Documente Academic
Documente Profesional
Documente Cultură
Proceedings
International Session
Organized by:
Department of Electrical Engineering
Brawijaya University
Indonesia
PUBLISHED BY:
Department of Electrical Engineering
Faculty of Engineering
Brawijaya University
eeccis@ub.ac.id
LAYOUT EDITOR
COORDINATOR
Wijono
MEMBERS
Angger Abdul Razak
Eka Maulana
Renie Febriyanti
Marina Dicarara
Firman Triyanto
Fahad Arwani
Erny Anugrahany
All papers in this book have been selected by the reviewers and technical committee.
All authors have signed the copyright declaration of their papers.
ii
ORGANIZING INSTITUTION
STEERING COMMITTEE
Prof. Ir. Harnen Sulistio, M.Sc., Ph.D.
Dr. Ir. Sholeh Hadi Pramono, M.S..
REVIEWER
Asc.Prof. Dr. Mamdouh (Aswan University, Egypt)
Asc. Prof. Dr. Mahrus (Aswan University, Egypt)
Dr. Corina Martineac (Rumania)
Ishtiaq R. Khan, Ph.D (Singapore)
Hazlie Muslikh, Ph.D (UM, Malaysia)
Dr. Hamzah Arouf (Malaysia)
Prof. Dr. Kaharudin Dimyati (Malaysia)
Md. Atiqur Rahman Ahad, B.Sc.,M.S.,M.S.,PhD (Bangladesh)
Prof. Adi Susanto, MSc. Ph.D (UGM, Indonesia)
Prof. Thomas Sri Widodo, DEA (UGM, Indonesia)
Prof. Dr. Ir. Arif Djunaidy, MSc (ITS, Indonesia)
Dr. Aris Triwiyatno (UNDIP, Indonesia)
Dr. Ir. Son Kuswadi (ITS, Indonesia)
Purnomo Sidi Priambodo, Ph. D (UI, Indonesia)
Dr. Ir. Muhammad Nurdin (ITB, Indonesia)
Dr.-Ing. Ir. M. Sukrisno (STEI-ITB, Indonesia)
Dr. Ferry Hadary, ST, M. Eng (UNTAN, Indonesia)
Dr. Mashury Wahab (PPET-LIPI, Indonesia)
Dr. Rini Nurhasanah, M. Sc (UB, Indonesia)
Ir. Wijono, MT. Ph.D (UB, Indonesia)
Hadi Suyono, Ph.D (UB, Indonesia)
Dr. Sholeh Hadi Pramono (UB, Indonesia)
iii
TECHNICAL PROGRAM COMMITTEE
Muhammad Ary Murti (IEEE Indonesia Section)
Kuncoro Watuwibowo (IEEE Indonesia Section)
Arief Hamdani (IEEE Indonesia Section)
Ford Lumban Gaol (IEEE Indonesia Section)
Panca Mudjiraharjo (KIT - Japan)
Onny Setyawati (Universitat Kassel - Jerman)
M. Rusli (University of Wollongong - Australia)
Sholeh Hadi Pramono (UB - Indonesia)
Agung Darmawansyah (UB - Indonesia)
M. Aziz Muslim (UB - Indonesia)
Hadi Suyono (UB - Indonesia)
Rini Nurhasanah (UB - Indonesia)
Wijono (UB - Indonesia)
iv
SEMINAR PROGRAM
09.30 - 10.45 1ST KEYNOTE SPEECH BY DR. IR. UNGGUL PRIYANTO, M.SC (DEPUTY
CHAIRMAN FOR TECHNOLOGY OF INFORMATION AND COMMUNICATION,
ENERGY, AND MATERIALS OF THE AGENCY FOR THE ASSESMENT AND
APPLICATION OF TECHNOLOGY)
10. 45 - 12.00 2ND KEYNOTE SPEECH BY DR. EKO FAJAR PRASETYO (FOUNDER OF
VERSATILE SILICON TECHNOLOGY, FIRST IC DESIGN COMPANY IN
INDONESIA)
17.25 CLOSING
v
FOREWORD BY THE DEAN OF
FACULTY OF ENGINEERING,
BRAWIJAYA UNIVERSITY
Assalamualaikum warahmatullahi wabarakatuh
F lecturers, students, and all other people impossible to cite individually, for having
involved in the good achievement of the organization of the EECCIS 2012 Seminar.
I also would like to express my gratitude to Dr. Ir. Unggul Priyanto, M.Sc and Dr. Eko
Fajar Prasetyo for having accepted to become the keynote speakers of this EECCIS 2012
Seminar.
The EECCIS 2012 Seminar follows the success of the previously held seminars of
EECCIS 2000, 2004, 2006, 2008 and 2010. It becomes a part of scientific activity
programmes in the Faculty of Engineering to contribute to the creation of Brawijaya
University as a research university, and furthermore as an entrepreneurial university.
vi
PREFACE BY THE CHAIRMAN OF
THE ORGANIZING COMMITEE
Assalamualaikum warahmatullahi wabarakatuh
The hardwork of our technical program committee for the success of this seminar has
been indicated by the large number of the scientific paper received. There have been
received about 189 papers coming from Indonesia, Malaysia, Japan, and Australia. After a
very rigorous process of reviewing by reviewers coming from Switzerland, Egypt,
Malaysia, Singapore, Bangladesh, and Indonesia, only about 83% of the received papers
have been accepted to be presented in a series of oral-presentation sessions during the
seminar, and also to be documented and published in the Proceedings of EECCIS 2012.
Sincere thanks go to all members of the Steering Committee and reviewers, who have
worked hard to guarantee the good quality of papers presented in this seminar.
On the part of the chairman of the Organizing Committee, I also would like to convey my
very high appreciation on the enthusiasm and hardwork shown by the whole technical
program committee, and also to many other people who are involved directly or indirectly
in contributing to the good achievement of this seminar.
Finally, I would like to thank and welcome all researchers, lecturers, students, industry,
and all other participants to the EECCIS 2012 Seminar. We admit that there are still
numerous lacks in the organization of this seminar, however any suggestions are always
welcome for our improvement in the future.
vii
TABLE OF CONTENT
Cover i
Organizing Institution iii
Seminar Program v
Preface by the Dean of the Faculty of Engineering vi
Preface by the Chairman of the Organizing Committee vii
Table of Content viii
A. ELECTRICAL POWER
[149-EEA_28] Fractional Open Voltage Maximum Power Point Tracking Using ATMega8535
For Photovoltaic System
Gunawan Wibisono, Sholeh Hadi Pramono, M. Aziz Muslim
Student of Master Degree Program Lecturer of Brawijaya University A1
B. ELECTRONICS
[111-EEB_24] Automated Measurement of Haemozoin (Malarial Pigment) Area in Liver Histology Using
Image J 1.6
Dwi Ramadhani, Tur Rahardjo, and Siti Nurhayati
Center for Technology of Radiation Safety and Metrology, National Nuclear Energy Agency of Indonesia B1
C. COMMUNICATION
[034-EEC_05] Android Smartphone Based for the Local Directory Application of Public Utility
Arini, MT., Viva Arifin, MMSi., Chery Dia Putra, S.Kom.
Informatics Engineering Program State University (UIN) Syarief Hidayatullah, Jakarta C1
vii
[062-EEC_11] Tropical rain effects on Free-Space Optical and 30 GHz wireless systems
M. Derainjafisoa and G. Hendrantoro
Department of Electrical Engineering, Institut Teknologi Sepuluh Nopember C2
[172-EEC_35] Singled-fed Circularly Polarized Triangular Microstrip Antenna with Truncated Tip Using
Annular Sector Slot for Mobile Satellite Communications
Muhammad Fauzan Edy Purnomo, Sapriesty Nainy Sari
Department of Electrical Engineering, University of Brawijaya C4
D. CONTROLS
[051-EED_12] Maximum Power Point Tracking using Fuzzy Logic Control for Buck Converter in
Photovoltaic System
Mahendra Widyartono, Sholeh Hadi Pramono, M. Aziz Muslim
Student of Master Degree Program and Lecturers, Department of Electrical Engineering, Brawijaya University D1
[093-EED_21] A Computer Fluid Dynamics Study of 6.5 Micron AA 1235 Annealing Treatment in
Sided Blow Inlet-Outlet Furnace
Ruri A. Wahyuono, Wiratno A. Asmoro, Edy Sugiantoro, Muhamad Faisal
Department of Engineering Physics, Institut Teknologi Sepuluh Nopember, Surabaya
D2
viii
[132-EED_29] The Height Control Systems of Hydraulic Jack Using Takagi Sugeno Fuzzy
Logic Controller
Fitriana Suhartati, Ahmad Fahmi
Department of Electrical Engineering University of Brawijaya, Electrical Engineering of State University of Malang D3
[146-EED_32] An Application of Adaptive Neuro Fuzzy Inference System (ANFIS) with Substractive
Clustering for Lung Cancer Early Detection System
Mohamad Yusuf Santoso, Syamsul Arifin
Faculty of Industrial Technology, Institut Teknology Sepuluh Nopember, Surabaya
D4
[163-EED_36] PID Design for 3-Phase Induction Motor Speed Control Based on Neural Network
Levenberg Marquardt
Dedid Cahya H., Agus Indra G., Ali Husein A., Ahmad Arif A.
Politeknik Elektronika Negeri Surabaya
D5
[182-EED_39] Zelio PLC-Based Automation of Coffe Roasting Processs
M. Aziz Muslim, Goegoes Dwi N., Ali Mahkrus
Department of Electrical Engineering, Faculty of Engineering, Brawijaya University D6
[188-EED_40] Prediction of CO and HC on Multiple Injection Diesel Engine Using Multiple Linear
Regression
Bambang Wahono, Harutoshi Ogai
Graduate School of Information, Production and Systems, Waseda University D7
E. INFORMATICS
[048-EEE_07] Acceptance of Mobile Peyment Application in Indonesia
Hendra Pradipta
Informatics Management, State Polytechnics of Malang E1
[072-EEE_16] Analysis and Implementation of Combined Triple Vigenere Cipher and ElGamal Cryptography
using Digital Image as a Cryptographic Key
Komang Rinartha, Agung Darmawansyah, Rudy Yuwono
STMIK Stikom Bali, Department of Electrical Engineering, Faculty of Engineering, Brawijaya University
E4
ix
[083-EEE_19] Heart Rate Variability Analysis on Sudden Cardiac Death Risk RR Interval by
Using Poincare Plot Method
Ponco Siwindarto, I.N.G. Wardana, M. Aris Widodo, M. Rasjad Indra
Faculty of Engineering and Medical Faculty of Brawijaya University E5
[094-EEE_21] Lung Cancer Prediction in Imaging Test Based on Gray Level Co-occurrence Matrix
Sunngging Haryo W., Agus M. Hatta., Syamsul Arifin
Department of Engineering Physics, Faculty of Industrial Technology, Institut Teknologi Sepuluh Nopember E6
[184-EEE_38] Fast and Accurate Interest Points Detection Algorithm on Brycentric Coordinates
using Fitted Quadratic Surface Combinaed with Hilbert Scanning Distance
Tibyani Tibyani, Sei-ichiro Kamata
Graduation School of Iformation , Production, and System, Waseda University
E11
[186-EEE_39] Generating Security Keys From Combination of multiple Biometric Sources
Primantara Hari Trisnawan
Camputer Sciences, University of Malaysia
E12
x
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
AbstractPhotovoltaic (PV) systems are power source controlled by the duration of the switch on and off (ton and
systems that have non-linear current voltage toff). This method is known as pulse-width modulation
characteristics (I-V) under different environments condition. (PWM) switching [3].
The system consists of PV generator (cells, modules, PV
array), energy storage (batteries), buck converter, and
resistive load. The proposed maximum power point control
is based on fractional open voltage The controller must
maintain PV voltage at VMPP by changing the converter duty
cycle so that maximum power can be generated in varying
operating condition. Using proposed maximum power point
tracking (MPPT) method, average power is 11,8% higher
than direct load, and 21.9% higher than constant output
voltage scheme. The system have better accuracy and
stability even in dynamic operating conditions.
Index TermsPhotovoltaic system, Maximum Power
Point Tracking, Fractional Open Voltage.
I. INTRODUCTION
Photovoltaic systems (PV) is a system to convert the
sun's energy directly into electrical energy. PV system is
one of the renewable energy alternatives. The power of
sunlight received by the earth outside the atmosphere is
about 1300 watt/m2 [1].
Simple PV system consists of a PV cell unit, a solar
charger, and a battery. The battery is an electro-chemical
devices. Various types of batteries have different
characteristics, and different ways of charging. The battery
charger must adjust according to these characteristics so Figure 1. Ideal characteristic of PV cell showing MPP
that the battery can last long. On the other hand, the PV Maximum Power Point Tracking (MPPT) is a sub-
cell output also has certain characteristics that may not be system designed to extract maximum power from power
suitable to charge the battery. source [2]. The maximum power point is shown in Figure
A solar charger should be able to bridge between the (1) above. In the case of solar power source, the
PV cell output voltage and current is varied according to maximum point varies due to the influence of changes in
the level of solar lighting with a battery that must be electrical characteristics as function of temperature, solar
charged with a certain voltage for optimum power transfer iradiation, heating and others. With the change of
and not damage the battery. temperature and solar iradiation, the voltage and current
Conversion efficiency of solar energy into electric output of the PV modules are also changing and reducing
energy via PV cells are low, only around 15-20%. One of efficiency of PV systems. MPPT maximizes power output
the effort to improve energy conversion efficiency of of the panels in different conditions to detect the best
photovoltaic cells is using the Maximum Power Point working point of the power characteristics and then
Tracking (MPPT) method [2]. controls the current or voltage on the panel.
DC-DC converter is used to convert the DC input General requirement for MPPT is simple and low cost,
voltage that varies into controlled DC output voltage at the fast tracking the changing conditions, and fluctuations of
desired voltage level. The basic form of DC-DC converter small output. Due to the nature of PV system is non-
is buck converter, are also called step-down converter. As linear, i.e. current and voltage that varies depending on
the name implies, step-down converters produce a dc environmental conditions, it isvery important to operate
output voltage of the average lower than the input dc
voltage. In DC-DC converter, average output voltage is
A1- 1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
the PV system at the condition of maximum power point. Figure 2. Proposed system
This will improve the efficiency of PV systems. The Buck converter used here is a simple common
There are various methods and ways to implement the Buck converter as is Figure (3)
MPPT control, e.g., by perturb and observe, incremental
conductance, fractional open voltage, parasitic capacitance
[2].
A1- 2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
PV Power
45
40
35
30
PV Power
25
20
10
Output voltage ripple is shown in Figure (6) below.
5
The ripple is satisfactory low at 125 mV.
0
OC 40 35 30 25 20 15 10 9 8 7 6 5 4 3 2 1 0
The input voltage graph can be seen in Figure (7) Resistance (Ohm)
below. As seen from the figure, the PV voltage can be
Constant output MPPT direct load
maintain around VMPP at high load. Average voltage using
MPPT at high load is 14.99 V, while direct load gave Figure 8. PV Power over resistance
10.22 V, and constant output voltage gave 12.43 V. To
maintain VMPP at low load, we must use buck-boost
converter to increase apparent load. With only a buck
converter, the algorithm only work at high load. VI. CONCLUSION
From the results and analysis, it can be concluded that
As seen from Figure (8) below, the algorithm also
PV system using fractional open voltage can maintain
maintain power at high load. Average power using MPPT
voltage at VMPP and PV power at high load. Using MPPT,
at high load is 28,47 W, evidently higher than direct load
panel voltage is maintained closer to VMPP at 14.99 V. The
at 25.45 W, and constant output voltage gave 23,35 W.
average power also 11,8% higher than direct load, and
21.9% higher than constant output voltage scheme.
A1- 3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
REFERENCES
Sholeh Hadi Pramono received Bachelor
Degree from Electritrical Engineering
[1] Markvart, T dan Castaner, L. Solar Cells: Materials, Manufacture Department, Brawijaya University in 1986. He
and Operation. Elsevier Amsterdam (2005) received his Master Degree and Doctoral Degree
both from University of Indonesia, in 1995 and
[2] Khaligh, A dan Onar, O.C. Energy Harvesting: Solar, Wind and
2010, respectively. Since 1987 he is with
Ocean Energy Conversion System. CRC Press Florida (2010)
Electrical Engineering Department, Brawijaya
[3] Mohan N, Undeland T, M, and Robbins, W, P. 1995. Power University. His current research interest including
Electronics. Converters, Applications, and Design. (2nd Edition). fiber optics, telecommunication and renewable
John Wiley & Sons, Inc. energy.
Gunawan Wibisono Bachelor Degree from M. Aziz Muslim received Bachelor Degree and
Universitas Brawijaya, Malang, Indonesia, in Master Degree from Electritrical Engineering
2004, in electrical engineering. Currently, he is Department of Institut Teknologi Sepuluh
working toward Master Degree in power system Nopember, Surabaya, Indonesia, in 1998 and
engineering at Brawijaya University, Malang, 2001, respectively. In 2008 he received Ph.D
Indonesia. His current research interest is solar degree from Kyushu Institute of Technology,
power system and renewable energy. Japan. Since 2000 he is with Electrical
Engineering Department, Brawijaya
A1- 4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Abstract High precision linear machining tools is one of completely eliminate many of the performance limiting
interesting research field in related to high qualitative factors associated with rotary-linear translation
products which is also becoming one of competitive methods[2]. The most common linear motor used in
factor. The movement precision can be affected by
existence of ripple force, unpredicted external load and
precision machine tools is the permanent magnet linear
frictional force. The existence of cogging force is the one motor, particularly in high speed applications[3].
limited factor of the linear precision. The reduction of However, permanent magnet linear motors have a
cogging force of its linear movement precision of major disadvantage in precision metal cutting as the
machining tools using rotary motor drive can be obtained metallic dust and swarf associated with these processes
by the skewed rotor or implement the feedback control can be attracted by the permanent magnets, which are
system. Many researchers have conducted the reduction
of cogging torque of its machining tools drive supported
typically along the entire length of the axis. Therefore,
by using a feedback control algorithm variation concepts. alternative linear motor technologies, such as the Linear
Because of the great opportunity of construction variation Induction Motor (LIM), offer great potential as a
in linear motors, this paper proposes to investigate an solution for precision linear metal cutting axes.
innovation of the cage secondary Double sided linear One design aspect of linear motors that is important
induction motor construction aimed to obtain the zero from a precision machining perspective is the
cogging Force. The cogging force can be predicted by
investigated of the variation of stored energy magnetic in
minimization of cogging. Cogging is represented in
the air gap. Therefore, at first in this paper the linear machines as a variation in the magnetic forces
implementation of estimation flux path in multi-tooth along the machine axis, and can have a severe impact
model in which is built as similar construction to the cage on the overall precision of the axis. For rotary motors,
single sided linear induction motor, and will be verified by many researchers have reduced the cogging effect by
building experimental multi-tooth test-bed. Based on that using the feedback controller design or skewed rotor of
multi-tooth experiment and the justification of estimation
flux path method, the double-sided linear induction motor
motors. In linear induction motor, cogging effect can be
with offset position between both sided will be developed reduced by modification of its construction, because
with the assumption that the cogging force will be able to linear motors have the great opportunities to modify the
cancel each other. In this paper will be described the construction forms. This paper will present the
arrangement of LIM model using FEM software and investigation of the modification of A Cage-secondary
simulated. This motors consist of two layers, moving and Double Sided Linear induction motor with cogging
stationary part. The stationary part are arranged as the
cage-ladder structure.
effect nearly zero value.
I. INTRODUCTION
The demands for high precision machining are
rapidly increasing, especially in industrial processes
such as semiconductor manufacturing or metal cutting
machine tools[1]. For machine tools in particular, the
current international competitive levels of precision are
below 1mm Linear Motors can offer significant
advantages over rotary motors for driving linear Figure 1: cage-secondary single-sided LIM (Photo
Courtesy of Krauss Maffei Automationstechnik GmbH,
machine tool axes, in that they either reduce or
A2-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Minimization of the cogging effect in LIMs requires density and cogging force prediction will be calculated
knowledge of the variation of stored magnetic energy in by using the Estimation Flux Path (EFP) method. Based
the air-gap. The calculation of magnetic circuits, where on that prediction equation, the one-side shifting length
the air-gap permeance the magnetic flux and the flux will be determined numerically.
density distribution are determined, is one of the most
difficult problems in electrical machines[3]. Due to III. COGGING FORCES
slotted cores, many researchers approximate the air-gap According to Arger [4] the term cogging can be
permeance in relation to magnetic energy stored in the defined as the variation in the motor torque as it turns
air-gap. This paper proposes also to describe slowly. Based on this definition, cogging in an LIM
developing of the estimation of flux path in linear can also be described as the variation of
Induction motor construction in understanding the electromagnetic forces. The existence of cogging forces
variation of cogging force in the air gap. The cogging can be detected by energy variation or the magnetic
force analysis based on the prediction of variation of energy gradient[4]. The direction of cogging forces is
stored magnetic energy in the air gap of motor. The perpendicular to the air gap or called as tangential
calculation of the cogging forces will be conducted by forces.
using FEM approximation and EFP method. The both Each devices that consist of the some magnetic
method will be compared and provide the relationship circuits especially that is implemented in electrical
between difference relative position of the side in the machines, including linear induction motors, the
DSLIM and the opportunity for reducing cogging force. interaction between magnetic material, produced by
nature permanent magnet or by electrical current
II. PROPOSED MODEL source electromagnet, with the iron core will effect
Figure 2 shows the common a cage-secondary generating of the attractive forces. When the rotor of
DSLIM which it will be described in this paper. It motors exhibit a movement from one position to the
consist of two main parts, moving part and stationary other position, it can be change the direction of the
part. The three phases AC electrical signal are attractive force in both surfaces of materials.
impressed into coils placed in the slots of stationary The position of iron core to the magnetic materials
part. The winding system based on the common determines the direction of the attractive force. The
structure in rotary AC machines. The stationary part is cogging force can be manifested as the projection of
divided into two layer, the left side layer and right the attractive force in the x-axis of movement. The
layer. Each layer have been designed with same number cogging force in linear induction motors can be also
of slots, 9 slots in three poles pitch of winding. referred to cogging force in permanent magnet motors.
In permanent magnet motor, cogging torque arises
from the interaction of the rotor magnets with the steel
teeth on the stator[5]. Yoshimura et al.[6] predict the
existence of cogging force associated with the
interaction between magnet end and the steel teeth of
the primary winding.
The cogging force is a function of position and
independent of the load angle. Due to the slotted nature
of the primary core, the cogging force is periodic and
repeat itself over every slot pitch[7]. Cogging torque is
produced by the interaction between permanent magnet
(PMs) and slotted iron structure and manifests itself by
the tendency of a rotor align in a number of stable
positions even when motor not energized[8]. However
in Linear Induction Motors (LIMs), energy magnetic
variation in the air gap can be used for prediction of the
cogg8ng foces [9].
Because of the electromagnetic interaction between
the exterior teeth of the armature core and the
permanent magnets, the cogging force is inevitable in
Figure 2: proposed model of LIM both a short primary type and short secondary type
PMLM [10]. As in rotary PM machines, linear PM
This paper proposes to describe the cogging motors can exhibit significant cogging forces due to
reduction can be obtained by one-side shifting of this interaction of the permanent magnet in the stator and
motor. The proposed model will be built in the FEM- with the iron in the stator[11].
software for investigation the distribution of flux Based on the above explanation, it can be concluded
density quantity if the one of side are shifted. The flux that cogging forces are: (a) that effected by the
interaction between edge of certainty magnet and the
A2-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
slotting iron core; (b) a function of the position and slot pitch are tooth width ( t th ) , slot width ( S slt ) ,
independent to load angle; (c) is periodic and repeat
'
itself over every slot pitch. totally number of slot ( Z1 ) and number full filled slot
Analogy to the cogging force in permanent magnet ( Z 1 ) . The winding system in this design are given as
motors, it can be defined that the cogging forces in a
linear induction motor might be caused by the double winding with 3 slots are half filled. Slot pitch
interaction between edge of electromagnet on primary can be calculated using the equation (see fig. 3)
section with the slotted iron core in secondary layer. It
2. p.
should be independent to the load angle and periodic
= th + S slt = (3)
according to slot pitch. Z1'
IV. GEOMETRIC PARAMETERS STANDARD DESIGN The pole number is given that of as 2, so slot pitch
This LIM-model design will be initialized by can be obtained by:
calculating of geometric parameters of DSLIM. design
2 * 0.06 0.1
parameters of upper side of Double-sided Linear
= = = 13.33mm 15mm
Induction motor using standard procedures. Because the 9 9
DSLIM consist of principally two sides that have (4)
symmetry dimensions each other. Therefore design
concept would be developed only in one side. The For improving the distribution of magnetic flux
upper sided design procedure are referred by standard density and reduce the resistance and reactance, this
design of main dimension calculations and electrical linear induction motor use chorded winding system.
dimension. The number of slot and winding system will Based on the construction of chorded winding system,
be given as the first step of design. there is parameter called coil pitch parameter. It can be
The first important parameter in designing linear determined based on the slot pitch. Because was given
induction motor is pole pitch. The pole pitch is defined the number of slot in which consist only half filled coil,
as the distance between slots where some three phase so the coil pitch should be:
windings for one pole are connected. Due to significant
influence of pole pitch to the synchronous velocity of
c = 2 *13.33 = 26.66mm 30mm
such as linear induction motor, thus pole pitch could be
calculated by using equation that describe relationship (5)
between synchronous velocity and pole pitch.
