
Al2O3 10-20%
SiO2 3-50%
Na2O 2-10%
CaO 2-8%
TiO2 Trace-25%
Mineralogical Phases:
Mineralogical phases of red mud are listed below [5]
Hematite Fe2O3
Goethite FeO(OH)
Gibbsite AlOH3
Diaspore AlO(OH)
Quartz SiO2
Cancrinite (NaAlSiO4)6CaCO3
Kaolinite Al2O3 2SiO2 2H20
Calcilte [CaCO3]
Engineering Properties.
Permeability test is carried out as per the IS: 2720 (Part XVII). The coefficient of permeability of the
red mud specimen is found out using falling head method. Coefficient of permeability was found to be
5.786e-7cm/s.
Triaxial Compression Test is best suited for Clayey soil. The sample of size 38mm dia. x 76mm
height. After applying confining pressure (e.g. 0.5, 1.0 or 1.5 kg/cm2) deviator stress is applied till
failure. Having minimum two readings Mohrs stress circles are plotted. A line tangent to the Mohrs
circles is failure envelope and shear parameters: Cohesion and Angle of Internal Friction. Results are
shown in the table 1
Unconfined Compressive strength: The samples of sizes 38 mm diameter and height of 76 mm were
prepared by static compaction method to achieve maximum dry density at their optimum moisture
contents. Unconfined compressive strength tests were conducted at a strain rate of 1.25 mm/min. The
results obtained are tabulated in table 1
1056 Vol. 7, Issue 3, pp. 1053-1059

California Bearing ratio test: The sample of nearly 4.5 to 5 kg was compacted in a mould of volume
2250cc with 5 layers and 56 blows were given for each layer. For soaked CBR value, the different
sample of identical size is prepared and kept soaking for 4 days with the surcharge. This test was
conducted as per IS: 2720 (Part XXXI). The test results are entered in the table 1
V. COMPREHENSIVE UTILIZATION OF RED MUD IN CONSTRUCTION
A. Red mud in cement replacement
Dicalcium silicate in red mud is also one of the main phases in cement clinker, and red mud can play
the role of crystallization in the production of cement clinker. Fly ash is mainly composed of SiO2 and
Al2O3, thus can be used to absorb the water contained in the red mud and improve the reactive silica
content of the cement. Scientists conducted a series of studies into the production of cement using red
mud, fly ash, lime and gypsum as raw materials. Use of red mud cement not only reduces the energy
consumption of cement production, but also improves the early strength of cement and resistance to
sulfate attack [6]
B. Concrete industry
Red mud from Birac Alumina Industry, Serbia was tested as a pigment for use in the building material
industry for standard concrete mixtures. Red mud was added as a pigment in various proportions
(dried, not ground, ground, calcinated) to concrete mixes of standard test blocks (ground limestone,
cement and water) [7]. The idea to use red mud as pigment was based on extremely fine particles of
red mud (upon sieving: 0.147 mm up to 4 wt%, 0.058 mm up to 25 wt% and the majority smaller than
10 microns) and a characteristic red colour. Compressive strengths from 14.83 to 27.77 MPa of the
blocks that contained red mud between 1 and 32% were considered satisfactory. The reported tests
have shown that neutralized, dried, calcined and ground red mud is usable as pigment in the building
materials industry. Red oxide pigment containing about 70 % iron oxide was prepared from NALCO
red mud by [8] after hot water leaching filtration, drying and sieving.
C. Red mud in the brick industry
D. Dodoo- Arhin, et al [9] have been investigated bauxite red mud-Tetegbu clay composites for their
applicability in the ceramic brick construction industry as a means of recycling the bauxite waste. The
initial raw samples were characterized by X-ray diffraction (XRD) and thermo gravimetric (TG)
analysis. The red mud-clay composites have been formulated as 80%-20%, 70%-30%, 60%-40%,
50%-50% and fired at sintering temperatures of 800C, 900C and 1100C. Generally, mechanical
strengths (modulus of rupture) increased with higher sintering temperature. The results obtained for
various characterization analyses such as bulk densities of 1.59 g/cm3 and 1.51 g/cm3 compare very
well with literature and hold potential in bauxite residue eco-friendly application for low-cost
recyclable constructional materials. Considering the physical and mechanical properties of the
fabricated brick samples, the batch formulation which contained 50% each of the red mud and
Tetegbu clay is considered the best combination with optimal properties for the construction bricks
application and it could be employed in lighter weight structural applications.
VI. UTILIZATION OF RED MUD AS FILLING MATERIAL
A. Road Base Material
High-grade road base material using red mud from the sintering process is promising, that may lead to
large-scale consumption of red mud. Qi [10] suggest using red mud as road material. Based on the
work of Qi, a 15 m wide and 4 km long highway using red mud as a base material was constructed in
Zibo, Shandong. A relevant department had tested the sub grade stability and the strength of road and
concluded that the red mud base road meets the strength requirements of the highway [11].
B. Mining
Yang et al. [12], from the Institute of Changsha Mining Research, have studied the properties,
preparation and pump pressure transmission process of red mud paste binder backfill material. Based
1057 Vol. 7, Issue 3, pp. 1053-1059

on this study, a new technology named pumped red mud paste cemented filling mining has been
developed by the Institute of Changsha Mining Research, in cooperation with the Shandong
Aluminum Company. They mixed red mud, fly ash, lime and water in a ratio of 2:1:0.5:2.43, and then
pumped the mixture into the mine to prevent ground subsidence during bauxite mining. The tested 28-
day strength can reach to 3.24 MPa. This technology is a new way not only for the use of red mud, but
also for non-cement cemented filling, successfully resolving the problem of mining methods in the
Hutian bauxite stop. Underground exploitation practice on the bauxite has proved that cemented
filling technology is reliable and can effectively reduce the filling costs, increase the safety factor of
the stop and increase the comprehensive benefits of mining [13].
VII. RECOVERY OF COMPONENTS FROM RED MUD
Red mud primarily contains elemental compositions such as Fe2O3, Al2O3, SiO2, CaO, Na2O and K2O.
Besides, it also contains other compositions, such as Li2O, V2O5, TiO2 and ZrO2. For instance, the
content of TiO in red mud produced in India can be as much as 24%. Because of the huge amount of
red mud, value elements like Ga, Sc, Nb, Li, V, Rb, Ti and Zr are valuable and abundant secondary
resources. Therefore, it is of great significance to recover metals, especially rare earth elements, from
red mud.
Due to the characteristics of a high iron content, extensive research into the recovery of iron from
Bayer process red mud have been carried out by scientists all over the world. The recycling process of
iron from red mud can be divided into roasting magnetic recovery, the reducing smelting method, the
direct magnetic separation method and the leaching extraction method, according to the different
ways of iron separation. Researchers in Russia, Hungary, America and Japan have carried out iron
production experiments from red mud. Researchers from the University of Central South have made
steel directly with iron recovered from red mud [14]. The Chinese Metallurgical Research Institute
has enhanced the iron recovery rate to 86% through making a sponge by red mud-magnetic separation
technology. Sun et al. [15] researched magnetic separation of iron from Bayer red mud and
determined the process parameters of the magnetic roasting-magnetic selecting method to recover
concentrated iron ore.
VIII. SUMMARY AND CONCLUSION
Specific gravity of the red mud is 3.04 which is very high compared to the soil solids. So the
density of red mud will be more and so the strength is more.
From the fig 1 graph showing particle size distribution mud indicated grains are fine and it is
well graded. So the soil can be used as an embankment material, backfill material etc.
From the Atter bergs limits it is concluded that the plasticity Index of the red mud is13.2. So,
according to the IS classification based on plasticity A-line, the soil falls under ML. Means it
is silt with low compressibility.
The maximum dry density and optimum moisture content of the red mud is 1.53gm/cc and
33.5% Respectively
Co-efficient of permeability of red mud is 5.786e-7cm/s which shows that permeability is
very low. Low permeable materials can be used for construction of earthen dams, road
embankments etc.
The cohesive strength and the angle of shear resistance obtained from the triaxial test are
0.123kg/cm
2
and 26.8
0
. The strength value of the red mud is higher than the conventional clay
material.
CBR value of the red mud in soaked condition is 4.2% which is greater than the 3%, so we
can use the red mud as a road material in village side.
By seeing all these properties of the red mud we can utilize the red as geotechnical material
like Backfill material, road sub-grade material embankment material. Red mud is further
stabilized to enhance the more strength with lime, gypsum, fly ash etc
Utilization of red mud is established in brick manufacturing, partial cement refilling, in
concrete industry and stabilization process.
1058 Vol. 7, Issue 3, pp. 1053-1059

A wide variety of potential uses of red mud have been reviewed, yet there is no economically
viable and environmentally acceptable solution for the utilization of large volumes of red
mud.
There is urgent need to undertake research and development for studying the metal speciation
and the changes associated with red mud reuse in the construction purposes and during the
wet storage of red mud in ponds.
IX. SCOPE FOR FUTURE WORK
In the present work, all the tests are worked on the unstabilized red mud. After these tests, it is
concluded that, the red mud can be used as a geotechnical material for varies purposes. In future, one
can also work with the red mud by stabilizing by adding lime, fly ash, gypsum.
REFERENCES
[1]. R. K. Paramguru, P. C. Rath, and V. N. Misra, Trends in red mud utilization - a review, Mineral
Processing & Extractive Metallurgy Review, vol. 26, no. 1, pp. 129, 2005.
[2]. U. V. Parlikar, P. K. Saka, and S. A. Khadilkar, Technological options for effective utilization of
bauxite residue (Red mud) a review, in International Seminar on Bauxite Residue (RED MUD),
Goa, India, October 2011.
[3]. Suchita Rai, K.L. Wasewar, J. Mukhopadhyay, Chang Kyoo Yoo, Hasan Uslu Neutralization and
utilization of red mud for its better waste management ARCH. ENVIRON. SCI. (2012), 6, 13-33
[4]. A. R. Hind, S. K. Bhargava, Stephen C. Grocott, The surface chemistry of Bayer process solids: a
review, Colloids and Surfaces A : Physicochem. Eng. Aspects, 146 (1999) 359374
[5]. E. Balomenos, I. Gianopoulou, D. Panias, I. Paspaliaris A Novel Red Mud Treatment Process :
Process design and preliminary results TRAVAUX Vol. 36 (2011) No. 40.
[6]. Qiu XR, Qi YY. Reasonable utilization of red mud in the cement industry. Cem. Technol.
2011;(6):103105.
[7]. Cablik V (2007). Characterization and applications of red mud from bauxite processing. Gospodarka
Surowcami Mineralnymi (Mineral Resource Management) 23 (4): 29-38.
[8]. Satapathy BK, Patnaik SC, Vidyasagar P (1991). Utilisation of red mud for making red oxide paint.
INCAL-91, International Conference and Exhibition on Aluminium at Bangalore, India 31st July-2nd
Aug. 1991 (1): 159-161.
[9]. D. Dodoo-Arhin*, D. S Konadu, E. Annan, F. P Buabeng, A. Yaya, B. Agyei-Tuffour Fabrication
and Characterisation of Ghanaian Bauxite Red Mud-Clay Composite Bricks for Construction
Applications. American Jour of Materials Science 2013, 3(5): 110-119.
[10]. Qi JZ. Experimental Research on Road Materials of Red Mud; University of Huazhong Science and
Technology: Wuhan, China; 2005.
[11]. Yang JK, Chen F, Xiao B. Engineering application of basic level materials of red mud high level
pavement (In Chinese). China Munic. Eng. 2006;(5):79.
[12]. Yang LG, Yao ZL, Bao DS. Pumped and cemented red mud slurry filling mining method (In Chinese).
Mining Res. Develop. 1996;(16):1822.
[13]. Wang HM. The comprehensive utilization of red mud (In Chinese). Shanxi Energy Conserv.
2011;(11):5861.
[14]. Li, WD. New Separation Technology Research of Iron from Bayer Progress Red Mud; Central South
University Library: Changsha, China; 2006.
[15]. Sun YF, Dong FZ, Liu JT. Technology for recovering iron from red mud by Bayer process (In
Chinese). Met. Mine. 2009;(9):176178.
AUTHORS BIOGRAPHY
Kusum Deelwal was born in Karnal, Haryana, India, in 1971. She received the Bachelor
degree in Civil Engineering from the Barkatullah University Bhopal, in 1996 and the Master
in Environmental engineering degree from MANIT University, Bhopal, in 2003, both in
Civil engineering. She is currently pursuing the Ph.D. degree with the Department of Civil
Engineering, Bhopal. Her research interests is in Environmental engineering.

1059 Vol. 7, Issue 3, pp. 1053-1059

Mukul Kulshreshtha was born in Uttar pradesh, in 1967. He received both the Bachelor
degree in Civil Engineering and Masters Degree in Environmental from the IIT Kanpur. He
completed his Doctors degree from IIT Delhi in Environmental Engineering. Presently he
is working as a professor in MANIT, Bhopal, Madhya Pradesh.

Kishan Dharawath was born in Hyderabad, Andra Pradesh, India, in 1974. He received the
Bachelor degree in Civil Engineering from the JNTU Hyderabad, Andra Pradesh, in 1998
and the Master in Geotechnical Engineering from IIT Madras in 2001, Civil engineering. He
received Ph.D. from MANIT, Bhopal in 2013. He is currently working as an assistant
professor in MANIT, Bhopal. His area of interest is Geotechnical and Geoenvironmental.
1060 Vol. 7, Issue 3, pp. 1060-1066

PERFORMANCE ANALYSIS OF IEEE 802.11E EDCA WITH
QOS ENHANCEMENTS THROUGH ADAPTING AIFSN
PARAMETER
Vandita Grover and Vidusha Madan
Department of Computer Science, University of Delhi, New Delhi, 110007, India

ABSTRACT
Enhanced Distributed Channel Access is a priority mechanism which supports access category (AC) wise QoS
by differentiating initial contention window (CW) size and arbitration inter-frame space (AIFS) time. EDCA has
fixed range of CW and AIFS that has been chosen initially. In contemporary EDCA model AIFS parameter for
each AC remains static, irrespective of number of stations in a network. We propose a scheme where we vary
AIFS time of each access category, depending on current congestion on network. The proposed model
incorporates a mechanism where AIFS wait is smaller if there are fewer stations contending and moves to more
AIFS wait time in case of many stations willing to transmit. Simulations are also conducted to validate our
suggested enhancements for AIFSN parameter.
KEYWORDS: Enhanced Distributed Channel Access (EDCA), Access Category (AC), Arbitration Interframe
Spacing Number (AIFSN), Contention Window (CW)
I. INTRODUCTION
IEEE 802.11 [6] WLAN is a widely used, robust, scalable communication network which transmits
information over wireless links. It is a protocol for best effort service designed to work in two modes.
Distributed Coordination Function (DCF) [6] employs carrier sense multiple access with collision
avoidance (CSMA/CA) with binary back-off for asynchronous data transmission.
Point Coordination Function (PCF) [6] is optional mechanism and uses a centrally controlled channel
access mechanism to support time-sensitive traffic flows. PCF has not been implemented in most
current products.
With evolving customer needs like requirement of video/audio streaming, wireless phone over IP,
real-time applications Quality of Service (QoS) is particularly important. DCF doesnt provide QoS,
PCF not being widely implemented and these applications do not differentiate between different data
streams with different QoS requirements. To support such applications the IEEE 802.11e [5] MAC
employs contention based channel access EDCA and centrally controlled channel access function
Hybrid Coordination Function (HCF).
This paper has been organised in five sections. We discuss legacy DCF and EDCA models for
contention and transmission in Wireless LANs in sections II, III. In section IV we propose an adaptive
model which will react according to present load conditions in the network by attuning AIFS
parameter. Section V studies simulation results of the proposed scheme.
II. DISTRIBUTED COORDINATION FUNCTION
2.1. The Basic Access Mechanism [6]
A station (ST) with a new frame to transmit monitors the channel activity. When the channel is idle
for distributed interframe space(DIFS) period, it backs-off for random number of slots. The backoff
counter is frozen, if during backoff the medium is sensed busy. When the medium is free; the backoff
counter resumes and ST transmits when the backoff counter reaches zero. Otherwise, if the channel is
sensed busy (either immediately or during the DIFS), the station persistently monitors the channel
1061 Vol. 7, Issue 3, pp. 1060-1066

until it is measured idle for DIFS time. ST generates the random backoff interval before transmitting
to avoid collision with frames being transmitted by other stations. To evade channel capture, ST waits
a random backoff time between two consecutive new frame transmissions, even if the medium is
sensed idle in the DIFS time.
DCF employs a discrete-time backoff scale. The time immediately following an idle DIFS is slotted,
and a station is allowed to transmit only at the beginning of each slot time (size ). Slot time is set to
the time needed at any station to detect the transmission from any other station.
The back off procedure is exponential. During each transmission, the backoff time is uniformly
chosen in the range (0, w-1) where w is called the contention window, and depends on the number of
failed transmissions for the packet. At the first transmission attempt, w is initialized to CWmin
(minimum contention window). After each unsuccessful transmission, w is doubled, up to a maximum
value CWmax = 2
m
CWmin.
When the channel is sensed idle, the backoff time counter is decremented; frozen, when a
transmission is detected on the channel, and reactivated when the channel is sensed idle again for
DIFS period. The station transmits when the backoff time reaches zero.
2.2. RTS/CTS Mechanism [6]
This is an optional four way handshake technique to avoid the hidden terminal problem. Hidden
terminals are those stations which are sensing the channel for being idle but are far away from the
sending/receiving channel. Since they cannot sense the channel activity (being at a distance), they
transmit and hence resulting in collision. To address this problem the following mechanism is
employed.
A station (ST) willing to transmit a frame, waits until the channel is sensed idle for a DIFS, and
follows the backoff conditions, and transmits a special short frame called request to send (RTS) before
sending the frame. When the receiving station (DST) detects an RTS frame, it responds, after a SIFS,
with a clear to send (CTS) frame. ST transmits its frame only if it receives CTS frame correctly and
CTS frames carry information of the length of the frame to be transmitted. This information can be
read by any listening station, which can update its network allocation vector (NAV). NAV has
information about the period of time when the channel will remain busy. Therefore, when a terminal is
hidden from either the transmitting or the receiving station, by detecting just one frame among the
RTS and CTS, it can delay further transmission accordingly, and thus avoid collision.
The RTS/CTS mechanism reduces the frames involved in contention process and hence is very
effective in terms of system performance. If stations wanting to transmit at the same time employ the
RTS/CTS mechanism, collision occurs only on the RTS frames, and it is detected easily by the
transmitting stations by the lack of CTS responses making the system adapt accordingly.
III. ENHANCED DISTRIBUTED CHANNEL ACCESS[5]
DCF provides the best effort service and real-time multimedia applications (like voice, video) are not
differentiated with data applications. EDCA is designed to enhance the DCF mechanism to provide
prioritized QoS and to provide a distributed access method to support service differentiation among
traffic classes. Traffic is categorised in four Access Categories (ACs). Smaller CWs are assigned to
ACs with higher priorities so that probability of successful transmission is biased towards high-
priority ACs. The initial CW size can be set differently for different priority ACs, yielding higher
priority ACs with smaller.
Table 1. Contention Window Boundaries [5] for different ACs.
AC CWmin[AC] CWmax[AC] AIFSN
0 CWmin CWmax 7
1 CWmin CWmax 3
2 (CWmin+1)/2-1 CWmax 2
3 (CWmin+1)/4-1 (CWmin+1)/2-1 2