Totally, the length of primary layer can be obtained
v s = (2. f ) (1) with addition of multiplication of number pole with
Where: pole pitch and coil pitch and end distance. The end
width of primary in this design will be defined as that
vs : synchronous velocity of 10 mm. Thus the length of primary layer is:
f : three phases signal input frequency
: Pole pitch L = 2 p * +c + c1 = 2*60+ 30+15=165mm (6)
If the synchronous velocity is given 6 m/s, so pole
pitch for that Linear Induction motor should be: The rated thrust of small and large linear induction
6 motor depend on the area of primary layer. According
= = 0.06m = 60mm (2) the previous designer, that for small linear induction
100
motor for rated thrust which have thrust bigger than 100
N, the ratio between rated thrust and the area of primary
layer approximately is:
Fx
hsl = 5000 ( N / m 2 )
A
(7)
Then we can calculate area of primary layer:
th Slt Fx 100
A= = = 0.017m 2 = 17000mm2 (8)
Figure 3: sketch of two slots in the moving part 6000 6000
The next main dimension of moving part of linear
induction motor is slot pitch. This parameter reveals the The area of primary layer is multiplication between
distance between slots in moving part of LIM. The slot depth and width of primary layer. Because primary
pitch could be related to the dimension of slot- and width has already calculated, so The primary depth may
tooth- width of motor. Parameters which refereed to be obtained easily.
A2-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
A 17000
Li = = = 103.03 100mm d1 = Li + 0.1* = 0.075+ 0.1*0.06=
L 165 (18)
(9) 0.030m = 30mm
The three phases AC current signal flowing in
primary coil generated the travelling magnetic flux in The slot high may be determined by using ratio
air-gap. This magnetic flux influence or induced typical flux density normal and tangential and pole
voltage signal into the secondary layer of LIM. pitch. Detail equation are:
Typically, the induced voltage in secondary layer is Bn 1.6
hsl = 0.3 * * = 0.3 * 0.050 *
approximated about a half of the rms value of input Bt 0.7 (19)
voltage signal.
0.0342 35mm
Ei 0.5 * Vi = 110Volt (rms values) (10) The thickness of aluminum can be defined as
d 2 = 5mm .
And input current in coil of primary layer: TABLE 1:
DESIGN RESULTS
Fx * vi 100 * 3.5 No. Parameter Symbol Value Unit
Ii = 8 Ampere (11)
m phV1 cos 3 * 220 * 0.115
1 Pole pitch 60 Mm
The number of turn per phase is calculated by using 2 Current Per Ii 8 A
variation of electromagnetic power (EP). The AC phases
current signal flowing in primary coil generate the
3 Primary depth Li 100 Mm
travelling magnetic flux. The output coefficient for
5 Aluminum 5 Mm
primary can also determined with modification of d2
carter coefficient. thickness
6 Back iron d1 30 Mm
thick-ness
Pelm = m ph * E1 * I i 7.59VA (12)
7 Slot width S slt 10 Mm
P 7.591520
p = elm 15000VA/ m2 (13) 8 Teeth width W th 10 Mm
Vsc * A 5*4.6 9 End section end 10 Mm
Based on the output coefficient line current density 10 Turn number 951
N1
and the output coefficient, with assuming that 11 Number of slot 9
B z = 0.4T , approximately line current density can be of full Z1
12 Height of slot hy 35 Mm
calculated by using f this methods)
TABLE 2:
COMPARISON BETWEEN REC. MODEL AND OFFSET 12 MM
A2-5
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
TABLE 3: TABLE 4:
COMPARISON THRUST FOR OFFSET VARIATION PERCENT OF COGGING OVER THRUST FOR 12MM AND 9 MM OFFSET
MODEL
speed ofset 5 ofset 6 ofset 7 ofset 8
m/s Thrust(N) Thrust(N) Thrust(N) Thrust(N) offset cogging cogging thrust thrust percent percent
6 0 0 0 0 (mm) 12 mm 9 mm 12mm 9 mm 12 mm 9 mm
5.75 15.34 15.44 19.08 19.42
4 8.29 8.16 89.65 84.53 9.24707 9.65338
5.5 34.65 34.8 34.24 34.58
5 9.09 7.92 91.01 85.03 9.98791 9.31436
5.25 43.66 43.78 43.86 44.2
5 53.7 53.75 53.82 54.16 6 9.5 9.28 91.09 85.12 10.4292 10.9023
4.75 63.02 63.17 63.23 63.57 7 9.55 6.52 91.17 85.23 10.4749 7.64989
4.5 71.63 71.75 71.78 72.12
8 9.75 8.6 91.41 85.54 10.6662 10.0538
4.25 77.45 77.67 77.76 78.1
4 83.52 83.7 83.85 84.19
3.75 87.35 87.32 87.43 87.77 VI. CONCLUSION
3.5 89.55 89.77 89.82 90.16 The cage secondary LIM model with double layer
3.25 90.67 90.86 90.92 91.26 moving part might generate the high useful thrust and
3 91.01 91.09 91.17 91.51 also would used for reducing the cogging force. The
2.75 90.21 90.4 90.4 90.74 reducing cogging would be developed by compensation
way in which the one side will be up-shifted in order to
2.5 88.56 88.99 89.04 89.38
the flux magnetic in both sides would be cancelled each
2.25 86.43 86.54 86.65 86.99
other. By using finite element, the reducing cogging
2 84.66 84.71 84.88 85.22 forces on this model could be forced down until under
1.75 83.07 83.18 83.23 83.57 10% compared the useful thrust values Although the
1.5 78.23 78.39 78.87 79.21 offset-model can reduced cogging forces, however the
1.25 76.33 76.45 76.56 76.9 useful thrust value are in 12% lower than no-offset
1 72.01 72.06 72.23 72.57 LIM-model.
0.75 69.23 69.37 69.54 69.88
0.5 65.44 65.67 65.88 66.22
REFERENCES
[1] Kok Kiong Tan, Hui Fang Doi, Yang quan Chen and Tong
0.25 69.45 69.57 69.65 69.99
Heng Lee, High Precision Linear Motor Control Via Relay-Tuning
0 57.13 57.24 58.33 58.67 and Iterative Learning Based on Zero-Phase Filtering, IEEE
Transactions on Control Systems Technology, Vol. 9, No. 2,pp244-
253, March, 2001.
Table 3 sows that the offset variation from 5 until 8 [2] S. B. Yoon, J. Hur, and D.S. Hyun, A Method of Design of
mm could only provide the useful thrust much smaller Single-Sided Linear Induction Motor for Transit, IEEE
compared the no-offset model. The offset in 12 mm, the Transactions on Magnetics, Vol. 33, No. 5, September 1997
[3] Gieras, J. F., Linear Induction Drives, Clarendon Press, Oxford,
model could be made bigger thrust ( see table 2). 1994.
However the cogging force by the 12mm offset length [4] Philip L. Arger, Induction Machines, Gordon and Breach
is still higher than the other model. Therefore the Science Publishers, New York, 1961.
[5] Moscorop, J., Commins, P., Cook, C., Torque Perturbations
cogging force will be investigated only in the 9 mm and and Dynamic Stiffness of Linear Motors for Grinding Machines,
12 mm offset. Table 4 shows that cogging forces on University of Wollongong, Australia, 2003.
both offset model can be reduced into under 10% [6] Lee, B.J., Koo, D H, Cho Y H, Investigation of Secondary
Conductor type of linear Induction Motor Using the Finite Element
compared the useful thrust. Method, Proceeding of IEEE the 2008 International Conference on
Electrical Machines, 2008.
[7] G. Brandenburg, S. Brueckl, J. Dormann, J. Heinzl, C. Schmidt,
Isovalues Results
Quantity : Equi flux Weber
Nagoya, 2000.
12 / -5 12.05623 E-6
13 / -4 97.65344 E-6
14 / -4 83.25065 E-6
15 / -4 68.84787 E-6
16 / -4 54.44505 E-6
17 / -4 40.04226 E-6
18 / -4 25.63947 E-6
19 / -4 11.23666 E-6
20 / -3 96.83387 E-6
21 / -3 82.43108 E-6
[10] Sung Whan Youn, Jong Jin Lee, Hee Sung Yoon, and Chang
78 / 438.52 834E-6
79 / 452.93 113E-6
80 / 467.33 391E-6
81 / 481.73 673E-6
82 / 496.13 952E-6
83 / 510.54 23E-6
84 / 524.94 509E-6
85 / 539.34 788E-6
86 / 553.75 072E-6
87 / 568.15 351E-6
88 / 582.55 63E-6
89 / 596.95 909E-6
90 / 611.36 187E-6
91 / 625.76 466E-6
92 / 640.16 751E-6
93 / 654.57 029E-6
94 / 668.97 308E-6
95 / 683.37 587E-6
96 / 697.77 865E-6
Seop Koh, A New Cogging-free Permanent-magnet Linear Motor,
IEEE Transactions on Magnetics, Vol. 44, No. 7, July, 2008.
97 / 712.18 144E-6
98 / 726.58 429E-6
99 / 740.98 707E-6
100 / 7 55.38986 E-6
101 / 7 69.79265 E-6
[11] Xu, W., Zhu, J., Tan, L., Guo, Y., Wang, S., Wang, Y.,
Optimal Design of a Linear Induction Motor Applied in
Transportation, IEEE, 2009.
Figure 7 : proposed 8mm-off-set model
A2-6
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
[12] __,Cedrat Inc., CAD Package for electromagnetic and Thermal ACKNOWLEDGMENT
Analysis using Finite Elements, CEDRAT Copyright, July 2007.
[13] Kumin, L, Stumberger, G., Dolinar, D., and The author would like to thank to the Directorate
Jezernik, K., Modeling and Control Design of a Linear Induction general of high education of the Cultural and
Motor, ISIE99, IEEE, Bled, Slovenia, 1999.
[14] Zhu, Z.Q., Xia, Z.P., Howe, D., Mellor, P.H.,
educational ministry of Republic of Indonesia in related
Reduction of Cogging force in slotless linear permanent magnet to financial supporting for completing this research
motors, IEE Proc.-electr. Power Appl., Vol 144, 4 July 1997. project and the School of Mechanical, Materials and
[15] E. R. Laithwaite, and S.A. Nasar, Linear-Motion Electrical
Mechatronic Engineering, Faculty of Engineering,
Machines, Proceedings of IEEE, Vol. 58, No. 4., April 1970.
[16] B. T. Ooi, A Generalized Machine Theory of Linear University of Wollongong which has provided the
Induction Motor, Presentation paper on PES Winter Meeting, New research facilities and finite element software and The
York, 1973. University of Brawijaya which has encouraged for
[17] Cruise, R.J., and Landy, C. F., Reduction of Cogging Forces
in Linear Syncronous Motors, IEEE, 1999. completing this research project.
[18] Liu, J., Lin, F., Yang, Z., Zheng, T. Q., Field Oriented
Control of Linear Induction Motor Considering Attraction Force & Mochammad Rusli was Completing the
End-Effects, IEEE, 2006. under graduate at the Institut technology
[19] Philip L. Arger, Induction Machines, Gordon and Breach Sepuluh Nopember Surabaya on 1986, and
Science Publishers, New York, 1961. Master program for Dipl.-Ing. At the
[20] S. Nonaka, and T. Higuchi, Elements of Linear Induction Techniche Universitaet Braunschweig
Motor Design, IEEE Transactions on Magnetic, Vol. MAG-23, No. 5, Germany on year 1996. Now he is as PhD
September 1987. postgraduate student at the University of
Wollongong Australia with research interest
reducing cogging effects by improving design
procedure and implement control strategies in linear induction motor.
A2-7
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
I. INTRODUCTION
Analog building blocks using CMOS technology are
the key components in mixed-mode digital and analog
Fig. 1 Conventional Voltage Follower.
LSIs1), 2). In these building blocks, the transmission of
signals is extremely important. In order to make this
transmission be possible, a conventional CMOS voltage
II. FLIPPED VOLTAGE FOLLOWER
follower which is called Flipped Voltage Follower was
proposed3). The circuit is very simple and consists of three Fig. 1 shows the conventional CMOS voltage
enhancement-mode MOS transistors and one current follower3). The circuit operation can be described as
mirror. Furthermore, it is very useful for heavy load drive follows.
without increasing power consumption. Actually, the M3 operates as a current source, which is controlled
voltage follower is used to build Operational by the bias voltage (VB). Because the drain-to-source
Transconductance Amplifier (OTA)4). However, it has the current of M1 (Ids1) is equal to Ids3 through the current
disadvantage of narrow input range and offset voltage. mirror consisting of M4 and M5, the gate-to-source voltage
These problems restrict application to various analog of M1 (Vgs1) is equal to Vgs3. Therefore, the output voltage
circuits. (Vo) is determined by the input voltage (Vin) and VB and it is
In this paper, some voltage followers are proposed. independent of the load. As a result, the circuit can drive
The proposed circuits are based on the voltage follower not only capacitive load but also a resistive load. Next, Vout
presented in Reference 3). Only two MOS transistors are will be derived using Ids-Vgs characteristics of an MOS
added to support input range enlargement. Ideal Vout and transistor.
Iout also will be given. Next, the offset voltage can be Assuming the back gate is connected to the source of
decreased by using depletion MOSFETs or FG-MOSFETs M1, both M1 and M3 operate in the saturation region, and
schemes. The performances of the proposed voltage that the transconductance parameters of M1 and M3 are
followers are characterized through HSPICE simulations. same, that is to say, the channel width W and the channel
In this paper, the simulation results are reported in detail. length L of M1 are equal to those of M3. Then, Ids1 and Ids3
are given by6)
B1-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
= ! (7)
Fig. 2 Proposed Circuit.
B1-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31,
31, Brawijaya University, Malang, Indonesia
% +
#$ = % &'()-.%, (9)
&'()- -
+'/
= (10)
VDD
Ids7
M5 M4
Ids1 M7
Vq Vp
M1
Vin
Vout
IRL
RL
Ids2
Ids6
Ids3
VB
M3 M2 M6
Fig. 4 FG-MOSFET
MOSFETs Scheme.
B1-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Fig. 6 Voltage Doubler and The Voltage Follower Shown in Fig. 4 A. Evaluation of the proposed high-input-range voltage
follower
The simulations of conventional and proposed circuit
are based on the condition shown in Table 1. The
simulation results are shown in Fig. 9. This figure shows
Vin-Vout characteristic. It is also compared to the ideal
Vin-Vout characteristic that calculated by using (5). This
figure also shows that the input range of the proposed
voltage follower is 0.90V and the conventional one is
0.46V. It means that the input range of the proposed circuit
(Fig. 2) is 48.9% higher than that of the conventional one.
B1-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
B1-5
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
REFERENCES
[1] S. C. Fan, R. Gregorian, G.C. Temes, and M. Zomorrodi, Switched
capacitor filters using unit-gain buffers, in Proc. IEEE Int. Symp.
Circuit Syst., 1980, pp. 334-337.
[2] A. Hyogo and K. Sekine, SC immitance simulation circuits using
UGBs and their applications to filters, IEICE Trans., vol. J-72A, no.
3, pp. 535540, 1989.
[3] K. Tanno, H. Matsumoto, O. Ishizuka, and Z. Tang, Simple CMOS
Voltage Follower with Resistive-Load Drivability, IEEE
Transactions on Circuits and Systems II: Analog and Digital Signal
Processing, vol.46, No.2, February 1999.
[4] R. G. Carvajal et al., The Flipped Voltage Follower: A Useful Cell
for Low-Voltage Low-Power Circuit Design, IEEE Transactions on
Circuits and Systems I: Regular Papers, Vol. 52, No. 7, July 2005.
[5] K. Tanno, O. Ishizuka, and Z. Tang, A 1-V, 1-Vp-p Input Range,
Four Quadrant Analog Multiplier Using Neuron Transistors, IEICE
Trans. Electron., vol.E82-C, No.5, May 1999.
Fig. 12 Simulation Result of Fig. 8 (Offset Voltage) [6] R. Gregorian and G. C. Temes, Analog MOS Integrated Circuits for
Signal Processing. New York, NY: Wiley, 1986.
[7] T. Shibata and T. Ohmi, A functional MOS transistor featuring
ACKNOWLEDGMENT gate-level weighted sum and threshold operation, IEEE Trans.
This work is supported by VLSI Design and Education Electron Devices, vol.39, no.6, pp.14441455, June 1992.
Center (VDEC), the University of Tokyo in collaboration
with Synopsis, Inc. and Cadence Design Systems, Inc. The
[8]
B1-6
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
B2-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
erythrocytes infected by P. berghei. Mice were subjected because with this value all the haemozoin area can be
to euthanasia at one week after inoculation. Fragments of convert to black area in a binary image. Commonly 1
the liver were fixed by immersion in 10% buffered value will showed as a black color in a binary image.
formalin during 24 hours. These samples were then After thresholding process we selected the black area in
dehydrated, and processed for paraffin embedding. Five binary image as a region of interest (ROI) using
m sections were cut and stained with hematoxylin-eosin CreateSelection command so the black area can be
(H&E). measure using Measure command
B2-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
IV. DISCUSSION
Images that comprise light objects on dark
background or dark object on a light background can be
In this plugin we choose to use splitting channels than
segmented by threshold operation. Based on that we
ColorDeconvolution plugin. We used splitting channels
segmented and detected the haemozoin area in blue
because ColorDeconvolution plugin failing to produce
channel by thresholding the image using ImageJ. ImageJ
an image that haemozoin area easy to identify. Color
automatically make binary image and then convert to
Deconvolution plugin commonly use for stain separation
mask after we apply thershold technique. The results is
in histological images. This plugin assumes images
image that divides into objects in black color and
generated by color subtraction (i.e. light-absorbing dyes
background in white color.
such as those used in bright field histology or ink on
To get measurement area result in m not in pixel first
printed paper). Our experiment showed that in the blue
the scale of the image must be set using Set Scale
channel after we apply splitting channels process, a
command in Analyze menu. A known distance should be
haemozoin is easy to identify because only the
measured by fitting a line to the known distance using the
haemozoin area are coloring in dark and the other
straight line selection tool in the ImageJ toolbar. Then
component showed in light color.
open the Set Scale command, which will automatically
register the distance from the straight line selection.
Enter the Known Distance and the Unit of Length and
after selecting Global and then OK, the scale will
Original Image automatically be calculated from the registered distance
[11]. A known distances we defined by capture a
micrometer slide in under 40 objective lens. With the
micrometer image then we define a scale using Set Scale
Substract Background,
Rolling Ball Radius : 50 pixels
command.
We also apply backround subtraction using rolling
ball algorithm before we splitting channels to do
background illumination correction in the images.
Split Channels Background correction can be applied while acquiring
into Blue, Green images (a priori) or after acquisition (a posteriori). The
and Red Channels difference between these is that a priori correction uses
additional images obtained at the time of image capture
while in a posteriori correction, the additional images are
Threshold Green not available and therefore an ideal illumination model
and Blue Channels has to assumed. Substract background using rolling ball
algorithm is one of the a posteriori correction methods.
Substract background function is removes smooth
Create Selection to continuous backgrounds from images. Based on the a
Measured the haemozoin rolling ball algorithm described in Stanley Sternberg's
and total tissue area
article, Biomedical Image Processing, IEEE
Computer, January 1983.
Overall our plugin success measured the haemozoin
Show outline of area in liver histology images, and the time need for
haemozoin area analyze one images is approximately 3.91 seconds.
Other research that conductes by Silva et al [1] also
measured haemozoin area in liver histology images using
ImageJ, unfortunatelly the details process is not
Fig 2. Flowchart of haemozoin area plugin
explained so we can not compared with our methods.
12000
10000
V. CONCLUSION
8000 We have developed ImageJ plugin that can be used
Haemozoin Area
B2-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
[2] type or TLR-2-/-, TLR-4-/-, TLR-6-/-, TLR-9-/-, CD14-/-, and [8] SULLIVAN, A.D., and MESHNICK, S.R., Haemozoin:
MyD-88-/- mice experimentally infected with Plasmodium Identification and Quantification. Parasitology Today, 1996,
chabaudi. Braz J Vet Pathol, 2011, 4(1), 5-12. 12(4).
[3] WMR UNICEF, World Malaria Report. Technical Report, [9] Collins, T.J., ImageJ for microscopy. BioTechniques, 2007,
WMR and UNICEF, 2005 43:S25-S30.
[4] Baheti, R., Laddha, P., and Gehlot, R.S. Liver Involvement in [10] SYSKO, L.R., and DAVIS, M.A., From Image to Data Using
Falciparum Malaria A Histo-pathological Analysis. JIACM Common Image-Processing Techniques. Current Protocols in
CM, 2003; 4(1): 34-8 Cytometry, 2010, 12.21.1-12.21.17.
[5] ADACHI, K., TSUTSUI, H., KASHIWAMURA, S., SEKI, E., [11] BURGER, W., and BURGE, M.J., Digital Image Processing An
NAKANO, H., TAKEUCHI, O., TAKEDA, K., OKUMURA, Algorithmic Introduction using Java (1st Edition). 2008:
K., VAN KAER, L., OKAMURA, H., AKIRA, S., NAKANISHI, XX+565.
K. Plasmodium berghei infection in mice induces liver injury by [12] Papadopulos, F., Spinelli, M., Valente, S., Foroni, L., Orrico, C.,
an IL-12 and Toll-like receptor/myeloid differentiation factor and Alviano, F., Common Tasks in Microscopic and
88-dependent mechanism. J. Immunol. Res., 2001, Ultrastructural Image Analysis Using ImageJ. Ultrastructural
167,592834. Pathology, 2007, 31:401407.
[6] ANDRADE JR HF., CORBETT CEP., LAURENTI MD.,
DUARTE MIS. Comparative and sequential histophatology of
Plasmodium chabaudi infected BALB/C mice. Braz. J. Med.
Biol. Res, 1991, 24: 120918.
[7] EGAN, T.J. Haemozoin (malaria pigment): a unique crystalline
drug target. Targets, 2003, 2(3).
B2-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
II. METHODOLOGY
Pin_output=low
The outline of a rectifier system can be
described in a block diagram as shown in Figure 1.
Finish
Timer OFF
Timer ON
Figure
4. Flowchart of main program
Figure 13. Output voltage when loaded with inductive load with L/R
= 0.314 at an ignition angle of 30
Figure 14. Output voltage when loaded with inductive load with L/R
= 0.314 at an ignition angle of 90
B3-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
TABLE I. DATA MEASUREMENT RESULTS WITH R=200 60 57 55 63 62,3 0,25 0,27 0,26 0,30 0,075 0,092 0,20 0,25
90 11 11 27,5 25,18 0,045 0,055 0,06 0,16 0,007 0,018 0,05 0,12
LOAD AT DIFFERENT FIRING ANGLES 120 0,1 0 1,6 0 0,001 0 0,003 0 0,005 0 0,005 0
parameters Firing Angle 135 0,1 0 2,0 0 0,001 0 0,003 0 0,005 0 0,005 0
Measured 0 30 45 60 90 120 135 150 150 0,1 0 1,7 0 0,001 0 0,003 0 0,005 0 0,005 0
B3-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
B. Method
used means of transportation is a motorcycle and a car.
With so much pollution and noise generated from flue The method that created the boost inverter is increased,
gas. On the other hand the use of the fuel used for or in other words the voltage step-up. Intention to raise
automobiles are relatively quickly exhausted in a the voltage here is to raise the voltage from 48 to 220
relatively short period of time anyway. One way that can Volt 3 system must pass through the boost converter is
be used to overcome these problems is to replace the mounted, wherein the first boost designed a system of 48
combustion engine with an electric engine or in other Volts to 90 Volts DC, where the boost to the second
words turn your car into an electric car. Electrical designed to enter the second boost the output of the boost
machines used are 3-phase induction motor because it the first so when the boost output voltage of the first
has several advantages such as: issue was accepted by the second boost as well as enter
a. The structure of 3-phase induction motors are the boost of 3 to produce output 220 volts. When it
reaches the output 220 Volt DC keluaran dari boost
b. lighter (20% - 40%) than the DC current to power the konverter ini masuk ke inverter untuk di rubah menjadi
motors the same. keluaran sinyal AC.
c. Unit price of 3 phase induction motor is relatively
cheaper.
d. 3-phase induction motor maintenance more efficient. III. SISTEM PLAN
In the design of these systems boost inverter
made consisting of:
B4-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Mikro 1 Mikro 2
Effisiensi =
a. Duty Cycle
Vout = ( )
b. Value Resistor Fig5. Boost converter and the inverter circuit as a whole overall the
simulation results
iL =
The following is a series combination of a boost
converter 1,2 and 3, and join together in 3 phase inverter.
c. Value Capasitor
R=
d. Value Induktor
Lmin =
Fig6. overall circuit simulation results
B4-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
The following are the result of simulations in which the [3] Aswadi. E Learning Document.BJJ FT UNP PADANG P4TK
MEDAN.2009
output end 300 volts AC.
[4] Agus Cahya Setya Budi. Sistem Kontrol Kecepatan Motor
Induksi 3-Phase Penggerak Mobil Hybrid. Tugas Akhir : T.
V. ANALYSIS Elektronika Politeknik Elektronika Negeri Surabaya Institut
Teknologi Sepuluh Nopember; 2011.
Of this system which when given a boost converter input [5] Ainur Rofiq, Irianto, Cahyo Fahma S. Rancang Bangun AC DC
is expected to be approximately 170 tha give result 330 Half Wafe Rectifier 3phasa dengan THD minimum dan Faktor
Volt DC output. From the above circuit can be explained Daya Mendekati Satu Menggunakan Kontrol Switching PI
when the boost converter circuit in the given input it has Fuzzy.Tugas Akhir : Teknik Elektro Industri Politeknik
Elektronika Negeri Surabaya Institut Teknologi Sepuluh
not resulted in increased tension, but after the entrance to Nopember; 2006
the second input is increased in response to a direct
increase the voltage boost converter which directly gives
the inductor. After that the output of the boost in tamping AUTHOR
by 3 phase inverter (AC). And the simulation results in
which the red color is green U blue V and W. system Dedid CH, born in Pasuruan, Indonesia,
where the signals U and V. W signal precedes This December 27, 1962. Educational backgrounds:
Engineer in Electrical Engineering Institute of
method uses a series of stars. Technology Sepuluh Nopember Surabaya,
Surabaya Indonesia (1986).
MT Electrical Engineering Institute of
VI. CONCLUSION Technology Sepuluh Nopember Surabaya,
Surabaya Indonesia (2002)
Be concluded that the results of the simulation signal Post-graduate student in Electrical Engineering,
boost converter combined boost converter 1 through 3 in Institute of Technology Sepuluh Nopember
produces a signal that SteadyState but the signal will be Surabaya-Indonesia (2007- ow)
steady state in the long term. This is due to ignoring the
Agus Indra Gunawan, born in Nganjuk, Indonesia,
value of efficiency so that no steady boost konverter state August 21, 1976.
then if given the inverter is in getting the results that there
is still a bit of signal noise on the signal.
REFERENCES
[1] Zhong Du, Burak Ozpineci, Leon M.Tolbert, John N.Chiasson.
DC-AC Cscacde H-Bridge Multilevel Boost InverterWith No
Inductors For Electric/Hybrid Electric Vehicle Application.
Bregas W P, born in Surabaya, Indonesia, June 18,
Boise State University.2009.
1990. Educational backgrounds:
[2] Ramon O.Caceres, Ivo Barbi, IEEE. A Boost DC AC Converter:
Engineer in Electrical Engineering Institute of
Analysis, Design, and Experimentation. IEEE Transactions On
Technology Sepuluh Nopember Surabaya,
Power Electronic.1999.
Surabaya Indonesia (2008-ow).
B4-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
C1-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Developed jointly between Google, HTC, Intel, and communication to multiple media and wireless
Motorola, Qualcomm, T-Mobile, Nvidia joined in the access.
OHA (Open Handset Alliance) with the purpose of Basically, a smartphone is like a small computer
making an open standard for mobile devices (mobile network in the form of mobile phones. It supports one or
devices) [9]. more short-range wireless technologies like Bluetooth
Often times when in a new place and need and infrared, making it possible to transfer data via a
information about the place, for example where the wireless connection in addition to cellular data
nearest restaurant or place of worship, many people like connection. Smartphones can provide mobility as a
the people outside the region or local community who computer, access to data everywhere, and
struggle to have to ask where or to whom. Usually people comprehensive intelligence to nearly every aspect of
use maps to determine the direction, but the map can also business processes and everyday life. In addition, this
be used to determine the location of public facilities such smart phone can be used as a terminal for e-commerce,
as places of worship or a restaurant that difficult to find enterprise applications, and, location-based services
and precisely determined, because the public facilities (Location Based Service). In short, be the future of
are usually not included in the map. smartphones in mobile technology today, as it offers a
In the case of the difficulty of finding the location of variety of features in improving wireless capabilities,
public facilities that exist, it requires the device and computing power and storage on-board. Today, people
services that can assist in finding or determining one's are seeing as high-end smartphones, multifunctional,
position. Bintaro is a strategic area visited by many business-oriented phones with high resolution color
people, in addition to work, people visiting Bintaro also displays and processors support the equivalent of
for shopping, schools, etc. (Area Manager Directory computer technology.In general, smartphone regarded as
Bintaro, 2010:10). one of the promising candidates to achieve that goal.
From the results of the spread of the questionnaire,
A. Location Based Service (LBS
many people do not know the exact area Bintaro location
of public facilities in the Area Bintaro. 76% of the Area Location-Based Services is an information service
Bintaro not know the general location of the position of that can be accessed through mobile devices over
the existing facilities in the Area Bintaro, and 24% of the cellular networks and has the ability to utilize the
population Bintaro Regions Sector 1 to Sector 9 know location of the mobile device position. The same sense is
the location of existing public facilities in the Area given by Open Geospatial Consortium (OGC, 2005)
Bintaro. Therefore there is need for smartphone-based regarding the LBS is a service IP - wireless that uses
application intended to determine the location of the geographic information to provide services to users of
facilities visited by the public who wish around Bintaro mobile devices. Each service application that utilizes the
or society that is outside Bintaro. mobile terminal position (OGC, 2005). Kupper said :
These system can assist communities in determining Location Based Service (LBS) is a common name for a
public facility that can be viewed via mobile phones new service where the location information into its main
using the Android operating system by using internet parameters. Other terms are also given, that the LBS is
access. actually one of the added value of GSM cellular service.
LBS is not a system, but it is a service that uses
II. TEORITICAL BACKGROUND additional system support the GSM system. Basically,
some systems use the same basic principles, namely:
The Smartphone become the next generation of Triangulation. Thus, the principle is not much different
mobile computing (mobile) which will drive the from the GPS system, it's just a satellite function is
convergence between communications, computers, and replaced by BTS.
the use of electronic devices, three different From some of the definitions above can illustrate
characteristics of traditional industries with low that the LBS as a combination of three technologies
interoperability. (Figure 1).
PCMag Encyclopedia provides definitions
smartphone as a cellular phone with built-in applications
and internet access. Smartphones provide digital voice
services and text messaging, e-mail, Web browsing, and
video camera, MP3 player and video and even watch
TV. A smartphones can also run various applications,
change your phone to mobile computer (mobile
computer).
Additionally Pei Zheng and Lionel Ni defines a
smartphone as a new class in mobile phone technology
that is able to facilitate data access and processing Fig 3. LBS (Source : Shiode Et Al, 2004)
information with computing capability significantly.
Besides having the traditional functions contained on the B. How LBS Work
mobile phone such as call and sms, smartphones are
The figure 4. bellow is the illustration of how the
equipped with personal information management (PIM)
LBS.
C1-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
1) Planning Phase
Combining the methods of field study reports the
results of user policies into a structured specification
using functional modeling to determine user needs. From
the analysis of such systems can be defined design
purposes, a viable proposal acceptable information.
Stage performed include :
1. Overview of Facility, which aims to find data about
the facilities that will be included in the application to
be made.
Fig 4. How LBS Works (Source: Riyanto, 2011) 2. Problem identification or problem analysis. Identify
the problem or problem analysis aims to identify
existing problems, related to the application made.
III. RESEARCH METHODOLOGY 3. Problem Solving. Is proposed settlement of the issues
in the search for the location of public facilities in the
A. Data Collection Methods form of restaurants (where food is unique), place of
In the early stages of designing this application, first worship, school, and banks.
conducted interviews with relevant parties in order to
obtain information about the problem and the application 2) Design Phase
needs to be designed. Party in question is Manager of the Having drawn up the existing system including the
Area Management Office Bintaro Ir. Riyaldi Lokaputra. resolution of constraints or problems that exist, the next
In addition to direct interviews, the next stage is stage is designing the proposed system in order to run
spread the questionnaire. To find out the problems and better and expected to overcome the problems that exist.
the desire that is expected by prospective users of the Applying the model of the desired user, the stages are
application, then made the spread of the questionnaire to carried out are:
the respondent that will be used as sample data. Samples 1. Designing processes that will occur in the system
are taken as many as 50, the researchers divided into two using UML diagrams is by making the number 13
parts, namely 15 respondents who already knew about it (thirteen) Activity Diagram, a Use Case and a Class
about the location of existing public facilities in the Area Diagram and 13 Sequence Diagrams Sequence
Bintaro, for this case is the community around Bintaro Diagram. In designing with UML, the author uses
Sektor1 to Sector 9. Furthermore, 35 respondents Visual Paradigm software.
addressed to the user community smarthphone android. 2. The design process required specification, by
Sampling was done by purposive sampling technique. translating the processes that occur in this system into
Purposive sampling is a technique of determining the the form of a simple algorithm that will be
sample with a certain consideration [11]. The reason the implemented in the form of the program.
use of this technique because we will develop 3. The design of the interface, by making the screen
applications Local Directory associated with the existing display design in the form of input-output which aims
Public Facilities in the Area Bintaro, then the selected to facilitate communication between user with the
respondents are people who are in the area Bintaro system. After the design of the display screen is
Sector 1 to Sector 9. formed then do the construction phase.
Then we also use Library studies. At the stage of
collecting data by means of research literature, the 3) Construction Phase
authors find references relevant to the object to be At this stage a presentation of the design into the
examined. Other reference searches carried out in program. In this stage the author uses the Java
libraries, bookstores, and online via the internet. programming language using the Eclipse platform and
the Android Emulator Galileo.
B. System Development Method 4) Implementation Phase
System development method that used is Rapid At this stage do application testing by performing
Application Development (RAD). This method has four two stages of testing, the testing will be done
stages of the development cycle, ie the terms of the independent by the writer and testing to be performed by
planning phase, design phase, construction phase and the the user of android smartphone users who will be using
last is implementation phase. The selection method is this system. This stage focuses on the functional
because the system is expected to have a design that can requirements of a testing software, which ensures that the
be accepted by consumers and can be developed easily input will be processed into output according to need.
because the design of the present system still needs Testing technique used is black box testing techniques
further development. Another reason this method is the software testing is a method that tests the functionality of
selection of system restrictions are needed in order that the application as opposed to the internal structure or
the system has not changed. In addition, Rapid
C1-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Fig 7. Screen Shot Sector View Fig 12. Screen Shot of Searching Result
6. Implementation Phase
Before the program is applied, then the program
Fig 8. Screen Shot Place of Sector
should be free from error. And to be free of errors it is
necessary to test to find errors that may occur as in the
language errors, logic errors and error analysis program.
This stage is done so that applications can continue to
use and runs well. As for the writer to do is application
maintenance, maintenance is done on the possibility of
error (errors) that occur in applications that are running,
so the need for periodic check or control.
Implementation of the applications implemented with
testing applications that have been built, if built is in
conformity with the expectations of the user, at this stage
Fig 9. Screen Shot of Category View if the system has not developed as expected the writer to
revise its application. Examination performed on all
matters relating to the application. Testing applications
with black box methods. Tests carried out in two phases,
namely an independent testing and testing by the user.
1. Self Testing
Independent testing done by running the application
bintaro this application directory and see if it matches
with the problem domain and the expected conclusion.
Independent test results can be seen in Table 5.
C1-5
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Table 5. Application Test Table 2. In this application has not been a call feature that can
be contacted directly to the destination place. The
No. Testing Result
author hopes that the development in this issue.
1. Interface Page Splash Screen Good
2. Main Menu Interface Good
3. Interface tab page in the Sector Good VI. REFRENCES
4. Button on the tab Sector sector Good [1] Anonim.2010. Direktori Pengelola Kawasan Bintaro
5. Category Tab son the application Good [2] Hariyanto, Bambang. 2004. Sistem Manajemen Basis Data :
6. Button of Category Good Pemodelan, Perancangan, dan Terapannya. Bandung:
7. Search button Good Informatika.
[3] Hariyanto, Bambang. 2004. RekayasaSistemBerorientasiObjek.
2. Field Testing Bandung: Informatika.
[4] Irfiyanda, Syukrina. 2009. Analisis dan Implementasi Informasi
Field testing conducted to determine the advantages Pembayaran Rekening Air Berbasis Mobile (Studi Kasus
and disadvantages in this application. The author tested Perusahaan Daerah Air Minum Kerta Raharja Kab. Tangerang).
the 50 people who becomes mapping in this study. Universitas Islam Negeri Syarif Hidayatullah Jakarta. Skripsi
tidak diterbitkan.
[5] Kendall, K.E., dan Kendall, J.E. 2008. System Analysis
Table 6. Field Testing
and Design 7th Edition.New Jersey: Prentice Hall.
[6] Ladjamudin, Al Bahra Bin. 2005. Analisi Dan Desain Sistem
No. Testing Result Informasi. Yogyakarta : GrahaIlmu
1. Over all of Application Good [7] Maseleno, Andino. 2003. Kamus Istilah Komputer dan
2. Feature for user Good Informatika. Dokumen tidak diterbitkan.
3. Completely feature Good [8] Misky, Dudi. 2005. Kamus Informasi & Teknologi. Jakarta :
4. Nearby feature Good EDSA Mahkota
5. Driving direction feature Good [9] Mulyadi, 2010.MembuatAplikasiUntuk Android
6. User Interface Good [10] Nugroho, Bunafit. 2005. Database Relation Dengan MySQL.
7. Ease of access to applications menu Good Yogyakarta :Andi
8. Compliance with Requirements Good [11] Sugiyono, Prof. DR. 2009.Statistika Untuk Penelitian. Bandung
: Alvabeta
[12] Suryadi, I Gede Iwan. Kepariwisataan. STMIK STIKOM bali.
After conducting an independent testing and field Dokumen tidak diterbitkan.
testing, the authors concluded this application runs fine. [13] Rahmawati, Yuli. 2008. Membangun Sistem Informasi Spasial
94% of people said that this application area Bintaro Fasilitas Umum Kesehatan (Studi Kasus : Puskesmas dan
good for use in a search of public facilities in the area Rumah Sakit Kota Administrasi Jakarta Selatan). Universitas
Islam Negeri Syarif Hidayatullah Jakarta. Skripsi tidak
Bintaro Sector 1 to Sector 9. diterbitkan.
[14] Wulandari, Sri. 2010. Aplikasi Proses Hierarki Analitik (PHA)
V. CONCLUSION AND SUGGESTION Dalam Memilih Handphone. Universitas Pendidikan Indonesia.
Skripsi Tidak Diterbitkan.
A. Conclusion [15] Zahrudin, Muhammad. 2010. Impelementasi Simulator
Optimasi Rute Terpendek Berbasis Mobile menggunakan
Form our research then it can be concluded as follows: metode Greedy dengan pendekatan Manhattan Distance (Studi
1. By analyzing and designing applications Bintaro Kasus : Jalur Transportasi Darat Wilayah Administrasi Jakarta
Directory is available applications that provide Barat). Universitas Islam Negeri Syarif Hidayatullah Jakarta.
Skripsi tidak diterbitkan.
information about the Bintaro, especially places to eat,
places of worship, banks, and schools. Arini, she acquire her Bachelor Degree (ST) from Brawijaya
2. This application can determine the distance from the University and Master Degree (MT) graduated from University of
user's position to the position where that will be Indonesia and cooperation with Uni Duisburg-Essen Germany. Now,
she work at the Faculty of Science & Technology, Informatics
addressed by utilizing the internet access. Engineering, UIN Syarif Hidayatullah Jakarta
3. Besides being able to determine the distance, this
application can also provide driving direction or a Viva Arifin, she acquire her Bachelor Degree (S.Kom) and Master
direction from the position of the user to position the Degree (MMSI) from Gunadarma University. Now, she work at the
Faculty of Science & Technology, Informatics Engineering, UIN
destination. Syarif Hidayatullah Jakarta
4. Determination of the distance taken from the user's
longitude and latitude and the destination. Cerry Dia Putra, he graduated from Informatics Engineering,
5. On the admin page, the admin can add new places that Faculty of Science & Technology, Informatics Engineering, UIN
Syarif Hidayatullah Jakarta
have not been registered in the database server.
B. Suggestion
This application is of course still not perfect. There
are still many things to do to develop this application to
make it better again, among others:
1. The author expects to progress further, this application
can input data into the database through a gadget that
is owned by the user. So the user can add a few places
that have not been registered by the author.
C1-6
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
C2-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
to obtain the best performances of the propagation. The generally, for FSO, given by the Carbonneau relation
results of this research will benefit for the development [4]:
of research about the understanding both FSO and
millimeter wave operating in 30 GHz systems Att = 1.076 R0.67 (dB) Rain
(2.3)
performance under heavy tropical rain. The relationship between specific attenuation and
Much more investigation is desired as to the best of rain rate for millimeter wave communication is given by
our knowledge comprehensive channel models for [6]
optical wireless communications which still have many
unanswered questions. The organization of the paper is R = kR (2.4)
as follows: the first part presents the background of the Constant k and , according to ITU-R model, is
propagation in free space. The second part describes the given as:
methodology adopted to predict the rain attenuation (k H +k H +(k H -k H )cos 2 cos2)
using SST. Finally, a detailed analysis was done to k=
compare the performances for both Millimeter wave and
2
(2.5)
FSO systems considering operation and deployment.
(kH H +kV V +(kH H -kV V )cos2cos2)
= (2.6)
II. METODOLOGY 2k
A. Rain rate Where k H , kV , H and H for horizontal and vertical
Wind speed and orientation data range was provided polarization are given in [5].
by the meteorology station at Juanda - Surabaya, a
division of the Department Of Climatology and C. Attenuation for FSO
Geophysics Meteorology. The measurement of rain fall The attenuation of the rain rate events from January
was done in ITS campus of Surabaya. A disdrometer 2008 to January 2010 was computed. Maximum rain
optic combined with a rain gauge was used, and put on rate of 416.53 mm/hr and 351.66 mm/hr was recorded
roof of the mechanical engineering building. From this respectively on January and February events 2009.
calculation, it could found the rain fall value on the Fig.1 and Fig.2 below show the rain attenuation for
ASDO software since January 01, 2008 until January 31, both North and East direction link. The attenuation
2010. difference among the 0.5, 0.75, 1, 1.5 and 2 Km single
links increases linearly with the distance. The link
B. Synthetic Storm Technique (SST) orientation shows a considerable difference. For the 2
The synthetic storm Technique describes the value km link length, the attenuation reaches up to 90.2 and
of rain fall that moved on the line because of wind with 100.5 dB respectively for North and East direction link
particular speed. The data range concern about some at 0.001% outage probability. Rain attenuation
interval of time which are January, February, November estimation results show that the North direction link has
and December 2008, January and February 2009, and the largest attenuation. This is caused by the main wind
also January 2010 where the rain falls were heavy. directions in Indonesia which are from West and East.
A rain event for a time minute sample was made. CCDF Rain Attenuation Link North-South
1
The rain intensity records were done for each 10 seconds 10
0.5 km
and a sampling for each minute time was performed for 0.75 km
the calculation of the rain attenuation using the SST 10
0
1 km
1.5 km
method. The frequency operation is about the wireless
2 km
Pb.[Attenuation > Absciss]%
optical systems. 10
-1
-4
10
Fig. 1 CCDF rain attenuation for 90o direction link at 0.5, 0.75, 1, 1.5
and 2 km for FSO
A (k ) = nN=0 aR (bk-n) L (dB) (2.2)
i N A SST multilink is used to measure the rain
where Ai is the rain attenuation for i=1, 2 ,, n. attenuation in Surabaya Indonesia. The rain attenuations
at 0.5 km link length, direction East (0o), for outage
L is the segment length, R is the rainfall intensity probability 0.1%, 0.01%, 0.001% are respectively
b
(mm/h), aR is the rain attenuation (dB/Km) and is 2.913, 15.26 and 24.79 dB (fig. 1). It is shown clearly in
C2-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
10
-1
10
-4
10
-2
10
-3 0 50 100 150
10
Rain Attenuation [dB]
-4 Fig. 4 CCDF rain attenuation for 90o link from 0.5-2 Km for 30 GHz
10
1
frequency
10
0.5 km
0 50 100 150 0.75 km
0
Rain Attenuation [dB] 10 1 km
1.5 km
Fig. 2 CCDF rain attenuation for 0o direction link from 0.5-2 km for Pb.[Attenuation > Absciss]% 2 km
FSO -1
1 10
10
link 0.5km
link 0.75km -2
0
10 link 1km 10
link 1.5km
Pb.[Attenuation > Absciss]%
link 2km
-1 -3
10 10
-2
10 -4
10
-3
10
0 50 100 150
-4
10 Rain Attenuation [dB]
Fig. 5 CCDF rain attenuation for 45o direction link from 0.5-2 Km for
30 GHz frequency
CCDF Rain Attenuation for 30GHz MM Wave
1
0 50 100 150 10
Rain Attenuation [dB]
link 0.5km
o o o o link 0.75km
Fig. 3 CCDF rain attenuation for 0.5 (0 ), 0.75 (45 ), 1 (90 ), 1.5 (135 ) 0
10 link 1km
and 2 km (180o) links for FSO
link 1.5km
link 2km
Pb.[Attenuation > Absciss]%
C2-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
-1
10 Availability 0.5 0.75 1 1.5 2
30Ghz 1km
FSO 1.5km 99% 1.63 2.39 3.11 4.5 6.08
-2 30Ghz 1,5km
10
FSO 2km
99.9% 1.58 2.34 2.86 4.68 5.47
30Ghz 2km
99.99% -2.95 -3.86 -5.38 -6.4 -6.3
-3 o
10 90 Direction Link Distance (Km)
Link
-4 Link
10
Availability 0.5 0.75 1 1.5 2
99% 1.66 2.5 3.33 4.99 6.65
0 20 40 60 80 100 120 99.9% 1.68 2.52 3.35 5.02 6.7
Rain Attenuation [dB]
99.99% -2.78 -4.17 -5.55 -8.33 -11.1
Fig. 7 Comparison of the rain attenuation for 0.5 (0o), 0.75 (45o), 1 o
(90o), 1.5 (135o) and 2 km (180o) links for FSO and 30 GHz frequency. 45 Direction Link Distance (Km)
1
CCDF Rain Attenuation Link North-South Link
10 Link
0.5 km FSO Availability 0.5 0.75 1 1.5 2
0.5 km 30GHz
0
0.75 km FSO
99% 1.63 2.44 3.19 4.68 6.02
10
0.75 km 30GHz 99.9%
1 km FSO
1.62 2.52 3.31 5.03 6.08
Pb.[Attenuation > Absciss]%
-1
10 1 km 30GHz 99.99% -2.44 -3.38 -4.9 -8.02 -7.84
1.5 km FSO
1.5 km 30GHz
-2
2 km FSO
TABLE.2 DIFFERENCE ATTENUATION (IN DB) BETWEEN FSO AND 30
10
GHZ FREQUENCY MULTILINK
2 km 30GHz
-3 Link Availability
10
Link Link
99% 99.9% 99.99%
Direction Length
-4 o
10 0 0.5 1.62 1.58 -2.61
o
45 0.75 2.42 2.52 -3.77
o
0 50 100 150 90 1 3.32 3.36 -5.49
Rain Attenuation [dB] o
135 1.5 4.67 5.03 -7.63
Fig.8 Comparison of the rain attenuation for 90o direction link from o
0.5-2 km for FSO and 30 GHz frequency 180 2 6.01 5.47 -6.3
C2-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
C2-5
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
C3-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
C3-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
3. Rich Client Platform analyzing some of first aid applications that have been
4. Rich Internet Applications made previously and to identify features based on
5. Application Frameworks application purposes.
6. Application Lifecycle Management (ALM) 2. Workshop Design
7. Service Oriented Architecture (SOA) Activity in the content by designing proposed
Generally, used Eclipse to build innovative, application to be run better and expected to overcome
industry-standard software, and tools along with its the problems which exist.
framework help job is easier. 3. Implementation
Activities at the contents with a presentation of the
design into the program using the Java JDK (Java
Development Kit) as a programming language
integrated into Eclipse, and Android SDK (Software
Development Kit) and continued with the installation to
Android handsets.
Some of the reasons researchers use RAD in application
development for first aid on android smart phone:
1. First aid application is a simple application that was
developed by researchers and require a short time. This
is because all components are provided in the Android
Fig. 2 Eclipse Galileo application framework. RAD is very precise so that the
method is applied because this method emphasizes an
III. RESEARCH METHOD extremely short development cycle.
Developing multimedia application divided into two 2. Existing needs can be well understood, the RAD
stages, data collection and application multimedia process enables to create a completefunctional system
methods. in a short period of time.
3. RAD can develop applications quickly and sustainable
3.1 Method of Data Collection implement the design and specification of user
1. Literature Studies requirements using tools such as Java.
Researchers conducted a study of literature by reading
and studying books and e-book related to first aid, IV. RESULTS
android-based programming as well as books and articles Refer to RAD stages, the application built based on
obtained from printed media and internet to support the following stages.
topics covered in preparation of this research. 4.1 Requirement
2. Field Study As described in the previous chapter, in this stage,
Observation data collection methods, researchers researcher identified objectives and system requirements
differentiate into three parts, namely: information arising from those goals.
a. Observation
b. Interview A. System Prototype Development Goals
c. Questionnaire Development of prototype system aims to help users
android in providing first aid in an accident that both
3.2 Method of Application Development happened to him or others.
A system development method that researchers use in
this study is method of Rapid Application Development
(RAD) which was introduced by James Martin in 1991. B. Finding Information Regarding First Aid
RAD is a development cycle that is designed to provide Information search purposes first aid aims to meet the
much faster development and higher quality results than data completeness measures first aid handling. Here the
those achieved with traditional life cycle (SHPS). The researchers conducted search first aid information through
selection method is because the system is expected to have books that discuss first aid, interviews to people who know
a design that can be accepted by consumers and can be
about first aid, and search for information on a certain
developed easily because the design of the current system
still requires further development. Another reason this website about first aid (www.gotoaid.com).
method is the selection of system restrictions are needed in
order that system has not changed. In addition, Rapid C. Learn About User Android
Application Development (RAD) was chosen because the Purpose of studying culture is Android smart phone
applications will be built an application that is built in a users to know the habits of Android smart phone users
fairly short period of time. when interacting with applications and to maximize the
Rapid Application Development (RAD), which design of user interface or application interface will be
researchers have used the following stages: [15] developed. Here researchers read a lot of good references
1. Requirements from printed books, e-book as well as supporting websites
Activities charged with finding a general overview of
first aid, learning the culture or cultures android users,
C3-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
A. System Design
Here, researchers conducted a system design that will
be applied in application. The developed application is
called "First Aid on Mobile" for first aid in an accident with
combined of several technologies such as Google's APIs, as
well as the GPS is implemented into the smart phone
android.
First aid measures data that existed at first aid on
Mobile derived from several sources such as providers of
books that discuss about first aid [2-7], also a site that
displays first aid tutorial [16]. From these sources the
researchers collect material that will be displayed in the
First Aid application on Mobile.
Fig. 5 Activity Diagram FIRST AID
B. User Interface Design
At this stage, the researchers designed user interface
E. Sequence Diagram Design
or interface view of application.
Sequence diagram is interaction diagram which
expressed with time, or the other word said with the
diagram from top to bottom. Sequence diagrams express
every user of few streams that pass through a use case.
TindakanFIR JenisTindakanFIR DetilTindakanFIR
: user ST AID ST AID ST AID
tindakan_FIRST
AID()
kategori_tindakan_FIRST
AID
jenis_tindakan_FIRST
AID()
show_list_jenis_tindakan_FIRS
T AID detil_tindakan_FIRST
AID()
show_detil_tindakan_FIRS
T AID
C3-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
2. Conducting the process of packaging by using facilities researchers. Based on the results of testing, all functions
provided by Android SDK. can be run well.