1062 Vol. 7, Issue 3, pp. 1060-1066

Differentiation is achieved by applying AIFS wait instead of using fixed DIFS as in the DCF. The
AIFS for a given AC is determined by the following equation:
AIFS = SIFS + AIFSN * slot time
AIFSN: AIFS Number and determined by the AC and physical settings
Slot time: Time slot duration.
Table 2. Default EDCA Parameters [5] for different ACs.
AC CWmin[AC] CWmax[AC] AIFSN
AC_BK 15 1023 7
AC_BE 15 1023 3
AC_VI 7 15 2
AC_VO 3 7 2
The AC with the smallest AIFS has the highest priority. The physical carrier sensing and the virtual
sensing methods are similar to those in the DCF. The countdown procedure when medium is sensed
idle is different in EDCA. After the AIFS period, the backoff counter decreases by one at the
beginning of the last slot of the AIFS. In DCF, this is done at the beginning of the first time slot
interval following the DIFS period.
For each station, different ACs have different queues for buffering frames. Each AC within a station
acts like a virtual station and contends for channel access. AC independently starts its backoff after
sensing the medium idle for at least AIFS period. During virtual collision: different ACs finish AIFS
wait simultaneously, the AC with higher priority has the opportunity for physical transmission, while
the lower priority AC follows the backoff rules.
A transmission opportunity (TXOP) limit has been defined which is the time interval for a station
when it can initiate transmissions. During a TXOP, a station may be allowed to transmit multiple data
frames from the same AC with a SIFS wait between an ACK and next frame. This is also referred to
as contention free burst (CFB).
IV. ADAPTING AIFSN PARAMETER IN EDCA
Joe Nauom Sawaya, Bissan Ghaddar in [1] and Joe Nauom Sawaya, Bissan Ghaddar and others in [3]
suggest a scheme where based on current load conditions CW value is estimated and the system
outputs CW value at which node will be able to transmit. However, the legacy EDCA, [1] and [3] and
many other proposed adaptive models do not take into account the time each AC has to wait before
transmitting. In EDCA, each access category has a fixed value of AIFSN using which it computes the
AIFS time for which that access category must wait. The drawback of this scheme is that load
conditions of the system are not taken into account. The load conditions include two main factors:
Number of stations in the system
Probability of collision
In this section we propose a scheme where AIFS wait time will vary according to the current
congestion in the network. The system is assumed to be in condition where there are many stations
contending for the slot and each station always has a frame to transmit. So the number of stations is an
important part when load is being determined. The probability of collision determines the amount of
congestion in the system at any time. It gives the number of frames that have to be retransmitted.
The stations must wait for a long time even when the load is less. In case the load is more the stations
wait for a predefined time leading to more collisions. For example, if there are only 2 stations in the
system and the AIFS for the access category with least priority is, say 7, it has to wait unnecessarily
for a long time. This will only in increase the load as collisions would increase. To reduce the delay
or number of collisions, depending on the system state, the AIFSN values may be adapted
dynamically.
To have variable AIFSN values we define a range of values for each access category from which the
actual AIFSN value is computed. We denote this computed AIFSN value by AIFSN'. The range of
values for different access categories must be non-overlapping so that an access category with higher
priority does not have to wait for a longer period than an access category with low priority. The
selected value of AIFSN' lies at the lower end of the range if load is less, thereby selecting a smaller
1063 Vol. 7, Issue 3, pp. 1060-1066

value. In case of greater load the value selected lies at the far end of the predefined interval. This
results in stations having a smaller waiting time when load conditions are light and larger waiting time
when the system is congested.
Let the range of values for each access category is denoted by [Mi, Ni]. For example, the interval
could be [2, 5] for some access category. Thus the possible values from which the access category can
choose its AIFN value lie in range 2 to 5. This range of values for each access category is
predetermined depending on the optimal values that would be obtained after simulation. The AIFSN'
values are chosen from amongst these values. The load condition is determined by the probability of
collision at that time. We take Pc since it includes the no of stations in the system at any point of time
as well as the amount of congestion in the system. We assume that each station always has a frame to
transmit. Each time before transmitting a frame, the station must compute the collision probability Pc
and the AIFSN' value to be used. The computation is illustrated as follows
Let the interval of values for access category i from which the AIFSNi' value will be chosen be [2, 5].
Let number of possible values to be chosen from be mi
mi =5-2+1 = 4
The number of slots in which the probability of collision can be divided is thus 4.
The length of each slot is
ki = 100/4 =25
i.e. we have four slots from 0-25%, 25-50%, 50-75% and 75-100%
If the probability of collision lies in the first slot the AIFSN value 2 is chosen. Similarly if the
collision probability lies in the range 50-75% AIFSN value 4 is chosen and so on.
Say the value of Pc is 0.3, then its quantified as 30%.
AIFSNi' = 2 + (0.3 X 100) / 25
AIFSi = SIFS + AIFSN'i * slot time
Generalising the above illustration
AIFSN range: [Mi, Ni]
The number of possible values AIFSN' can choose from is calculated as
mi = AIFSN[Ni-Mi+1] (1)
Let ki denote slot length corresponding to mi
ki = 100 / mi (2)
We have a one to one correspondence between the possible Pc values and the AIFSN values to be
chosen from. We choose that value of AIFSN which corresponds to the percentage slot that contains
the current Pc value. This we can obtain by adding the minimum AIFSN value of the range of values
to the corresponding slot of the probability.
AIFSNi' = minAIFSNi + (Pc X100)/ ki (3)
Using this AIFSNi' we compute the AIFSi values.
V. SIMULATION ANALYSIS OF SUGGESTED MODEL
5.1. configuration
We used Pamvotis Network Simulator [8] (a free and open source simulator) to test various scenarios
of dynamically adjusting AIFSN number. The source code of Pamvotis supports EDCA with static
values of AIFS number for each category that can be configured during set up using GUI. We
enhanced Pamvotis code with additional Java code to support AIFSN range which could be set-up
before the simulation run. Our suggested mathematical formula using Equations 1, 2 and 3 was
implemented to adapt AIFSN number depending on network congestion and many stations contending
for a slot to transmit.
To achieve load condition, we have varied the number of stations in the system from 10 to 35 and data
rate to be 2Mbps. Each station has a frame ready to send, with packet generation rate being constant
and packet length (8000 bits) being uniform for all packets. Each simulation run was of 300 seconds.
We had run simulations for various ranges for AIFS parameter. We discuss the scenario which reduced
delay significantly.

1064 Vol. 7, Issue 3, pp. 1060-1066

Scenario 1: Classic EDCA
AC0 : AIFS [7-7], CWmin : 15 CWmax: 1023
AC1 : AIFS[3-3], CWmin : 15 CWmax : 1023
Scenario 2: Adapted EDCA
AC0 : AIFS [6-9], CWmin : 15 CWmax: 1023
5.2. ANALYSIS OF DELAY AND THROUGHPUT GRAPHS
Pamvotis simulator randomly generates an Access Category for a station. We have calculated the
mean delay for each AC for stations varying from 10 to 35.
Following are the mean delay graphs for each Access Category.

Figure 1. Delay Comparison (AC0)



1011121314151617181920212223242526272829303132333435
0
100
200
300
400
Nodes
A
c
c
e
s
s

D
e
l
a
y
(
m
s
e
c
)
Scenario 1
Scenario 2
1065 Vol. 7, Issue 3, pp. 1060-1066

In both the models, in accordance with service differentiation DelayAC0 > DelayAC1 > DelayAC2 >
DelayAC3. As the number of stations increase the average delay per access category increases due to
system load and collisions. We observe noticeable decrease in delay for Scenario 2 with range of
AIFS parameters in comparison to fixed AIFSN in Scenario 1.
For throughput analysis we consider average throughput of each AC in both the scenarios, since
Pamvotis generates ACs randomly. We then compare average throughput of the entire system from 10
to 35 nodes for the adapted and legacy EDCF model.

Figure 5. Scenario 1
Figure 6. Scenario 2
We also see that the suggested model achieves service differentiation and ThroughputAC3 >
ThroughputAC2 > ThroughputAC1 >= ThroughputAC0 where the access category is present.

10 12 14 16 18 20 22 24 26 28 30 32 34 0
0
20
40
60
80
100
Nodes
T
h
r
o
u
g
h
p
u
t
(
k
b
p
s
)
AC3
AC2
AC1
AC0
10 12 14 16 18 20 22 24 26 28 30 32 34 0
0
20
40
60
80
100
Nodes
T
h
r
o
u
g
h
p
u
t
(
k
b
p
s
)
AC3
AC2
AC1
AC0
1011121314151617181920212223242526272829303132333435
0
20
40
60
80
100
Nodes
A
c
c
e
s
s

D
e
l
a
y
(
m
s
e
c
)
Scenario 1
Scenario 2
1066 Vol. 7, Issue 3, pp. 1060-1066

Figure 7. Average System Throughput
A marginal increase in average system throughput is also observed for Scenario 2.
Simulation results validate our model by maintaining service differentiation, showing a significant
decline in delay for each AC and an increase in average system throughput.
VI. CONCLUSIONS AND FUTURE WORK
In this paper we had accustomed the AIFS parameter based on the load conditions in the system. The
simulation results show improvement in system performance in terms noticeable decrease in media
access delay and increase in throughput of each access category in comparison to the static AIFS
values used in conventional EDCA algorithm.
Various scenarios were tested and we found through simulation that AC0: 6 9 AC1: 3 - 6
AC2: 2 2 and AC3: 1 1 are best suited for the model we proposed. We also conclude that using
this range of AIFS parameters will continue to give higher priority to voice/video traffic and low
priority ACs will not have to wait too long to transmit when there is less congestion in the system.
In future, it can be explored how adapting both CW[1] and AIFS on the network depending on current
load conditions will improve the quality of service.
ACKNOWLEDGEMENTS
This work was a part of M.Sc. (C.S.) curriculum of Department of Computer Science, University of
Delhi. The authors would like to thank Mr. Pradyot Kanti Hazra (Associate Professor,Deptt. of Comp,
Sc., University of Delhi) for his valuable guidance.
REFERENCES
[1]. Joe Naum Sawaya, Bissan Ghaddar, A Fuzzy Logic Approach for Adjusting the Contention Window
Size in IEEE 802.11e Wireless Ad Hoc Networks
[2]. Stefan Mangold, Sunghyun Choi, Peter May, Ole Klein, Guido Heirtz, Lothar Stibor, Performance
Analysis and Enhancements for IEEE 802.11e Wireless Networks
[3]. Joe Naoum-Sawaya, Bissan Ghaddar, Sami Khawam, Haidar Safa, Hassan Artail, and Zaher Dawy,
Adaptive Approach for QoS Support in IEEE 802.11e Wireless LAN.
[4]. Qiang Ni, IEEE 802.11e Wireless LAN for Quality of Service.
[5]. Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications, IEEE Std
802.11TM-2007-Revision of IEEE Std 802.11-1999
[6]. IEEE 802.11 WG, Reference number ISO/IEC 8802-11:1999 (E) IEEE STD 802.11, 1999 edition.
International Standard [for] Information Technology-Telecommunications and information exchange between
systems-Local and metropolitan area networks-Specific Requirements- Part 11: Wireless LAN Medium Access
Control (MAC) and Physical Layer (PHY) specifications 1999
[7]. Byron W Putman, 802.11 WLAN Hands-On Analysis: Unleashing the Network Monitor for
Troubleshooting a and Optimization
[8]. Pamvotis IEEE 802.11 WLAN Simulator - http://pamvotis.org/
AUTHORS
Vandita Grover has been associated with University of Delhi as an Assistant Professor since 2011. Prior to that
she worked with Aricent Technologies for around three years. She did M.Sc.(CS) from Delhi University in 2008.
Vidusha Madan did M.Sc.(CS) from Delhi University in 2008 and is presently working at CSC India.
10 12 14 16 18 20 22 24 26 28 30 32 34
0
100
200
300
400
Nodes
T
h
r
o
u
g
h
p
u
t
(
K
b
p
s
)
Scenario 1
Scenario 2
1067 Vol. 7, Issue 3, pp. 1067-1074

DATA DERIVATION INVESTIGATION
S. S. Kadam, P.B. Kumbharkar
Department of Computer Engineering, SCOE, Sudumbare, Pune
University of Pune, Pune, Maharashtra, India

ABSTRACT
Malicious software is a major issue in todays computer world. Such software can silently reside in users
computer and can easily interact with computing resources. It is necessary to improve the honesty of host and
its system data. For improvement in security and honesty of host, this work is introduced. This mechanism
ensures the correct origin or provenance of critical system information and prevents utilization of host
resources by malware. Using this mechanism the source where a piece of data is generated can be identified. A
cryptographic origin approach ensures system properties and system- data integrity at kernel level. A frame
work is used for restricting outbound malware traffic. This frame work identifies network activities of malware.
This frame work can be used as powerful personal firewall for investigating outgoing traffic of a host.
Specifically, our derivation verification scheme requires outgoing network packets to flow through a checkpoint
on a host, to obtain proper origin proofs for later verification
I NDEX TERMS Authentication, malware, cryptography, derivation, networking.
I. INTRODUCTION

Compared to the first generation of malicious software in late 1980s, modern attacks are more
stealthy and pervasive. Kernel-level root-kits are a form of malicious software that compromises the
integrity of the operating system. Such root-kits stealthily modify kernel data structures to achieve a
variety of malicious goals, which may include hiding malicious user space objects, installing
backdoors and Trojan horses, logging keystrokes, disabling firewalls, and including the system into a
botnet [2]. So, host-based signature-scanning approaches alone were proven inadequate against new
and emerging malware [6]. We view malicious software or malware in general as entities silently
residing on a users computer and interacting with the users computing resources. For example, the
network calls may be issued by malware to send outbound traffic for denial-of-service attacks, spam.
Goal of our work is to improve the reliability of the OS-level data flow; specifically, we provide
mechanisms that ensure the correct origin or derivation of critical system data, which prevents
antagonist from utilizing host resources [1]. We define a new security mechanism data-derivation
honesty. It verifies the source from which a piece of data is generated. For outbound network packets,
we deploy special cryptographic kernel modules at strategic positions of a hosts network stack, so
that packets need to be generated by user-level applications and cannot be injected in the middle of
the network stack. It gives low overhead. The implication of network-packet origin is that one can
deploy a sophisticated packet monitor or firewall at the transport layer such as [7] without being
bypassed by malware. The application of this system is for distinguishing user inputs from malware
inputs, which is useful in many scenarios.
Contribution Work: A new cryptographic derivation verification approach is presented here. And
its applications in understanding strong host-based traffic-monitoring are demonstrated.
The key exchange between the two modules is performed using asymmetric keys which is expensive
due of their storage and computation cost. This requires RSA algorithm for public key generation and
encryption which has high time complexity, so we replace this algorithm with general three-tier
security framework for authentication and pair wise key establishment. This three-tier security
architecture consists of three separate modules i.e. sign, verify, access module. Two polynomial
1068 Vol. 7, Issue 3, pp. 1067-1074

identifier pools of size M and S are created. Sign and access module are randomly given Km (Km>1)
and 1 identifiers from M respectively, similarly verify module and access module are randomly given
Ks and Ks-1 identifiers from S respectively. To establish a direct pair wise key between sign module
and verify module, a sign module needs to find a stationary access module in its neighborhood, such
that, access module can establish pair wise keys with both sign module and verify module. In other
words, a stationary access module needs to establish pair wise keys with both the sign module and the
verify module. It has to find a common polynomial m (from M) with the sign module and a common
polynomial k (from K) with the verify module.
Existing root-kit detection work includes identifying suspicious system call execution patterns,
discovering vulnerable kernel hooks, exploring kernel invariants, or using a virtual machine to enforce
correct system behaviors. For example, Christodorescu, Jha, and Kruegel collected malware behaviors
like system calls and compared execution traces of malware against benign programs [9]. They
proposed a language to specify malware behavior and an algorithm to mine malicious behaviors from
execution traces. A malware analysis technique was proposed and described based on hardware
virtualization that hides itself from malware. Although existing OS level detection methods are quite
effective, they typically require sophisticated and complex examination of kernel instruction
executions. To enforce the integrity of the detection systems, a virtual machine monitor (VMM) is
usually required in particular for root-kit detection. TPM is available on most commodity computers.
Information flow control has been an active research area in computer security. As early as in the 70s,
Denning et al [3][4] has proposed the lattice model for securing the information flow and applied it to
the automatic certification of information flow through a program. Data tainting, as an effective
tracking method, is widely used for the purposes of information leak prevention and malware
detection. Taint tracking can be performed at different levels.
Here use of TPM as a signature generator may be viewed as a special type of data tainting. In addition
to conventional taint tracking solutions such as hardware memory bit or extended software data
structure, here TPM-based solution uniquely supports the cryptographic operations to enforce data
confidentiality and the integrity of taint information. The important feature about TPM is its on-chip
secret key. Therefore, the client device can be uniquely authenticated by a remote server. Our paper
focuses on a host-based approach for ensuring system-level data integrity and demonstrates its
application for malware detection. In comparison, network trace analysis typically characterizes
malware communication behaviors for detection. Such solutions usually involve pattern-recognition
and machine learning techniques, and have demonstrated effectiveness against todays malware. Our
work provides a hardware-based integrity service to address that problem. In comparison to NAB
which is designed specifically for browser input verification, this work provides a more general
system-level solution for keystroke integrity that is application-oblivious.
The element of human behavior has not been extensively studied in the context of malware detection,
with a few notable exceptions including solutions by Cui, Katz, and Tan and Gummadi [5][8]. They
investigated and enforced the temporal correlation between user inputs and observed traffic. BINDER
describes the correlation of inputs and network traffic based on timestamps. It does not provide any
security protection against the detection system itself, e.g., how to prevent malware from forging
input events.
The work by Srivastava and Giffin on application aware blocking of malware traffic may bear
superficial similarity to our solution [10]. They used a virtual machine monitor (VMM) to monitor
application information of a guest OS without using any cryptographic scheme. Existing system use
root kit-detection work which includes identifying suspicious system call execution patterns,
discovering vulnerable kernel hooks, exploring kernel invariants or using a virtual machine to enforce
correct system behaviors. The lattice model for securing the information flow through a system as an
effective tracking method is widely used for the purposes of information leak prevention and malware
detection and can be performed at different levels, for example within an application, within a system,
or across distributed hosts. But this system lacks in cryptographic operations to enforce data
confidentiality and the integrity.
1069 Vol. 7, Issue 3, pp. 1067-1074

III. DESIGN
Introduced derivation verification mechanism has a essential difference from the traditional
cryptographic signature scheme. In most signature schemes the signer is assumed to be a person who
exercises judgment in signing documents and also in protecting his or her signing keys. In the
environment of malware detection, the signer and verifier are programs. Prevention against these
attacks is critical. For network-traffic monitoring, malware may attempt to send traffic by directly
raising functions at the network-layer, but not at the lower level data-link or physical layers. We
assume that the Trusted Platform Module (TPM) is corrupt opposing; the cryptographic operations are
applied suitably; and the remote server is trusted and secure. TPM provides the guarantee of load-time
code integrity. It does not provide any detection ability for run-time compromises such as buffer
overflow attacks [11]. Advanced attacks [12], [13] may still be active under this assumption,
indicating the importance of our solutions.
We describe three actions for data-derivation investigation on a host: setup, sign and verify.
Setup: the data producer sets up its signing key k and data consumer sets up its verification key k0 in
a secure fashion that prevents malware from accessing the secret keys.
Sign(D, k): the data producer signs its data D with a secret key k, and outputs D along with its proof
sig.
Verify(sig,D, k0): the data consumer uses key k0 to verify the signature sig of received data D to
ensure its origin, and rejects the data if the investigation fails. Although simple, the cryptographic
derivation investigation method can be used to ensure and impose correct system and network
properties and appropriate workflow under a trusted computing environment.
IV. INVESTIGATING DERIVATION OF OUTBOUND TRAFFIC
The cryptographic derivation verification technique in a network setting, for ensuring the reliability of
outbound packets, as they flow through the hosts network stack is illustrated here. Fig. 1 shows the
network stack. Genuine traffic origins from the application layer whereas malware traffic may be
injected to the lower layers.