3. Test the program by using the android smart phone
B. Beta Testing
handsets, as well as perform debugging or repair the
program if necessary. Beta testing is testing conducted objectively where
testing is done directly to the field of public and not
A. Software Implementation restricted to certain circles. Testing is done by creating a
The software used in the building of this application is questionnaire to find out the opinions of respondents to
as follows: First Aid application on mobile, and then distributed to
1. Windows 7 32 bit Operating System some users to be filled which will serve as a sample and will
2. Android SDK (Software Development Kit) do the calculation to be taken its conclusion on the result of
3. IDE (Integrated Development Environment) using making the application of this system. Based on testing
Eclipse Galileo performed, 90% of respondents understood the steps to
4. Android Developer Tools (ADT) help injured people in accident.
5. Android : Froyo 2.2 with Google API SDK Level 8
6. Java V. CONCLUSION
7. XML Based on testing of application and questionnaire, it
can be concluded that First Aid application based android
B. Hardware Implementation smartphone developed under the rules of medical action of
The hardware used in building this application is as First Aid as its knowledge base which has capacity of 5 MB
follows: memory. The application is able to provide information
1. Intel Pentium Dual Core 2.6 GHz about actions and supporting information which cover
2. Memori 3GB location of hospitals and pharmacies, emergency numbers
3. VGA 1GB and rescue techniques required in performing first aid in
4. Harddisk 250GB accident. For the future research, the application could be
developed as a web-based application as continue to
5. Monitor
function as mobile application. The web application is used
6. Mouse and Keyboard as a medium for knowledge base development system that
7. Handset Smartphone Android (Samsung GT-S5660 serves as a media updated. Moreover, its better to develop
Galaxy Gio) for cross platform applications in other mobile platforms,
like Symbian or Blackberry.
C. User Interface Implementation
Implementation is the stage where system is ready to REFERENCES
operate in actual stage, so it will be known whether the
[1] A. Nugroho. Analisis dan Desain Sistem Informasi, Yogyakarta:
system has been created completely according to plan or Andi, 2004.
not. In software implementation will be explained how this [2] N. Saubers, Semua yang Harus Anda Ketahui Tentang FIRST AID.,
system works, by providing system display. Yogyakarta: PallMall, 2011.
[3] M. Kartono, Pertolongan Pertama., Jakarta: PT Gramedia Pustaka
Utama, 2005.
[4] S. Sudirman, Panduan FIRST AID, Jakarta: Restu Agung, 2008.
[5] A. Thygerson, First Aid : Pertolongan Pertama Edisi Kelima.
Jakarta: Erlangga, 2011.
[6] A. Yunisa, Pertolongan Pertama Pada Kecelakaan, Jakarta:
Victory Inti Cipta: Jakarta, 2010..
[7] Peraturan Menteri Tenaga Kerja dan Transmigrasi
(Permenakertrans) Nomor: PER.15/MEN/VIII/2008 tentang
Pertolongan Pertama pada Kecelakaan di Tempat Kerja.
[8] R. Meier. Professional Android 2 Application Development.,
London: Willey Publishing, Inc, 2008.
[9] A. Mulyadi, Membangun Aplikasi Android, Yogyakarta:
Multimedia Center Publishing, 2010.
[10] M. Murphy, Beginning Android 2, Barkeley: APRESS, 2009.
[11] M. Murphy, The Busy Coders Guide to Android Development,
United States of America: CommonsWare, 2008.
[12] J. Steele, The Android Developers Cookbook: Building
Fig. 7 Display of Main Menu Applications with the Android SDK, New York: Addison Wesley,
2010.
[13] I.Horton, Beginning JavaTM 2: JDKTM 5 Edition, Indianapolis:
4.4 System Testing Wiley Publishing, Inc, 2005.
The testing was used to examine new system by black [14] J. Friesen, Beginning JavaTM SE 6 Platform: From Novice to
box method focuses on the functional requirements of Professional, USA: Apress, 2007.
software. [15] Kendall & Kendall. 2008. System Analysis And Design. London:
Pearson International Edition 7th Edition.
A. Alpha Testing [16] First Aid. [Online]. Available: http://gotoaid.com
C3-5
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
AbstractIn this research, a new model of the triangular function of Is or two of triple truncated tip canalso affect
patch antenna with truncated tip using annular sector slot the bandwidth. It makes the bandwidth become wider than
embedded on the ground plane.The antenna instead of single without it (see Fig.2). In the other hand, the slot antenna
layer, singly-fed, small, wide bandwidth and analyzed by both embedded on patch antenna or and on ground plane
using method of moments. The result for simulation among
cause the bandwidth antenna decreased, but its
of the new model antenna c3s, c3 and c1, in the case of
frequency characteristic, S-parameter, and input impedance advantaged isprobably to make small antenna. The
are good results. The bandwidth c3 is widest than the others. function of slot is decreasing the frequency
The bandwidth c3sis almost the same with c1. It is caused by operationwherein the current path or guide wavelength g
using double truncated tip Is on the below side of the patch of the TM10 modewith slot is more length than the current
antenna thus the total of vector current distributions become path without slot. The purpose of this research is to yield
increased are just around this area. In the case of antenna c3s, the optimized result between the small antenna and the
the bandwidth decreased due to by using annular sector slot slightly wide bandwidth.
embedded on the ground plane. Moreover, the bandwidth of
antenna c1 is also decreases, because of without using the
II. METHODOLOGY
truncated tip Is, but it is slightly wider than antenna c3s.
The methodology instead of literature study, qualitative
Keywords-singly-fed, truncated tip, annular sector slot, and quantitative analysis related microstrip antenna,
triangular-patch. especially truncated tip using annular sector slot antenna,
feeding probe, and mobile satellite communications.
Ensemble version 8.0by moment method software is
I. INTRODUCTION used to design antennafor mobile satellite communication.
To obtain circular polarization (CP)operation, some There are some advantages i.e. for industry and research
designs by embedding a cross slot of unequal slot lengths department to develop technology of telecommunication,
in the circular patch [1] or inserting slits of different especially for design antenna with the new one of
lengths at the edges of a square patch [2] or truncated tip technology.
of equilateral triangular antenna [3] or using proximity The conclusion is enclosed of this paper that it
feed embedded on below radiating patch antenna [4] have contributes the valuable analysis of singly-fedcircularly
been proposed recently. In the case of the equilateral polarized triangular microstrip antenna with truncated tip
triangular antenna with a truncated tip [3], for RHCP the using annular sector slot for mobile satellite
probe-fed is usually located in the right half of the communications.
triangular patch. Conversely, LHCP radiation can be
obtained in the left half of the triangular patch. In this III. ANTENNA CONFIGURATION
research, the new phenomena happen if the patch antenna Fig.1 shows the configuration of antenna design. The
changed become triple truncated tip with case Is> Ip(see triangularpatch has a side length of a = b and using a
Fig.1), probe-fed RHCP and LHCP located on the left and conventional substrate (relative permittivity 2.17 and loss
right half of the triangular patch, respectively. In the case tangent 0.0009). The antenna is fed by single probe which
Is<Ip, the RHCP and LHCP can be obtained with the rule located on right half for LHCP. Here, any three of small
above [3], but if case Is = Ip, both of RHCP and LHCP can triangular tip wherein two of all are the same side length
not happen, only linear polarization can be obtained. In Is.It is affected to excite more magnitude current path
this case, the function of two truncated tip with length Is around this area which moving to y direction and x
are as switch to move variation of polarization, if the direction, hence increasing the bandwidth. The others
probe-fed exist on the same place. In addition, the triangular tip has side length Ipowing to the truncated-tip
C4-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
perturbation, the effective excited patch surface current The bandwidth c3sis almost the same with c1. It is caused
path in the y direction is slightly more length than that in by using double truncated tip Is on the below side of the
the x direction, which gives the y-directed resonant mode patch antenna thus the total of vector current distributions
a resonant frequency slightly smaller than of the x- become increased are just around this area. In the case of
directed resonant mode. That is, the dominant mode antenna c3s, the bandwidth decreased due to by used
(TM10 mode) of the triangular patch can be split into two annular sector slot embedded on the ground plane.
near-degenerate orthogonal resonant modes of equal Moreover, the bandwidth of antenna c1is also decreases,
amplitudes and 900 phase difference for LHCP operation. because of without using the truncated tip Is, but it is
slightly wider than antenna c3s.
Fig.2 shows that the value of gain and axial ratio
(top view) (Ar)for simulation of new model antenna at the resonant
frequency. They are as followed that antenna c1 operates
at the frequency 2.76 GHz, gain RHCP= 6.66 dBic, Ar =
substrate
2.91 dB, antenna c3is frequency operation = 2.9 GHz, gain
LHCP = 6.98, Ar = 3.02 dB, antenna c3sis frequency
LHCP operation = 2.505 GHz, gain LHCP = 6.08, Ar = 1.75 dB.
RHCP
In addition, eachantennais fed by probe-fed at the same
loci on the patch antenna. It is clear that antenna c1 and c3
did not satisfy the targets yet, especially the axial ratio. It
is due to by the loci of feeding are still not maximize yet
on the surrounding of patch antenna.Moreover, peak gain
patch slot
probe_fed antenna c3s at the frequency resonant is lowest than the
others. It is caused by annular sector slot embedded on the
(side view) ground plane can decreased of gain. In addition, the
bandwidth of gain c3 is the widest than the others. It is due
substrate ground to used the truncated tip Is and without using annular
sector slot.
Aluminium plate
Fig.1. Configuration of simulated antenna, slot on the Gain-RHCPc1 Arc1 Gainc3 Arc3 Gainc3s Arc3s
ground plane 8 8
C4-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
area on the both of below patch antenna (Is) can enhance the purpose design. If we want to design small antenna,
the bandwidth of S-parameter. owning the enough bandwidth, antenna c3s will be better,
but if we want to design wideband antenna, so antenna c3
0 is should be chosen, especially for applicationmobile
satellite communication.
The future work will be done to design small antenna
S-parameter [dB]
REFERENCES
-20 [1] H. Iwasaki, A circularly polarized small-size
S-parameter c1 microstrip antenna with a cross slot, IEEE Trans.
S-parameter c3
S-parameter c3s Antenna Propagat.,vol.44, pp1399-1401, Oct. 1996
[2] K.L. Wong and J.Y.Wu, Single-feed small circularly
-30 polarized square microstrip antenna, Electron. Lett.,
vol. 33, pp.1833-1834, Oct.23,1997
2.4 2.5 2.6 2.7 2.8 2.9 3 [3] C.L. Tang, J.H. Lu, and K.L. Wong, Circularly
Frequency [GHz] polarized equilateral-triangular microstrip antenna
Fig.3. S-parameter with truncated tip, Electron. Lett., vol. 34, pp. 1227-
1228, June, 1998.
Fig.4 depicts the input impedance characteristic of Rx. [4] J. T. S. Sumantyo, K. Ito, D. Delaune, T. Tanaka, and
This figure shows that the real part of simulation is H.Yoshimura Simple satellite-tracking dual-band
difference of each others, but in the case antenna c3sreal triangular patch array antenna for ETS-VIII
impedance by closed50 at the frequency applications, Proc.IEEE Int. Symp. Antennas and
operation.Moreover, the reactance part of new model Propagation, pp. 25002503, 2004
antennac3s is the best than the others by closed 0 at the
resonant frequency.
Muhammad Fauzan Edy Purnomo was born in
150 150 Banjarmasin, Indonesia, in June 1971. He received the
B.E. and M.E. degrees in Electrical Engineering from
100 100 University of Indonesia, Jakarta, Indonesia in 1997 and
50 50 2000. He is presently with the Electrical Department
Xin[]
C4-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
AbstractThe Development of WLAN is growing rapidly less understood and has a lot of interdependence whereas
along with the possibility of multimedia services fuzzy logic is able to work based on human intuition and
transmission through the wireless network. Until now, logic, the combination of the two algorithms is expected
WLAN has been recognised as a most reliable wireless data to generate a mechanism that allows a better performance
communication and it has a transmission rate more speed
improvement.
than WiMAX as well as celluler system. To ensure this
application and to maintain the quality of service, WLAN Studies that use fuzzy logic and genetic algorithms
needs an adaptive admission control system to guarantee on admission control as in [1] fuzzy logic is used as an
the channel availability for each connection request in high admission control in high speed networks such as ATM
speed transmission rate. In this paper, we proposed a (Asynchronous Transfer Mode), this scheme is an
Genetic Fuzzy Algorithm as an admission control in WLAN improvement over conventional admission control
802.11e. Simulation will be done to investigated the ability schemes. In [2], fuzzy logic tuned by genetic algorithm is
of Genetic Fuzzy Algorithm as an admission control in used for call admission control in ATM networks. ATM
WLAN 802.11e in relation with Quality of Service technique that uses statical multiplexing complicate for
Guarantee.
the application of mathematical models, hence the fuzzy
Keywords ; WLAN 802.11e, Admission Control, Genetic system is very suitable to be applied. Furthermore, [3]
Fuzzy System Fuzzy Logic is used to control parameters such as cell
congestion status, load availability, and the total
I. INTRODUCTION interference. Fuzzy rules are used for admission criteria,
from the simulation results, obtained an improvement
Since the release of the IEEE 802.11 standard in when compared to classical admission control strategy.
1999, the applications running on the WLAN flatform Fuzzy logic is also used as an Adaptive Contention
increasingly diverse, such as voice to video streaming. Window (CW) based on the ambiguity of information
But the Medium Access Control (MAC) on the IEEE from the channel in [4],. This scheme then compared with
802.11 standard was originally designed for best effort the scheme of differentiation, it was found that this
applications (such as e-mail, web browsing) so it can not scheme has the capability of adaptation for streaming
meet Quallity of Service requirements for many types of applications. [5], the combination of fuzzy logic and
new applications develop. genetic algorithms used for optimal access network and
To support Quality of Service, Enhance Distributed promising in the heterogeneous wireless networks.
Channel Access (EDCA) was introduced in IEEE In this paper, we propose the application of Genetic
802.11e WLAN standard, which was built as a derivative Fuzzy System (GFS) as the admission control in WLAN
of the Distributed coordination function (DCF), equipped 802.11e. This is a new proposal because no one has
with a prioritization of the four access categories (AC). proposed this scheme before. GFS will be applied as an
This is achieved by variations in the size of contention admission control by considering the parameters of
window (CW) in back-off mechanism in each category. collition rate and network load so that decisions can be
The continued development of multimedia services led taken with proper admission.
to the need in the admission control is increasing as well,
so it introduced an adaptive admission control to improve
performance of WLAN 802.11e. The use of fuzzy logic II. IEEE 802.11E MAC LAYER [6]
and genetic algorithms as adaptive admission control has The IEEE 802.11e refer to the specifications
been widely used in research compared to other artificial developed by The IEEE for Wireless Local Area
intelligence algorithms, both algorithms are preferred to Networks (WLAN). MAC Layer of The IEEE 802.11e is
apply, it is because they do not require complex distinguished by the previous 802.11 standards by the
mathematical models and easy to implemented. Genetic availability of the Access Category (AC) aimed at
algorithm allows to find solutions to problems that are prioritization of data. Includes 802.11e Enhance
C5-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Distributed Channel Access (EDCA). The IEEE 802.11e higher priority AC gets smaller AIFS values.
has four transmission queues, each acting as a single Formulation of AIFS as follows:
entity Enhance DCF, which is an Access Category (AC). AIFS [AC] = SIFS + AIFS [AC] x Slot Time (2)
Therefore EDCA works with four AC, where each AC, SIFS = Short Inter-Frame Space
dealing with one different access channels b. CWmin, CWmax
This is the value of backoff counter which is uniformly
distributed random value between the contention
window CWmin and CWmax. The higher priority AC
getting smaller value of CWmin and CWmax.
c. TXOP (Transmission Opportunity) limit
This is the maximum duration of transmission after the
medium requested. TXOP obtained from the EDCA
mechanism called EDCA-TXOP. During the EDCA-
TXOP, a station can transmit multiple data frames from
the same AC, where SIFS time period, split between the
ACK and data transmission sequence. The higher
priority AC get the larger of it's TXOP limit. TXOP
for each AC-I is defined as:
TXOP [i] = (MSDU [i] /R) + ACK +SIF +AIFS [i] (3)
Figure 1. Enhance Distributed Channel Access (EDCA) [7]
MSDU [i] is the packet length to the AC-I, ACK is the
Each AC consists of a queue-free delivery and a time required to transmit an acknowledgment, R is the
channel access function with its own parameters, namely physical transmission rate, SIF is the time period are
the minimum and maximum Contention Window required by SIF, AISF [i] is the time of AIFS in AC-I.
(CWmin, CWmax), Arbitration Interframe Space (AIFS)
and the duration of the Transmission Opportunity III. IEEE 802.11E MAC LAYER [6]
(TXOP).
EDCA access mechanism can be described as This research consists of two major parts namely
follows, when the medium is busy before the backoff designing a Genetic Fuzzy System (GFS) to be applied to
counter reaches zero, then the backoff should be stopped the 802.11e WLAN admission control and testing of the
temporarily, and the station must wait for a period of scheme on a single WLAN system.
AIFS. When the medium is idle again, during the AIFS, At this early stage, to design a model of genetic
backoff counter minus one. algorithm scheme is applied to the WLAN 802.11e
After the transmission failed, a new CW value is EDCA admission control, as shown in Figure 2.
calculated by Presistence Factor (PF), which is also a
unique value according type of AC. CW value is
calculated based on the following provisions:
C5-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
In the figure 2 shown a modification of conventional The model for the input membership functions used
EDCA admission control system by adding the GFS as a is the triangular model, it is intended to avoid too much
device that responsible for determining whether a request overhead to the process of fuzzy on the MAC layer
will be served or not.
GFS output is obtained from the process of genetic Figure 4. Input Membership functions
algorithms and fuzzy rules by observing the condition of
the network. The following figure illustrates this idea, as Membership
Degre
a note that the genetic learning process aims to design or
optimize the KB (Knowledge Base). Consequently, a
GFS is a design method for fuzzy system of basic rules
that combine evolutionary techniques (genetic algorithms
to achieve automatic generation or modification to all or
part of the KB).
0. 001 0. 005 0.02 0.03
Network load
Membership
Degre
While the network load is used as the input fuzzy 1) Generate An Initial Population Of Encodec Rules
membership functions were also measured at the MAC Intial population can be generated from the output of
layer. Network load is represented as a time when the fuzification set. The code is obtained by concatenated
wireless medium busy represented by transmission rule using AND Operator. Location on the chromosome
apportunity (TXOP), the time needed to transmit an indicate the start and end of a particular rule. The total
MSDU, TXOP obtained from equation (5) number of fuzzy sets in the DB is L [11] :
Net_load = queue_ength[j][i] xTXOP [i] (6) n and m is the number of input and output variables. Ni is
the number Ni represents the number of linguistic terms
C5-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
associated to input variable Xi and Mj the number of Using this methodology, a stronger rule can evolve with
linguistic terms associated to output variable Cj. The every new generation.
general structure of a rule with AND operator is [16] :
3) Crossover and mutation
if X1 is Y1 AND X2 is Y2 then Z1 is C1 (8) According to the theory of Genetic Algorithm, a
crossover operator selects substrings of genes of the same
X1, X2 are input variables, Y1, Y2 is Linguistic Value, length from parent individuals which are known as off-
Z1 is output variables and C1 is value. springs from the same point, replaces them and generates
a new individual. This point can be selected randomly
For implementing fuzzy rules, we uses fuzzy mamdani [13]. For designing chromosome, we used binary
membership functions used in Term set1: {High, encoding .Different rules can be represented in form of
Medium, Low} while term set 2 consist of output label chromosomes labeled as individuals. Here, single point
set {Strong accept, Weak Accept, Strong Reject and crossover operator has been implemented.
Weak reject} for output variables. Different combinations
will be utilized for chromosome representation scheme.
IV. IMPLEMENTATION
2) Fitness Function This research will be implemented in Network
The encoding scheme which is discussed as follows : The Simulator-2 or The NS-2. The NS-2 used is NS 2.28
rule be in the form : If A then C, whre A is Antecedent because EDCA module is implemented in NS-2.28
and C is consequent: The predictive performance of a version made by Sven Wietholter and Christian Hoene
rule can be summarized by a 2 x 2 matrix, which from the Technical University Berlin Telecommunication
called a confusion matrix, as illustrated in table. 1. Networks Group, EDCA module was used because it has
The labels in each quadrant of the matrix have the been verified and used by many researchers by many
following meaning: [12] : researchers as well as having good documentation
[14,15,16].
Table 1. Confusion Matrix In the The NS-2.28, 802.11e WLAN is implemented
Actual Actual in the classes. Modifications done on a class-802_11e.h
Positive Negative mac, mac-802.11.cc, priq.h and priq.cc so it can measure
Positive the collision rate and network load.
TP FP
Prediction
Negative
FN TN V. CONCLUSION
Prediction
Fuzzy rule base is used for optimization or search
Where ; problems, and genetic algorithms are widely known and
TP = True Positives = Number of examples satisfying A used for global search techniques with the ability to
and investigate a global search space for solutions that fit
FP = False Positives = Number of examples satisfying A with a single scalar performance measurement. In
but not C addition to the ability to find the nearest optimal solution
FN = False Negatives = Number of examples not in the a complex search space, the structure of the genetic
satisfying A but satisfying C code and performance features of the genetic algorithm
TN = True Negatives = Number of examples not makes it suitable as a candidate to be combined with
satisfying A nor fuzzy system. The ability of the fuzzy system is expanded
with use of genetic algorithms to develop a broad range
Therefore, CF(Prediction)=TP/(TP+FP) (9) of approaches to designing a fuzzy system fundamental
rules.
Prediction accuracy is measured by (9) by looking for The use of soft computing especially fuzzy rule base
proportion of the examples that have predicted class C, for admission control in 802.11e WLAN has been done,
that is actually covered by the rule antecedent. Rule however the results still can not meet the expected level
Completeness (true positive rate) can be measured by the of QoS. Use of Genetic Fuzzy, is expected to optimize
following equation. the search for fuzzy membership functions so as to
improve QoS as expected..
Comp = TP / (TP + FN) (10) Previous research was conducted on a fix WLAN
networks, in this study will be conducted on wireless
By combining (9) and (10) we can define a fitness mesh networks with multimedia traffic so the proposed
function such as: algorithm can be tested more optimal.
[2] Barolli, Leonard et.all, A CAC Scheme for Multimedia [11] Cordon, O., Herrera, F., Hoffmann, F., Magdalena, L.:
Applications Based on Fuzzy Logic, Proceedings of the Genetic Fuzzy Systems, Evolutionary Tuning and
19th International Conference on Advanced Information Learning of Fuzzy Knowledge Bases, Advances in Fuzzy
Networking and Applications (AINA05) 2005 system-Applications and Theory, Vol. 19, pp.89-
[3] Piyaratna, Sanka et.all., A Genetic Algorithm Tuned Fuzzy 93,97,179-183,World Scientific, USA (2001)Electronic
Logic Based Call Admission Controller for ATM Publication: Digital Object Identifiers (DOIs):
networks, University of Adelaide, 1997 [12] Freitas, A.: A Survey of Evolutionary Algorithms for Data
[4] Dini, Paolo and Cusani,Roberto, A Fuzzy Logic Approach Mining and Knowledge Discovery, 31-36 , AAAI Press,
to Solve Call Admission Control Issues in CDMA Brazil (2003)Article in a conference proceedings:
Systems, EUSFLAT - LFA 2005 [13] Akerakar, R., Sajja, P.S., Knowledge-Based Systems,
[5] Naoum-Sawaya, Joe, Ghaddar, Bissan, A Fuzzy Logic Jones and Bartlett, Massachusetts (2010)
Approach for Adjusting The Contention Window Size in [14] Kevin Fall, Kannan Varadhan, The ns Manual, The
IEEE 802.11e Wireless Ad Hoc Networks, University of VINT Project, 2009
Waterlo, Canada, 2005 [15] S.McCane, S.Floyd, NS Network Simulator ,available at :
[6] IEEE 802.11e, Wireless LAN Medium Access Control www.isi. edu/nsnam/ns/
(MAC) and Physical Layer Extension in the 2.4 GHz [16] Wang, Yue, A Tutorial of 802.11 Implementation in ns-2 ,
Band, Supplement to IEEE 802.11 Standard, IEEE , MobiTab Lab.
September 1999
[17] Mankad, Kunjay, Sajja, Pritti Srinivas and Alkerkar
[7] IEEE 802.11e, Wireless LAN Medium Access Control Rajendra,Evolving Rules Genetic Fuzzy Approach-An
(MAC) and Physical Layer Extension in the 2.4 GHz education Case study, International Journal of Soft
Band, Supplement to IEEE 802.11 Standard, IEEE , Computing (IJSC), Vol. 2, No.1, February 2011
September 1999
[18] Kejik, Petr, Hanus, Stanislav, Comparison of Fuzzy Logic
[8] Oscar Cordon, et.al., Genetic Fuzzy Systems: Evolutionary and Genetic Algorithm Based Admission Control
Tuning And Learning Of Fuzzy Knowledge Bases, World Strategies for UMTS System, available at :
Scientific Publishing Co. Pte. Ltd., Singapore, 2004. http://www.radioeng.cz/fulltexts/2010/10_01_
[9] Munadi, Rendy, R.Rumani, Layla, Performance Analyze 006_010.pdf
of IEEE 802.11e WLAN For Mixed Traffics TCP-UDP [19] Alkhawlani,Mohammed, Ayesh, Aladdin, Access Network
Using Adaptif Admission Control Mechanism, IGCES Selection Based on Fuzzy Logic and Genetic Algorithms,
2008, Desember 2008, Malaysia Hindawi Publishing Corporation Advances in Artificial
[10] D. E. Goldberg, Genetic Algorithms in Search, Intelligence Volume 2008, Article ID 793058, 2008
Optimization, and Machine Learning. Addison-Wesley,
1989
C5-5
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
C6-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
The wavelength of core mode in the FBG ( ) period tfbg a long the longitudinal direction which
can be calculated by the following equation, follows the equation 3.