Fig. 1. Network stack. Genuine traffic origins from the application layer whereas malware traffic may be
injected to the lower layers. Traffic checkpoints are placed at the Sign and Verify modules.
Traffic checkpoints are placed at the Sign and Verify modules. The design of a lightweight traffic
monitoring framework is described. It can be used as a building block for constructing powerful
personal firewalls or traffic-based malware detection tools. We demonstrate the effectiveness of our
traffic supervising framework in identifying the network activities of quiet malware.
The network stack is part of the hosts operating system and consists of five layers, application,
transport, network, data link, and physical layers. User outbound traffic travels all five layers on the
stack from the top to the bottom before being sent out. System services are typically implemented as
applications, thus their network flow also traverses the entire Internet protocol stack. Specifically, our
1070 Vol. 7, Issue 3, pp. 1067-1074

derivation investigation scheme requires outgoing network packets to flow through a checkpoint on a
host, to obtain proper origin proofs for later investigation. Any traffic sent through disabling or
bypassing the firewall can be detected, as the packets are unable to provide their origin proofs. And
we can effectively prevent any traffic to be sent without passing through a certain checkpoint
appreciably improving the assurance of traffic-based malware detection on hosts. Such a simple yet
powerful traffic-monitoring framework can defer advanced detection on application-level traffic such
as [14]. Genuine outbound network traffic passes through the entire network stack in the hosts
operating system. We develop a strong cryptographic procedure for enforcing the proper origin of a
packet on a host.
A. Architecture of Traffic Provenance Verification
Here a general approach is described for improving the assurance of system data and properties of a
host, which has applications in preventing and identifying malware activities. The host-based system
security solutions against malware complement network-traffic-based analysis. We demonstrated
application in identifying quiet malware activities of a host, in particular how to distinguish
malicious/unauthorized data flow from valid one on a computer that may be compromised.

Fig. 2. System Architecture
Our design of the traffic-monitoring framework extends the hosts network stack and deploys two
kernel modules, Sign and Verify modules, as illustrated in Figure 2. Both signing and verification of
packets take place on the same host but at different layers of the network stack the Sign module is at
the transport layer, and the Verify module is at the network layer. The two modules sharing a secret
cryptographic key monitor the integrity of outbound network packets. All legitimate outgoing network
packets first pass through the Sign module, and then the Verify module. The Sign module signs every
outbound packet, and sends the signature to the Verify module on the same host, which later verifies
the signature with a shared key. The signature proves the provenance of an outgoing packet. If a
packets signature cannot be verified or is missing, then the packet is labeled as suspicious.
V. IMPLEMENTATION
The system follows three-tier architecture. Our main checkpoints are created by polynomial identifier
pools.
The three-tier security architecture consists of three separate modules i.e. sign, verify, access module.
Two polynomial identifier pools of size M and S are created. Sign and access module are randomly
given Km (Km>1) and 1 identifiers from M respectively, similarly verify module and access module
are randomly given Ks and Ks-1 identifiers from S respectively. To establish a direct pair wise key
between sign module and verify module, a sign module needs to find a stationary access module in its
neighborhood, such that, access module can establish pair wise keys with both sign module and verify
module. In other words, a stationary access module needs to establish pair wise keys with both the
sign module and the verify module. It has to find a common polynomial m (from M) with the sign
module and a common polynomial k (from K) with the verify module.
1071 Vol. 7, Issue 3, pp. 1067-1074

Summary:
Input: In this module we will design GUI and Database of the system and we complete input module.
Sign: In this module we create sign module and transfer data input module to sign module and we add
private key and public key of sign module and encrypt data.
Verify: In this module we create verify module, using this module and sign module we exchange the
key and decrypt the data. In this module we use UMAC.
Output: In this module we develop output module it is use to get data and in that we will some option
for save and reject data.

Fig. 3 Process flow.
RSA algorithm is used here for key generation, encryption and decryption.
RSA Key Generation Algorithm:
Begin
Select two large prime numbers p, q
Compute
n = p q
v = (p-1) (q-1)
Select small odd integer k relatively prime to v
gcd(k, v) = 1
Compute d such that
(d k)%v = (k d)%v = 1
Public key is (k, n)
Private key is (d, n)
End
RSA Encryption Algorithm:
Begin
Let Input: integers k, n, M
M is integer representation of plaintext message
Let C be integer representation of cipher text
Compute
C = (M
k
)%n
end
Output: integer C
cipher text or encrypted message
RSA Decryption Algorithm:
Begin
Let Input: integers d, n, C
C is integer representation of cipher text message
Let D be integer representation of decrypted cipher
text.
Compute
D = (C
d
)%n
end
Input Data Sign Module Verified Data Verify Module Data Data
key Exchange
add public and private key add public and private key
Generate signing key
and symmetric key
b0,b1 a0,a1
1072 Vol. 7, Issue 3, pp. 1067-1074

Output: integer D
- decrypted message
VI. STATISTICAL MODEL
Set Theory Analysis:
Identify the Input User Data:
IN= {in1, in2, in3.}
Where IN is main set of Input User Data like in1, in2, in3inn
Identify the public key:
PK= {pk1, pk2, pk3.}
Where PK is main set of public key User like pk1, pk2, pk3pkn
Identify the Private key:
PriK= {prik1, prik2, prik3.}
Where PriK is main set of private key User like prik1, prik2, prik3prikn
Identify the Key Exchange:
KE= {ke1, ke2, ke3.}
Where KE is main set of key Exchange like ke1, ke2, ke3ken
Identify the Key Generation:
KG= {kg1, kg2, kg3.}
Where KG is main set of Key Generation like kg1, kg2, kg3kgn
Identify the Symmetric key:
SK= {sk1, sk2, sk3.}
Where SK is main set of symmetric key like sk1, sk2, sk3skn
Identify the Signing Key:
SIK= {sik1, sik2, sik3.}
Where SIK is main set of singing key like sik1, sik2, sik3sikn
Process:
We define a security property data origin integrity. It states that the source from which a piece of data
is generated can be verified. We give the concrete illustration of how data-provenance integrity can be
realized for system-level data.
Identify the processes as P.
P= {Set of processes}
P = {P1, P2, P3, P4Pn}
P1 = {e1, e2, e3, e4, e5}
Where
{e1= input Data}
{e2= key generation}
{e2= key exchange}
{e3= create signature}
{e4= verify signature}
{e5=display Message}
Create Signature:
We use UMAC algorithm to generate signature for each packet.
UMAC signature = H_K(S) XOR F(nonce)
Where H = hash algorithm
K = Signing Key
S = Source
F = Pseudorandom number generator
Signature Encryption:
We use Advance cryptography algorithm for signature encryption.
CB1 = P (XOR) KB1
CB2= CB1 >> 3
CB3 = CB2 (XOR) KB2
CB4 = CB3 (XOR) KB3
1073 Vol. 7, Issue 3, pp. 1067-1074

CB4 is the encrypted data.
P Plain text
CB Cipher block
KB Key block
The Three Tier Security Architecture:
Polynomial Identifier pool M= {m1, m2, m3} of size m.
Polynomial Identifier pool S= {s1, s2, s3} of size s.
Sign and access module are randomly given Km (Km>1) and 1 identifiers from M respectively.
Verify and access modules are randomly given Ks and Ks-1 identifiers from S respectively.
VII. EXPECTED RESULTS
Table 1. Expected Results

The above table shows the expected results of system. The given system improves the security of data
as well as kernel module.
VIII. CONCLUSION
Here a general approach for improving the assurance of system data and properties of a host is
described, which has applications in preventing and identifying malware activities. Defined host-
based system security solutions against malware complement network-traffic based analysis. Here
application of derivation investigation mechanism is demonstrated in identifying quiet malware
activities of a host, in order to distinguish malicious/unauthorized data flow from genuine one on a
computer.
Technical contributions are, the model and operations of cryptographic derivation identification in a
host based security setting is proposed. Its important usage for achieving highly assured kernel data
and application data of a host, and associated technical challenges are pointed out. And, origin
investigation approach is demonstrated by a framework for ensuring the honesty of outbound packets
of a host. This traffic-monitoring framework creates checkpoints that cannot be bypassed by malware
traffic.
IX. FUTURE SCOPE
In this project key stroke integrity also can be verified.
REFERENCE
[1]. Kui Xu, Huijun Xiong, Chehai Wu, Deian Stefan, Danfeng Yao Data-Provenance Verification For Secure
Hosts In IEEE Transactions on Dependable and Secure Computing Vol.9 No.2 Year 2012
[2]. A. Baliga, V. Ganapathy, and L. Iftode. Automatic inference and enforcement of kernel data structure
invariants. In 24
th
Annual Computer Security Applications Conference (ACSAC), 2008.
[3]. D. E. Denning. A lattice model of secure information flow. Commun. ACM, 19:236243, May 1976.
[4]. D. E. Denning and P. J. Denning. Certification of programs for secure information flow. Commun. ACM,
20:504513, July, 1977.
[5]. W. Cui, R. H. Katz, andW. tian Tan. Design and Implementation of an extrusion-based break-in
detector for personal computers. In ACSAC, pages 361370. IEEE Computer Society, 2005.
Input Sign Module Verify Module Output
Any input file Enabled Enabled Verifies the origin and
generates result
Any input file Disabled Enabled Fails
1074 Vol. 7, Issue 3, pp. 1067-1074

[6]. M. G. Jaatun, J. Jensen, H. Vegge, F. M. Halvorsen, and R. W. Nergard. Fools download where angels
fear to tread. IEEE Security & Privacy, 7(2):8386, 2009.
[7]. H. Xiong, P. Malhotra, D. Stefan, C. Wu, and D. Yao. Userassisted host-based detection of outbound
malware traffic. In Proceedings of International Conference on Information and Communications Security
(ICICS), December 2009.

[8]. R. Gummadi, H. Balakrishnan, P. Maniatis, and S Ratnasamy. Not-a-Bot: Improving service
availability in the face of botnet attacks. In Proceedings of the 6
th
USENIX Symposium on Networked
Systems Design and Implementation (NDSI), 2009.
[9]. M. Christodorescu, S. Jha, and C. Kruegel. Mining specifications of malicious behavior. In ESEC-FSE
07: Proceedings of the 6th joint meeting of the European software engineering conference and the ACM
SIGSOFT symposium on the foundations of software engineering, pages 514, New York, NY, USA,
2007. ACM.
[10]. A. Srivastava and J. Giffin. Tamper-resistant, Application-aware blocking of malicious network
connections. In RAID 08: Proceedings of the 11
th
international symposium on Recent Advances in
Intrusion Detection, pages 3958, Berlin, Heidelberg, 2008. Springer-Verlag
[11]. S. Garriss, R. Caceres, S. Berger, R. Sailer, L. van Doorn, and X. Zhang. Trustworthy and personalized
Computing on public kiosks. In MobiSys 08: Proceeding of the 6th international conference on Mobile
systems, applications, and services, pages 199210, New York, NY, USA, 2008. ACM.
[12]. A. Baliga, P. Kamat, and L. Iftode. Lurking in the shadows: Identifying systemic threats to kernel data.
In IEEE Symposium on Security and Privacy, pages 246251. IEEE Computer Society, 2007.
[13]. J. Wei, B. D. Payne, J. Giffin, and C. Pu. Soft-timer driven transient kernel control flow attacks and
defense. In ACSAC 08: Proceedings of the 2008 Annual Computer Security Applications Conference,
pages 97107, Washington, DC, USA, 2008. IEEE Computer Society.
[14]. Z. Wang, X. Jiang, W. Cui, and X. Wang. Countering persistent kernel rootkits through systematic hook
discovery. In RAID 08: Proceedings of the 11th international symposium on Recent Advances in
Intrusion Detection, pages 2138, Berlin, Heidelberg, 2008. Springer-Verlag.

BIBLIOGRAPHY
Sharda Kadam is currently a student M. E. in the department of Computer Engineering at
Siddhant College of Engineering, Sudumbare, Pune under Pune University. She received
her B.E. degree from SVMEC, Nasik, under Pune university.

P.B. Kumbharkar is currently working as a H.O.D. He is a Ph.D. student at Pune University. He received
M.E. degree in the Department of Computer Engineering at Pune University

1075 Vol. 7, Issue 3, pp. 1075-1081

DESIGN AND IMPLEMENTATION OF ONLINE PATIENT
MONITORING SYSTEM
Harsha G S
Department of Electronics & Communication
Channabasaveshwara Institute of Technology, Gubbi, 572216, India

ABSTRACT
Patients condition is continuously monitored and web access functionality is embedded in a device to enable
low cost widely accessible and enhance user interface functions for the device. A web server in the device
provides access to the user interface functions for the end user through the device web page. This will be
dynamically updated with the recent monitored patient values from the module. Upon which the end user in the
network can avail live values on the webpage.
KEYWORDS - Embedded; Ethernet; TCP/IP Protocols; LPC1768; Cortex M3.
I. INTRODUCTION
Online data monitoring system is one of the promising trends in the era of computing in todays
system automation industry. The proposed project is one such attempt of signing online patient
condition monitoring system using Cortex-M3 core [2]. In this project we will develop Ethernet device
drivers or Cortex-M3 core to transmit the monitored sensor data (patient condition) to internet [1]. The
system can complete the remote monitoring and maintenance operations of equipment through the
network using Web browser [7]. By introducing Internet into control network, it is possible to break
through the spatial temporal restriction of traditional control network and effectively achieve remote
sensing, monitoring and real time controlling of equipment. The main essence of this project is to
design and implement online data monitoring system using ARM CORTEX M3 CORE and
TCP/IP Ethernet connection for data monitoring applications [3].
II. ORGANIZATION OF THE PAPER
Introduction tells about the problem statement, existing and proposed systems, and disadvantages of
old system as well as advantages of newly designed system is given. Related work explains the
different terminology which are quite useful to understand Ethernet protocol which is the main
communication protocol used in the system. Requirement lists the hardware and software
specification for this project. It also describes the overall description of project, product perspective,
user characteristics and specific requirements. Here different design constraints, interface and
performance requirements explained. System Definition deals with the clear definition of the
properties and characteristics of the embedded system prior to starting hardware and software
development essential to achieve a final result that matches its target specifications. System Design
deals with the advanced software engineering where the entire flow of the project is represented by
professional data flow diagrams and sequence diagrams. System Implementation section explains
coding guidelines and system maintenance for the project. Results explains in details about the
outcome of the experiment and compares it with the result obtained in existing system. Conclusion
and Future work section describe the summary of the related work and future enhancements of the
proposed system. Reference section contains the papers and other documents referred in the research.
Author Profile gives a brief introduction about the author.
1076 Vol. 7, Issue 3, pp. 1075-1081

III. RELATED WORK
Traditional method of deployment of Monitoring devices is based on the simple circuitry or low level
controllers. This makes them inefficient to cope up with the speed they are needed to be processing the
tasks and lacks functionalities like Ethernet support, SD card support, UART, Timers and Counters [6].
So they fail to communicate with the hospital personnel or the doctor directly and results in serious
situations. But automating this process is one of the modern approaches for this problem. This is
already implemented in many hospitals. Although the data is being monitored automatically in those
systems, it is not capable of reaching the Doctor or concerned person or to bring it to their notice[8].
IV. REQUIREMENTS
These requirements are required in order to carry out the execution of project.
A. Hardware Requirements
Hardware requirements for this project are listed as follows:
LPC 1768 H-Plus Ex Header Board: Cortex M3 Development board brought up by Coinel
Technologies Ltd., Bangalore.
Graphical LCD: 128*68 Pixels display with white backlit LED.
Sensors: Pulse Rate (NS13), Temperature (LM35) and Gas sensor used to obtain the patient
body conditions.
Buzzer: Used as an actuator to produce alarm when the threshold is reached.
GSM module SIM300: To send SMS alert to the Doctor/concerned person.
CooLinkEx J-Tag Debugger: To debug the program or to burn the program hex code into the
processor.
Ethernet Cable: To connect the network with the board.
USB-Power Supply Adapter: To power up the LPC1768 HPlus ex board.
D-Link Wi-Fi Router: To create wireless network.
B. Software Requirements
Software requirements for this projects are listed as follows:
Keil IDE : Keil development environment to edit , compile and debug code. It can even burn
the code to the processor using plugins required for J-Tag code burner.
Flash Magic : To burn the Hex file into the microcontroller.
Coocox CoLinkEX J-Tag Debugger : To debug code live on the hardware with step by step
feature. We can also use software breakpoints.
WebConverter V1.0 : To convert the webpage from HTML format into array of strings.
MyPublicHotspot : To create WLAN using the Wi-Fi modem of the Laptop.
V. SYSTEM DEFINITION
Defining a system based on the hardware and software is an important stage. This contains four steps in
defining a system with Ethernet connectivity for an embedded product. In these four stages, we
consider the embedded system as a black box. Which has a clear definition of the properties and
characteristics of the embedded system prior to starting hardware and software development is
essential to achieving a final result that matches its target specifications.
A. Specifying Required Functionality
The first thing is to focus on the aim of our system being designed. This is related to the application
and can encompass virtually anything that can be done using a MCU with considerably high speed.
One such example can be seen from the figure 1.

1077 Vol. 7, Issue 3, pp. 1075-1081

Figure 1. System Functionality Example.
B. Specifying Access Method
After we specify required functionality the next thing is to specify the accessing method of the
requirement. That is how we can access the output being generated from the controller.

Figure 2. User Interface Options.
Figure 2 shows the black box representation of the embedded system and how the various access
methods can be used to monitor and control the embedded system.
In our project we can go for one or more of the commonly used access methods used, they are:
Using a web browser.
Using HyperTerminal.
Having the embedded system send e-mail.
Using a custom application.
C. Specifying Configuration Method
Each and every device which is connected within a network will comprise of both a MAC address as
well as IP address which will be unique for that system. These are used to identify, locate and
communicate each other. Embedded systems with LPC1768 IC needs to obtain only the IP address
because the MAC address of the same will be preprogrammed in the flash memory of the IC.
There are four common configuration methods available to choose from:
Automatic Network Configuration.
Automatic Network Configuration with Netfinder.
Static Network Configuration.
Static Network Configuration with Netfinder.
D. Specifying Field Re-Programmability Requirements
The final part of the system definition is determining the field re-programmability requirements of the
embedded system.
1078 Vol. 7, Issue 3, pp. 1075-1081

The options for field re-programmability are:
No support for field re-programmability.
Re-programmability using a 3 or 5 pin header.
Re-programmability using a 10-pin header.
Re-programmability using a bootloader.
VI. SYSTEM DESIGN
System design is a transmission phase from a user oriented documented system to a purely
programmatic oriented system for programmers database personnel. The system design makes the
high level decision about the overall architecture of the system. The system design phase provides the
understanding and procedure details necessary for implementing the system recommended study. The
target system is arranged into subsystems based on the analysis structure and the proposed architecture.
A. System Architecture
The figure 3 represents the functional architecture of the system. Here the pulse rate, temperature and
gas sensors are the inputs for the system which will be used to obtain the parameters which are the
input for the system. Using the analog input from these sensors the processor computes for the data
required and handles the situation based on the requirements of our system.

Figure 3. Block Diagram of the System.
After obtaining digital values for the sensor outputs, they are checked for the threshold levels based on
the normal values expected and if the threshold is passed beyond then an alert is sent to doctor using
GSM modem. Then the web page which is dynamically updated with the values shall be pushed to the
network. This can then be accessed on a device within that network using its web browser pointed to
the defined IP address.
B. System Flow Diagram

Figure 4. Flowchart of the Project.
1079 Vol. 7, Issue 3, pp. 1075-1081

The process starts with initialization of the input outputs and configuring system parameters like
frequency. Then the peripheral devices are powered and configured for the purpose. Declaring and
initialization of the variables are to be made along with the prototype inclusion. After initialization of
the peripherals like external interrupt, timer interrupts, Ethernet, ADC etc. The sensor data is acquired
in analog form and are converted into digital values as per the unit of the measured variable and are
checked with threshold levels.

Figure 5. Flowchart of the Project (Cont..).
The values are updated on the local Display. If the threshold level is reached then the SMS alert is
sent to the Doctor number using GSM modem then dynamically updated webpage with recent values
is pushed to the Ethernet. This can be accessed from browsers of the devices in the same network.
VII. SYSTEM IMPLEMENTATION
Implementation is the realization of the application or execution of the plan given in the design. This
section deals with detailed description of how the project goal is achieved.
This phase continues until the system is operating in production accordance with the defined user
requirements. Successful implementation may not guaranteed improvement but it will prevent
improper installation.
The whole Implementation part of this project is divided into three different levels, they are:
Hardware Design
Software Generation
Application Development
A. Hardware Design
With a system definition in place, it is now time to start designing the hardware. The hardware design
flow consists of 5 steps corresponding to the 5 sections of a schematic for an embedded system with
Ethernet connectivity.
They are listed as follows:
Custom application circuitry - sensors, indicators, and other application-specific circuitry.
MCU - the main system controller.
Ethernet Controller - provides the MCU with the capability to send and receive data over a
network.
Ethernet Connector - the RJ-45 connector, magnetics, and link/activity LEDs.
Power circuit - provides the embedded system with regulated 3.3 V power.
B. Software Generation
In this step of the system design process, we will be generating the software that interacts with the
LPC1768 Ethernet module to provide the embedded system with Ethernet connectivity. Using the
TCP/IP Configuration Wizard, this step is one of the easiest steps in the entire system design process.