, (1) , (3)
where, neff.core is effective refractive index of the core,
and is the grating period. where, pm and ext are grating period of the phase
FBGs are intrinsically insensitive to the mask and the external tilt angle between the grating and
environment, because the core modes are the sample fiber, respectively. A phase mask of 10 mm
well-screened from incident of the light from outside long was used in this experiment.
due to the presence of the cladding7). On the other
hand, in the case of the TFBG which has a tilt angle ,
not only the core mode but also a number of cladding
modes are observed in the reflection spectrum. The
reflection wavelengths of the cladding modes
( ) are calculated using the following
equation,
, (2)
where, neff.cladding is the effective refractive index of
the cladding which is corresponding to each ext
C6-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
adjusted in wide index range by simple mixing of the the core mode compared to 7.3-TFBG was observed.
glycerine and water9). The concentration of glycerine The peak wavelength and the intensity were 1580 nm
solution used in the experiment was from 12% to 84%. and 2 dB, respectively. The cladding mode were
The detailed information for index of sample was observed in a range form 1520-1578 nm. The maximum
summarized in Table 1. intensity of the cladding mode was 11.5 dB at 1548 nm.
Table 1. Refractive index of sample We observed two coupling modes in the cladding mode
Sample Ref.Index as well.
Air 1.0003
Water 1.3255
LP11
Ethanol 1.3539
Glycerine 12% 1.3417
Glycerine 24% 1.3579
Glycerine 35% 1.3727
Glycerine 46% 1.3876 LP1n
LP2n
Glycerine 66% 1.4146
Glycerine 84% 1.4389
We observed the core and the cladding modes in the
transmission spectra and investigated the relationship
Figure 5(a). transmittance spectra 7.3-TFBG
between the spectral change and to the change in
refractive index.
LP11
C6-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
V. DISCUSSION
Two modes have been investigated in the TFBGs
transmittance spectra. These modes are the core mode
and the cladding mode. The amplitude of transmittance
spectra become larger by irradiation time. When the
tilted angle of TFBG became larger, the amplitude of
core mode getting smaller and its wavelength shift to the
Figure 7. Relative wavelength shift of TFBG with different tilt angle
longer. In cladding mode, transmittance spectra shows
Similar experiments have been done for other TFBG two coupling modes. At 9.9-TFBG, the core mode
with different tilt angles. Figure 7 shows the relative almost disappeared, it has a cladding mode only. The
wavelength shift to the refractive index of the liquids and amplitude of 9.9-TFBG transmittance spectra after 20
air. The maximum correlation of relative index is minutes different with others. It has small amplitude.
achieved at 8-TFBG. As the tilted angle become larger, We investigated one peak in cladding mode for liquid
the correlation between the index and obtained data measurement. The results show that wavelength shifts to
become stronger. the longer when the refractive index of liquid was
C6-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
REFERENCES
[1] X Dong, et.al.: Tilted Fiber Bragg Grating; Principle and
Sensing Applications, Photonic Sensor, 1,6-30, 2011.
[2] Yin S et.al.: Fiber Optic Sensors. Second Edition, CRC
Press, New York, 2008.
[3] Hill K.O and Meltz G: Fiber Bragg Grating Technology
Fundamentals and Overview, Journal of Lightvawe
Technology, 15(8), 1263-1276, 1997.
[4] Othonos A and Kalli A: Fiber Bragg Gratings, Artech
House, Boston, 1999.
Figure 10. Peak wavelength compared by Laffont method in 8-TFBG [5] Laffont G and Ferdinand G: Tilted Short-period Fibre
Resolution of refractive index measurement depend Bragg Grating Induced Coupling to Cladding Modes for
on optical spectrum analyzer resolution, it was 20 pm. Accurate Refractometry, Meas.Sci. Technol. 12,
The resolution was measured by this method estimated to 765-770, 2001.
-2
be 9.2x10 to 3.0x10 .
-3 [6] R Kasyhap: Fiber Bragg Gratings, Academic Press, San
Diego, 1999.
[7] Erdogan E: Cladding-mode Resonance in Short- and
VI. CONCLUSION Long-period Fiber Grating Filters, J. Opt. Soc. Am. A, 14
Several TFBGs have been fabricated and its (8), 1760-1773, 1997.
transmittance spectra during fabrication has been [8] A Cusano, et.al.: Single and Multiple Phase Shifts Tilted
Fiber Bragg Grating, Research Letters in Optic,1-4, 2009.
investigated. We have also measured the refractive index [9] Rheims J: Refractive Index Measurement in the near-IR
of liquid using wavelength shift monitoring in a cladding using Abbe Refractometer, Meas. Sci. Technol, 8,
mode and transmittance response of TFBGs by liquid 601-605, 1997.
droplet. Based on our experiment, we conclude that:
TFBG could be fabricated using phase mask
C6-5
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Abstract-In the recent years, more and more conditions can be met by the WiMAX.
information services require high-speed data access. Video WiMAX is a basic standardized IEEE 802.16
streaming is a real-time service with high-speed data technology that allows transfer of data to access
access which conveys information such as audio and video wireless broadband access as an alternative to cable or
networks using Internet Protocol (IP). Using the streaming
DSL (Digital Subscriber Line). WiMAX can provide
technology clients can play the video in real time
condition. However, its strongly influenced by the access to the type of fixed, nomadic, portable and
bandwidth. Insufficient bandwidth for the streaming mobile wireless broadband to the condition of LOS and
process will cause losses and greater delay[1]. Therefore, NLOS. Just with one Base Station, the theoretical
in order the video streaming service to approach its ideal coverage of the cell radius could reach 50 km. WiMAX
conditions it is necesary to be applied on a network which also includes QoS features that enable services such as
has a high speed data access and large bandwidth. Such voice and video with low delay. According to the
conditions can be fulfilled by the WiMAX network WiMAX Forum, the system can transmit data at speeds
802.16d, because it is the network technology based on up to 75 Mbps per carrier for the type of fixed and
international standard IEEE 802.16 which enable to
portable access. In a network with mobile access types,
transfer data to wireless broadband access as an
alternative to cable or DSL. WiMAX can provide the based on its specifications, it can generate speeds of
folllowing types of access : fixed, nomadic, portable and more than 15 Mbps with a radius up to 3 km. This
mobile wireless broadband on the line of sight (LOS) and indicates that WiMAX technology can be used through
non line of sight (NLOS) conditions[2]. Based on the notebooks and PDAs which can be implemented on
calculations, by varying the distance 1 km - 15 km between mobile phones.
transmitter and receiver for LOS and 1 km - 5 km for In this paper, the calculation of video streaming
NLOS, the value of the propagation losses on NLOS is parameters on the network of WiMAX 802.16 analyzed
found to be much larger than on LOS. In LOS conditions, were pathloss, bit energy to noise ratio (Eb / No), bit
the value of bit error probability is smaller than the NLOS
error rate (BER), packet loss probability of streaming
conditions for all types of modulation. The best conditions
occur in LOS using QPSK modulation with 2.6 Mbps data video, delay end-to-end throughput as well.
rate with bit error probability 2.6184x10-45 and packet
loss probability of video streaming is 9.1200x10-4. II. METHODOLOGY
C7-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Based on that model Performance parameters of video With the same calculations for different distances
streaming on the 802.16d WiMAX network being between the BS to the SS, the obtained results as shown
analyzed include the end to end delay, propagation in Fig.2.
losses, energy bit per noise, bandwidth, probability of
bit error, packet loss, and throughput. Performance is
reviewed from several conditions, namely LOS, NLOS
outdoor and indoor NLOS.
C 12.6 17.1 20
C7-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
On the other hand, the calculation of the signal level Conditions Non Line of Sight (NLOS)
at the receiver side with indoor NLOS CPE and the In NLOS conditions two types of CPE,outdoor and
distance between BS and SS as far as 1 km can be indoor were used. Value of Eb / No for outdoor NLOS
calculated as follows: using QPSK modulation techniques with a data rate (R)
used = 2.6 Mbps is:
C7-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
C7-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
There for,
Based on the Fig III., it can be seen that the value of bit
error probability is very small in LOS conditions. While
in NLOS conditions, the probability of bit error in the
NLOS outdoor larger than the indoor NLOS. This can
occur because of differences in CPE specifications are
On the other hand, according to equation (21), the used in both circumstances.
probability of bit error on QAM modulation can be
calculated as: Table V.
Probability Packet Loss Video Streaming
Probability of Packet Loss
For 16-QAM, Type of
Outdoor Indoor
Modulation LOS
M 1 / 2 1
1/ 2
2 3 log 2 M Eb NLOS NLOS
Pb16 QAM == 1 erfc ...(21) QPSK 2,6
log 2 M M 1 / 2 2( M 1) No 9.1200e-004 9.1200e-004 9.1200e-004
Mbps
3
( )
2 3 log 2 16 1/ 2 QPSK 4
Pb16 QAM == 1 erfc 5.7289 1018 Mbps
9.1200e-004 9.1200e-004 9.1200e-004
log 2 16 4 30 16-QAM
9.1200e-004 9.2000e-004 9.1705e-004
5,3 Mbps
16-QAM
9.1200e-004 9.1705e-004 9.1519e-004
7,9 Mbps
Then the probability of packet loss in streaming 64-QAM
video 802.16d WiMAX network with 16-QAM 9.1200e-004 9.1346e-004 9.1292e-004
11,9 Mbps
modulation is calculated using equation (22) namely: 64-QAM
9.1200e-004 9.1330e-004 9.1282e-004
13,2 Mbps
For 64-QAM,
M 1 / 2 1
1/ 2
2 3 log 2 M Eb
Pb 64 QAM == 1 erfc
log 2 M M 1 / 2 2 ( M 1) No
7
Pb16 QAM ==
2
log 2 64
1 erfc
8
3 log 2 64
126
(8 . 8969 10 19 ) 1/2
C7-5
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Fig. IV shows the relationship between the packet loss [4] Datasheet CPE Easy ST dan Pro ST Airspan
probability of video streaming with data on LOS and Network Inc.
NLOS conditions. In QPSK modulation type, ie the data [5]Kwang-Cheng Chen and J. Roberto B. de Marca,
rate of 2.6 Mbps and 4 Mbps, the packet loss 2008 Mobile WiMAX. London : John Wiley &
probability is very small streaming video since the Sons
value of bit error is also small. While on the QAM [6]Forouzan, Behrouz. 2000. Data Communication and
modulation type, the higher the data rate, the lower the Networking. United States : McGraw-Hill
probability of packet loss video streaming. [7]Freeman, Roger L. 1994. Reference Manual for
Telecommunications Engineering 2nd Edition.
IV. CONCLUSIONS AND RECOMMENDATIONS Toronto : John Wiley & Sons
Based on the calculation and analysis of the video
streaming performance on the 802.16d WiMAX
network, the conclusion is obtained as follows: The First author, Dwi Fadila Kurniawan received the Master Degree
packet loss probability of video streaming in LOS in CDMA (Code Division Multiple Access) Multimedia from the
conditions have the same value on all modulation Institute of 10 November, Surabaya, in 2001. He worked as a lecturer
in electrical engineering departement the University of Brawijaya,
techniques, ie 9.1200 x 10-4. This can occur because
Malang, Indonesia. His research has been in the areas of microwave,
the value of bit error probability is very small. On the
antenna propagation, and mobile communication.
other hand, in NLOS conditions by using QPSK
modulation technique the value is 9.1200 x 10-4. Using The second author, Muhammad Fauzan Edy Purnomo was born
QAM techniques, the value of packet loss probability is in Banjarmasin, Indonesia, in June 1971. He received the B.E. and
inversely proportional to data rate. M.E. degrees in Electrical Engineering from University of Indonesia,
Jakarta, Indonesia in 1997 and 2000. He is presently with the
Electrical Department University of Brawijaya, Malang, Indonesia
References where he is working toward as lecturer. His main interests are in the
[1]Schwartz, Mischa. 1987. Telecommunication areas of microwave, mobile communication, microstrip antennas,
Network. Addison-Wesley. array antenna for mobile satellite communications, and Synthetic
Aperture Radar (SAR). He has been ever be a student member of the
[2]Andrews, Jeffrey G., Arunabha Gosh, Rias
IEICE and IEEE.
Muhamed. 2007. Fundamental of WiMAX :
Understanding Broadband Wireless Networking. The third author, Widya Rahma received the engineering degree
Massachusetts : Pearson Education Inc. in telecommunication from Electrical Department University of
[3]Datasheet Base Station MicroMaxd Airspan Network Brawijaya in 2010. His research has been in the areas of the mobile
Inc. communication and microwave.
C7-6
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
AbstractWe present an experimental of Free-Space charactheristics of light for example light attenuation
Optical Communication (FSOC) system characteristical caused by the O-H absorption such as rain, fog, and
performance in a turbulence medium. The FSOC system is snow. Other degradation are caused by Rayleigh and Mie
fiber detection via TPS (Tube Propagation Simulator)and scattering and difraction as well. The physical properties
using 1550 nm optical modem as the main source of of atmosphere are random fluctuation in temperature,
communication and EDFA with output of +23 dBm).The
pressure and wind speed. These charactheristics cause the
index structure of 10-15 - 10-13as representation atmosphere
index turbulences areused for calculateintensity distribution medium of the atmosphere behave random fluctuation
model (scintillation) using gamma-gamma. The results of index of refraction. The random fluctuation of the
experiment shows that in the weak to moderate scale of propagation medium is called turbulence, which the size,
turbulence, highest value of mean SNR and high quality dimension and density of the air change randomly in
mean BER are achieved for spherical waves. While from space and time. Due to index of refraction fluctuates
measurements the <BER> is in the range of 10-6 10-11. randomly, the intensity of light that propagate along the
Keywords-component; free-space optical communication medium suffer attenuation and scintillation. Those all
(FSOC), turbulence media, Scintillation,< Prfade>,< BER>. phenomena finally degrade the strength of signal
performance of FSOC in the receiver system.
I. INTRODUCTION Some research works have been devoted to study
Free-Space Optical Communication (FSOC) system these phenomena intensively in order to enhance the
has been implemented widely in many contries so rapidly FSOC performances and mitigate the effect of
because its provide high link of capacity, free-license, turbulence. The signal strength suffers from degradation
low cost of deployment, easy of maintenance, and could and also has fluctuation of intensity at the receiver then
be integrated with existing communication system [1]- the FSOC system is designed to be amplified by EDFA.
[3]. The latest development of FSOC is used as an The EDFA is configured in a saturated regime condition
integrated space-terestrial network e.g. to enhance to boost the signal strength at the receiver, in order to
communication links for satelite to satelite crosslinks, up- mitigate signal successfully[10] [11]. The effect of
and-down between space platforms and aircraft, ships, turbulence also cause the beam wandering in receiver
and other ground platforms, and among mobiles and side. To overcome this wandering problem, an array
stationary terminals terestrial [4]. The potential of link detector system has been implemented at receiver and
capacity of FSOC have been achieved at the scale of 4.10 have been reported succesfully [12]. Another work also
Gb/s with the length of transmission 2.4-Km amplified by studied intensively in transmitter system in order to
Erbium Doped Fiber Amplifier (EDFA) [5]. Enormous mitigate the effect of turbulence. The multi input multi
bandwidth also have been investigated by modulated output (MIMO) method is implemented, where the
32x40 Gbit/s of WDM system in FSOC over 1,2-Km multiple laser diodes are applied at the transmitter and
using laser diode 1550-nm [6]. FSOC also has been multiple photodetectors are applied at the receiver
integrated with the broadband network in Japan with the system. The MIMO system is reported succesfully
length transmission of 2-Km by implementing 800-nm overcoming the turbulence effect and enhance the FSOC
laser diode and tested on natural environment such as rain performance [13].
and fog [7]. Another work also showed that FSOC has In this work we use the FSOC of fiber detection
many advantages of free of EMI, inexpensive method in order to analize its performance in turbulence
deployment and more faster, while RF signal was medium. The turbulence media is using Tube Propagation
transported through the link [8]. Evenmore FSOC have Simulator (TPS) that designed to capable simulate the
been used as wireless broadband in order to support the turbulence as well as at atmosphere. We use optical
optical fiber system in metropolitan area by using LED modem of 1550 nm as the main source of optical
transmitter [9]. communication equiped with EDFA with output of +23
dBm. The output beam of EDFA is collimated and
The major problem of FSOC system is the media of
transmitted through TPS, passing the turbulence media.
propagation is atmosphere, which its natural
C8-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
=
TPS is modelled in laboratory by a simulation of using of
plastic tube filled with the flow of hot water vapour (1)
(steam) and mixed with the cold air at temperature of Where is the quantum eficiency of photodetector, Ps is
160C. This turbulence media is called random media of the received signal intensity, is the light frequency, h is
tube propagation simulator. The motivation of this work is planck constant, and B is the bandwidth. Eq. 1 is the
to investigate the characteristical performance influenced value of signal to noise ratio (SNR) by ignoring the
by scale of turbulence that is designed from weak to
background illumination, circuit and thermal noise which
strong turbulence.
called limited shot-noise. The parameter of BER (bit
error rate) in the form basic modulation OOK (on-off
II. BASICS THEORY
keying) is represented as [15]:
Free-Space optical communication system is
lightwave communication that using free-space for =
(2)
propagation medium rather than optical fiber. The
fundamental difference of FSOC and optical fiber Eq. 2 represents the value of BER by considering the
communication systemis that FSOC is not guided at free- random noise in the photodetector that lead to mistaken
space, whereas the optical fiber is guided and immune bit from 0 to be 1 or vice versa [15].
from the surrounding noise. The FSOC are consists of In the case of turbulence in weak scale that lead to
three main parts TX, Free-space terestrial medium, and irradiance fluctuation or scintillation the governing
RX [14]. FSOC consist of laser diode as an optical equation PDF (probability of density function) on the
photodetector is the lognormal and modelled in Eq. 3:
~ +
Pr =< " >= $ %& ( ,((3)
transmitter (TX) and photodiode as the optical detector
(RX) gathered with the optical system to direct the
optical beam [14]. The laser beam is collimated via Where-& .0, 1 + 13 4 = -& 56 is the flux variance and
optical lens configuration system, which called telescope
dependent upon the diameter of aperture. Due to the
transmitter to direct the optical beam through air to reach
nature of turbulence is random fluctuation, the value of
the receiver lens and focused into the photodetector as as
SNR is no longer of deterministic but rather than mean
shown in Fig. 1.
value and can be expressed as [15]:
Atmosphere
Direct Detection Method
= (4)
8 9: < = <
T Beam R
; >
X Expander X
(5)
%& ( is the gamma-gamma distribution of unit mean as
There are several advantages of FSOC system which
are listed as implementing smaller antenna (telescope),
the representation of PDF:
?@ ABC
( = ( I?H@ .2KLM(4, ( > 0 (6)
smaller size and weight of the components, power
?9@ / H
concentration in a very narrow beam, and enormous & D E D F
bandwitdh. FSOC, furthermore is considered to be more
compact, simple configuration device, and inexpensive And for the case of spherical wave the parameter of and
compared to its technological competitor such as optical as the representation of atmospheric trubulences, are:
L=
fiber and microwave communication. For that reason, it
S b
is now being developed so vastly for many areas of (7)
R .UVC< a
communication. The simple configuration of FSOC NOP R _] aH
X<] \
system is the direct detection method, which means at the RWXB .XYZ< B .[\C [ ^ a
receiver side the optical beam is directly collimated by a Q `
M=
lens onto photodetector, as shown in Fig. 1. The benefit
of Direct Detection method are simple and unnecesary to c[] (7)
S X<] \b
< [^
R .[XC WXB .\VC a
use the optical fiber as the point of focus spot from the
NOPR aH
X<]
receiver lens. This method also reduce the effect of beam
RWXB .VZ< B .\<Z< C [ ^ a
wandering. However there is a disadvantage of direct Q `
detection method due to shot noise caused by influenced
of environment temperature outdoor. Also a mandatory
requirement to locate the detector outside the door. III. METHOD OF EXPERIMENT
Moreover it requires the optical filter to reduce the
background noise of another optical sources such as
The FSOC system on the experiment is shown in
comes from the sun or another that may be detected by
Fig.2. And the TPS (Tube Propagation Simulator) is also
the photodetector.
shown in Fig. 3. While the the scheme of turbulence scale
When the turbulence is assumed absence on the
is shown in Table 1. Optical modem of 1550 nm is used
medium of propagation or in the atmosphere the SNR is
in FSOC system, equiped with EDFA with gain of +23
represented as [15]:
C8-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
dB. Beam collimator is collimating the beam output of a. <Prfade> of strong turbulence
EDFA transmitted through turbulence media in TPS and
reach the lens focuser at Unit RX. Focused beam from Plane Wave
lens focuser directed into fiber (SMF/MMF) to measure *Spherical Wave
Gaussian Wave
the scintillation by Power Meter.
Plane Wave
*Spherical Wave
Gaussian Wave
Plane Wave
*Spherical Wave
Gaussian Wave
C8-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
c. <Prfade> of moderate turbulence From the the results of simulation and measurements
can be well understood that:
The order <Prfade> and <BER> is getting lower
ModerateTurbulence
as the rise of <SNR>
The order <Prfade> and <BER> is getting lower
as the rise of scale turbulences.
The difference results of simulation and
measurements is caused by the index
scintillation in gamma-gamma model is the
d. <BER> of moderate turbulence means from three beam waves model.
The results from measurements is more exact
value from the scintillation distribution of the
Weak Turbulence received power.
The gamma-gamma simulation by all means is
approaching the results of measurements.
The highest characteristical performances of
FSOC system is well achieved at strong
turbulences with the order of <Prfade> and
e. <Prfade> of weak turbulence <BER> are 10-1, 10-8, 10-10, and 10-10,
respectively.
Weak Turbulence
The effect of intensity deterioration caused by the
random absorption, diffraction, and scattering of beam
wavesthat cause a random mean SNR. The higher
intensity fluctuation on the photodetector causes the
probability of fade become more higher also. The fade
probability means that the profile intensity is fluctuated
f. <BER> of weak turbulence
Figure 5. <Prfade> and <BER>measurement in three scales of by the beam wandering at the receiver as well. The beam
turbulences. spot also moves randomly, hence the photodetector
receive a random wandering at the same time. This can
From the results of simulation using gamma-gamma degrade the performance of the FSOC system in fiber
which the distance of propagation is L = 1000 m can be detection method. The higher of scale of turbulence will
summarized as the Table 2 as follows: rise up the fade probability, hence the mean BER is going
to lower quality as the low of mean SNR. The high value
C8-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
C8-5
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
D1-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31,
31, Brawijaya University, Malang, Indonesia
open circuit voltage andd short circuit current is a key This point represents
epresents the maximum efficiency in
parameter [6]. convertering
ing sunlight into electricity [7].
[7
/
1 "1$ Figure 2. I-V characteristic for Wuhan Rixin MBF75 PV module
The purpose of fuzzy control is to extract
ln 1
1* "3$ maximum power from PV modules at a certain certa level
of solar iradiation [8]. Fuzzy logic control has several
advantages such as suitable for use on systems that are
I-V
V characteristic of PV cells can be defined as not linear and cann work with imprecise inputs.
inputs Fuzzy
follows : control (Fig. 3) using input voltage (V) and current (I)
and generates a duty cycle (D) output
outp that used as
!" # $ # buck converter input. The input voltage and current for
exp ' 1( "4$
%& # fuzzy control is from the output voltage and power
from PV module. Fuzzy logic control consist of three
stages : fuzzification, inference method and
Where IPH is the photovoltaic current, ID is the diode defuzzification.
current, RS is the series resistance and RP is paralel
resistance of PV cells.
B. I-V curve of PV cells
The relationship of current-voltage
voltage is used to
measure the electrical characteristics of PV devices.
devices
The I-V curve describes the flow of voltage through
the imposition of a short-circuit
circuit current ISC to open
Figure 3. Fuzzy logic controller.
controller
circuit voltage VOC. This curve is used to obtain the
level performance of PV systems (cells,cells, modules, PV
array). I-VV curve is obtained by performing A. Fuzzification
experiment with exposing PV cells or modules at the Fuzzification stage is a stage where the input
level of constant iradiation, maintaning the cell variable was changed into the language (linguistic)
temperature, varying the load resistance, and then based on the membership
mbership function. Triangular and
measuring the resulting current and voltage
volt [7]. I-V trapesium membership function with seven fuzzy
curve for Wuhan Rixin MBF75 PV modules can be subsets VVS (very veryy small), VS (very small), S
seen in Figure 2. Horizontal axis is for voltage and (small), M (medium), B (big), VB (very big), VVB
vartical axis is for current. (very very big) is used (Fig. 4). Variable V and I are
PV cells can operate on a wide range of areas of used as input, and variable D as output.
current and voltage. Simply by varying the load
resistance from zeroero (short circuit) to infinity (open B. Inference method
circuit), it is possible to determine the highest At this stage, Mamdani method is used to control
efficiency PV cells deliver maximum power. Because the generated output.
utput. The design of basic rules (rule
power is the result of voltage multiplied by current, base) consists of 49 fuzzy control rules. This rule
then the point of maximum power (Pm) appears in the implemented by computer and used to control the duty
I-V curve where the outcome of the current (Imp) cycle of buck converter in order to obtain maximum
multiplied by voltage (Vmp) is maximum. No power is power from PV modules in different conditions. These
generated on the condition of short circuit or open rule were expressed as IF-THEN
THEN statements as follows
circuit conditions, so that maximum power is
generated only at one point on the curve called knee. R1 : IF V is VVS and I is VVS THEN D is VVB.
D1-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
I VVS VS S M B VB VVB
M VVB VB B M S VS VVS
C. Defuzzification
At this stage, the outputs of fuzzy logic control is
changed from linguistic variables into numeric
variables using membership function. With (b)
defuzzification, fuzzy logic control can generates
analog output signal that can be converted into digital
signals and control the power converter of MPPT
system. Centroid type of defuzzification is used for
this research.
IV. SIMULATION OF FUZZY MPPT
The MATLAB/Simulink software is used for the
simulate fuzzy MPPT with PV module and resistive
loads. The system consist of :
A. PV Module Block (c)
Simulate the non-linear I-V characteristic of Figure 4. Membership function : (a) first input V, (b) second input
Wuhan Rixin MBF75 PV module. Table 2 I, (c) output D.
summarized specifications of the PV module.
B. Fuzzy Controller Block
TABLE II. PV MODULE SPECIFICATION Simulates the fuzzy MPPT process and computes
Brand Wuhan Rixin the desired duty cycle of the buck converter using
Model MBF75 solar panel voltage and current. The fuzzy controller
Material Polycrystalline Silicon
block performs the fuzzification, inference method,
and defuzzification process.
Power output (max) 75 W
Voltage output (max) 17,5 V C. Buck Converter Block
Current output (max) 4,29 A DC DC converter is used to convert the DC input
Open circuit voltage 21,5 V voltage that varies into controlled DC output voltage at
Short circuit current 4,72 A the desired voltage level. Buck converter (Fig. 6)
Open circuit voltage -0,35% / C produces a dc output voltage of the average lower than
temperature coefficient the input DC voltage. A capacitor (C) with a value of
Short circuit current +0,036% / C 2200 F is used to reduced module ripple voltage.
temperature coefficient The equation of the buck converter circuit is as
1
Working temperature - 40 ~ 90C follows :
2 3 . - ./ #/ 5 6 (5)
-./
-0 1
- 1
= (./ .59: ) (6)
-0 8
< = + #=>? (./ .59: ) (7)
D1-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Cu
rre
nt
(A)
V 1000 W/m2
200 W/m2
Figure 6. Buck converter circuit (L=1mH, C=2200F,
RL=80m, RC=5m).
t (s)
D. PWM Block
Generates the pulse signals for the buck converter
based on the desired duty cycle. Figure 7. Load current without fuzzy MPPT and with fuzzy MPPT
E. Load
A 15 ohm resistive load is connected to PV
module via buck converter.
V. RESULT AND ANALYSIS Vo
This chapter contains the results and analysis of lta
fuzzy MPPT simulation system (Fig. 5). Fuzzy MPPT ge
(V)
simulation uses two operating conditions. Case 1
without the fuzzy MPPT (e.g., direct connection of 1000 W/m2
module PV and the load). Case 2 with fuzzy MPPT. 200 W/m2
The solar iradiation used for simulation is vary
between 200 ~ 1000 W/m2. Figure 7 and 8 show the t (s)
load current and voltage characteristic of the two
conditions.
With the change of solar iradiation, the current Figure 8. Load voltage without fuzzy MPPT and with fuzzy MPPT.
through the load R = 15 ohms also changed. Figure 7
shows that in case 1 (without fuzzy MPPT) by VI. CONCLUSION
changing of solar iradiation from 1000 W/m2 to 200
From the simulation results and analysis, it can be
W/m2 resulted the load current drop from 1,38 A to
concluded that PV system using fuzzy MPPT has
0,81 A. While case 2 (using fuzzy MPPT), the load
better performance than the system without MPPT.
current drop from 1,10 A to 1,06 A.
This can be seen on the load current and voltage
This condition is also hold for the load voltage
curves which are more stable than the system without
(Fig. 8). In the condition without fuzzy MPPT (case
MPPT. System with fuzzy MPPT have better stability
1), the load voltage drop from 20,74 V to 12,18 V.
even in dynamic operating conditions.
While using the fuzzy MPPT (case 2) the load voltage
down from 16,60 to 15,95 V. These results indicates
that PV system using fuzzy MPPT has better
performance than PV system without using fuzzy
MPPT.
D1-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
D1-5
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
D2-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
applies heat by convection through the atmosphere shock. The thermal shock effect can be reduced by
inside an annealing furnace. To avoid oxidizing any applying graded pre-heating.
un-evaporated lubricant residues or forming magnesium
oxide on magnesium-bearing alloys, annealing may be III. METHOD
carried out in a dry, inert (low O2) atmosphere such as First analysis of the annealing problem is temperature
nitrogen gas. A large integrated aluminum rolling plant and airflow distribution since decrease in temperature in
may have its own nitrogen generating plant for this some volume of chamber can induce incomplete
purpose [2], [6]. evaporation. This has caused to some rolls of aluminum
Non-heat-treatable aluminum alloy, commonly foil is still in worse wet-ability and/or sticky. This
heated for 1.5 2 hours in the range of operating analysis is conducted based on the simulation results of
temperature 635 765oF or equivalent to 335 445oC Computational Fluid Dynamics (CFD) simulation
[6], [7]. The heat released to the rolled aluminum has intended to analyze temperature and air flow
another objective. It also evaporates the carrying coolant distribution in the empty and filled chamber of Foil
oil inside rolled aluminum. So thats why, annealing Annealing Furnace (FAF). The furnace is distinguished
chamber must be dry (very low oxygen intensity) to as FAF A and FAF B which the aluminum is in specific
avoid oxidizing of coolant oil in the surfaces. orientation inside the chamber.
Annealing process in SAI is batch annealing. It
means loading a furnace with a batch of metal, roll of A. Computational Fluid Dynamics
aluminum foil, and holding it there until the annealing The computational fluid dynamics, usually
process is complete. Rolls of aluminum foil are annealed abbreviated as CFD, is a branch of fluid mechanics
as a single batch, depending on the size of the foils and using numerical methods to analyze and solve problems
the size and shape of the furnace [2]. In batch annealing, that involve flows of fluid. Numerical method is built by
heat conveyed by the furnace atmosphere to the outside employs the governing equations such as conservation
surfaces of the foils must be conducted through the of energy, momentum and continuity. Energy
metal to the innermost layers, and sufficient time must conservation is determined as equation shown below
be allowed for all parts of each foil to absorb enough [8], [9].
heat to achieve the planned anneal [1], [3]. Batch
annealing is an efficient approach and is the most (E ) + .(r (E + p )) = .keff T
commonly used method in high-production aluminum t (1)
+ .( eff . ) + S h
foil mills. r
160
For the solid region (i.e. newborn body), energy
140 transfer is calculated by employing equation as follow
Pre-heating 2 (160oC ~ 8 hr)
120 [8], [9]:
100 Pre-heating 1 (130 oC ~ 8 hr)
r
80 ( h) + .(v h) = .( kT ) + S h (2)
60
t
40
where is solid density, h is sensible enthalpy, k is
0 10 20 30 40 50 60 70 80 90 conductivity constant of newborn, T is newborn skin
Time (hours)
Fig. 1. Heat treatment scheme of batch annealing in SAI
temperature, and Sh is volumetric heat source.
The equation (1) and (2) are complemented by
The annealing scheme is describe as three stages of continuity and conservation of momentum defined
thermal treatments (See Fig. 1). They are heating, below:
soaking and cooling. In heating stage, aluminum foil is
heated to particular temperature up to 5 hours.
. u = 0 (3)
Temperature setting depends on the thickness of du
aluminum foil. Soaking stage holds annealing chamber = F p + 2 u (4)
temperature to particular value for 15 20 hours. This dt
step also includes evaporating and drying. The last where p is normal pressure (N/m2), F is body force on
stage, cooling, chills amount of rolled aluminum for two solid region.
hours inside the annealing chamber. Itll be pulled out For natural-convection flows, faster convergence of
later if the temperature reaches 70oC [2]. numerical calculation can be retained with the
Evaporation of carrying coolant oil of 6.5 micron Boussinesq model. It sets the fluid density as a function
aluminum foil is estimated to occupy 15 hours of of temperature. The Boussinesq model is represented by
heating with temperature setting 225oC. Since the equation below:
presence of receding fold (RF) on aluminum foil after = 0 (1 (T T0 )) (5)
annealing, it is identified that the mechanism of
preheating (big temperature different) induces thermal
D2-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
dE cv pn
V
2
= Q& cv W& cv + m& i hi + i + gz i
dt p =1 2 (9)
qn
V
2
Fig. 2. Temperature distribution of FAF A in (a) right-
m& e he + e + gz e left view and (b) front-rear view
q =1 2
This Fig. 2 expresses temperature distribution of FAF
There are no work applied in the control volume so that A chamber. It shows that the chamber has better
the value of W& cv equals to zero. Potential energy temperature distribution in bottom to half of chamber
difference of inlet and outlet can be neglected since the height. The upper stage reaches temperature setting in
value is too small. Steady state analysis results as the center.
follow:
D2-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
D2-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
150 1 1 0.97
2 1 0.93
100 Metal Z1 1 0.97
Air Temperature at Zone 1 Metal Z2 ~ ~
50 Air Temperature at Zone 2
Air Temperature at Zone 3
Metal Temperature Both of FAF A and FAF B have a quite big different
0 of conduction time theoretically and actual, taken from
0 10 20 30 40 50 60 70 80 90
Time (hours)
response of temperature control. As usual, this data
Fig. 8. Response of temperature control of annealiang analysis indicates that improper heat treatment is
process in FAF A occurred while annealing. Two conditions that might
become the cause of this condition are undesired heat
The steady conduction time of aluminum foil rolls in process/heat transfer and inappropriate control mode for
FAF A is theoretically obtain by dividing conductive temperature annealing.
heat transfer by accumulative evaporation heat. Based
on thermodynamic calculation, the conduction time of C. Discussion
aluminum foil in FAF A is 0.152 hour. This value is
Considering the result of temperature and airflow
lower than the actual conduction time which needs 1 4 distribution from CFD simulation, FAF A has good
hours to settle. The detail data of settling/conduction thermal distribution than FAF B. However, the settling
time of aluminum foil in FAF A is as in Table 1. time to evaporating phase in FAF A is about 1 4 hours.
Table 1. Settling time for evaporation temperature The FAF B has average settling time to evaporation
set point of FAF A phase about 1 hour. It fit to transition setting time of
pre-heating and evaporating. The temperature and
Zone T Set 0.5oC Tinit evap. airflow distribution for FAF B is worse than FAF A.
1 1 0.96 That is caused by the profile of airflow inside the
2 1 0.96 chamber is different. Comparing to furnace that has
inlet-outlet in the side of chamber, the airflow of FAF B
3 1 0.96
is worse than the FAF A. This condition is caused by the
Metal Temp. ~ 4 geometry of blade sticked in the inlet and outlet zone is
different to FAF A. The orientation of aluminum foil roll
The recorded temperature response of annealing in is also affect to airflow distribution. It is recommended
FAF B is follow. to change the blade of FAF B as the FAF A has or
change the orientation of aluminum roll annealed in the
250
chamber to get better het treatment.
Comparing settling time of metal/aluminum in to
200 conduction time of aluminum oil rolls in each FAF,
there is big different value. Theoretically in FAF A, it
only need about 0,016 0,152 hour to reach the set point
Temperature (K)
150
temperature. However in fact, aluminum foils need 4 to
reach the set point temperature. Moreover aluminum foil
100 in FAF B cant achieve the set point temperature. It is
Air Temperature at Zone 1
probably caused by two problems explained in the
Air Temperature at Zone 2 previous point. They are improper tuning controller and
50
Metal Temperature at Zone 1 the condition of insulating material. Due the FAF A and
Metal Temperature at Zone 2 B still using PID controller, it should be check that the
0 value of proportional constant, time derivative and time
0 10 20 30 40 50 60 70 80 90
Time (hours)
integral setting is based on transient response. Since the
settling time is too much longer then the proportional
Fig. 9. Response of temperature control of annealiang constant should be substituted to higher value to get fast
process in FAF B response. The other problem may be caused by the
insulating material (e.g. glass wool, grafite, gypsum,
The evaporation settling time based on control
etc) inside the annealing chamber doesnt work properly
response in FAF B (see Table 2) is about 1 hour for
so that there is so much heat losses during the annealing
aluminum in zone 1 and cant be reached for aluminum
process. In order to reduce the heat losses, it is
in zone 2 (see Fig. 9). The conduction time for annealing
recommended to check the condition or thermal
aluminum foil in FAF B is 0,042 hour theoretically.
conductivity of the insulating material.
D2-5
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
V. CONCLUSIONS REFERENCES
Annealing treatment of AA 1235 in foil annealing [1] The Aluminum Association. 2007. Rolling Aluminum: From the
Mine Through the Mill. The Aluminum Association, Inc.
furnace has been analyzed. FAF A has quite better [2] Visual Quality Characteristic of Aluminum Sheet and Plate, the
temperature and airflow distribution which is set point Aluminum Association Inc., 4th Edition February 2002
temperature and higher airflow magnitude is distributed [3] Annisa Kesy Garside, Penentuan Setting Parameter Proses
Finishing Rolling untuk Aluminium Foil dengan Thickness Exit
in upper side of chamber. The worst temperature and
7 Mikron di PT. Supra Aluminium Industri. Laporan Magang
airflow distribution is possessed by FAF B. The Dosen, Program Hibah A1, Jurusan Teknik Industri FT,
upper-right side of chamber gets lower temperature so Universitas Muhammadiyah Malang, 2005.
that it may induce weta. There is heating problem due [4] Smith, W. F. 1990. Principle of Materials Science and
Engineering 2nd Edition. New York: McGraw-Hill Publishing
the big difference between conduction time of real Company.
process and theoretical calculation. It is probably caused [5] Jing Zhang, Fusheng Pan, Rulin Zuo, Chenguang Bai. The low
by improper control tuning of PID controller in the temperature precipitation in commercial-purity aluminium
sheets for foils. Journal of Materials Processing Technology.
furnace and heat losses by under works of insulation 2008; 206: 382 387.
material. [6] Ozgul Keles, Murat Dundar. Aluminum foil: Its typical quality
problems and their causes. Journal of Materials Processing
Technology. 2007; 186: 125 137.
VI. ACKNOWLEDGMENT [7] R. J. Vidmar. (1992, August). On the use of atmospheric plasmas
Thanks to Dr.-Ing. Doty D. Risanti and Dyah Sawitri, as electromagnetic reflectors. IEEE Trans. Plasma Sci. [Online].
21(3). pp. 876880. Available:
M.T. for the helpful comments on the analysis of http://www.halcyon.com/pub/journals/21ps03-vidmar
annealing treatment. This study was supported by PT. [8] J. Blazek. Computational Fluid Dynamics: Principle and
Supra Aluminium Industri Pasuruan for giving Applications. ELSEVIER SCIENCE ltd. 2001.
[9] Fluent manual. Modeling Heat Transfer. Fluent Inc.. September
measurement data of FAF and Indonesia-Germany Fast 29. 2006.
Track Scholarship from Directorate of Higher Education [10] Incropera, F. P. and D. P. DeWitt. 1996. Fundamentals of Heat
for the grant. and Mass Transfer 4th Edition. U.S.A.: John Wiley & Sons, Inc.
D2-6
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
I. INTRODUCTION
The development of automotive rapidly that increase
the number and variety of vehicles make maintenance
and repairs are done differently for each type of car. One
variation of the car is different height of each type. Such
as jeep would be higher than a sedan and car
modification cars that usually lower. Figure 1 Bottle Type Hydraulic Jack Schematic
The various height of the car raises a matter of Courtessy: www.hyjack.net
convenience for people who will do the repair parts are
located at the bottom (under the car) and ultimately will Where:
affect the time taken and the results of repair. This a. Reservoir
problem is necessary to drive automatic hydraulic jack b. Main Cylinder
can be adjusted to the person who will fix the car. c. Piston
This research designed a height control systems of d. Pump
hydraulic jack model based on Takagi Sugeno fuzzy e. Screws relief
algorithms which is simple and reliable to be applied as a f. Oil channel to the piston
control on a wide range of systems using only input g. Check Valves
variables and output without having to know the system
D3-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Based on the design of the hardware block diagram in a. Determine the input variables and output variables.
fig. 2, there are two ways of working tools, which Input variables for fuzzy controller are error and
controls the initial speed and control of the opening and error, while the output variable is deltaOutput,
closing the valve. Height that was desired and that has with
been achieved perform through the LCD monitor. Err(n)=SP(n)PV(n) (1)
Variable used in the design can be seen in Fig. 3 deltaErr(n)=Err(n)Err(n-1) (2)
below. Output(n)=Output(n-1)+deltaoutput(3)
where:
Hreq = height total desired
H2 = Height of objects that affect Hreq
Hreff = Set point as a result of the difference between
H2 and Hreq
Hcurr = output of jack, from condition has not lifted up
to Hreff
Figure 6 Diagrammatic of Membership Function of Error
DC motors drive a prototype lever hydraulic jack
rated voltage source of +12 volt dry batteries as a power
c. Fuzzy logic controller rules is based on experience
supply.
and in the form of If-Then. After the crisp input is
The desired height is determined by the A (65mm), B
converted into fuzzy input, according to Takagi
(75mm) and C (85mm), height of the car with a sliding
potentiometer. Microcontroller as a fuzzy logic Sugeno method it is processed based on 9 standard
controller receives input from the difference in height of rules below:
one of the buttons with sliding potentiometer and 1. If error = PS and error = PS, then output =
calculate the height of the jack of potentiometer that has 1.21V.
been converted by the ADC. 2. If error = PS and error = M, then output =
After the weight and speed are involved in each range, 1.21V.
, so that fuzzification process result membership degree 3. If error = PS and error = PB, then output =
of each input value. 1.21V.
Further evaluation rule, where the entries have been 4. If error = M and error = PS, then output =
involved in the rule base, defuzzification using Takagi 1.35 V.
Sugeno method generate the output (control signal). 5. If error = M and error = M, then output =
1.35 V.
IV. FUZZY LOGIC CONTROLLER 6. If error = M and error = PB, then output =
1.21 V.
Fuzzy logic controller algorithm as shown in fig. 4
D3-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
7. If error = PB and error = PS, then output = Table 5 Testing Result of Hreq 65 mm and Href 53
1.55 V. Href (mm) LCD Display Actual Height
8. If error = PB and error = M, then output = (mm) (mm)
1.55 V. 52 52.4
9. If error = PB and error = PB, then output = 52 51.5
1.55 V. 53 51 51.1
52 51.5
V. TEST AND ANALYSIS 51 51.4
D3-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
[6] Mihajlov,Miroslav,Vlastimir Nicolic, Dragan Antic, Position Fitriana Suhartati was born in Sidoarjo on 17th October 1974. She
Control of electro-hydraulic Servo System Using Sliding Mode hasgraduated from electrical engineering magister program at Sepuluh
Control Enhanced by Fuzzy Controller, Facta Nopember Institute of Technology, Surabaya, Indonesia, in 2003. Her
UniversitatisSeries:Mechanical Engineering Vol I No. 9, Serbia major field of study is control systems.
Montenegro, 2002. Since 1998, she has teached in electrical engineering department of
[7] Reznik,Leonid, Fuzzy Controllers,Victoria University of Brawijaya University, Malang, Indonesia. She has many publications,
Technology, Newnes, Melbourne, 1997. and she written a book Adaptive Control Systems (TEUB, Malang,
[8] Ross, Timothy J. Fuzzy Logic With Engineering Aplication. 2009). Her research interest is in the field of intelligent control systems
McGraw-Hill Inc., 1995. and adaptive control systems with application to motor and solar
[9] Sullivan, James A, Fluid Power:Theory and Applications. energy systems.
Prentice-Hall,Virginia, 1975.
[10] Zuhal, Dasar Teknik Tenaga Listrik dan Elektronika Daya, PT. Ahmad Fahmi was born in Malang on 31st July 1973. He
Gramedia Pustaka Umum, Jakarta, 1993. hasgraduated from electrical engineering magister program at Sepuluh
[11] www.hyjack.net/animation Nopember Institute of Technology, Surabaya, Indonesia, in 2005. His
major field of study is control systems.
Since 1998, he has teached in electrical engineering department of
State University of Malang, Indonesia. He has many publications, and
he written a book Robotics (UM, Malang, 2009). His research interest
is in the field of control systems and fuzzy logic with application to
motor and renewable energy.
D3-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Abstract - Cancer is a disease that related with work for helping doctor to make decission.
uncontrolled cell growth. To date, lung cancer is one of the
most deadly disease. An application of ANFIS with 1.1 Lung Cancer Identification
subtractive clustering for lung cancer early detection system Lung cancer is a disease characterized by
was developed in this study. Characteristic and chest x-ray uncontrolled cell growth in tissues of the lung. It is also
datas were used in this study. The data used to build best the most preventable cancer. Cure rate and prognosis
ANFIS model, that will be applied in the software. Results depend on the early detection and diagnosis of the
from the software was validated with doctors decission.
disease. Lung cancer symptoms usually do not appear
Parameters that used to determine system performance are
RMSE, VAF, and the succes rate. The best ANFIS model until the disease has progressed. Thus, early detection is
for characteristic data was obtained in ra = 0,4; RMSE for not easy. Many early lung cancers were diagnosed
training = 0,1193, RMSE for testing = 0,2030, VAF for incidentally, after doctor found symtomps as a results of
training = 93,34%, VAF for testing = 82,28%, the success test performed for an unrelated medical condition4].
rate of software for training data = 96 % and for testing There are two major types of lung cancer: non-small
data = 96%. While for chest x-ray data, the best model was cell and small cell. Non-small cell lung cancer (NCLC)
obtained in ra = 0,4; RMSE training = 0,0185, RMSE testing comes from epithelial cells and is the most common type.
= 0,1063, VAF training = 99,85%, VAF testing = 94,84%, the Small cell lung cancer begins in the nerve cells or
success rate of software for training data = 95,56 % and for
hormone-producing cells of the lung. The term small
testing data = 88,46%.
cell refers to the size and shape of the cancer cells as
Index Terms: ANFIS, characteristic, chest x-ray, lung
seen under a microscope. It is important for doctors to
cancer, subtractive clustering distinguish NSCLC from small cell lung cancer because
the two types of cancer are usually treated in different
I. INTRODUCTION ways. Lung cancer begins when cells in the lung change
and grow uncontrollably to form a mass called a tumor (or
Cancer is a disease associated with the uncontrolled a lesion or nodule). A tumor can be benign
cell growth. Today, lung cancer is one of the most deadly (noncancerous) or malignant (cancerous). A cancerous
desease. According to World Health Organization (WHO), tumor is a collection of a large number of cancer cells that
every year there are more than 1.3 million cases new of have the ability to spread to other parts of the body. A
lung cancer and bronchitis in the world, and the mortality lung tumor can begin anywhere in the lung [
rate approximately 1.1 million [ HYPERLINK \l "rhd11" HYPERLINK \l "Per03" 5 ].
1 ]. In Indonesia, 1 of 1000 persons is a new sufferer of
lung cancer, it means that more than 170.000 new 1.2 Adaptive Neuro Fuzzy Inference System (ANFIS)
sufferers annually2]. Both in Indonesia and other
Adaptive Neuro Fuzzy Inference System (ANFIS)
developed countries, reported that most of cases diagnosed
is combination of fuzzy inference system (FIS) which
when its were in advance stage (stage III and IV).
illustrated in neural network architecture. Fuzzy
An artificial intellegent system for lung cancer early
inference system which used is first order Takagi-
detection based on characteristic and chest x-ray was
Sugeno-Kang (TSK) model, for computaion simplicity
designed by [ HYPERLINK \l "Ari11" 3 ]. But, this
and convenience6].
software has the best performance around 66,6%. Then,
ANFIS structure consists of five layers represent
the aim of this study is to develop these artificial
neural network architecture is shown in fig1. The square
intellegent with deifferent clustering method, subtractive
node is an adaptive node, it means parameters value can
clustering. It be expected to improving system
change in the midst of training process. The circle node is
performance. The used subtractive clustering, for making
non adaptive node with fixed value. There are different
decicion based doctors experties consistently. This system
equation for each layer.
D4-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
, = 234 -./0"-01 /
following equation:
where 5 =
6
(7)
Fig. 1. ANFIS structure [ HYPERLINK \l "Jan93" 6 ]
78
Layer 1 Pi is the potential of ith data point, n is the total number
Mathematical equation for this layer dependent on of data points, xi and xj are data vectors in data space
type of membership function. For example, if gaussian including both input and output dimensions, is a
membership function: positive constant and is selected as 4, and ra is a positive
, = = = 1,2
constant defining the neighborhood range of the cluster or
(1)
, = = = 1,2
simply the radius of hypersphere cluster in data space.
(2) Each time a cluster center is obtained, the data
points that are close to new cluster center are penalized in
Fig 1 illustrated an ANFIS with two inputs (x and
y). The output ( , is inputs membership degree.
order to facilitate the emergence of new cluster centers.
The revising of the potential is done by subtraction as
Membership function that used is gaussian with parameter
shown in the following equation:
and c, called premis parameter. Its value can be
/=" = # /
<- > A
determined from ANFIS training in MATLAB software.
, = , ,; . ?@
where B = C. D
(8)
Layer 2
Layer 2s fuction is to multiply every input signal
, = = . , = 1,2
Pi* is the i-ths new potential value, is squash factor, a
which comes from layer 1s output. The equation is:
positive constant greater than 1. The positive constant rb
(3)
is somewhat greater than ra and it helps avoiding closely
The node in layer 2 is non adaptive node (fixed spaced cluster centers.
parameter). Node number in this layer show the created To accept or reject new cluster center, the following
rule number. criteria was suggested by [7]:
> H
Layer 3 E"
E F
Every node in layer 3 is non adaptive node that If
show normalized firing strength, the i-th node output ratio
<H
E"
accept as a cluster center and continue
, = = , = 1,2
!"
with all node output. The equation is:
E F
else if
!# $!
(4)
reject and end the clustering process
+ 1
MN"O E"
Layer 4 all previously found cluster centers
78 E F
Every node in this layer is adaptive node with if
%, = = & + ( +
equation: accept as a cluster center and continue
(5) else
In layer 4, there are normalized firing strength from layer reject set the potential at to 0. Select the
D4-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
result in an over-defined system. The suggested values for compared with doctors decicion. The validations result
and ra are 1.25 1.5and 0.15 ra 0.30 [7]. show the success rate of software for making decision.
output
0.5
In this early system development, characteristic and
0
chest x-ray datas were used in training for ANFIS
-0.5
modeling. Characteristic data provides information about 0 5 10 15 20 25 30 35 40 45 50
data ke-
normal and suspected patient. There four kinds of ra = 0,4 testing
1.5
characteristic data for identification: amount of cigarette
1
consumed per day, duration of smoking, occupation, and
output
0.5
cough. Chest x-ray data that used for training was
0
obtained from [8].