1080 Vol. 7, Issue 3, pp. 1075-1081

C. Application Development
The application code that will implement the required system functionality specified in the first
question of the system definition must co-exist and share resources with the TCP/IP Library. To
develop the code, we will need a good understanding of how the TCP/IP Stack operates.

Figure 6. Adding Dynamic HTML Content.
The figure 6 shows how we can add the HTML code on to the loop and it also tells how we can
update the values dynamically onto the webpage.
VIII. RESULTS
The output is seen on any of the devices within the same network just by accessing the IP address
assigned for the webpage from any web browser. Snapshot of the result webpage can be seen in the
figure 7.

Figure 7. Adding Dynamic HTML Content.
1081 Vol. 7, Issue 3, pp. 1075-1081

IX. CONCLUSION & FUTURE WORK
A. Conclusion
Three efficient sensors are interfaced with the controller & seen working fine.
TCP-IP libraries are used to communicate through Ethernet.
Cortex controller is proved to be highly flexible & efficient.
Tasks are efficiently managed in the loop so as to perform all functions properly using
available CPU time.
Software sanity is obtained on interfacing with all the peripherals.
WLAN is created and found to be working fine with all the clients in the network.
Outputs are checked on GLCD, mobile phones, tablets, laptops and desktop pcs.
B. Future Work
Actuators are automatically controlled based on the body conditions measured.
Provisions to be made for the doctor to control these actuations from the webpage.
REFERENCE
[1] Dhruva R. Rinku, Mohd arshad , Design and implementation of online data acquisition and controlling
system using cortex m3core. IJESAT - Volume-3, Issue-5,259-263 ,2013.
[2] S.Navaneethakrishnan, T.Nivethitha, T.Boobalan, Remote patient monitoring system for rural population
using ultra low power embedded system. ISSN: 2319-5967 ISO 9001:2008 Certified ,2013.
[3] IEEE Standards for Local Area Networks Carrier Sense Multiple Access With Collision Detection
(CSMA/CD) Access Method and Physical Layer Specifications, ANSI/IEEE Std 802.3-1985, 1985.
[4] Jean J Labrosse, Embedded Systems Building Blocks R&D Books, pp 61-100.
[5] An Axelson Embedded Ethernet and Internet Complete Lakeview Research LLC, 2003.
[6] Josephy Yiu The Definitive Guide to ARM Cortex M3 Newness,2007.
[7] Pont, Michael J Embedded C, Pearson Education 2007.
[8] LPC 17xx User Manual, NXP Semiconductors.
[9] Arul Prabahar A, Brahmanandha Prabhu, Development of a Distributed Data Collection System based on
Embedded Ethernet 978-1-4244-9799-71111$26.00 2011 IEEE.
[10] GSM basics History on Tutorial point, Wikipedia and private line.
AUTHOR PROFILE
Harsha G S (M23) was born in Karnataka, India in 1990. He received the B.E. and
M.Tech. Degrees from Visvesvaraya Technological University, Belgaum, India, in 2012
and 2014 respectively on Electronics. He is the founder of many technology blogs
www.GoHarsh.com & www.TecLogs.com, and he also founded a webdesign company in
the early 2012, www.WebsCrunch.com. His main areas of research interest are Embedded
systems, VHDL, Image Processing and Web Design.

1082 Vol. 7, Issue 3, pp. 1082-1090

COMPARISON BETWEEN CLASSICAL AND MODERN
METHODS OF DIRECTION OF ARRIVAL (DOA) ESTIMATION
Mujahid F. Al-Azzo
1
, Khalaf I. Al-Sabaawi
2

Department of Communications Engineering,
Faculty of Electronics Engineering, Mosul University, Iraq

ABSTRACT
In this paper, a comparison between the classical method; Fast Fourier Transform (FFT), and modern
methods; MUltiple SIgnal Classification (MUSIC) method and EigenVector method (EV) for direction of arrival
estimation are investigated. The simulation results of single source and two sources DOA estimation are realized
for three methods (FFT, MUSIC, and EV). After DOA estimation is realized an investigation is making to show
the performance of the (FFT, MUSIC, and EV) methods for DOA estimation of min. difference between angles of
sources versus SNR at fixed number of elements. The experimental results are achieved for single source DOA
estimation using set of ultrasonic transducers.
KEYWORDS: Array Signal Processing, DOA, MUSIC, EV.
I. INTRODUCTION
Increase of demand for the wireless technology service have spread into many areas such as, sensor
network, public security, environmental monitoring, mobile in smart antenna, and search and rescues.
All these applications can be counted the main reason for determine the direction of arrival (DOA)
estimation of incoming signal in wireless systems. The DOA also, used in other applications such as
radar, sonar, seismology, and strategy of defense operation. In smart antenna technology, a DOA
estimation algorithm is usually incorporated to develop systems that provide accurate location
information for wireless services [1]. The DOA technique one branch of array signal processing [2].
This paper based on uniform linear array (ULA) of multiple sensors that deals with array receiving
antenna to extract a useful information from it.
Many algorithms are founded to solve the problem of DOA [3]. Initially Beam forming, ESPRIT,
Maximum likelihood algorithm, subspace methods (Pisarenko Harmonic Decomposition (PHD)[4],
MUltiple SIgnal Classification (MUSIC)[5], and Eigenvector method (EV) [6]) and other algorithms.
We are used in this paper classical method (FFT) [7] and modern methods (MUSIC and EV) and
compare between them. This paper include the problem formulation of DOA estimation. After that the
theoretical and mathematical expression for MUSIC and EV methods are introduced. Then the
simulation and experimental results has been realized. The conclusions and suggestion for future work
are the last sections of this paper.
II. PROBLEM FORMULATION
We assume that a system consist of uniform linear array (ULA) with N-elements and M-sources and
the distance between elements is d. The first element of the array consider as a reference element. The
scenario for this system is the sources in the far field and the incoming data is plane wave. This system
are shown in figure (1).

1083 Vol. 7, Issue 3, pp. 1082-1090

The data received at the array antenna is
= + (1)
Where
= [(
1
), (
2
), (
)] Array steering vector.

: Signal source.
: is an additive noise term whose mean is 0 and variance is
2
I.
The algorithms will be using in this paper are based on the autocorrelation matrix of the received data
[8]. These algorithms are MUltiple SIgnal Classification (MUSIC) and EigenVector method (EV), and
compare these with classical Method Fourier Transform (FT).
III. MULTIPLE SIGNAL CLASSIFICATION (MUSIC)
MUSIC method is a high-resolution algorithm that based on eigen-decompostion of the
autocorrelation matrix. This method is decompose the covariance matrix into two subspaces, signal
subspace and noise subspace. Estimation the direction of arrival of incoming signal is determined from
steering vectors that orthogonal to the noise subspace, which is by finding the peak in spatial power
spectrum.
Suppose that there are M sources, the receiving signal of N elements uniform linear antenna array is
given by
= (
=1
+ = + (2)
The signal auto covariance matrix can be written as the average of N array output samples:
= {
} The eigen-decomposition is
=

=1
(3)
Where = (
1
,
2
,,
) it is eigenvalues and sorting in ascending sequence:

1

2
, ,
>
+1
= =
. That is the first M eigenvalues are in connection with the signal and their numeric
value are all more than
2
. The signal divided into two subspace signal subspace and noise subspace.
The signal subspace is the eigenvector (
1
,
2
, ,
) corresponding to the largest eigenvalues (

1

,
2
,,
), so the signal subspace is:
= [
1
,
2
, ,
].
is the diagonal matrix consist of the m

larger eigenvalues. The later N-P eigenvalues are totally depended on the noise and their numeric value
are
2
. The noise subspace is the eigenvector corresponding to the remaining eigenvalues (
+1

,
+2
,,
), so the noise subspace
= [
+1
,
+2
, ,
].
is the diagonal matrix consist of

the m larger eigenvalues. So could be divided into:
=

(4)
Duo to each column vector is orthogonal to noise subspace:
), = 1,2, , the spectrum

of the MUSIC are derived:
()
=
1
()
()

(5)
From eq. (5), we can estimating the DOA by searching the peak value [9].

Fig. 1. Uniform Linear Array (ULA) and the Direction
of Arrival (DOA).

1 2 N
4
3
()

1084 Vol. 7, Issue 3, pp. 1082-1090

IV. EIGEN VECTOR METHOD (EV)
In addition to the MUSIC algorithm, a number of other eigenvector methods have been proposed for
estimation the DOA. One of these, the EigenVector (EV) method. The EigenVector is closely relate to
the MUSIC algorithm. Specifically, the EV method estimates the exponential frequencies from the
peaks of the eigenspectrum:
()
=
1
()
|
2
=+1
(6)
Where
is the eigenvalue associated with eigenvector
.
(): Array steering victor
The only difference between the EV method and MUSIC is the use of inverse eigenvalue (the
are the
noise subspace eigenvalues of R) weighting in EV and unity weighting in MUSIC, which causes EV to
yield fewer spurious peaks than MUSIC. The EV Method is also claimed to shape the noise spectrum
better than MUSIC [9].
V. SIMULATION RESULTS
The DFT, MUSIC, and EV are simulated for estimation the (DOA) using software MATLAB program.
We are using in this paper ULA 10 elements, the distance between the elements is half wavelength
(0.5), and SNR is (30 dB), and snapshot 1000. The figures (2, 3, and 4) shows the estimation of angle
of arrival for single source for three methods.
Fig.2 shows the result of using FFT algorithm. It is estimation the DOA for single source at 20
degree.
A high sidelobe is clear and this is one of the disadvantages of using the FFT method. The peak of main
beam at 20
and its very wide (5
) to(38
). The wide beam causes an ambiguity in estimation the DOA.

This ambiguity causes problem in locate the accurate angle, especially in military application that need
accurate angle. Another disadvantage for the wide beam is causes loss power.

The result shown in fig.3 is that of using MUSIC algorithm. It is able to estimation the DOA for single
source at 20
degree. There is very small (negligible) sidelobe and these are the advantages of MUSIC
method. The main beam of MUSIC method narrow than the main beam in FFT method so it is overcome
the ambiguity in estimation the DOA and loss of the power.

-80 -60 -40 -20 0 20 40 60 80
0
5
10
15
20
25
30
35
40
45
50
FFT
Angles of Arrival

M
a
g
n
i
t
u
d
e

(
d
B
)

Figure 2. DOA Estimation for one source (20) degree using FFT method. (SNR=30dB, N=10, d=0.5)
1085 Vol. 7, Issue 3, pp. 1082-1090

The result shown in fig.4 is that of using EV algorithm. It is capable to estimation the DOA for single
source at 20
degree. There is very small (negligible) sidelobe and these are the advantages of EV
method. The main beam of EV method is very sharp and narrow than the beams in FFT and MUSIC
method so it is overcome the ambiguity in estimation the DOA and give accurate estimation.

The figures (5, 6, and 7) shows the estimation of angle of arrival for two sources for three methods. We
are using the same parameters of the single source. Fig.5 shows the result of using FFT algorithm. It is
able to resolve between two adjacent sources but the difference between them is high more than (20
)
degree. Additionally, a high side lobe is clear and this is one of the disadvantage of the FFT method.
The falling between peak and vale of two sources not enough and equal to about 5dB because the wider
beam of the FFT method.

-80 -60 -40 -20 0 20 40 60 80
0
5
10
15
20
25
30
35
40
MUSIC
Angle of Arrival AOA(degree)

m
a
g
n
i
t
u
d
e
(
d
B
)

-80 -60 -40 -20 0 20 40 60 80
0
5
10
15
20
25
30
35
40
45
Eignvector

m
a
g
n
i
t
u
d
e
(
d
B
)

Figure 3. DOA Estimation for one source (20) degree using MUSIC method. (SNR=30dB, N=10,
d=0.5)
Figure 4. DOA Estimation for one source (20) degree using EV method. (SNR=30dB, N=10, d=0.5)
1086 Vol. 7, Issue 3, pp. 1082-1090

Fig.6 shows the result of using MUSIC algorithm. It is able to recognize between two adjacent sources
(20
, 26
) the deference between them more small than FFT about (6)
degree. There is very small

(negligible) sidelobe and these are the advantages of MUSIC method. The falling between peak and
vale of two sources better than FFT and equal to about 10dB.

Fig.7 shows the result of using EV algorithm. It is able to recognize between two adjacent sources
(20
, 24
) the deference between them smaller than FFT and MUSIC about (4)
degree. There is very

small (negligible) sidelobe and these are the advantages of EV method. The falling between peak and
vale of two sources the best over than FFT and MUSIC methods and equal to about 18dB because it is
very narrow beamwidth.

-80 -60 -40 -20 0 20 40 60 80
0
5
10
15
20
25
30
35
FFT
Angle of Arrival (AOA)

M
a
g
n
i
t
u
d
e

(
d
B
)

-80 -60 -40 -20 0 20 40 60 80
0
5
10
15
20
25
30
35
40
MUSIC

m
a
g
n
i
t
u
d
e
(
d
B
)

Fig.5 DOA Estimation for two sources (20, 40) degree using FFT method. (SNR=30dB, N=10, d=0.5)
Fig.6 DOA Estimation for two sources (20, 26) degree using MUSIC method. (SNR=30dB, N=10, d=0.5)
1087 Vol. 7, Issue 3, pp. 1082-1090

Fig.8 shows the performance of the (FFT, MUSIC, and EV) to recognize between sources when the
SNR was changed at fixed number of elements (10 elements). In FFT, method for low SNR needs high
difference between angles to resolve between sources because of low resolution for this method, but
when SNR high it needs small difference to resolve between sources. For the same parameters, the
result of the MUSIC method is the best when compared with FFT methods. The result of EV, method
is the best when compared with both MUSIC and FFT methods.

Fig.8. DOA Estimation for minimum difference between angles versus SNR.
VI. EXPERIMENTAL RESULTS
The ultrasonic transducers are used in DOA estimation experiment set for single source. The FFT,
MUSIC, and EV methods are used for DOA estimation. Then a comparison is made between high-
resolution and classical methods for different values of the system parameter. The system parameters
are N (number of samples), x (distance between samples and equal to d in equation), Zo (distance
between transmitter and receiver).
In this experiment, the parameters used are N=20 samples Zo=72 cm, x=0.2cm, f=40 MHz, = 0.8
cm, =22
0
(Direction of arrival).
The results shown in figures (9, 10, 11)
-80 -60 -40 -20 0 20 40 60 80
0
5
10
15
20
25
30
35
40
45
50
Eignvector

m
a
g
n
i
t
u
d
e
(
d
B
)

0
5
10
15
20
25
30
35
40
0 5 10 15 20 25 30 35
SNR
Investigation on FFT ,MUSIC ,and EV
FFT EV MUSIC
Fig.7 DOA Estimation for two sources (20, 24) degree using EV method. (SNR=30dB, N=10, d=0.5)
1088 Vol. 7, Issue 3, pp. 1082-1090

Fig.9 DOA estimation for single source using FFT method
Figure (9) demonstrate the FFT method. It can estimation the DOA. High sidelobe level is clear about
-8 dB. In addition, the error in resolution is high and equal to 45 %. The beamwidth of the main beam
are wide compared to other methods and equal to (8
0
). The wide beamwidth give error in estimation
and if there is two sources in this range it is difficult to distinguish between them and consider as one
source.

Fig.10 DOA estimation for single source using MUSIC method

-80 -60 -40 -20 0 20 40 60 80
0
5
10
15
20
25
30
FFT
Direction of Arrival (DOA)
P
o
w
e
r
(
d
B
)
-80 -60 -40 -20 0 20 40 60 80
0
5
10
15
20
25
30
MUSIC

P
o
w
e
r
(
d
B
)

1089 Vol. 7, Issue 3, pp. 1082-1090

Fig.11 DOA estimation for single source using EV method
Figures (10& 11) are shows the results for using MUSIC and EV methods. It is capable to estimate the
DOA with small sidelobe equal to -9.3 dB in MUSIC and -9.5 dB in EV. there is no error in resolution
of estimation. These methods are the best. The performance for the above method can be indicated
using the table (1).
Table (1) comparison between the performances of the DOA estimation methods for single source

VII. CONCLUSIONS
From the simulation results and experimental results about the performance of classical method (FFT)
and modern method (MUSIC, and EV). We can conclude that the classical method is work properly at
high SNR and long data but this work began falling when the data or the SNR is decreasing. In short
data the (FFT) method need high difference between angles of sources to resolve between them because
it is low resolution and high sidelobe level. The modern method (MUSIC and EV) is better than (FFT)
method. For same parameter that used in (FFT) method the (MUSIC and EV) need much smaller
difference between angles of sources to resolve between them and negligible sidelobe because it high
resolution algorithms and the beam width of the main is very narrow and this beam give accurate
estimation of DOA.
VIII. SUGGESTION FOR FUTURE WORK
We are suggestion for future work to realize the experimental results of two sources DOA estimation.
Make investigation performance of three methods for two sources DOA estimation. Compare between
the simulation and experimental results.
-80 -60 -40 -20 0 20 40 60 80
0
5
10
15
20
25
30
Eignvector

P
o
w
e
r
(
d
B
)

Method Estimation Error SLL
FFT 12
0
45% -8 dB
MUSIC 22
0
0% -9.3 dB
EV 22
0
0% -9.5 dB
1090 Vol. 7, Issue 3, pp. 1082-1090

REFERENCES
[1]. Z.Chen, G. Gokeda, Y.Yu, Introduction to Direction of Arrival Estimation, Artech House, 2010
[2]. D.H. Johnson and, D.E. Dudgeon, Array Signal Processing: Concepts and Techniques, Prentice-Hall,
Englewood Cliffs NJ,1993.
[3]. S. Chandran, Advances in Direction-of-Arrival Estimation, Artech House2006.
[4]. V. F. Pisarenko, The retrieval of harmonics from a covariance function Geophysics, J. Roy. Astron.
Soc., vol. 33, pp. 347-366, 1973.
[5]. R. Schmidt, Multiple emitter location and signal parameter estimation. IEEE Transactions on Antenna
and Propagation.Vol. 34, p 276 280, 1986.
[6]. D. H. Johnson and S. R. DeGraaf, Improving the Resolution of Bearing in Passive Sonar Arrays by
Eigenvalue Analysis, IEEE Transactions On Acoustics, Speech, And Signal Processing, Vol. Assp-30,
No. 4, August 1982
[7]. W. Zhu, DOA estimation for broadband signal that use FFT interpolation method, IEEE 4th
International Conference on Software Engineering and Service Science (ICSESS), 2013.
S. F. Cotter, A Two Stage Matching Pursuit Based Algorithm for DOA Estimation in Fast Time-Varying
Environments, IEEE Proc. of the 15th Intl. Conf. on Digital Signal Processing, 2007.
[8]. Z. Xiaofei1, L. Wen, S. Ying, Z. Ruina, and X. Dazhuan, A Novel DOA estimation Algorithm Based
on Eigen Space, IEEE International Symposium on Microwave, Antenna, Propagation, and EMC
Technologies For Wireless Communications,2007.
[9]. M. H. Hayes, "Statistical Digital Signal Processing and Modeling", John Wiley & Sons, Inc., Georgia
Institute of Technology, 2007.

Authors:

Mujahid F. Al-Azzo received the B.Sc. M.Sc. in electrical engineering /electronic and
communication in 1982, and 1985 respectively, and the Ph.D. in communication engineering
in 1999 all from Mosul University/ Iraq. His interest is in the fields of signal processing,
spectral analysis, and direction of arrival estimation.

Khalaf I. Al-Sabaawi received the B.Sc. in electronics engineering / communication
department in 2011, an M.Sc. student from 2012 and until now all from Mosul University/
Iraq. His interest is in the fields of signal processing, spectral analysis, and direction of arrival
estimation.