-0.5
Subtractive clustering was employed in training 0 5 10 15 20 25
data ke-
process. This method will generate data to make its
Fig 4: Validation graph
natural membership function. One of the subtractive
clusterings parameters is ra, becomes varibael for
obtaining some ANFIS models. Fig 2 shows the III. RESULTS AND DISCUSSIONS
subtractive clusterings parameters in MATLAB and fig 3 3.1 ANFIS Model
shows the ANFIS training process for ra = 0,4. 3.1.1 Characteristics Data
The result of ra variation in design for ANFIS model
for characteristics data shown in table 1. Its also show
the result from ANFIS model validation for training and
testing datas, represented in RMSE and VAF.
D4-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
With ra = 0,4, it was resulted 19 clusters. It means Table 4 VAF and RMSE ANFIS Model Comparison for Chest X-ray
Data
for each inputs, the are 19 membership functions in
Data Cluster RMSE VAF (%)
gaussian form. Then, this model used to design early
2 MF 0,364194 48,8421
detection software for characteristics data. Training 3 MF 0,29113 56,5538
The comparison for RMSE and VAF result for SUBTRACTIVE 0,0185 99,85
characteristics data between this study and [3] shown in 2 MF 0,25199 65,45689
table 2. From table 2, it was obtained that the result of this Testing 3 MF 0,275235 62,49139
study better than [3]. It was caused by subtractive SUBTRACTIVE 0,1063 94,84
clustering method can produces membership function
naturally, so that its more suittable with the system.
3.2 Early Detection Software
3.1.2 Chest X-ray Data The software have been created shown in fig 5 and
The result of ra variation in design for ANFIS model fig 6. Tabel 5 and table 6 give the comparison of
for characteristics data shown in table 1. Its also show softwares success rate. Form those tables, it was obtained
that either this study or [3]s study resulting sotfware with
the result from ANFIS model validation for training and
success rate more than 90%. It was caused by success rate
testing datas, represented in RMSE and VAF. calculation based linguistic variable only. In creating
software code, there is a value which used for making
Table 3 Chest X-rays Data Validation Result For ra decicion. For both studies, its value are not fixed.
Variation
RMSE VAF (%)
ra MF
Training Testing Training Testing
0,1 2 0,1550 0,3229 89,20 52,74
0,2 5 0,1131 0,0646 89,20 98,05
0,3 5 0,0210 0,0702 99,80 97,75
0,4 5 0,0185 0,1063 99,85 94,84
0,5 4 0,0316 0,1210 99,55 93,13
0,6 4 0,0572 0,1963 98,53 81,95
0,7 3 0,1689 0,2233 87,17 77,76
0,8 3 0,1584 0,2760 88,71 65,03
0,9 2 0,1546 0,1837 89,25 84,73
1 2 0,1754 0,2012 86,16 81,74
Table 4 shows that ra variation not always resulting Fig. 5. Early detection characteristics data software
different number of membership function. Althougt there
are any same number for several ra value, all of it RMSE
and VAF value are different. It shows that change of ra
Table 5 Characteristics data softwares success rate comparison
will resulting the change of ANFIS parameters, premis
Data Cluster Success rate
and consequent. 2 MF 96%
From the training result, minimum error and best Training 3 MF 98%
VAF were obtained for ra = 0,4. While for testing, SUBTRACTIVE 96%
minimum error and best VAF were obtained for ra = 0,4. 2 MF 91,43%
So, the model with ra = 0,4 was choosed to the best Testing 3 MF 97,15%
ANFIS model. This model then used to design early SUBTRACTIVE 95,56%
detection software for chest x-rays data. With ra = 0,4, it
was resulted 5 clusters. It means for each inputs, the are 5
membership functions in gaussian form.
The comparison for RMSE and VAF result for chest
x-rays data between this study and [3] shown in table 4.
From table 2, it was obtained that the result of this study
better than [3]. It was caused by subtractive clustering
method can produces membership function naturally, so
that its more suittable with the system. Beside that,
image processing in [3] was done for all chest x-ray
region. In this development, chest x-ray was processed
with GLCM, take only the region that contain chest
Fig. 6. Early detection chest x-ray software
image.
D4-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Table 6 Characteristics data softwares success rate comparison [5] Perhimpunan Dokter Paru Indonesia, "KANKER
Data Cluster Success rate PARU : PEDOMAN DIAGNOSIS &
2 MF 91,43% PENATALAKSANAAN DI INDONESIA,"
Training 3 MF 97,15% 2003.
SUBTRACTIVE 95,56% [6] Jyh-Shing Roger Jang, "ANFIS: Adaptive-
2 MF 90% Network-Based Fuzzy Inference System," IEEE
Testing 3 MF 85% TRANSACTIONS ON SYSTEMS, MAN, AND
SUBTRACTIVE 88,46% CYBERNETICS, VOL. 23, NO. 3, pp. 665-685,
1993.
IV. CONCLUSIONS AND FUTURE WORKS
[7] S. L. Chiu, "A Cluster Estimation Method with
A new method of lung cancer detection system has Extension to Fuzzy Model Identification," in
been developed by using ANFIS subtractive clustering. IEEE Internat. Conf. on Fuzzy Systems, 1994, pp.
Its can improve systems performance. The best ANFIS 1240-1245.
model for characteristics data was obtained in ra = 0,4;
[8] Sungging Haryo Wicaksono, "Design of Lung
RMSE for training = 0,1193, RMSE for testing = 0,2030,
Cancer Prediction System Based On Image
VAF for training = 93,34%, VAF for testing = 82,28%,
Pattern Recognition," Surabaya, 2011.
the success rate of software for training data = 96 % and
for testing data = 96%. While for chest x-ray data, the [9] Sylvia Ayu Pradanawati, "Pengembangan Sistem
best model was obtained in ra = 0,4; RMSE training = Kecerdasan Buatan Berbasis Adaptive Neuro
0,0185, RMSE testing = 0,1063, VAF training = 99,85%, Fuzzy Inference System Untuk Diagnosa
VAF testing = 94,84%, the success rate of software for Penyakit Kanker Paru-paru," ITS Surabaya, 2011.
training data = 95,56 % and for testing data = 88,46%. [10] Badan Pusat Statistik, "Perkembangan Beberapa
For further study, this system can be improved not Indikator Utama Sosial-Ekonomi Indonesia,"
only for detection, but alsocan determine lungs stage. Jakarta, 2010.
Moreover, this system could be applied in a website that [11] Sri Kusumadewi, Analisis dan Desain Sistem
could be accessed by everyone. Fuzzy Menggunakan Tool Box MATLAB.
Yogyakarta: Graha Ilmu, 2002.
ACKNOWLEDGEMENT
[12] Ratih Setyaningrum, "Kemampuan Expert System
The study described in this paper has been - ANFIS untuk Diagnosa Kesehatan Pekerja
developed in Instrumentation and Control Laboratory, Industri dan Mencari Solusinya," in Seminar
Engineering Physics Department, Faculty of Industrial Nasional Aplikasi Teknologi Informasi,
Technology, Institut Teknologi Sepuluh Nopember. The Yogyakarta, 2007, pp. L15-L20.
authors acknowledge the Fast Track DDIP (Double [13] Agus Priyono et al., "Generation of Fuzzy Rules
Degree Indonesia Prancis) Grants from DIKTI with Subtractive Clustering," Jurnal Teknologi,
pp. 143-153, 2005.
(Directorate General of Higher Education Indonesia) for
having supported. [14] Tarig Faisal, Mohd Nasir Taib, and Fatimah
Ibrahim, "Adaptive Neuro-Fuzzy Inference
System for diagnosis risk in dengue patient,"
V. REFERENCES Elsevier: Expert System with Application, pp.
[1] rhd. (2011, Maret) FAJAR Online. [Online]. 4483-4495, 2011.
http://www.fajar.co.id/read-20110302235529- [15] Adi Purnomo and Sulistyo Puspitodjati, "Aplikasi
merokok-dan-kanker-paruparu Pemrograman C# untuk Analisis Tekstur Kayu
[2] Tjandra Yoga Aditama, "Situasi Beberapa Parquet dengan Menggunakan Metode Grey
Penyakit Paru di Masyarakat," Cermin Dunia Level Co-occurrence Matrix (GLCM)," Fakultas
Kedokteran, pp. 28-30, 1993. [Online]. Teknik Industri Universitas Gunadarma, 2009.
http://www.kalbe.co.id/files/cdk/files/10SituasiPe [16] Arna Fariza, Afrida Helen, and Annisa Rasyid,
nyakitParu084.pdf/10SituasiPenyakitParu084.htm "PERFORMANSI NEURO FUZZY UNTUK
l PERAMALAN DATA TIME SERIES," pp. D77-
[3] Syamsul Arifin, "Design of Artificial Intellegence D82, Juni 2007.
Software for Lung Cancer Diagnosis using [17] Mesothelioma Health Information. [Online].
Adaptive Neuro Fuzzy Inference System," in 3rd http://www.mesothelioma-
International Conferences and Workshops on health.org/images/normlungX.jpg
Basic and Applied Sciences, Surabaya, 2011, p. [18] Elisna Syahruddin. (2006, Juni) kankerparu.org.
T002.
[4] American Society of Clinical Oncology, "Guide
to Lung Cancer," Alexandria, 2011.
D4-5
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
I. INTRODUCTION
D5-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
One or two of the three constants can be tuned more x = the weights and biases in the network
prominent than others. Constant that stands out that will
contribute to the effect on the response of the system. Jacobi matrix is a matrix of first derivatives of the
weights and the bias error in the network. Jacobi matrix
between input layer and hidden layer is a matrix that
contains the error derivative of the weights between
input layer and hidden layer along with the bias. While
the Jacobi matrix between the hidden layer and output
layer contains the error derivative of the weights
between the hidden layer and output layer along with the
bias.
Fig. 2. Block diagram PID controller
Jacobi matrix element between the input layer and
hidden layer
e. Neural Network (6)
Neural network or artificial neural networks (ANN) is Jacobi matrix element in bias hidden layer
a distributed information processing structure in the (7)
form of directed graph. The advantages of this neural Jacobi matrix element between the hidden layer and
network is a network can learn where there are two output layer
stages in the operation of the ANN. At this stage of (8)
learning to adjust to any provision of input connections Jacobi matrix element in bias output layer
to the network produces the desired output with the (9)
structure and parameters of the optimal ANN. There are
two stages of learning, namely supervised learning (with
III. SISTEM PLAN
supervision) and unsupervised learning (without
supervision). While the initial testing phase input in the At this stage the hardware design and make the
form of the unknown information is given as input the Levenberg Marquardt algorithm for neural network PID
network. Each cell will perform computation by tuning that aims to control the plant to fit the desired.
activation function connection with the influence of Here is a block diagram of control system Neural
weight gained during the learning process. Network with Levenberg Marquardt training method.
f. Levenberg Marquardt
Levenberg Marquardt algorithm can be performed NNOut
Output
using the second derivative approach without having to PLANT
calculate the hessian matrix. If the feed forward neural Input NNErr
D5-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
provisions on how much the value of and , but many be carried out continuously until the same error with the
studies using = 0.1 and = 10. error limit.
While the overall LevMar NN algorithm is the
b. Step 2 following can be summarized as follows:
The second step is performed by calculating the START
function forward (feed forward)
~ Each hidden node input signal summing weights as
Inisialisasi Jaringan
follows: Random bobot V dan W
Set dan
Where:
Zinh = signal input for hidden node h Hitung maju pada simpul
Xi = value input for node i Hidden dan Ouput
Vih = value weight between input node i and hidden Zh = f(v,in), Yo = f(w,zh)
node h
V0h = value bias at hidden nodeh
Hitung sse
~ Calculate the value of hidden nodes based on the
1
sigmoid activation function. = ( )2
2
=1
c. Step 3
The third step is done by calculating the value of END
error where the error function approximated by Sum Fig 4. Flowchart NN LevMar
Square Error (SSE). Where to compare the value of each
output (Yo) with a target (tk) with the following
equation IV. SIMULATION RESULT
In the Sub is the result of simulation by the method of
training neural networ LevMar NN to be compared with
simulations by using this NN BackPro where the
d. Step 4 comparison of the results can be seen which one is better
The fourth step is done by calculating the Jacobian for NN training.
matrix (J) which contains the first derivative of the
weights and bias error. A. NN simulation LevMar
e. Step 5
The fifth step is done by calculating the difference in
weight
f. Step 6
The sixth step is done by calculating a new weight
Having obtained the new weights are calculated as the
error back to the step 3. If the error is reduced, the new
do ( / ) and return to step 2 to step 7. If a new error is
not reduced then do (x) and return to step 5. This will
D5-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Fig 8. Graph of velocity with the period 2000 to NN LevMar Set Point = 1500 RPM
D5-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
V. ANALYSIS
From the simulation results performed breaking can be
done to analyze the results of these simulations which
are categorized into 3 pieces, namely the fixed speed,
the speed varies according to period, and administration
expenses
Fixed speed
On testing the speed of response is obtained that
LevMar NN NN LevMar takes about 1 second rise time
and began to stabilize an average of 20 seconds. Its
Figure 12 Graph of the speed of 1500 RPM for NN BackPro
steady state error of not more than 20 RPM. However
Overshoot produced quite large for large yag RPM
From the graph it was found that rise time of about 1 about 500 RPM
second. For the overshoot of about 600 RPM. Stabilized While the testing of the response rate was obtained
within 25 seconds and the steady state error of about 20 that BackPro NN NN BackPro takes about 1 second rise
RPM. time and began to stabilize an average of 30 seconds. Its
steady state error of not more than 50 RPM. However
Set Point altered by the period 2000 Overshoot produced quite large for a large RPM about
500 RPM
VI. CONCLUSION
Once the testing is done by performing a simulation on
visual basic 6.0 obtained the following conclusion:
~ Overshoot LevMar NN generated by nearly half from
a set point given
~ Time to reach set point with a range of 1 second with
the assumption that the provisions of the motor is used
in accordance with the motor simulation
~ NN LevMar achieve faster convergence which is
about 20 seconds while the NN BackPro takes about 30
Fig 14. Image speed with load of 50% for the NN BackPro seconds
~ The process of adaptation is performed for the NN
From the graph the response to load change 50% found LevMar tend to be slow compared to NN where NN
that the system is able to restore to its original set point BackPro BackPro able to respond to a change of pace
while the open loop system can not return to its original well with the relatively short time is about 60 seconds.
state. ~ The oscillations that occur in less than LevMar NN
NN BackPro for high RPM, while about 10 RPM to low
D5-5
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
D5-6
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
D6-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
This process is conducted by means of tempering including a monitor (display) small, integrated on the
machine, such as depicted in Figure 3. module and extension modules are either direct
conversion of the temperature sensor to the voltage or
the addition of discrete I/O, as well as direct monitoring
capabilities to facilitate the Human Machine Interface
(HMI). In general, Zelio also has analog and discrete I
/O. We used Zelio PLC with the type of SR3-B261BD. It
has 16 input ports, which are can be set as 16 discrete
input or 10 discrete input with 6 analog inputs. This PLC
has 10 ports of discrete output. Figure 4 shown the Zelio
SR3-B261BD.
D6-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
A. General Description
Figure 11 shows the overall process controlled by our
proposed automation system. A block diagram of our
proposed process control schema is depicted in Figure
12.
Figure 6. Function Block Diagram
D6-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Starting from the initial heating, cylindrical roaster temperature of the coffee bean is already under
was switched on. This is done to avoid expansion 30oC (read by temperature sensor of thermostat
of part of the cylinder. When the temperature of the type), the beans will be issued through the holes
hot cylinder has reached the desired temperature in the tempering and go into the container of
(150o-170oC), the coffee beans are automatically roasted coffee beans.
inserted into the cylinder by using a vacuum Program of proposed control strategy in FBD is given
suction machine with a capacity of 36.85 kg of in Figure 13. The program is then loaded into the PLCs
coffee beans. However, in practice, the engine memory to implement the control strategy.
vacuum is not constantly suck all the coffee beans,
but with PLC control, a vacuum is sucking an
average of 10 Kg in 45 seconds, then release the
suction that eventually entered into a cylindrical
roaster in 20 seconds. The process was repeated 5
times so it can suck an estimated 50 kg in 5 minutes
and at the same time put the coffee beans into rotary
cylinder. The cylinder is then rotated constantly by
the motor, to make roasting process evenly occurs
on the coffee beans. A temperature sensor is placed
to measure real time in-cylinder temperature and
sending this data to the PLC.
The heating process of the cylinder is continued to
reach the temperature of 180oC to 220oC ( 10% of
the 200o C) according to the cylinder heat rise and
their normal temperature in roasting. The
difference in reference temperature is caused by the
difference of water content of wood used in
combustion and water content of roasting coffee.
The smaller the water content of coffee and wood,
the heat will rise faster than the high water content. Figure 13. Function Block Diagram of The Proposed Control Strategy
To supply oxygen into the heating system we used a
DC motor as fan / blower. Motor speed can be set V. EXPERIMENTAL RESULTS
according to the analog signal sent to the fan. If the In automatic mode control, duration of roasting
combustion temperature exceeds 200oC, then the process varies according to the water content of fuel
fan will be turned off. When the temperature wood, water content of roasting coffee and weight of
decreases from the desired value (from 170oC to coffee. The higher the water content of wood the longer
130oC), then the fan will rotate with a maximum roasting duration. Table 1 summarized the results of
speed. When the fan is spinning at a maximum, but three experiments.
still cannot raise the combustion temperature, and Table 1. Experimental Results
even drops below 115 oC, then the PLC will send a
Coffee Water Water Roating
signal to the siren failure as a sign that there has Trial
Roaster
Weight Content of Content of Duration
been a fuel shortage, this is followed the turning Type
(Kg) Coffee (%) Wood (%) (minutes)
off of a fan if the temperature is below the 1 10 10 12 15,33 25
minimum fan system. Once the wood fuel is 2 50 30 11,8 12,69 34
3 50 30 12 13,82 35
inserted, the temperature will rise to the desired
point.
Table 1 shows satisfactory results since it still less
After the roasting process goes according to plan,
than the maximum roasting time (40 minutes). The most
and the set temperature and time is reached, the
important point actually is not in the reducing of roasting
cylinder remain rotates while the door opening.
time, but in the ease work of the operator. Using the
Roasted coffee will go out to the tempering
proposed control system, in automatic mode, all
machine. Before the doors of roaster open, the
processes can be controlled easily from control panel .
temperer mixer and cooling blower are activated
first.
VI. CONCLUSSIONS
1. The Tempering Process. This process is the final
step. Within 15 seconds before coffee bean set to In this study, we developed an efficient centered
tempering, a mixer has been activated, followed automation systems for overall processes of producing
by the fall of all coffee bean in the temperer. This roasted coffee. This system can be operated in automatic
is done to prevent the buildup of the hot coffee mode as well as in manual mode. A Zelio PLC of type
beans into one side of the temperer. When the SR3-B261BD (as the main PLC) and SR3-XT43BD (as
an extended PLC) were employed as process controller.
D6-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
PLC programming is done using Function Block M. Aziz Muslim received Bachelor Degree and
Master Degree from Electrical Engineering
Diagram (FBD) model. Pt-100 and Thermostat are used Department of Institut Teknologi Sepuluh
to sense temperature at roaster cylinder and temperer. Nopember, Surabaya, Indonesia, in 1998 and
Experiments showing satisfactory results compared to 2001, respectively. In 2008 he received Ph.D
degree from Kyushu Institute of Technology,
previous system. In the term of roasting time, maximum Japan. Since 2000 he is with Electrical Engineering
roasting time is reduced from 40 minutes 35 minutes. Department, Brawijaya University. His current
research interest are control systems and computational intelligence
And using the proposed control system, in automatic including its wide applications in electronics, power systems,
mode, all processes can be controlled easily from control telecommunications, control systems and informatics.
panel . Goegoes Dwi Nusantoro received Bachelor
Degree from Brawijaya University and Master
ACKNOWLEDGMENT Degree from Gajah Mada University, Indonesia,
in 1999 and 2005, respectively. Since 2006 he is
Authors thank to Pusat Penelitian Kopi dan Kakao with Electrical Engineering Department,
Indonesia at Jember, for giving us facility and Brawijaya University. His current research
opportunity to conduct this research. interest are electronics control systems (embedded
control system) and robotics.
D6-5
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
D7-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
n
SSR = ( y
i =1
i yi ) 2 (5)
c ij [ SSE ]
( n p 1)
cij=(XT X)-1 i,j=0,1,2,p (10)
Fig. 2. Single cylinder diesel engine schematic view
Where, F is the statistic value, tj is relationship
parameter R and t statistic value. The correlation The experimental device of this research is Yanmar
coefficient R indicated that the matching level of the TF70 V-E diesel engine with 4 cycle horizontal type
calculation datum by the regression equation and the water-cooling and equipped with a turbocharger.
original datum, the result is the better when R is more
close to 1. Statistic values indicate the significance of the Table I Specification of a diesel engine
Engine type 4-cycle, 1cylinder, DI
multiple linear regression equation, whose values obey F 78 mm 80 mm
Bore Stroke
distribution. On the condition of less effective regression Top clearance 0.98 mm
analysis result, the statistics values of t correspond to non Con-rod length 115mm
significant variables should be rejected in turn according Compression ratio 21.4
to the value of tj. Then the regression analysis will be Cylinder capacity 0.382L
carried out again with the remaining significant factors. Maximum output 5.5/2600 KW/min-1
Full-length 640 mm
Finally, the prediction model of output is identified.
Full-height 474 mm
Full-width 330.5 mm
III. EXPERIMENT
In this research, we used the single cylinder diesel
engine experimental device to get the experiment data.
We used experiment data to build the model.
A. Experimental Device
Combustion model could be applied to calculating
many engine indices, BSFC, gas emission, pressure, etc.
In the current research, multiple linear regression model
is constructed, and only CO and HC are taken as
optimization objectives. These objectives are formulated
from groups of experiment data. The experiments with
multiple injections are performed on a diesel engine
experimental device (in Fig. 1) included the diesel Fig. 3. Multiple injection pattern
engine schematic view (in Fig. 2) whose specifications
are listed in Table I. The multiple injections include two B. Experimental Condition and Result
pilot injections and main injection, as shown in Fig.3. In this research, the diesel engine is set with three
stage injection pilot1 injection, pilot2 injection and main
injection. Then, the rotation speed 1500 rpm, EGR rate
0% and the engine temperature is set about 90 degrees.
The engine control parameters are set as Table II and the
D7-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
engine optimization objectives are listed as table III. Then, the regression equation for CO and HC can be
Table IV shows the data was obtained from Diesel expressed as:
engine. In Table IV, x1, x2, x3 and x4 as control y1= 24.6057+0.0060x10.0041x20.0030x3+0.0174x4
parameters (input). Then, y1 and y2 represents the y2= 60.35840.1760x10.1421x2+0.0543x3+0.1951x4
characteristic value of the optimization objective The predicted results of CO and HC are evaluated
(output). using correlation coefficient. The predicted result,
Table II Diesel engine control parameters experiment data and absolute error of CO and HC is
Control
Meaning Unit
Variation showed in Table VI, Fig. 4 and Fig. 5
Parameter Range
Table VI The predicted result, actual data, absolute error of CO and HC
Pilot 1 injection deg. -70,-50,-40,-3
x1 CO (%) HC (ppm)
timing ATDC 0
No
Pilot 2 injection deg. actual predict error actual predict error
x2 -40,-20,-15
timing ATDC
1 1.64 1.599228 2.49% 257 254.5442 0.96%
Main injection deg.
x3 -8~0
timing ATDC 2 1.6 1.631564 1.97% 253 254.4677 0.58%
x4 Engine Speed rpm 1500
3 1.62 1.613605 0.39% 253 254.7935 0.71%
Table III Optimization objectives 4 1.72 1.705824 0.82% 253 250.7749 0.88%
Optimization Meanin Uni
Objective g t 5 1.75 1.75494 0.28% 252 251.4144 0.23%
y1 CO % 6 1.75 1.786686 2.10% 252 251.8588 0.06%
pp
y2 HC 7 1.71 1.71181 0.11% 250 250.6663 0.27%
m
8 1.73 1.708817 1.22% 251 250.7206 0.11%
Table IV Data from input and output
engine 9 1.72 1.723193 0.19% 249 250.97 0.79%
piot1(T) piot2(T) main(T) CO HC
speed 10 1.73 1.68546 2.57% 252 250.6342 0.54%
deg deg deg r.p.m. % ppm
x1 x2 x3 x4 y1 y2 11 1.64 1.665097 1.53% 249 250.4934 0.60%
-70 -40 -3 1523 1.64 257 12 1.65 1.686789 2.23% 248 246.1185 0.76%
-70 -40 -8 1524 1.6 253
13 1.65 1.683796 2.05% 246 246.1728 0.07%
-70 -40 -2 1524 1.62 253
-50 -40 -4 1522 1.72 253 14 1.69 1.698172 0.48% 247 246.4222 0.23%
-50 -40 -3 1525 1.75 252 15 1.7 1.720926 1.23% 245 244.3264 0.27%
-50 -40 -2 1527 1.75 252
-50 -40 -6 1522 1.71 250 16 1.77 1.785008 0.85% 245 244.6943 0.12%
-50 -40 -5 1522 1.73 251 17 1.71 1.729905 1.16% 244 244.1635 0.07%
-50 -40 -4 1523 1.72 249
18 1.76 1.744282 0.89% 243 244.4128 0.58%
-50 -40 -3 1521 1.73 252
-50 -40 -2 1520 1.64 249 19 1.82 1.781749 2.10% 245 244.1788 0.34%
-40 -20 -5 1522 1.65 248 20 1.78 1.778756 0.07% 244 244.2331 0.10%
-40 -20 -4 1522 1.65 246
-40 -20 -3 1523 1.69 247 21 1.73 1.723653 0.37% 244 243.7023 0.12%
-30 -20 -2 1521 1.7 245 22 1.68 1.72066 2.42% 243 243.7566 0.31%
-30 -20 -6 1524 1.77 245
23 1.75 1.71168 2.19% 243 243.9195 0.38%
-30 -20 -5 1521 1.71 244
-30 -20 -4 1522 1.76 243 24 1.8 1.767373 1.81% 244 243.9295 0.03%
-30 -15 -6 1525 1.82 245
25 1.78 1.74701 1.85% 243 243.7887 0.32%
-30 -15 -5 1525 1.78 244
-30 -15 -4 1522 1.73 244 26 1.73 1.744016 0.81% 244 243.843 0.06%
-30 -15 -3 1522 1.68 243 average absolute average absolute
1.32% 0.37%
-30 -15 0 1522 1.75 243 error error
-30 -15 -7 1524 1.8 244
-30 -15 -6 1523 1.78 243
-30 -15 -5 1523 1.73 244
IV. RESULT
Based on the multiple linear regression equation, the
regression coefficients for each objective are listed as
Table V:
Table V regression coefficients
y1-CO(%) y2-HC(ppm)
j o o
-0.176
1 0.0060
0
-0.004 -24.605 -0.142 -60.358
2
1 7 1 4
-0.003 Fig. 4 The prediction and experiment of CO
3 0.0543
0
4 0.0174 0.1951
D7-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
V. CONCLUSIONS
In order to control the large number of control parameter
appropriately in consideration of CO and HC as the
engine output objectives, the model construction which
Fig. 5 The prediction and experiment of HC
reproduces the characteristic value of CO and HC from
control parameter is needed. In this research, the
Differences of the predicted values with the multiple linear regressions were applied to construct the
experiment data are shown in Fig. 6 and Fig. 7. engine model. The accuracy of predictions made using
multiple linear regression models depends on how well
the regression function fits the data, there should be
regular checks to see how well a regression function fits
a given data set. This can be done through regular
updates or monitoring to ensure that the error values are
always below a pre-specified error threshold. A pre
analysis of the control parameter is necessary for
successful MLR predicting as the result of the analysis
showed. In this paper, we have reported our predicting
model of CO and HC in multiple injection diesel engines
by multiple linear regression. It can be regarded that the
multiple linear regression method can effectively
Fig. 6 The difference of predicted value and experiment data of CO estimate the objectives.