1091 Vol. 7, Issue 3, pp. 1091-1108

MODELLING LEAN, AGILE, LEAGILE MANUFACTURING
STRATEGIES: AN FUZZY ANALYTICAL HIERARCHY
PROCESS APPROACH FOR READY MADE WARE
(CLOTHING) INDUSTRY IN MOSUL, IRAQ
Thaeir Ahmed Saadoon Al Samman
Department of Management Information System, MOSUL University, Iraq

ABSTRACT
The aim of this research was to develop a methodology for test whether an existing strategies can perform as
lean, agile, or leagile manufacturing strategies. The research the available factors and characteristics should
be defines from the literature to build the model based on was sent to company to a quire their responses, the
AHP was developed to aid decision maker or sort information based on a number of criteria . More after,
questionnaires was built to distribute it to internal and external experts in clothing industry according to their
qualifications. We identifying the manufacturing feature that where particularly by independent variable. By
preparing based on the conditions and characteristics that improved solutions to manufacturing strategies in
the clothing manufacturing company in Mosul and examines the three strategies of lean ,agile, leagile by
considering certain features The paper provide evidence that the choice of manufacturing strategies should be
based upon a careful analysis and characteristics The case study and empirical research reported in this paper
are specific to the clothing manufacturing and fashion industries, ,and there would be benefit in extending the
research into other sector given the increasing trend to the global sourcing and high level of price and high
level competition in clothing manufacturing has market characteristic ,such as short product life cycle ,high
volatility ,low predictability and high level of impulse purchase, highly diverse and heterogeneous making such
issues as quick response of paramount importance whilst there is a growing recognition of the need to match
the competitive advantage to the market ,there is still limited research into what criteria should be aid the
choice of manufacturing strategies ,this paper attempt to extend our understanding of issues .
KEY WORDS: lean, agile, leagile, analytical hierarchy process, fuzzy logic
I. INTRODUCTION
The changes occurring in the nature and level of international competition, many companies have
been resorting to new ways of manufacturing. This phenomenon has been called as new wave
manufacturing strategies [74]. During the development of these various strategies, many kinds were
given to these strategies by the companies which started the development of these strategies and
began implementing them. Some of these strategies were Lean Manufacturing, Agile manufacturing
and leagile manufacturing [6].
1.1. Lean manufacturing conceptual ideology
The term Lean was first used by Krafick[45] in 1988, a principle researcher in the International Motor
Vehicle Program (IMVP) at Massachusetts Institute of Technology (MIT), to describe what today is
known as the Lean Manufacturing or Lean Production paradigm [66]. A major output of the IMVP
research efforts was the publication of the book, The Machine that Changed the World: The story of
lean production [82]. The book chronicled the operations found in the automotive industry, capturing
the dramatic differences in approach and ensuing performance found among the world's leading
1092 Vol. 7, Issue 3, pp. 1091-1108

automakers. In particular, the book examined how the techniques employed by Japanese automakers,
namely Toyota, outpaced the performance achieved by U.S. and European competitors. Much has
been written in the academic and popular business press about Toyota's much envied competitive
weapon, the Toyota Production System (TPS)[27].
Every industrialized country has now fully recognized the importance and benefits of lean ideology
[52]. Lean manufacturing (LM) is it uses half of everything human effort in the factory,
manufacturing space, investment in tools, engineering hours to develop a new product [30] . It makes
use of its tools to strive for zero inventories, zero downtimes, zero defects, and zero delays in the
production process.
The implementation of lean principles in any organization begins by identification of the value
stream, Value is all the aspects of a product that a customer is willing to spend his/her money on it
[27]. i.e., all those activities, required to manufacture a product(goods, services) to a customer. The
numerous activities performed in any organization can be categorized into the following three types
[19]
(i) Value adding Activities (VAA)which include all of the activities that the customer
acknowledges as valuable.
(ii) Non Value Adding Activities (NVAAType II Muda) -These include all the activities that the
customer considers as non-valuable, either in a manufacturing system or in the service system. ,
Waste can be described as the opposite side of value on a Lean coin. These are pure wastes and
involve unnecessary actions that should be eliminated completely.
(iii) Necessary but Non Value Adding Activities (NNVAA - Type I Muda)
These include the activities that are necessary under the current operating conditions but are
considered as non-valuable by the customer[30].
Lean manufacturing focuses on elimination of waste (or "muda" in Japan) Taiichi Ohno believed
that fundamental for any company's success was the elimination of waste. Ohno (1988) developed a
list of seven basic forms of muda
(i) Overproduction (production ahead of demand).
(ii) Transportation (unnecessary transport of products).
(iii) Waiting (waiting for the next production step).
(iv) Inventory (all components, work-in-progress and finished product not being processed).
(v) Motion (unnecessary movement of operator equipment).
(vi) Over processing (unnecessary processing).
(vii) Defects in production (the effort involved in inspecting for and fixing defects)
(viii) Unused creativity[30] .
Womack and Jones (1996) added to this list with the muda of goods and services that fail to meet the
needs of customers.
Lean protest the need for an organization to develop a culture of continuous improvement in quality,
cost, delivery and design [6] Production is agile if it efficiently changes operating states in response
to uncertain and changing demands placed upon it. Production is lean if it is accomplished with
minimal waste due to unneeded operations, inefficient operations, or excessive buffering in
operations. So it is concluded that while agility presumes leanness, leanness might not presume
agility[30]. Lean production can be effectively utilized to remove wastes to improve business
performance. Emphasis on the elimination of loss and the waste of resource factors is largely
associated with lower inventory which is very clearly shown by analyzing the "rock and ship"
analysis. In this analysis, as the stock (water level) is reduced, sources of waste (rocks) appear in the
form of latency, poor quality conformity, prolonged preparations, unreliable processes etc. the
removal of these losses is caused to lower inventory without a negative effect on the flow of materials
(ships). [30][67]. Womack, Jones, and Roos add that in comparison to a mass production approach, a
lean company calls for far less inventory and incurs fewer defects while providing greater variety in
products [27][82[49].
1.2. Agile manufacturing conceptual ideology
The term agile manufacturing refers specifically to the operational aspects of a manufacturing
company which accordingly, try to translate into the ability to produce customized products at mass
production prices and with short lead times.[63,72].
1093 Vol. 7, Issue 3, pp. 1091-1108

There are a number of research reports available in the literature that discuss the concept of agile
manufacturing [27],[19],[7][8],[22],[54], [83],[15], [66],[9],[27]
Agile manufacturing is a new idiom that is used to represent the ability of a producer of goods and
services to survive and flourish in the face of continuous change. These changes can occur in markets,
technologies, business relationships and all other facets of the business enterprise [13].
Agile manufacturing can be defined as the capability of surviving and prospering in a competitive
environment of continuous and random change by reacting quickly and effectively to changing
markets, driven by customer-designed products and services. [42]
According to Naylor et al. (1999), agility means applying market knowledge and a vital corporation
to exploit profitable opportunities in a rapidly changing market place. The relation between agility
and flexibility is extensively discussed in the literature [13]. It has been proposed that the origins
of agility lie in flexible manufacturing systems [21]
Agility can be obtained by systematically developing and gaining capabilities that can make the
supply chain reflect rapidly and diversely to environmental and competitive changes [41].
Consequently, these firms need a number of distinguishing attributes to promptly deal with the
changes inside their environment. Such attributes include four main elements [70]: responsiveness,
competency, flexibility/adaptability and quickness/ speed. The base for agility is the amalgamation of
information technologies, staff, business process organization, innovation and facilities into main
competitive attributes. The main points of the definition of various authors may be summarized as
follow:
-High quality and highly customized products.
- Products and services with high information and value-adding content.
- Recruitment of core competencies.
- Responsiveness to social and environmental issues.
- Combination of diverse technologies.
- Response to change and ambiguity demand.
- Intra-enterprise and inter-enterprise integration [18][16].
The implementation of agile strategies has some benefits for firms, including quick and efficient
reaction to changing market demands; the ability to customize products and services delivered to
customers, the capability to manufacture and deliver new products in a cost-efficient mode [76],
decreased producing costs, enhanced customer satisfaction, elimination of non-value-added activities
and increased competitiveness.
The most publications on agility strategies can classified into four categories:
1. Conceptual models and framework for achieving agility, this mainly includes, dimension [19]
, enablers[23], aspects[85],theoretical model[70] . As well as methodology to support the
implementation agility through identifying drivers and providers[87].
2. Paths to agility , which considered the flexibility practices in terms of "volume flexibility
,modification flexibility and delivery flexibility " and responsiveness from three facets,
volume , product, process are vital paths to agility.[69][35][55][81].
3. Measuring and assessing the performance of agility these include the exploration of rules for
assessment ,the identification of criteria and the establishment of agility index, which was
proposed as a way of measuring the intensity level of agility attributes was
developed.[85][46][84]
4. The development of agility in a supply chain context.[64][34][71][5][86]
Therefore, agility has been advocated as the commerce paradigm of the century and in addition
agility is considered the winning strategy for becoming a general leader in an increasingly competitive
market of quickly changing customers requirements [1][38][17].
Agile manufacturing aims to meet the changing market requirements by suitable alliances based on
core-competencies, by organizing to manage change and uncertainty, and by leveraging people and
information.
[22],[54].
agile manufacturing does not represent a series of techniques much as it represents a elementary
change in management philosophy [48] It is not about small-scale improvements, but an completely
different way of doing business [42] with a primary emphasis on flexibility and quick response to the
changing markets and customer needs.
1094 Vol. 7, Issue 3, pp. 1091-1108

1.3. LEAGILE manufacturing conceptual ideology
Agility in concept comprises; responding to change (predictable or unexpected) in the appropriate
ways and due time, and exploiting and taking advantage of changes, as opportunities. Harrison and
Van Hoek (2005)argue that where demand is volatile, and customer requirements for variety is high,
the elimination of waste becomes a lower priority than responding rapidly to the turbulent
marketplace A similar view is shared by Desai et al., [12]that the lean production philosophy with its
current set of tools will not be able to tackle the increasing demand for customer specific products,
favoring , organizations to move towards more agile production philosophies, considered to be better
suitable to handle customer specific requirements with flexibility and responsiveness. According to
Desai et al , Sharifi and Zhang, [63] [12]
With increasingly customized products, the mass markets will be split into multiple niche markets in
which the most significant requirements will tend to move towards service level. [43]Recognized cost
as the market winner for systems operating on the lean manufacturing philosophy. While Mason Jones
et al. in Kim et al [13]identified service level as the market winner for agile manufacturing
philosophies. In all cases named above costs, quality, and lead time are market qualifiers where they
are not market winners. Rapid changes in the business environment and uncertainty have been part of
management studies and research for a long time, so managing uncertainties still remains one of the
most important tasks for organizations.
Production is agile if it efficiently changes operating states in response to uncertain and changing
demands placed upon it. Production is lean if it is accomplished with minimal waste due to unneeded
operations, inefficient operations, or excessive buffering in operations. So it is concluded that while
agility presumes leanness, leanness might not presume agility.
Lean production is a broad concept and encompasses terms such as flexible manufacturing, mass
customization and even agile manufacturing When lean tools are effectively applied taking into
consideration agility, one can increase flexibility by further introducing safety stocks, or operating
with some free capacity This will ensure that the supply chain is robust to changes in end consumers
requirements [57]. However, Naylor et al [57]warns that leanness and agility are mutually exclusive
and cannot be simultaneously applied at the same point in a supply chain. While leanness operates
best when planning horizons are long and products variants few, agility requires reactivity to customer
orders in short uncertain planning horizons and highly customized product variants. This has resulted
to the coining of a new production philosophy; leagile. [57].
According to Agarwal et al [1], leagility, is a philosophy best suited for an entire supply chain and not
for a single point in the supply chain. Leagile blends the lean and agile. Christopher and Towill
expanded the discussion in Naylor et al. (1999) and Mason-Jones et al. (2000a, 2000b). supported the
concept of hybrid manufacturing strategies and identified three practical ways of combing the lean
and agile paradigms: (Christopher, 2000; [10].
The first is via the Pareto curve approach; adopting lean for the 20% class high volume products
having 80% of the demand and agile for the 80% having 20 % of the demand.
The second is the use of a decoupling point or postponement principle. A de coupling point is a
position in the supply chain where one production paradigm takes over from another [57][40][1].
Since the lean philosophy focuses on cost efficiency along the whole value chain its tools can because
to run operations up to the decoupling point [31], [1]in a cost efficient way. While the agile
production principles are applied on the other side of the decoupling point. But then, there are still
some challenges like determining the position of a decoupling point such that the burden is rightfully
divided across the participants in the supply chain. At the same time it is important to have the
decoupling point closer to the customer so that lean practices can be applied to a greater portion of the
value chain. Since its position depends on end user, lead time sensitivity, and further, on where the
variability is greatest in the supply [57].
The third approach is to separating demand into base and surge demand, using lean for base demand
and agile for surge demand .the term Agile or leagile should remind us that we must be responsive to
change and uncertainties. We may need to come up with a number of other metrics required of our
management systems and processes, all of which may exist within the companies already working at
the level of best practice and based on the same basic good manufacturing practices, but the lean tools
still remain the foundation[11]
1095 Vol. 7, Issue 3, pp. 1091-1108

These three strategies were complementary rather than mutually exclusive, and yet it was likely that
each would work better than others in certain contexts. In addition to decoupling point identification,
there are other approaches to achieve a leagile supply chain. One such method is transshipment,
naturally led to coordinated replenishment policies across locations.
Another approach was proposed by Stratton and Warburton (2003). They argued that leanness and
agility were in practice caught in a trade-off conflict. However, if it were possible to define explicitly
the trade-offs, then a solution to resolve this contradiction was possible. In this way, they interpreted
the conflict nature in terms of dependency, fluctuation, inventory and capacity, and then
systematically linked such trade-offs to develop another approach. This approach was combined with
other approaches: the theory of inventive problem solving (TRIZ for short) and the theory of
constraints (TOC for short). TRIZ of a number of principle-based solution systems was applied to
solve the physical contradictions between leanness and agility while TOC was resolving trade-offs for
highly complementing to TRIZ. However, although there may exist various ways to combine them, it
is significantly important that lean production is a necessary but not a sufficient condition for
achieving agility (Kidd, 1994; Robertson and Jones, 1999; Mason-Jones et al., 2000b;[10]
Furthermore, agility cannot be achieved without experiencing relevant stages of leanness. Mason-
Jones et.al., (2000a) presented two reasons for this fact.
First, lean and agile supply chains share many common features that help speed up the achievement of
leagility. Hence, agility may be initiated by building on the relevant features of leanness.
Second, agility requires control of all processes in the supply chain. It is difficult, if not impossible, to
see how agility can be acquired without having first gone through the process enhancement stage of
lean production.[36].
1.4. Analytical hierarchy process ideology
The Analytic Hierarchy Process method (AHP) was developed by Thomas Saaty in the beginning of
1870s and it represents a tool in the decision making analysis [12].
The author of AHP Thomas L. Saaty called a process, and not a method probably because of the
process character of its elements [38].
Analytic Hierarchy Process (AHP) is originally introduced by Saaty in [4] as a excellent MCDM
(multi criteria decision making) tool which tries to satisfy several conflicting criteria[59]
The AHP technique can evaluate qualitative, quantitative and intuitive criteria comprehensively, and
it is possible to raise the level of confidence of it through carrying out consistency testing. The AHP
technique resembles the structure of human brain, and obtains quantitative results by transforming the
comparative weight between elements to ratio scale. The AHP technique is based on three principles;
hierarchical structuring, weighting, logical consistency.[43]
Analytic hierarchy process (AHP) is a methodological approach which implies structuring criteria of
multiple options into a system hierarchy, including relative values of all criteria, comparing
alternatives for each particular criterion and defining average importance of alternatives.
1.5. FUZZY LOGIC conceptual ideology
The philosophy of Fuzzy Logic (FL) may be traced back to the diagram of Taiji created by Chinese
people before 4600 B.C. But the study of Fuzzy Logic Systems (FLS) began as early as the 1960s. In
the 1970s, FL was combined with expert systems to become a FLS, which imprecise information
mimics a human-like reasoning process. FLS make it possible to cope with uncertain and complex
agile manufacturing systems that are difficult to model mathematically [23]. By opinion of [2], in
managerial practice, there are often situations when it is not enough for managers to rely on their own
instincts. With specific fuzzy programs it is even possible to choose suppliers, service providers or to
buy necessary goods.
II. RESEARCH METHODOLOGY
2.1. Modeling the Manufacturing Strategies Using AHP
Since several strategies can structure a particular manufacturing system which in turn provides certain
strategies (lean, agile, or leagile manufacturing), a value should be obtained based on measuring
1096 Vol. 7, Issue 3, pp. 1091-1108

factors and characteristics factors for this particular manufacturing system in order to identify the
strategies. Therefore for a proper decision to be made, these factors modeling using AHP as shown in
figure (1)

Figure 1. A Propose Model
Elimination of
Waste
Manufacturing
Capability
Market
Sensitiveness
Information Technology
(Information Driver)
Flexibility
The Most appropriate
Manufacturing System
Lead Time Cost Quality Service Productivity

Delivery Speed (DS)
New Product Introduction
(NPI)
Customer Responsiveness (CR)

Electronic Data Interchange (EDI)
Mean of Information (MOI)
Data Accuracy (DA)
Data and Knowledge Bases (DKB)

Source Flexibility (SF)
Manufacturing Flexibility (MF)
Delivery Flexibility (DF)
Overproduction (OP)
Inventory Waiting (IT)
Knowledge Misconection (KM)

Lean Manufacturing
Agile Manufacturing
LeAgile Manufacturing
1097 Vol. 7, Issue 3, pp. 1091-1108

The main measuring factors for lean, agile and leagile strategies are depended on five measures (lead
time, cost, quality, productivity, service level) and A characteristic can be defined as the feature of
the property which is obtained by considering several parameters(Elimination Of waste(Over-
production , Inventory transportation waiting , Knowledge Misconnects) , Flexibility (Manufacturing
flexibility, Delivery flexibility , Source flexibility), Information Technology(Electronic data
interchange, Means of information and data accuracy , Data and knowledge bases ) Market
Sensitivity(Delivery speed , New product introduction, Customer responsiveness). Hence the
manufacturing system in a described state performs under closely specified conditions that produce a
metric value. [41]
2.2.Data collection
We was chosen General Company for Readymade Wear Manufacturing in Mosul, Iraq, a field of
study, were selected by A committee composed of three directors of departments productivity in the
General Company for the manufacture ready-made clothes in city Mosul in Iraq are the Director of
the Section productive first (line productive - dishdasha) and Director of the Section productive
second (line productive Aalghemsalh) and Director of the Section productive third line productive
(Altruakh) "They evaluates certified manufacturing methods in the company through the scaling
factors of (Lead time, Cost, Quality, Productivity, Service )Where the questionnaire was designs to
seek expert opinion about the requires rating for implementing lean, agile and leagile manufacturing
strategies in industries. The opinions provide the necessary data which are captured from internal and
external experts according to their qualifications. For this purpose to indicate the relative importance
of each criterion according to standard corresponding giving weight relative (1-9) was unloading The
composed data is adjusted using Expert Choice software which is a multi-objective decision support
tool based on the Analytic Hierarchy Process (AHP). and use CGI program ,the feedback data input of
the five measuring factors that were filled out in the questionnaire is shown in table(1)
Table(1):feedback data input of the five measuring factors
metric Production line
1
Production
line2
Production
line3
mean
Lead time 5,7,9 3,5,7 5,7,9 .833 .633 .433
cost 3,5,7 1,3,5 3,5,7 .633 .43 .233
Quality 1,3,5 3,5,7 1,3,5 .566 .366 .166
productivity 5,7,9 3,5,7 5,7,9 .633 .43 .233
Service level 3,5,7 1,3,5 3,5,7 .833 .633 .43
Then , the feedback data input of characteristics factors as shown in tab(2),(3),(4),(5),(6) which
demonstrates the manufacturing performance for the( lead time, cost, quality ,productivity, service
level) with respect to (lean, agile and leagile) manufacturing strategy. a consistency ratio was
calculated by the software to check the applicability of the paired comparisons The value consistency
ratio should be 10 percent or less.
Therefore, all the consistency ratio of the below table is less than 10 %.
III. ANALYSIS AND DISCUSSION
They assess the manufacturing methods adopted in the company through the measurement criteria of (
lead time, cost, quality, productivity, level service) according to the logic FUZZThe linguistic terms
are used to assess the performance rating and importance weights of the integration values since it is
difficult for experts to determine the score of vague values such as training level of personnel [80].
Most of the time linguistic scales include 2 scales [46]. The linguistic variables {Excellent [E], Very
Good [VG], Good [G], Fair [F], Poor [P], Very Poor [VP], Worst [W]} were selected to assess the
performance rating of the integration capability (these are the examined coefficients). Then the
linguistic variables {Very High [VH], High [H], Fairly High [FH], Medium [M], Fairly Low [FL],
Low [L], Very Low [VL]} were selected to assess the importance weights of the integration
capabilities. Using the previous studies [80, 46], the table of fuzzy numbers for the linguistic variable
values was created .
1098 Vol. 7, Issue 3, pp. 1091-1108

The results were obtained as shown in the table (1) which represents the input data feedback, it is
worth noting here that he has been using the program ready minitab v23. For the year 2013 in getting
the results of multiplying matrices while the program is used gci to get the values of functions
belonging according to the logic FUZZY and calculate the eigine value, as well as index consistency,
as shown in Tables (2), (3), (4), (5), (6), (7)
2.2.1. lead time
Tab(2) demonstrates the characteristics factor for the measuring factor of lead time
Tab.(2) Evaluation model to lead time
Measuring
factors
Characteristics
Factors
Sub
characteristics
D1 D2 D3 LEAN AGILE LEAGILE
Lead time Elimination
waste

Over-production 1,3,5 3,5,7 1,3,5 .566 .366 .166
Inventory
transportation
waiting
1,3,5 1,3,5 1,3,5 .5 .3 .1
Knowledge
Misconnects
1,3,5, 3,5,7 5,7,9 .7 .5 .3
Flexibility
Manufacturing
flexibility,
1,3,5 3,5,7 1,3,5 .566 .366 .166
Delivery
flexibility
1,3,5 1,3,5 3,5,7 .566 .366 .166
Source
flexibility
3,5,7 1,3,5 3,5,7 .633 .433 .433
Information
technology

Electronic data
interchange
5,7,9 3,5,7 5,7,9 .566 .633 .233
Means of
information and
data accuracy
3,5,7 1,3,5 3,5,7 .633 .366 .166
Data and
knowledge bases
1,3,5 3,5,7 1,3,5 .633 .366 .166
Market
sensitivity

Delivery speed 1,3,5 3,5,7 3,5,7 .633 .166 .233
New product
introduction
1,3,5 1,3,5 1,3,5 .166 .3 .1
Customer
responsiveness
1,3,5 1,3,5 1,3,5 .5 .3 .1
2.33,4.33,6.33 2,4,6 1.58,3.41.5.25
2.2.2. COST
Moreover, tab(3) demonstrates the characteristics factor for the measuring factor of cost
Tab.(3) Evaluation model of the cost
Measuring
factors
Characteristics
Factors
Sub
characteristics
cost Elimination
waste

Over-production 5,7,9 1,3,5 5,7,9 .766 .566 .366
Inventory
transportation
5,7,9 3,5,7 5,7,9 .833 .633 .433
1099 Vol. 7, Issue 3, pp. 1091-1108

waiting
Knowledge
Misconnects
5,7,9 3,5,7 5,7,9 .833 .633 .566
Flexibility
Manufacturing
flexibility,
5,7,9 3,5,7 3,5,7 .766 .566 .366
Delivery
flexibility
3,5,7 1,3,5 3,5,7 .633 .433 .233
Source
flexibility
3,5,7 1,3,5 1,3,5 .566 .366 .166
Information
technology

Electronic data
interchange
3,5,7 3,5,7 1,3,5 .566 .433 .233
Means of
information and
data accuracy
3,5,7 1,3,5 3,5,7 .566 .433 .233
Data and
knowledge bases
3,5,7 1,3,5 1,3,5 .566 .366 .166
Market
sensitivity

Delivery speed 3,5,7 1,3,5 1,3,5 .566 .366 .166
New product
introduction
5,7,9 3,5,7 3,5,7 .766 .566 .366
Customer
responsiveness
3,5,7 1,3,5 1,3,5 .566 .366 .166
3.83,5.83,7.83 4.08,3.83,5.83 2.66,4.66,6.66
2.2.3. quality
Furthermore, tab(4) shows the characteristics factor for the measuring factor of quality.
Tab.(4) Evaluation model of the quality
Measuring
factors
Characteristics
Factors
Sub
characteristics
Quality Elimination
waste

Over-production 3,5,7 1,3,5 3,5,7 .633 .433 .233
Inventory
transportation
waiting
1,3,5 1,3,5 1,3,5 .5 .3 .1
Knowledge
Misconnects
3,5,7 3,5,7 1,3,5 .633 .433 .233
Flexibility
Manufacturing
flexibility,
1,3,5 5,7,9 3,5,7 .7 .5 .3
Delivery
flexibility
3,5,7 5,7,9 1,3,5 .7 .5 .366
Source
flexibility
1,3,5 3,5,7 3,5,7 .633 .433 .3
Information
technology

Electronic data
interchange
3,5,7 5,7,9 3,5,7 .766 .566 .366
Means of
information and
data accuracy
3,5,7 5,7,9 1,3,5 .7 .5 .3
Data and 1,3,5 1,3,5 5,7,9 .633 .433 .233
1100 Vol. 7, Issue 3, pp. 1091-1108

knowledge bases
Market
sensitivity

Delivery speed 5,7,9 3,5,7 5,7,9 .833 .633 .433
New product
introduction
5,7,9 5,7,9 3,5,7 .823 .633 .433
Customer
responsiveness
3,5,7 5,7,9 3,5,7 .766 .566 .366
2.66,4.66,6.66 3.5,5.5,7.5 2.66,4.66,6.66

2.2.4. Productivity
In addition ,tab(5) demonstrates the characteristics factor for the measuring factor of productivity.
Tab.(5) Evaluation model of the productivity
Measuring
factors
Characteristics
Factors
Sub
characteristics
Productivity Elimination
waste

Over-production 3,5,7 5,7,9 3,5,7 .766 .566 .366
Inventory
transportation
waiting
3,5,7 5,7,9 3,5,7 .766 .566 .366
Knowledge
Misconnects
3,5,7 3,5,7 3,5,7 .7 .5 .3
Flexibility
Manufacturing
flexibility,
1,3,5 5,7,9 5,7,9 .766 .566 .366
Delivery
flexibility
5,7,9 3,5,7 5,7,9 .833 .633 .433
Source
flexibility
3,5,7 5,7,9 3,5,7 .766 .566 .366
Information
technology

Electronic data
interchange
5,7,9 3,5,7 5,7,9 .833 .633 .433
Means of
information and
data accuracy
3,5,7 5,7,9 3,5,7 .766 .566 .366
Data and
knowledge bases
5,7,9 1,3,5 5,7,9 .766 .566 .366
Market
sensitivity

Delivery speed 3,5,7 5,7,9 3,5,7 .766 .566 .366
New product
introduction
5,7,9 3,5,7 3,5,7 .266 .566 .366
Customer
responsiveness
1,3,5 1,3,5 3,5,7 .566 .366 .166
3.66,5.66,7.66 3.66,5.66,7.33 3.33,5.33,7.33
2.2.5. Service level
In addition, tab(6) demonstrates the characteristics factor for the measuring factor of service level
Tab.(6) Evaluation model of the service level
Measuring
factors
Characteristics
Factors
Sub
characteristics
1101 Vol. 7, Issue 3, pp. 1091-1108

Service
level
Elimination
waste

Over-production 1,3,5 3,5,7 1,3,5 .566 .366 .166
Inventory
transportation
waiting
1,3,5 1,3,5 3,5,7 .566 .366 .166
Knowledge
Misconnects
3,5,7 5,7,9 3,5,7 .766 .566 .366
Flexibility
Manufacturing
flexibility,
5,7,9 5,7,9 3,5,7 .833 .633 .433
Delivery
flexibility
3,5,7 1,3,5 3,5,7 .633 .433 .233
Source
flexibility
3,5,7 1,3,5 3,5,7 .633 .433 .233
Information
technology

Electronic data
interchange
5,7,9 3,5,7 5,7,9 .833 .633 .433
Means of
information and
data accuracy
5,7,9 5,7,9 3,5,7 .766 .633 .433
Data and
knowledge bases
1,3,5 1,3,5 3,5,7 .566 .366 .166
Market
sensitivity

Delivery speed 3,5,7 1,3,5 3,5,7 .633 .433 .233
New product
introduction
5,7,9 3,5,7 1,3,5 .7 .5 . 3
Customer
responsiveness
1,3,5 3,5,7 1,3,5 .566 .366 .166
2.66,4.66,6.66 2.66,4.66,6.66 3,5,7

After entering the input ,the data output of the minitab program is shown in table(7) normalization all
the above means of characteristics factors by dividing by 10.(saaty,t.l.,1980) After data processing
methods adopted in the analytical hierarchy analysis, the outputs appear in the table (7) where the
data is converted to a standard format as shown in the table(7).
Tab(7)normalized of the measure factors means of the three decision maker
Service level productivity quality

.366 .566 .766 .266 .466 .666 .266 .466 .666
.366 .566 .733 .35 .55 .75 .2 .4 .6
.333 .533 .733 .266 .466 .666 .158 .341 .525

Cost lead time

.233 .433 .633 .266 .466 .666
.200 . 4 . 6 .266 .466 .666
.158 .341 .525 .383 . 583 .783

1102 Vol. 7, Issue 3, pp. 1091-1108

The proportion of consistency 10% between the vertebrae variables was found accurate comparisons
of marital and described in tables (8) (9) (10) Matrix comparisons that show the importance of marital
reliability of lean, agile, leagile manufacturing between each of the criteria.
Tab (8) Pairwise Comparison Matrix which appear lean manufacturing between index factor
Lead
time
cost quality productivity Service level
Lead time 1 0.142857 0.25 0.2 3
Cost
7 1 1 9 7
Quality

4 1 1 3 5
Productivity

5 0.111111 0.333333 1 5
Service level 0.333333 0.142857 0.2 0.2 1
Tab (9)Pairwise Comparison Matrix which appear agile manufacturing between index factor

Lead
time
Cost Quality Productivity Service level
Lead time 1 9 9 7 7
Cost 0.111111 1 7 5 3
Quality 0.111111 0.142857 1 5 5
Productivity 0.142857 0.2 0.2 1 0.142857
Service
level
0.142857 0.333333 0.2 7 1

Tab (10) Pairwise Comparison Matrix which appear reliability for leagile manufacturing between index
factor

Lead
time
Cost Quality productivity
Servic
e level
Lead time 1 5 7 5 3
Cost
0.2 1 9 7 5
Quality
0.142857 0.111111 1 5 7
Productivity

0.2 0.142857 0.2 1 0.142857
Service level

0.333333 0.2 0.142857 7 1

1103 Vol. 7, Issue 3, pp. 1091-1108

Then, we calculate standard pair wise comparison matrix for lean, agile, leagile manufacturing, which
appear in tables (11),(12),(13)
Tab(11) Synthesized matrix for lean manufacturing
Lead
time
Cost Quality Productivity

Service
level
Priority weight
Eigen value
Lead time .o57 0.058 .089 .014 .142 .072
Cost .403 .418 .359 .671 .333 .436
Quality .230 .418 .359 .223 .238 .293
productivity .288 .046 .118 .074 .238 .152
Service level .019 .058 .071 .014 .047 0.041
The principal eigenvector is necessary for representing the priorities associated with that matrix,
providing that the inconsistency is less than or equal to a desired value [62].The value which appear in
the last Colum represent relative weight to lean manufacturing to each criteria's (lead time ,cost
,quality, productivity, service level),now estimating the consistency ratio, is as follow:
Tab(12)Synthesized matrix for agile manufacturing
Lead time Cost Quality Productivity

Service
level
Priority
weight
Eigen value
Lead time .666 0.843 .517 .28 .433 .547
Cost .073 .093 .402 .2 .185 .190
Quality .073 .013 .057 .28 .309 .146
productivity .093 .018 .011 .04 .008 .034
Service level .093 .030 .011 .28 .061 0.095
Tab(13) Synthesized matrix for leagile manufacturing
Lead
time
Cost Quality Productivity

Service
level
Priority weight
Eigen value
Lead time .534 0.775 .403 .2 .185 .419
Cost .106 .155 .519 .28 .309 .273
Quality .074 .017 .057 .2 .433 .156
productivity .106 .021 .011 .04 .008 .037
Service level .176 .031 .008 .028 .061 0.111

The result in the tab(14 ) summarize the relative weights of different types of manufacturing strategies
according to each criterion factor measurements (lead time, cost, quality, productivity, service),
according to lean , agile, leagile manufacturing strategies.
Tab (14) Relative weight of different type of manufacturing strategies
Measurement factor Lean Agile Leagile
Lead time .072 .547 .419
Cost .436 .190 .273
Quality .293 .146 .156
Productivity .152 .034 .037
Service level .041 .095 .111
sum .994 1.012 .996
Mean .1988 .2024 .1992
The normalized means of Tab (7) is multiplied by the means of feedback data input of the five
measuring factor Tab(1) to obtain the following: the result of multiplication is tab(15)

1104 Vol. 7, Issue 3, pp. 1091-1108

Tab(15) Result of multiplication Tab(1) and Tab(1)
average
lean 2.539 1.786 1.156 .376 .593 1.284
agile 2.492 1.753 1.135 .369 .582 1.266
leagile 2.391 1.682 1.089 .354 .558 1.210

then applying the equation :
=
1.2530.18
3.520.18
= 0.321

Where 1.253 is the result of multiplication and .18 ,3.52are constant . The resulting (.321) is
multiplied by the certainty constant(.70) to get (.224)
Fig (2) is consulted to conclude that readymade ware in Mosul is below (.2-.10) which
represents the lean and bellow (.2924-.1124) agile and bellow(.293-.109) leagile baseline the
readymade ware in Mosul is traditional manufacturing. In order to facilitate its evolution to
lean manufacturing a should implement the following tools (cellular manufacturing, TQM,
value stream mapping ,5s, kaizen )

1

.818

1.818 3.106 7.332
- cut
2.67 4.159
Triangular fuzzy number
Fig(2) fuzzy number to conclude manufacturing strategy
The results above summarizes measurements factors (lead time, cost, quality, productivity, service
level) in addition to convert all values to the standard according to expert opinion .for the experts
opinion appear in the tab (16).
Tab (16) mean fuzzy number of opinions experts
Expert opinion Lean agile leagile traditional
Total mean .2888-.1088 .1124-.2924 .1092-.293 0
we can calculate the consistency ratio which may be less 10% for comparison the fuzzy expert
opinion number the tab(16)appear the mean fuzzy number of the opinion experts
IV. CONCLUSIONS
1- Enhanced manufacturing strategy performance implies that a manufacturing is quickly
responding to the market volatile the customer demand with effective cost reduction.
1105 Vol. 7, Issue 3, pp. 1091-1108

Leanness in a manufacturing maximizes profits through cost reduction while agility
maximizes profit through providing exactly what the customer requires. The leagile strategy
enables the upstream part of the chain to be cost-effective and the downstream part to
achieve high service levels in a volatile marketplace.
2- The AHP methodology adopted here arrives at a synthetic score, which may be quite useful
for the decision-makers. The purpose of the present work is to analyze the relative impact of
different enablers on three strategic paradigms considered for manufacturing.
3- It integrates various criteria, enablers and alternatives in decision model. The approach
also captures their relationships and interdependencies across and along the hierarchies. It is
effective as both quantitative and qualitative characteristics can be considered
simultaneously without sacrificing their relationships.
4- The manufacturing system strategy of readymade ware company is traditional
manufacturing.
5- To resolve the manufacturing system in order to become lean, agile or leagile; a lot of tools
will help in becoming lean like Cellular Manufacturing, Total Quality Management ,Poka
yoke, Kaizen , Value Stream Mapping, 5 S, increase its focus on customer service and
improve the quality of its. Also, some tools will help in becoming agile like Customer Value
Focus, IT.
6- The paper provide evidence that the choice of manufacturing strategies should be based
upon a careful analysis and characteristics.
REFERENCES
[1]. Agarwal, A., Shankar, R. & Tiwari, M. K., 2006, Modeling the metrics of lean, agile and leagile supply
chain: An ANP-based approach. European Journal of Operational Research, 173, pp. 211-225.
[2]. Agbor, Tanyi Emmanuel, 2008, An evaluation of production operations at Supply Chain Based on
Lean Concept, A thesis Report submitted to Department of Production Engineering
and Management, partially fulfilling the requirements for the award of the degree of Master of Sciences
in Production Engineering and Management, Royal Institute at Technology Stockholm, Sweden.
[3]. Association for Manufacturing Excellence, 1994, Forward to the Future: The Transition to Agile,
Quick Response Manufacturing Textile/Clothing Technology Corporation [TC]2 shows how to step
sprightly, IJ Textile/Clothing Technology Corporation_ National Apparel Technology Center, Target
Vol. 10, Number 4, Illinois, pp. 41-44.
[4]. Baker, J., 1996, Less lean but considerably more agile. Special section mastering management.
Financial Times, 10 May, Systems 2, pp. 8386.
[5]. Baramichai, M., Zimmer's, E. W. & Marangos, C. A., 2007, Agile supply chain transformation matrix:
an integrated tool for creating an agile enterprise, Supply Chain Management-an International Journal,
12, pp. 334-348.
[6]. Bhasin ,S. and P. Burcher, 2006, Lean Viewed as a Philosophy, Journal of Manufacturing Technology
Management, 17 (1), pp. 56-72.
[7]. Booth, R., 1996, Agile manufacturing, Engineering Management Journal, 6 (2), pp. 105-112.
[8]. Bunce, P. & Gould, P., 1996, From lean to agile manufacturing, IEE Colloquium (Digest), No. 278, pp.
3/1-3/5.
[9]. Cho, H., Jung, M. & Kim, M., 1996, Enabling technologies of agile manufacturing and its related
activities in Korea, Computers & Industrial Engineering, 30 (3), pp. 323-334.
[10]. Christopher, M. & Towill, D. R., 2001, An integrated model for the design of agile supply chains, An
International Journal of physical distribution and logistics management, 31(4), pp. 235
246.
[11]. Dean, Steven P., &Henderson, Ian, 2003, Lean ,Agile and Simple . iomnet.org.uk.
[12]. daiDes. S., Irani, Z. & Sharp, J. M., 1999, Working Towards agile manufacturing in the UK
Industry, International Journal of production and economics,62,155-169.
[13]. Devor, R., Graves, R.& Mills, J. J., 1997,Agile Manufacturing Research: accomplishments and
opportunities. IIE Transactions, 29 (10), pp. 813-823.
[14]. Esmail, K. & Saggu, J. A., 1996, Changing paradigm, Manufacturing Engineer December, pp. 285
288.
[15]. Forsythe, C., Ashby, M. R., 1996, Human factors in agile manufacturing, Ergonomics in Design 4 (1),
pp. 1521.
1106 Vol. 7, Issue 3, pp. 1091-1108

[16]. Gharakhani, Davood. Maghferati, Amid Pourghafar. Farahmandian, Arshad and Nasiri, Rasol.( 2013)
Agile manufacturing, Lean production, Just in Time systems and products quality improvement, Life
Science Journal;10(3s) . http://www.lifesciencesite.com
[17]. Ghatari,Ali,Rajabzadeh,GholamhosseinMehralian&ForouzandehZarenezhad, 2012, Developing a
model for Agile Pharmaceutical manufacturing: Evidence from Iran, Australian Journal of Basic and
Applied Sciences, 6(9), pp. 753-762.
[18]. Goldman, S.L. Nagel, R.N. (1993) Management, technology and agility: The emergence of a new era
in manufacturing, International Journal of Technology Management 8 (1/2) 18-38.
[19]. Goldman, S. L., Nage R. N. &Preiss K., 1995, Agile Competitors and Virtual Organizations: Strategies
for Enriching the Customer, Van Nostrand Reinhold Company, New York.
[20]. Goldshy, Thomas J., Stanley E. Griffiths & Adthony, S. Roath, 2006, Modeling Lean, Agile, and
Leagile supply chain strategy, Business Logics, Vol. 27, No. 1, pp. 57-80.
[21]. Gosling, J., Purvis, L., & Naim, M., 2010, Supply chain flexibility as a determinant of supplier
selection, International Journal of Production Economics, pp. 11-21.
[22]. Gould, P., 1997, What is agility, Manufacturing Engineer 76 (1), pp. 2831.
[23]. Gunasekaran, A., 1998, Agile manufacturing: Enablers and an implementation framework,
International Journal of Production Research, 36 (5), pp. 1223-1247.
[24]. Gunasekaran, A., 1999a, Agile manufacturing: A framework for research and development,
International Journal of Production Economics, 62 (1-2), pp. 87-105.
[25]. Gunasekaran, A., 1999b, Design and implementation of agile manufacturing systems, International
Journal of Production Economics, 62 (1-2), pp. 1-6.
[26]. Gunasekaran, A., 2001, Agile manufacturing: The 21
st
Century Competitive Strategy, Elsevier Science
Ltd, Oxford, U.K.
[27]. Gunasekaran, A., Tirtiroglu, E. &Wolstencroft, V.,2002, An investigation into the application of agile
manufacturing in an aerospace company, Technovation, 22 (7), pp. 405-415.
[28]. Gunasekaran, A., Yusuf, Y. Y., 2002, Agile manufacturing: A taxonomy of strategic and technological
imperatives, International Journal of Production Research, 40 (6), pp. 1357-1385.
[29]. Gunasekaran, A., K., Lai & T. C. E., Cheng, 2008, Responsive supply chain: A competitive strategy in
networked company, Omega, 36(4): pp. 549.
[30]. Gupta, Anil, & Kundra, T. K., 2012, A review of designing machine tool for leanness, Sadhana, Vol.
37, Part 2, April, Indian Academy of Sciences, pp. 241259.
[31]. Harrison, A. & Van Hoek, R., 2005, Logistics Management and strategy, London, Prentice Hall.
[32]. Hillman-Willis, T., 1998, Operational competitive requirements for the 21
st
century, Industrial
Management & Data.
[33]. Hitomi, K., 1997, Manufacturing strategy for future production moving toward manufacturing
excellence, International Journal of Technology Management 14 (68), pp. 701711.
[34]. Hoek, R. I., Harrison, A. & Christopher, M., 2001, Measuring agile capabilities in the supply chain,
International Journal of Operations & Production Management, 21, pp. 126-147.
[35]. Holweg, M., 2005, The three dimensions of responsiveness, International Journal Operations &
Production Management, 25, pp. 603-622.
[36]. Huang, Yu-Ying &Shyh-Jane Lib, 2009, Tracking the Evolution of Research Issues on Agility, Asia
Pacific Management Review 14(1), pp. 107-129.
[37]. Ishizaka, Alessio & Labib Asharaf, 2011, Review of the main developments in the analytic hierarchy
process, Expert Systems with Applications, 38 (11), pp. 14336-14345.
[38]. Ismail, H., I. Raid, J. Mooney, J. Pool ton & I. Arokiam, 2007, How small and medium enterprises
effectively participate in the mass customization game, IEEE transactions on engineering management,
54 (1), pp. 86-97.
[39]. Jones, D. T., 2003, The Beginner's Guide to Lean, Chairman Newsletter Lean Enterprise Academe,
November.
[40]. Kay, J., M & Prince J., 2003, Combining lean and agile characteristics: Creation of virtual groups by
enhanced production flow analysis, International Journal of production economics 85, pp. 305-318.
[41]. Khan, Arif, K., Bakkappa, B., Bhimaraya, A.M., and Sahay, B.S. , 2009 .Impact of agile supply chains
delivery practices on firms performance: cluster analysis and validation. Supply Chain Management:
An International Journal, 14(1): 41-48
[42]. Kidd, P. T., 1996, Agile manufacturing: a strategy for the 21
st
century, IEE Colloquium (Digest) 74,
6IEE, Steven age, England.
[43]. Kim, Dong Jun & Chung, Sung Bong, 2005, Development of an assessment model using AHP
technique for rail road projects experience service conflicts in Korea, Proceedings of the Eastern Asia
Society for Transportation Studies, Vol. 5, pp. 2260 2274.
1107 Vol. 7, Issue 3, pp. 1091-1108

[44]. Kim, S. W., Narasimhan, R. &Swink, M., 2006, Disentangling Leaness and agility: An empirical
investigation. Journal of operations management, 24, pp. 440-457.
[45]. Krafcik, J., Triumph of the Lean Production System. Sloan Management Review, 30(1), pp. 4152.
[46]. Lin, C. T., Chiu, H. &Chu, P. Y., 2006, Agility index in the supply chain. International Journal of
Production Economics, 100, pp. 285-299.
[47]. Maskell, B. H., 1994, Software and the Agile Manufacturer: Computer Systems and World Class
Manufacturing, Oregon Productivity Press, Portland.
[48]. Maskell, B. H.,1991, Performance Measurement for World Class Manufacturing, Productivity Press,
Cambridge, MA.
[49]. Mason-Jones, R., Naylor, B. &Towill, D. R., 2000. Lean, agile or leagile? Matching your supply chain
to the marketplace, International Journal of Production Research, 38, pp. 4061-4070.
[50]. Masoud, 2007, Decision Support System for Lean, Agile and Leagile Manufacturing, Master of
Science in the Industrial Engineering Department with the College of Engineering, King Saud
University, p. 27-30.
[51]. Meier, R. L. & Walker, H. F., 1994, Agile manufacturing, Journal of Industrial Technology, 10 (4), pp.
4143.
[52]. Meyer A. De & Wittenberg-Cox A., 1992, Creating product value: Putting manufacturing on
the strategic agenda, London: Pitman.
[53]. Monden, Y., 1998, Toyota production system: An integrated approach to just-in-time, 3
rd
ed., Norcross,
Georgia: Engineering and Management Press.
[54]. Moore, James S. M. R., 1996, Agility is easy, but effective agile manufacturing is not, IEE Colloquium
(Digest), No. 179, 4.
[55]. Narasimhan, R. & Das, A., 1999, Manufacturing agility and supply chain management practices,
Production and Inventory Management Journal, No. 40, pp. 4-10.
[56]. Narasimhan, R. & Das, A., 2000, An empirical examination of sourcing's role in developing
manufacturing flexibilities, Intl Journal of Prod Research, 38, pp. 875-893.
[57]. Naylor, J. B., M. M. Naim& D. Berry, 1999, Leagility: integrating the lean and agile manufacturing,
International Journal of Production Economics, 62, pp. 155-169.
[58]. Papadopoulou, T. C. & M. zbayrak, 2005, Leanness: Experiences from the Journey to Date, Journal
of Manufacturing Technology Management, 16 (7).
[59]. Pogarcic, Ivan, Francic, Miro, &Davidovic, Vlatka, 2008, Application of AHP method in traffic
planning, ISEP, Croatia.
[60]. Preiss, K., Goldman, S. & Nagel, R., 1996, Co-operate to Compete. Van Nostrand Reinhold, New
York.
[61]. EZ, J., 2006, Modern management, Computer Press, Brno: 2009. 397.
[62]. Saaty, T. L., L. Vargas, 1984, Inconsistency and rank preservation, Journal of Mathematical
Psychology 28 (2).
[63]. Schonberger, R. J., 1982, Japanese Manufacturing Techniques: Nine Hidden Lessons in Simplicity,
Free Press, New York.
[64]. Schonsleben, P., 2000, With agility and adequate partnership strategies towards effective logistics
networks, Computers in Industry, 42, pp. 33-42.
[65]. Sen, Rahul, Kumar, Saurabh& S., Chandrawat, 2011, The total supply chain: Transportation System in
mine planning, Faculty of mining and geology, International journal of production economics,62, pp.
107-118.
[66]. Sen, Rahul, Kumar, & Saurabh S. Chandrawat R. , 2012, The Framework for Manufacturing Methods
and Techniques, International Journal of Engineering Research and Applications (IJERA),
www.ijera.com Vol. 2, Issue 6, November- December, pp.697-702.
[67]. Seyedi, Seyed Nima, 2012, Supply Chain Management, Lean Manufacturing, Agile Production, and
their Combination, International Journal of Contemporary Research in business, Institute of
Interdisciplinary Business Research, Vol. 4, No. 8, ijcrb.webs.com, pp. 648-653.
[68]. Sharifi, H. & Zhang Z., 1999, A methodology for achieving agility in manufacturing organizations, An
introduction International Journal of production economics, No. 62, pp. 7-22.
[69]. Sharifi, H. & Zhang, Z., 2001, Agile manufacturing in practice - Application of a methodology,
International Journal of Operations & Prod Management, No. 21, pp. 772-794.
[70]. Sharp, J. M., Irani, Z. & Desai, S., 1999, Working towards agile manufacturing in the UK industry,
International Journal of Production Economics, No. 62, pp. 155-169.
[71]. Shaw, N. E., Burgess, T. F., De Mattos, C. &Stec, L. Z., 2005, Supply chain agility: the influence of
industry culture on asset capabilities within capital intensive industries, International Journal of
Production Research, No. 43, pp. 3497-3516.
[72]. Slack, N., 1991, The Manufacturing Advantage, Mercury Books, London.
1108 Vol. 7, Issue 3, pp. 1091-1108

[73]. Slack, N., Chambers, S., Harland, C., Harrison, A. & Johnston, R., 1995, Operations Management,
Pitman, London, England.
[74]. Story, John, 1994,Flexible manufacturing systems, New wave manufacturing strategies, Paul chapman
publishing Ltd., London.
[75]. Swafford, P. M., Ghosh, S. & Murthy, N., 2006, The antecedents of supply chain agility of a firm:
scale development and model testing, Journal of Operations Management, Vol. 24, pp. 170-188.
[76]. Swafford, P. M., S., Ghosh & N., N. Murthy, 2006, A framework for assessing value chain agility,
International Journal of Operations and Production Management, Vol. 26, No. 2, pp. 118-
140.
[77]. Swafford, P. M., S., Ghosh & N., N. Murthy, 2008, Achieving supply chain agility through IT
integration and flexibility, International Journal of Production.
[78]. Swafford, P., 2003, Theoretical development and empirical investigation of supply chain agility, Ph. D.
thesis, The Georgia Institute of Technology, Georgia, U.S.A.
[79]. Syamsuddin, Irfan&Junseok, Hwang, 2009, The Application of AHP Model to Guide Decision
Makers: A Case Study of E-Banking Security, Fourth International Conference on
Computer Sciences and Convergence Information Technology.
[80]. Ustyugova, Tatiana, Darja Noskievicova, CSc., 2013, fuzzy logic model for evaluation of lean and
agile manufacturing integration, 15. - 17. 5., Brno, Czech Republic, EU.
[81]. Van Hoek, R. I., Harrison, A. & Christopher, M., 2001, Measuring agile capabilities in the supply
chain, International Journal of Operations & Production Management, No. 21, pp. 126-147.
[82]. Womack, J. P., D. T. Jones & D. Roos, 1990, The Machine that Changed the World: The Story of Lean
Production, Rawson Associates, New York, NY.
[83]. Womack, J. P., Jones, D. T., 1996, Lean Thinking: Banish Waste and Create Wealth in Your
Corporation, Simon and Schuster, New York.
[84]. Yusuf, Y. Y., Gunasekaran, A., Adeleye, E. O. & Sivayoganathan, K., 2004, Agile supply chain
capabilities: determinants of competitive objectives, European Journal of Operational Research, No.
159, pp. 379-392.
[85]. Yusuf, Y. Y., Sarhadi, M. & Gunasekaran, A., 1999, Agile manufacturing: The drivers, concepts and
attributes, International Journal of Production Economics, No. 62, pp. 33-43.
[86]. Zhang, David Z. & Wang, Rundong , A Taxonomical Study of Agility Strategies and Supply Chain
Management, www.google.com.
[87]. Zhang, Z. & Sharifi, H., 2000, A methodology for achieving agility in manufacturing organizations,
International Journal of Operations & Prod Management, No. 20, pp. 496-512.
[88]. Zhang, D. Z. W. & Sharifi, H., 2007, Towards theory building in agile manufacturing strategy - A
taxonomical approach, IEEE Trans on Engineering Management, 54, pp. 351-370.
AUTHOR
THAEIR AHMED SAADOON AL SAMMAN was born in MOSUL, IRAQ in 1963. He
received the Bachelor in 1984 degree from the university of MOSUL, IRAQ and the
master in year 1987 degree from the university of MOSUL, IRAQ both in business
management and the Ph.D. in year 2008 degree from the university of MOSUL,IRAQ .He
is currently Ass. Prof. in management information system in the college of Business
administration in university of MOSUL. His research interests include operation
management and operations research.

.

International Journal of Advances in Engineering & Technology.
pg. A

MEMBERS OF IJAET FRATERNITY
Editorial Board Members from Academia
Dr. P. Singh,
Ethiopia.
Dr. A. K. Gupta,
India.
Dr. R. Saxena,
India.
Dr. Natarajan Meghanathan,
Jackson State University, Jackson.
Dr. Syed M. Askari,
University of Texas, Dellas.
Prof. (Dr.) Mohd. Husain,
A.I.E.T, Lucknow, India.
Dr. Vikas Tukaram Humbe,
S.R.T.M University, Latur, India.
Dr. Mallikarjun Hangarge,
Bidar, Karnataka, India.
Dr. B. H. Shekar,
Mangalore University, Karnataka, India.
Dr. A. Louise Perkins,
University of Southern Mississippi, MS.
Dr. Tang Aihong,
Wuhan University of Technology, P.R.China.
Dr. Rafiqul Zaman Khan,
Aligarh Muslim University, Aligarh, India.
Dr. Abhay Bansal,
Amity University, Noida, India.
Dr. Sudhanshu Joshi,
School of Management, Doon University, Dehradun, India.
Dr. Su-Seng Pang,
Louisiana State University, Baton Rouge, LA,U.S.A.
Dr. Avanish Bhadauria,
CEERI, Pilani,India.
Dr. Dharma P. Agrawal
University of Cincinnati, Cincinnati.
Dr. Rajeev Singh
University of Delhi, New Delhi, India.
Dr. Smriti Agrawal
JB Institute of Engineering and Technology, Hyderabad, India
Prof. (Dr.) Anand K. Tripathi
College of Science and Engg.,Jhansi, UP, India.
Prof. N. Paramesh
pg. B

University of New South Wales, Sydney, Australia.
Dr. Suresh Kumar
Manav Rachna International University, Faridabad, India.
Dr. Akram Gasmelseed
Universiti Teknologi Malaysia (UTM), Johor, Malaysia.
Dr. Umesh Kumar Singh
Vikram University, Ujjain, India.
Dr. A. Arul Lawrence Selvakumar
Adhiparasakthi Engineering College,Melmaravathur, TN, India.
Dr. Sukumar Senthilkumar
Universiti Sains Malaysia,Pulau Pinang,Malaysia.
Dr. Saurabh Pal
VBS Purvanchal University, Jaunpur, India.
Dr. Jesus Vigo Aguiar
University Salamanca, Spain.
Dr. Muhammad Sarfraz
Kuwait University,Safat, Kuwait.
Dr. Xianbo Qui
Xiamen University, P.R.China.
Dr. C. Y. Fong
University of California, Davis.
Prof. Stefanos Gritzalis
University of the Aegean, Karlovassi, Samos, Greece.
Dr. Hong Hu
Hampton University, Hampton, VA, USA.
Dr. Donald H. Kraft
Louisiana State University, Baton Rouge, LA.
Dr. Veeresh G. Kasabegoudar
COEA,Maharashtra, India.
Dr. Nouby M. Ghazaly
Anna University, Chennai, India.
Dr. Paresh V. Virparia
Sardar Patel University, V V Nagar, India.
Dr.Vuda Srinivasarao
St. Marys College of Engg. & Tech., Hyderabad, India.
Dr. Pouya Derakhshan-Barjoei
Islamic Azad University, Naein Branch, Iran.
Dr. Sanjay B. Warkad
Priyadarshini College of Engg., Nagpur, Maharashtra, India.
Dr. Pratyoosh Shukla
Birla Institute of Technology, Mesra, Ranchi,Jharkhand, India.
Dr. Mohamed Hassan Abdel-Wahab El-Newehy
King Saud University, Riyadh, Kingdom of Saudi Arabia.
Dr. K. Ramani
K.S.Rangasamy College of Tech.,Tiruchengode, T.N., India.
Dr. J. M. Mallikarjuna
pg. C

Indian Institute of Technology Madras, Chennai, India.
Dr. Chandrasekhar
Dr.Paul Raj Engg. College, Bhadrachalam, Andhra Pradesh, India.
Dr. V. Balamurugan
Einstein College of Engineering, Tirunelveli, Tamil Nadu, India.
Dr. Anitha Chennamaneni
Texas A&M University, Central Texas, U.S.
Dr. Sudhir Paraskar
S.S.G.M.C.E. Shegaon, Buldhana, M.S., India.
Dr. Hari Mohan Pandey
Middle East College of Information Technology, Muscat, Oman.
Dr. Youssef Said
Tunisie Telecom / Sys'Com Lab, ENIT, Tunisia.
Dr. Mohd Nazri Ismail
University of Kuala Lumpur (UniKL), Malaysia.
Dr. Gabriel Chavira Jurez
Autonomous University of Tamaulipas,Tamaulipas, Mexico.
Dr.Saurabh Mukherjee
Banasthali University, Banasthali,Rajasthan,India.
Prof. Smita Chaudhry
Kurukshetra University, Kurukshetra, Harayana, India.
Dr. Raj Kumar Arya
Jaypee University of Engg.& Tech., Guna, M. P., India.
Dr. Prashant M. Dolia
Bhavnagar University, Bhavnagar, Gujarat, India.
Dr. Dewan Muhammad Nuruzzaman
Dhaka University of Engg. and Tech., Gazipur, Bangladesh.
Dr. Hadj. Hamma Tadjine
IAV GmbH, Germany.
Dr. D. Sharmila
Bannari Amman Institute of Technology, Sathyamangalam, India
Dr. Jifeng Wang
University of Illinois, Illinois, USA.
Dr. G. V. Madhuri
GITAM University, Hyderabad, India.
Dr. T. S. Desmukh
MANIT, Bhopal, M.P., India.
Dr. Shaikh Abdul Hannan
Vivekanand College, Aurangabad, Maharashtra, India.
Dr. Zeeshan Ahmed
University of Wuerzburg, Germany.
Dr. Nitin S. Choubey
M.P.S.T.M.E.,N.M.I.M.S. (Shirpur Campus), Dhule, M.S., India.
pg. D

Dr. S. Vijayaragavan
Christ College of Engg. and Technology, Pondicherry, India.
Dr. Ram Shanmugam
Texas State University - San Marcos, Texas, USA.
Dr. Hong-Hu Zhu
School of Earth Sciences and Engg. Nanjing University, China.
Dr. Mahdi Zowghi
Department of Sharif University of technology, Tehran, Iran.
Dr. Cettina Santagati
Universit degli Studi di Catania, Catania, Italy.
Prof. Laura Inzerillo
University of Palermo, Palermo, Italy.
Dr. Moinuddin Sarker
University of Colorado, Boulder, U.S.A.
Dr. Mohammad Amin Hariri Ardebili
University of Colorado, Boulder, U.S.A.
Dr. S. Kishore Reddy
Swarna Bharathi College of Engineering, Khammam, A.P., India.
Dr. V.P.S. Naidu
CSIR - National Aerospace Laboratories, Bangalore, India.
Dr. S. Ravi
Nandha Engineering College, Erode, Tamilnadu, India.
Dr. R. Sathish Kumar
K L University, Andhra Pradesh, India.
Dr. Nawfal Jebbor
Moualy Ismail University, Meknes Morocco.
Dr. Ali Jasim Mohammed Al-Jabiry
Al-Mustansiriyah University, Baghdad, Iraq.
Dr. Meyyappan Venkatesan
Mekelle University, Ethiopia.
Dr. A. S. N. Chakravarthy
University College of Engineering, JNTUK - Vizianagaram Campus, A. P., India.
Dr. A. Abdul Rasheed
Valliammai Engineering College, Kattankulathur, Tamil Nadu, India.
Dr. A. Ravi Shankar
Indian Institute of Technology, Guwahati, Assam, India.
Dr. John Kaiser S. Calautit
University of Leeds, Leeds, UK.
Dr. S. Manikandan
R.M.D. Engineering College, Kavaraipettai, Tamil nadu, India,
pg. E

Dr. Sandeep Gupta
Noida Institute of Engineering and Technology, Gr.Noida, India.
Dr. Kshitij Shinghal
Moradabad Institute of Technology, Moradabad, Uttar Pradesh, India.
Dr. Ankit Srivastava
Bundelkhand University,Jhansi, U.P., India
Dr. Abbas M. Al-Bakry
Babylon University, IRAQ
Dr. K. Narasimhulu
National Institute of Technology Warangal, Warangal, Andhra Pradesh, India.
Dr. Messaouda AZZOUZI
University of Djelfa, Algeria.
Dr. Govindaraj Thangavel
Muthayammal Engineering College, Rasipuram, Tamil Nadu, India.
Dr. Deep Kamal Kaur Randhawa
Guru Nanak Dev University Regional Campus, Jalandhar, Punjab, India.
Dr. L. Mary Immaculate Sheela
Professor at R.M.D. Engineering College, Chennai, India.
Dr. Sanjeevi Chitikeshi
Murray State University, Murray, KY, USA
Dr. Laith Ahmed Najam
Mosul University, College of Science, Iraq.
Dr. Brijesh Kumar
School of Engineering & Technology, Graphic Era University Dehradun,
Uttarakhand, India

Editorial Board Members from Industry/Research Labs.
Tushar Pandey
STEricsson Pvt Ltd, India.
Ashish Mohan
R&D Lab, DRDO, India.
Amit Sinha
Honeywell, India.
Tushar Johri
Infosys Technologies Ltd, India.
Dr. Om Prakash Singh
Manager, R&D, TVS Motor Company, India.
Dr. B.K. Sharma
Northern India Textile Research Assoc., Ghaziabad, U.P., India.
pg. F

Mr. Adis Medic
Infosys ltd, Bosnia.
Mr. M. Muralidharan
Indian Oil Corporation Ltd., India.
Mr. Rohit Kumar Malik
Oracle India Pvt. Ltd., Bangalore, India.
Dr. Pinak Ranade
Centre For Development of Advance Computing, Pune, India.
Dr. Jierui Xie
Oracle Corporation, Redwood City, CA, U.S.A.
Dr. R. Guruprasad
Scientist, CSIR-National Aerospace Laboratories, Bangalore, India.
Ms. Priyanka Gandhi
Goken America, Honda Research and Development North America, U.S.A.
Mr. Ang Boon Chong
eASIC, Malaysia

Advisory Board Members from Academia & Industry/Research Labs.
Prof. Andres Iglesias,
University of Cantabria, Santander, Spain.
Dr. Arun Sharma,
K.I.E.T, Ghaziabad, India.
Prof. Ching-Hsien (Robert) Hsu,
Chung Hua University, Taiwan, R.o.C.
Dr. Himanshu Aggarwal,
Punjabi University, Patiala, India.
Prof. Munesh Chandra Trivedi,
CSEDIT School of Engg.,Gr. Noida,India.
Dr. P. Balasubramanie,
K.E.C.,Perundurai, Tamilnadu, India.
Dr. Seema Verma,
Banasthali University, Rajasthan, India.
Dr. V. Sundarapandian,
Dr. RR & Dr. SR Technical University,Chennai, India.
Mayank Malik,
Keane Inc., US.
Prof. Fikret S. Gurgen,
Bogazici University Istanbul, Turkey.
Dr. Jiman Hong
Soongsil University, Seoul, Korea.
Prof. Sanjay Misra,
Federal University of Technology, Minna, Nigeria.
pg. G

Prof. Xing Zuo Cheng,
National University of Defence Technology, P.R.China.
Dr. Ashutosh Kumar Singh
Indian Institute of Information Technology Allahabad, India.
Dr. S. H. Femmam
University of Haute-Alsace, France.
Dr. Sumit Gandhi
Jaypee University of Engg.& Tech., Guna, M. P., India.
Dr. Hradyesh Kumar Mishra
JUET, Guna , M.P., India.
Dr. Vijay Harishchandra Mankar
Govt. Polytechnic, Nagpur, India.
Prof. Surendra Rahamatkar
Nagpur Institute of Technology, Nagpur, India.
Dr. B. Narasimhan
Sankara College of Science And Commerce, Coimbatore, India.
Dr. Abbas Karimi
Islamic Azad University,Arak Branch, Arak,Iran.
Dr. M. Munir Ahamed Rabbani
Qassim University, Saudi Arabia.
Dr. Prasanta K Sinha
Durgapur Inst. of Adva. Tech. & Manag., Durgapur, W. B., India.
Dr. Tole H. Sutikno
Ahmad Dahlan University(UAD),Yogyakarta, Indonesia.
Dr. Anna Gina Perri
Politecnico di Bari, BARI - Italy.
Prof. Surendra Rahamatkar
RTM Nagpur University, India.
Dr. Sagar E. Shirsath
Vivekanand College, Aurangabad, MS, India.
Dr. Manoj K. Shukla
Harcourt Butler Technological Institute, Kanpur, India.
Dr. Fazal Noorbasha
KL University, Guntur, A.P., India.
Dr. Manjunath T.C.
HKBK College of Engg., Bangalore, Karnataka, India.
Dr. M. V. Raghavendra
Swathi Institute of Technology & Sciences, Ranga Reddy , A.P. , India.
Dr. Muhammad Farooq
University of Peshawar, 25120, Khyber Pakhtunkhwa, Pakistan.
Prof. H. N. Panchal
L C Institute of Technology, Mehsana, Gujarat, India.
Dr. Jagdish Shivhare
ITM University, Gurgaon, India.
Prof.(Dr.) Bharat Raj Singh
pg. H

SMS Institute of Technology, Lucknow, U.P., India.
Dr. B. Justus Rabi
Toc H Inst. of Sci. & Tech. Arakkunnam, Kerala, India.
Prof. (Dr.) S. N. Singh
National Institute of Technology, Jamshedpur, India.
Prof.(Dr) Srinivas Prasad,
Gandhi Inst. for Technological Advancement, Bhubaneswar, India.
Dr. Pankaj Agarwal
Samrat Ashok Technological Institute, Vidisha (M.P.), India.
Dr. K. V. L. N. Acharyulu
Bapatla Engineering College, Bapatla, India.
Dr. Shafiqul Abidin
Kalka Inst. for Research and Advanced Studies, New Delhi, India.
Dr. M. Senthil Kumar
PRCET, Vallam, Thanjavur, T.N., India.
Dr. M. Sankar
East Point College of Engg. and Technology, Bangalore, India.
Dr. Gurjeet Singh
Desh Bhagat Inst. of Engg. & Management, Moga, Punjab, India
Dr. C. Venkatesh
E. B. E. T. Group of Institutions, Tirupur District, T. N., India.
Dr. Ashu Gupta
Apeejay Institute of Management, Jalandhar, India.
Dr. Brijender Kahanwal
Galaxy Global Imperial Technical Campus, Ambala, India.
Dr. A. Kumaravel
K. S. Rangasamy College of Technology, Tiruchengode, India.
Dr. Norazmawati Md. Sani
Universiti Sains Malaysia, Pulau Pinang Malaysia
Dr. Mariateresa Galizia
University of Catania, Catania, Italy.
Dr. M V Raghavendra
Adama Science & Technology University, Ethiopia.
Dr. Mahdi Moharrampour
Islamic Azad University Buin zahra Branch, Iran.
Dr. S. Sasikumar
Jayaram College of Engg. and Tech.,Trichy, T.N., India.
Dr. Jay Prakash Verma
Banaras Hindu University, Varanasi, India.
Dr. Bensafi Abd-El-Hamid
Abou Bekr Belkaid University of Tlemcen, Tlemcen, Algeria.
Dr. Tarit Roychowdhury
pg. I

Supreme Knowledge Foundation Group of Institutes, Mankundu, Hooghly, India.
Dr. Shourabh Bhattacharya
Madhav Institute of Technology & Science (M.I.T.S.), Gwalior, India.
Dr. G. Dalin
SNMV College of Arts and Science, Coimbatore, India.
Dr. Syed Asif Ali
SMI University, Karachi, Pakistan.
Dr. Ushaa Eswaran
Velammal Engineering College, Chennai, India.
Dr. Ashok Dargar
Sir Padampat Singhania University, Udaipur,Raj., India.
Dr. Mohammad Hadi Dehghani
Tehran University of Medical Sciences, Tehran, I.R.Iran.
Dr. K. Nagamalleswara Rao
Bapatla Engineering College, Bapatla (PO), Andhra Pradhesh, India.
Dr. Md Shafiq Alam
Punjab Agricultural University, Ludhiana, Punjab, India.
Dr. D. Raju
Vidya Jyothi Institute of Technology, Hyderabad, India.

Research Volunteers from Academia
Mr. Ashish Seth,
Ideal Institute of Technology, Ghaziabad, India.
Mr. Brajesh Kumar Singh,
RBS College,Agra,India.
Prof. Anilkumar Suthar,
Kadi Sarva Viswavidhaylay, Gujarat, India.
Mr. Nikhil Raj,
National Institute of Technology, Kurukshetra, Haryana, India.
Mr. Shahnawaz Husain,
Graphic Era University, Dehradun, India.
Mr. Maniya Kalpesh Dudabhai
C.K.Pithawalla College of Engg.& Tech.,Surat, India.
Dr. M. Shahid Zeb
Universiti Teknologi Malaysia(UTM), Malaysia.
Mr. Brijesh Kumar
Research Scholar, Indian Institute of Technology, Roorkee, India.
Mr. Nitish Gupta
Guru Gobind Singh Indraprastha University,India.
pg. J

Mr. Bindeshwar Singh
Kamla Nehru Institute of Technology, Sultanpur, U. P., India.
Mr. Vikrant Bhateja
SRMGPC, Lucknow, India.
Mr. Ramchandra S. Mangrulkar
Bapurao Deshmukh College of Engineering, Sevagram,Wardha, India.
Mr. Nalin Galhaut
Vira College of Engineering, Bijnor, India.
Mr. Rahul Dev Gupta
M. M. University, Mullana, Ambala, India.
Mr. Navdeep Singh Arora
Dr B R Ambedkar National Institute of Technology, Jalandhar, Punjab, India.
Mr. Gagandeep Singh
Global Institute of Management and Emerging Tech.,Amritsar, Punjab, India.
Ms. G. Loshma
Sri Vasavi Engg. College, Pedatadepalli,West Godavari, Andhra Pradesh, India.
Mr. Mohd Helmy Abd Wahab
Universiti Tun Hussein ONN Malaysia, Malaysia.
Mr. Md. Rajibul Islam
University Technology Malaysia, Johor, Malaysia.
Mr. Dinesh Sathyamoorthy
Science & Technology Research Institute for Defence (STRIDE), Malaysia.
Ms. B. Neelima
NMAM Institute of Technology, Nitte, Karnataka, India.
Mr. Mamilla Ravi Sankar
IIT Kanpur, Kanpur, U.P., India.
Dr. Sunusi Sani Adamu
Bayero University, Kano, Nigeria.
Dr. Ahmed Abu-Siada
Curtin University, Australia.
Ms. Shumos Taha Hammadi
Al-Anbar University, Iraq.
Mr. Ankit R Patel
L C Institute of Technology, Mahesana, India.
Mr.Athar Ravish Khan Muzaffar Khan
Jawaharlal Darda Institute of Engineering & Technology Yavatmal, M.S., India.
Prof. Anand Nayyar
KCL Institute of Management and Technology, Jalandhar, Punjab, India.
Mr. Arshed Oudah
UTM University, Malaysia.
Mr. Piyush Mohan
pg. K

Swami Vivekanand Subharti University, Meerut, U.P., India.
Mr. Mogaraju Jagadish Kumar
Rajampeta, India.
Mr. Deepak Sharma
Swami Vivekanand Subharti University, Meerut, U.P., India.
Mr. B. T. P. Madhav
K L University, Vaddeswaram, Guntur DT, AP, India.
Mr. Nirankar Sharma
Subharti Institute of Technology & Engineering, Meerut, U.P., India.
Mr. Prasenjit Chatterjee
MCKV Institute of Engineering, Howrah, WB, India.
Mr. Mohammad Yazdani-Asrami
Babol University of Technology, Babol, Iran.
Mr. Sailesh Samanta
PNG University of Technology, Papua New Guinea.
Mr. Rupsa Chatterjee
University College of Science and Technology, WB, India.
Er. Kirtesh Jailia
Independent Researcher, India.
Mr. Abhijeet Kumar
MMEC, MMU, Mullana, India.
Dr. Ehab Aziz Khalil Awad
Faculty of Electronic Engineering, Menouf, Egypt.
Ms. Sadia Riaz
NUST College of E&ME, Rawalpindi, Pakistan.
Mr. Sreenivasa Rao Basavala
Yodlee Infotech, Bangalore, India.
Mr. Dinesh V. Rojatkar
Govt. College of Engineering, Chandrapur, Maharashtra State, India.
Mr. Vivek Bhambri
Desh Bhagat Inst. of Management & Comp. Sciences, Mandi Gobindgarh, India.
Er. Zakir Ali
I.E.T. Bundelkhand University, Jhansi, U.P., India.
Mr. Himanshu Sharma
M.M University, Mullana, Ambala, Punjab, India.
Mr. Pankaj Yadav
Senior Engineer in ROM Info Pvt.Ltd, India.
Mr. Fahimuddin.Shaik
JNT University, Anantapur, A.P., India.
Mr. Vivek W. Khond
G.H.Raisoni College of Engineering, Nagpur, M.S. , India.
Mr. B. Naresh Kumar Reddy
K. L. University, Vijayawada, Andra Pradesh, India.
Mr. Mohsin Ali
APCOMS, Pakistan.
Mr. R. B. Durairaj
pg. L

SRM University, Chennai., India.
Mr. Guru Jawahar .J
JNTUACE, Anantapur, India.
Mr. Muhammad Ishfaq Javed
Army Public College of Management and Sciences, Rawalpindi, Pakistan.
Mr. M. Narasimhulu
Independent Researcher, India.
Mr. B. T. P. Madhav
K L University, Vaddeswaram, Guntur DT, AP, India.
Mr. Prashant Singh Yadav
Vedant Institute of Management & Technology, Ghaziabad, India.
Prof. T. V. Narayana Rao
HITAM, Hyderabad, India.
Mr. Surya Suresh
Sri Vasavi Institute of Engg & Technology, Nandamuru,Andhra Pradesh, India.
Mr. Khashayar Teimoori
Science and Research Branch, IAU, Tehran, Iran.
Mr. Mohammad Faisal
Integral University, Lucknow, India.
Prof. H. R. Sarode
Shri Sant Gadgebaba College of Engineering and Technology, Bhusawal, India.
Mr. Rajeev Kumar
KIRAS, GGSIP University, New Delhi, India.
Mr. Mamilla Ravi Sankar
Indian Institute of Technology, Kanpur, India.
Mr. Avadhesh Kumar Yadav
University of Lucknow, Lucknow, India.
Ms. Mina Asadi
Independent Researcher, Adelaide, Australia.
Mr. Muhammad Naufal Bin Mansor
Universiti Malaysia Perlis, Malaysia.
Mr. R. Suban
Annamalai University, Annamalai Nagar, India.
Mr. Abhishek Shukla
R. D. Engineering College, Ghaziabad, India.
Mr. Sunil Kumar
Adama Science and Technology University, Adama, Ethiopia, Africa.
Mr. A. Abdul Rasheed
SRM Valliammai Engineering College, Chennai, T. N., India.
Mr. Hari Kishore Kakarla
K L University, Guntur, Andhra Pradesh, India.
Mr. N. Vikram
Jayaprakash Narayan College of Engineering, Dharmapur, A.P, India
Mr. Samir Malakar
MCKV Institute of Engineering, Howrah, India.
Ms. Ketaki Solanki
pg. M

Vishveshwarya Institute of Engineering and Technology, Ghaziabad, India.
Mr. Ashtashil V. Bhambulkar
Rungta Group, Raipur, Chhattisgarh State, India.
Mr. Nitin K. Mandavgade
Priyadarshni College of Engineering, Nagpur, Maharashtra, India
Mr. Kamal Kant Sharma
M.M. University Mullana, Ambala, India.
Mr. Chandresh Kumar Chhatlani
JRN Rajasthan Vidyapeeth University, Udaipur, Rajasthan, India.
Mr. A. Ravi Shankar
Sarada Institute of Technology & Science, Khammam, A.P., India.
Mr. Balajee Maram
GMRIT, Rajam, A.P., India.
Mr. Srinivas Vadali
Kakinada Institute of Engineering and Technology KIET, Kakinada, A.P., India.
Mr. T. Krishna Kishore
St. Anns College of Engineering And Technology, Chirala, A.P., India.
Ms. Ranjana Rajnish
Amity University, Lucknow, India.
Mr. Padam Singh
J. P. Institute of Engineering & Technology, Meerut, India.
Ms. Sakshi Rajput
Maharaja Surajmal Institute of Technology, GGSIPU, Delhi, India
Mr. Jaware Tushar H.
R.C. Patel Institute of Technology, Shirpur, MS, India
Mr. R. D. Badgujar
R C Patel Institute of Technology, Shirpur Distt., Dhule, MS, India
Mr. Zairi Ismael Rizman
Faculty of Electrical Engineering, Universiti Teknologi MARA, Malaysia.
Mr. Ashok Kumar Rajput
Radha Govind Engineering College, Meerut, India.
Mr. Nitin H. Ambhore
Vishwakarma Institute of Information Technology, Pune, India
Mr. Vishwajit K. Barbudhe
Agnihotri College of Engineering, Nagthana, Wardha, India
Mr. Himanshu Chaurasiya
Amity University, Noida, U.P., India.
Ms. Rupali Shelke
Marathwada Mitra Mandal's Polytechnic, Pune, M.S., India
Mr. Pankaj Manohar Pandit
JDIET, Sant Gadgebaba Amravati University (M.S.), Yavatmal, India
Mr. Gurudatt Anil Kulkarni
Marathwada Mitra Mandal's Polytechnic,Pune, M.S., India
Ms. Pragati Chavan
Marathwada Mitra Mandal's Polytechnic,Pune, M.S., India
Mr. Mahendra Umare
pg. N

Nagpur Institute of Technology, Nagpur, India
Prof. Ajay Gadicha
P.R. Pote (Patil) College of Engineering, Amravati, India.
Mr. M. Pradeep
A.S.L.Pauls College of Engineering and Technology, Coimbatore, Tamil Nadu, India.
Mr. Susrutha Babu Sukhavasi
K.L. University, Guntur, A.P, India.
Mr. Vivek Kumar Singh
Sustainable Energy Systems, MIT Portugal Program, University of Coimbra, Portugal.
Dr. V. Bhoopathy
Vivekanandha College of Technology for Women, Namakkal, Tamilnadu, India.
Mr. Akash Porwal
Axis Institute of Technology and Management, Kanpur, U.P., India.
Mr. Ravindra Jogekar
RTM NU, Nagpur, India.
Md. Risat Abedin
Ahsanullah University of Science & Technology, Dhaka, Bangladesh.

OUR FAST REVIEWING PROCESS IS OUR
STRENGTH.

URL : http://www.ijaet.org
E-mail : editor@ijaet.org
ijaet.editor@gmail.com

IJAET Volume 7 Issue 3 July 2014

Încărcat de

Informații document

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

IJAET Volume 7 Issue 3 July 2014

Încărcat de

Drepturi de autor:

Formate disponibile

VOLUME-7, ISSUE-3

cases are chosen and these projections

inimage (x, y, r) < (x, y, g)

inimage(x, y, b) < (x, y, g)

is the inverse Darcy Parameter,

= is the material parameter related to relaxation time and

)] Array steering vector.

) it is eigenvalues and sorting in ascending sequence:

) corresponding to the largest eigenvalues (

), so the signal subspace is:

is the diagonal matrix consist of the m

), so the noise subspace

is the diagonal matrix consist of

), = 1,2, , the spectrum

is the eigenvalue associated with eigenvector

and its very wide (5

). The wide beam causes an ambiguity in estimation the DOA.

degree. There is very small

degree. There is very

S-ar putea să vă placă și