ACKNOWLEDGMENT
The author would like to thanks INPEX Scholarship
Foundation for financing the study in Waseda
University, Japan.
REFERENCES
[1] P.K. Karra, S.C. Kong, Diesel Engine Emissions Reduction
Using Particle Swarm Optimization, Combustion Science and
Technology, 182:7,Taylor and Francis, pp. 879903.
[2] J. Zhang, Y. Li, J. Cao, Sensor situation based on the multiple
linear regression forecast, 2011 IEEE International Conference
Fig. 7 The difference of predicted value and experiment data of HC on Computer Science and Automation Engineering (CSAE),
pp.47-50, China, 10-12 June 2011.
From table VI, Fig. 6 and Fig. 7, the maximum absolute [3] N. Amral, C.S.Ozveren, D. King, Short term load forecasting
error of CO is 2.57%, the minimum absolute error of CO using Multiple Linear Regression, 42nd International
is 0.07% and the average absolute error of CO is 1.32%. Universities Power Engineering Conference, pp.1192 1198,
The maximum absolute error of HC is 0.96%, the Brighton, 4-6 Sept. 2007.
[4]
D7-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Abstract - The Unified Theory of Acceptance and Use services such as mobile payment. Moreover,
of Technology (UTAUT) model with two additional development towards Cash Less Society also supported
variables (perceived of Risk and perceived of trust) is used the growth of this mobile service application. Yet, the
to assess the users perception towards the adoption of adoption of mobile payment has not been as good as
mobile payment application. The results of this research
expected. Users perceptions towards mobile payment
show that from the correlation analysis performance
expectancy, effort expectancy, social influence and still need to be observed further to tackle this issues.
perceived of trust have positive relationship towards The greatest problems in the adoption of technology
behavioural intention to use mobile payment application. such as mobile payment is lower respond from
Whereas, perceived of risk hasnegative relationship on users(Schierz et al., 2009). Therefore, this research
behavioural intention. From the results, it can be attempts to describe users perceptions towards the
concluded that although Indonesian consumers have adoption of mobile payment.
willingness to use mobile payment application, they are
still considering trust and risk as main determinant to use This research propose Unified Theory of Acceptance
mobile payment application. and Use of Technology (UTAUT) model combine with
trust and risk to investigate users behavioural intention
Keywords: mobile payment, UTAUT, risk and trust
towards technology adoption. So far, there is little
research regarding the acceptance of mobile payment in
Indonesia. Thus, this research will provide a
I. INTRODUCTION comprehensive study about factors that affecting
consumer to adopt the technology. The findings of this
Over the past several years the number of mobile research would be beneficial for both academics and
telephone users has increased significantly. Driven by business practitioners (vendors). From academic point
the sharply increasing of mobile usage and the of view, this research offers a framework to identify the
evolvement of its functionality (Viehland and Leong, determinants of users perception on mobile payment
2007), mobile telephone has become common devices application. On the other hand, from business
in our daily activities(Hwang et al., 2007). More and perspective, this research would give guidance in order
more people rely on mobile devices to conduct their improve the mobile payment usage in Indonesia.
business activities. Furthermore, this condition has
underpinned the shift of traditional commerce to II. LTERATURE REVIEW
electronic commerce, which utilizes technology to
support the performances. From several forms of A. Mobile Commerce
electronic commerce, mobile payment has turned into
one of the promising business model in the near future. The widespread usage of mobile telephone and the
Mobile payment enables consumers to performs their emerging of wireless technology have shifted the
activities such as paying goods and services using business activities in general. Many businesses have
mobile devices (Kim et al., 2009a). Mobility and wide utilized mobile telephone as an alternative instrument
reach characteristics which are addressed with mobile for delivering services (Medhi et al., 2009), such as
telephone has supported the penetration of mobile advertising, commerce and financial.
telephone adoption for commerce activities.It can be Many types of mobile financial application emerge
seen from several countries that has been utilized and are available in several countries such as MobilPay
mobile payment as alternative payment methods for (Germany), Paybox (UK), PayDirect (US) and many
commerce transactions, such as MobilPay (Germany), more (Norman, 2002).In line with the widespread of
Paybox (United Kingdom), PayDirect (United States) mobile commerce, mobile payment has appeared as an
and many more (Norman, 2002). Concerning interesting subject and become promising mobile
Indonesia, within five years time, the number of mobile services in thefuture(Karnouskos and Vilmos, 2004).
telephone subscriber has increased dramatically from 30 Several industries have been influenced by the emerging
million in 2004 to 180 million users by the end of 2009 of mobile payment applications, including financial
(Asia, 2010). This number shows that Indonesia has big services, retail, telecommunication, information
potential opportunity to get exploit in mobile telephone services, entertainment, and technology (Smith, 2006).
E1-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
It seems that business practitioners have assured that III. RESEARCH FRAMEWORK
mobile payment may generate competitive advantages
for their business.
Theoretically, mobile payment can be defined as a Performa
payment for services or goods which is conducted Perceived
through mobile devices instead of paying with cash
Effort Behaviour
(Karnouskos, 2004). In this business model, mobile
devices are employed for initiation, authorization and Perceived
completing the payment process.
Social
E1-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
influence users perception towards mobile payment respondents, 66 peoples (66%) are male and 34 of them
application. are female which represents 34% of total respondents.
Performance expectancy is used to measures how In addition, a reliability analysis was used for
much people perceive that using technology is useful to measured the questionnaire items. Validity is defined as
improve job performance. This determinant has three the extent to which the indicator really measures the
items used to investigate users perception regarding the model (Bryman and Bell, 2007). Because this research
technology adoption. Effort expectancy is defined as has adopted concept from previous research and
how ease of use is associated with the users experience literature reviews as constructive model (Kim et al.,
and their acceptance to adopt the system. Four items 2009b; Wu and Wang, 2005; Venkatesh et al., 2003),
will be performed in this determinant to examine users therefore, it can be considered as valid concept.As
perception regarding the technology adoption. Social expected the results of the measurement is shows good
influence refers to how people feel that others advices degree of reliability since each variables obtain
and recommendation to employ a certain systems is coefficient alpha greater than 0.5(Hair et al., 1995;
important. Three questions have been proposed in the Cronbach, 1951).
online survey to examine users perception regarding Table 1. Cronbachs Alpha
the technology adoption. These questions have been
developed based on items used in estimating UTAUT Constructs/ Variables Cronbachs N of
model. Alpha items
Performance 0.870 3
Expectancy (PE)
Effort Expectancy 0.905 4
(EF)
Social Influence (SI) 0.611 3
Perceived of Risk 0.859 4
(PoR)
Perceived of Trust 0.847 4
(PoT)
Behavioural Intention 0.855 3
(BI)
table below shows that all variables are significantly combination of two additional variables (perceived of
correlated with Behavioural Intention variables, which risk and perceived of trust) will provide in depth insight
is shown by the significance value (p<0.05 and p<0.01). regarding the successful and the failure of mobile
From all five variables in the model research, four payment adoption.
variables have positive correlation with behavioural
intention (i.e. PE= 0.453, EF= 0.259, SI= 0.350, and REFERENCES
PoT= 0.521). Whereas, perceived of risk has negative [1] Asia, D. M. A. (2010) Indonesia:Mobile Penetration
correlation (PoR= -0.274). Available at:
http://comm215.wetpaint.com/page/Indonesia%3A+Mobil
Table 2. Correlations e+Penetration (Accessed: 29 June 2010).
[2] Bank.Indonesia. (2010) 'Jumlah Uang Elektronik ',
[Online]. Available at:
BI PE EF SI PoR PoT http://www.bi.go.id/web/id/Statistik/Statistik+Sistem+Pem
Pearson 1 .453 .259 .350 -.274 .521 bayaran/Uang+Elektronik/JmlUang.htm (Accessed: 30
Correlation July 2010).
(BI) [3] Bryman, A. and Bell, E. (2007) Business research
Nonetheless, the correlation table indicates weak methods. Oxford Oxford Univ. Press.
[4] Christer, C. (2006) Proceedings of the 39th Annual Hawaii
association towards behavioural intention because it International Conference on System Sciences (HICSS'06)
shows value <0.8. Relation between perceived of trust Track 6.
variable with behavioural intention is more likely to be [5] Cronbach, L. (1951) 'Coefficient alpha and the internal
the highest value (0.521). In conclusion it can be structure of tests', Psychometrika, 16, (3), pp. 297-334.
[6] Euromonitor (2010) Mobile Telephone Subscribtions
validated that each variables (partially) have significant Available at:
correlation with behavioural intention, but, it is http://portal.euromonitor.com/Portal/Statistics.aspx
considered as weak relationship. (Accessed: 15 March 2010).
[7] Hair, J. F., Anderson, R. E., Tatham, R. L. and Black, W.
VI. CONCLUSION C. (1995) Multivariate Data Analysis
[8] Hwang, R. J., Shiau, S. H. and Jan, D. F. (2007) 'A new
mobile payment scheme for roaming services', Electronic
As mention previously, this study attempts to describe Commerce Research and Applications, 6, (2), pp. 184-191.
user perception towards mobile payment in terms of [9] Karnouskos, S. (2004) 'Mobile payment: A journey
applying UTAUT model combined with risk and trust. through existing procedures and standardization initiatives',
IEEE Communications Surveys and Tutorials, 6, (4), pp.
1) The results show positive relationship for 44-66.
performance expectancy, effort expectancy, social [10] Kim, C., Mirusmonov, M. and Lee, I. (2009a) 'An
influence, and perceived of trust towards empirical examination of factors influencing the intention
behavioural intention to use mobile payment. It to use mobile payment', Computers in Human Behavior,
26, (3), pp. 310-322.
indicates that if performance expectancy value [11] Kim, C., Tao, W., Shin, N. and Kim, K. S. (2009b) 'An
increase, the value for behavioural intention empirical study of customers' perceptions of security and
variable also increases. This condition also applies trust in e-payment systems', Electronic Commerce
to others variable. Research and Applications, 9, (1), pp. 84-95.
[12] Medhi, I., Gautama, S. N. N. and Toyama, K. (2009) 'A
2) Perceived of risk has negative relationship on Comparison of Mobile Money-Transfer UIs for Non-
behavioural intention. It shows that if user Literate and Semi-Literate Users', in Greenberg, S.,
interpretation concerning the risk towards mobile Hudson, S. E., Hinkley, K., RingelMorris, M. and Olsen,
payment application is low, user intention to use D. R.(eds) Chi2009: Proceedings of the 27th Annual Chi
Conference on Human Factors in Computing Systems,
mobile payment application is also low. Vols 1-4. pp. 1741-1750.
3) Fivevariables (performance expectancy, effort [13] Norman, M. S. (2002) M-Commerce:
expectancy, social influence, perceived of trust and Technologies,Services,and Business Models. John Wiley
perceived of risk) have significant relationship on &Sons, Inc.
[14] Oakshott, L. (2001) Essential Quantitative Methods for
the users acceptance of mobile payment Business Management and Finance.
application, thus, it can be used to predict the result [15] Park, J. K., Yang, S. J. and Lehto, X. R. (2007) 'Adoption
of behavioural intention variable of mobile technologies for Chinese consumers', Journal of
From the descriptive analysis, it appears that Electronic Commerce Research, 8, (3), pp. 196-206.
[16] Schierz, P. G., Schilke, O. and Wirtz, B. W. (2009)
Indonesia consumers are having willingness to use 'Understanding consumer acceptance of mobile payment
mobile payment as alternative payment for their services: An empirical analysis', Electronic Commerce
transactions. Yet, they still consider trust and risk as Research and Applications.
main determinant on their perception towards mobile [17] Smith, A. D. (2006) 'Exploring m-commerce in terms of
viability, growth and challenges', International Journal of
payment. From managerial implication, the research Mobile Communications, 4, (6), pp. 682-703.
confirms the important of these five variables as the [18] Venkatesh, V., Morris, M. G., Davis, G. B. and Davis, F.
main determinants in order to increase the adoption of D. (2003) 'User acceptance of information technology:
mobile payment application. Therefore, mobile payment Toward a unified view', MIS Quarterly: Management
Information Systems, 27, (3), pp. 425-478.
providers have to concerning their application to meet [19] Viehland, D. and Leong, R. (2007) 'Acceptance and Use of
the user needs and wants, particularly on five variables Mobile Payments', ACIS 2007 Proceedings. pp. Wu, J.-H.
above. This research has validated the use of UTAUT and Wang, S.-C. (2005) 'What drives mobile commerce?:
model to investigate users acceptance regarding the An empirical evaluation of the revised technology
acceptance model', Information & Management, 42, (5),
adoption of technology. Moreover, with the pp. 719-729
E1-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
a result in [1].
Abstract This paper deals with attitude consensus of
multiple spacecraft in a team where each spacecraft II. KINEMATIC AND DYNAMIC OF RIGID BODY
applies three-axis reaction wheels. Two control laws for SPACECRAFT
two different cases are presented. The attitude consensus
A. Vector Notation
Consider a vector as follows: = . Here, is
of multiple spacecraft is achieved under a connected
information exchange topology. Simulations run to verify
the effectiveness of the control laws in reaching the
1 0 0
controllers for single spacecraft case and attitude
= = = 0 1 0! = "
synchronization case under directed
0 0 1
information-exchange graph. C. G. Mayhew et al [5] properties: ,
and det = 1.
proposed a quaternion-based hybrid feedback controller
for attitude synchronization under connected and acyclic
information-exchange graph. All of these researches In a rigid body general case, a rotation matrix for
used dynamic model of spacecraft rigid body with transforming a vector expressed in the inertial frame to
external control torque. be expressed in the body frame, & , is defined as
In practice, an actuator that may generate above type follows:
= &
of control torque is, for example, thruster. Nevertheless,
there is another actuator that also usually taken into & (1)
Let & and are rotating w.r.t. (with respect to) each
account on the research field about spacecraft control
E2-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
- = .)*&0- )
*&/1
)
*&/2
- 3 =4
5 & & =
*+ / D. Kinematic Equation
Consider ) -
)
*, * +C
-
S78&& 9& ) - = & S78&& 9
*, *, *,
*+ ; *<
declared in (3); the time derivative
(2)
*<
of rotation matrix & can be represented in a unit
0 = & )?78&& 9 +)
* +C
--)
* +C
- = where H&* = H * H& . Note that map is a
& *< *<
?78&& 9 &
group homomorphism, [10]. For convenience in
& = H& .
Rotation matrix & represents the attitude history of
& w.r.t. . Practically, it may be obtainedeven
Consider &* = & * * = & &* , where
sun sensor and geomagnetic sensor. Meanwhile, 8&& = *& . Take time derivative of * w.r.t. * , then
indirectlyfrom reference sensor(s), e.g. star tracker,
&*
0 = & )?78&&* 9 +) - -.It implies ) - =
* +[ * +[
&* *< * *< *
may be measured by means of rate gyro sensor or
4
5* = & 78&&
8* 9 8&&* = 8&&
8* , then the
Noting
C. Attitude Representation Using Unit Quaternion & &
From the Eulers theorem, an angular displacement of unit quaternion representation of attitude kinematic for
&* is as follows:
a frame w.r.t. another frame can be obtained by a single
) - =X ) - Z
*H+[ *Y+[ *J+[ ;
respectively, whose properties: F = F and F F = 1. *< * *< *< *
The Euler parameter composed of F and G is defined
(12)
as follows: where
I I J J
The following assumptions are used in deriving the
H H =. 3
I J +I J +J J
equation of the rigid spacecraft dynamic with 3-axis
(5)
reaction wheels:
H H = H H = =1 0 0 0> = U
coincidence with ] -axis of & , ^ -axis of &
(7) The rotation axis of each reaction wheels is
Note that the rotation matrix can be related to H and z-axis of & , respectively.
through Rodriques formula whose the map : K
SO 3 , where H = H .
E2-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
` )) - +) --=
*8++C *8+aC
(15)
*< *<
by a directed graph as follows:
in & . In practice, a reaction wheel has a limited angular } } } is the edges set; and } = x z
reaction wheels absolute angular velocity decomposed
8&
lm & n q from spacecraft u ; x is the parent node and z is
` ` k ( p child node; and x z = 1 . Meanwhile, the edge
.?78& 97`8& + ` 8b 93 = .
& & &
3
` ` k 8&b p denoted by 7 x z 9 in the undirected graph
km np
j ( o corresponds to edges 7 x z 9 and 7 z x 9 ; and
x z = z x = 1 . If there is no edge between
) - spacecraft u and spacecraft v, then x z = z x = 0.
*8++C
` `
.?78& 97`8&& +` 8&b 93 = r
&
. 3 s
*<
` `
)
*8+aC
-
A balanced graph is a graph whose }z x z =
(16)
*<
z z x , where 1 i, j n . Every undirected
}
Using a fact 2.17.3 in [7], then (17) is equivalent to graph has symmetric } . Thus every undirected graph
(15) and (16). is a balanced graph. An undirected graph is connected if
there is an undirected path between every pair of
`` ) - = ?78&& 9`8&& ?78&& 9` 8&b
*8++C different spacecraft.
_ *<
c(17)
`` ) - = ?78&& 9`8&& + ?78&& 9` 8&b + ``
*8+aC
*<
IV. PROBLEM STATEMENT
As noted above, there is a redundancy in representing
III. SPACECRAFT FORMATION REPRESENTATION the same physical orientation using unit quaternion. The
t -spacecrafts would be denoted by subscript (and attitude in 3 would be represented using two values
Index of a spacecraft in a formation composed by
sometimes both sub- and superscript) u and its local of unit quaternion in K .
both sub- and superscript) v , where 1 u, v t. achieved, the agreement value on 3 is the
neighbor would be denoted by subscript (and sometimes Nevertheless, when the attitude consensus is
) 7H& ( = HF 9, 8& ( = 8F - as (
*8+
+ y & x
7` x ` 9m n = ? )8& - ` x 8&
& x & x
y C
& x x x
_
x x x
c
*<
(22)
E2-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
7H& ( = U9,_
Remark IV-2 Regarding the argument of globally
7 & x & z x & z asymptotically stable for quaternion-based attitude
_8&
control in [8], the attitude consensus designed to follow
( 8& ( = - as (
x & z
& x z (23) Definition IV-1 is necessary for globally asymptotically
stable guarantee.
7H& x & z ( = U,_
V. MAIN RESULT
_8& x
( 8&
& z
( = - as(
& x z
Lemma V-1 If for 1 u, v t and u v , x z =
First, it is worth to consider some following lemmas.
(24)
z x , then
where = =0 0 0> , 1 u, v t and u v.
}x }z u v )8&& x
x 8&
& z
z - & x & z
= }x }
z x H& U ,
team satisfy the following conditions:
H& x & z 0 U > 0
z x & z
u H& x & z ( U = 0
<
then
} }
8&& 0 8& 0 >0 = 8&& x
u &
x & z
x z x v x & z
u 8&& ( 8& ( = 0,
x & z x z
< x z
= 2 }x } _
z x 7I& 19
*Y+ y + {
U, . In addition, U, is an isolated point,
where
z x & z *<
where if the spacecrafts in a team is there at ( = 0, then
_+& x & z
*J+ y + {
n
*<
consensus is achieved since ( = 0. Note that this first & z
they will agree to stay there for all the time, i.e. attitude
} }
= x 8&& x & z & x & z
x
case corresponds to the leaderless consensus case.
z
x z
Now, consider the second case corresponding to the
as a root of the corresponding graph |} that spacecraft v that measured at its own frame, i.e. & z ,
spacecrafta virtual leader spacecrafthaving index
E2-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
= k }
z x & ( + 8 x 8& ( = }z & ( + 8 x 8& (
& x & x
x z x & z x (25) x x z x & z x (26)
where positive scalar , symmetric positive where positive scalar x z , symmetric positive
definite matrix 8 x > 0 , 7` x ` x 9 definite matrix 8 x > 0, 7` x ` x 9
> 0 and x z is the element of a corresponding > 0 and x z is the element of a corresponding
adjacency matrix of |} . If H& * K is a constant adjacency matrix of |} . Therefore, consensus is
and 8&& ** = =0 0 0> = then consensus is achieved with the agreement value 7 & x HF , 8F 9
achieved with the agreement value 7 & x HF , 8F 9 3 , where:
3 , where: HF = H& x ( = H& z ( ; and
HF = H& * = H& x ( = H& z ( ; 8F = 8&
& x
( = 8&
& z
( = 31
x z
and
8F = 8 = 8& ( = 8& (
& x & z Proof:
x z
= 31
The proof may be obtained through the same way as the
proof of Theorem V-3 above.
Proof:
Consider a corresponding Lyapunov candidate function VI. NUMERICAL EXAMPLES
= + tu=1 8 uu 7` x ` 98
as follows:
1 u
Simulations of some cases are presented here to verify
2 x u . Taking time the main results. All simulations are run use following
= + 8 u 7` x
u
` 9 1 0.1 0.1
x
x
(
u=1 ` = 0.1 0.1 0.1!
0.1 0.1 0.9
0.8 0.1 0.2
Considering Lemma V-2, (18) and (25), one has
_?78& x
9` x 8&b xx 8 x 8&& x
9 0.3 0.4 1.2
` = 0.01"
& x x
}
x ` = 0.015"
= 8&& 8 x 8&& x
0
` = 0.02"
x x
x
0 1 0 0
Noting that = 0 implies 8&& x
= 1 0 1 1
= x z =r s
x
(18) and (25), & x & z = , if 8&& xx = 0 1 0 0
. Considering
Clearly, & x & z = implies H& x & z = U . 0 0 0 0
.
= 0.05
Note that U, is an isolated point, then the 8 = ` 1 ; 8 = ` 2 ; 8 =`3
0
implies Therefore, -1
regarding LaSalles theorem, they have globally 0 100 200 300 400 500
Time(s)
600 700 800 900 1000
0
1
-1
0 100 200 300 400 500 600 700 800 900 1000
Remark V-4 The addition spacecraft having index Time(s)
in Theorem V-3 has the same property as virtual 1
(i)
0
leader in [1], i.e. transmits (and not receives) the
2
-1
desired agreement value. 0 100 200 300 400 500 600 700 800 900 1000
Time(s)
0
3
i=2
spacecraft in a team is as follows: Fig 1. Attitude of spacecrafts in a team in according to Theorem V-3
E2-5
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
1
ACKNOWLEDGMENT
(i)
0.5
0
0 50 100 150 200 250 300
The authors would like to gratefully acknowledge
Time(s) suggestions and comments of one of our draft paper on
0.5
the same subject from Prof. Wei Ren and his former
(i)
0
1
-0.5
student. This research is partly supported by Beasiswa
0 50 100 150
Time(s)
200 250 300
Unggulan, BPKLN, Kemendikbud, Indonesia.
1
REFERENCES
(i)
0.5
2
0
0 50 100 150 200 250 300 [1] W. Ren, Distributed Attitude Alignment in Spacecraft
Time(s) Formation Flying, In. J. Adapt. Control Signal Process, 2007,
1 21, pp. 95-113.
i=1
(i)
0
3
E2-6
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
updating frequencies.
AbstractAbstract based-classification of research Several earlier works have extensively studied the
publication topics, in the past has been very much done in naive Bayes text classification. They used English text
the manual way. Sometime, it could be done difficultly by to be classified. NBC also could be implemented to
operator who dont have any good knowledge on all of
research topics including its coverage. In this research, we
classifiy text using Bahasa Indonesia, for example:
develop an automatic Indonesian publication document Indonesian News Document Classification (Wibisono,
classifier using Nave Bayes Classifier (NBC). The first 2005).
step of classification process is preprocessing (including In this research, we implement NBC to classify
stemming process, removing stopword and feature research publication abstract document based on its
selection). That pre-processing processes will be applied topic11 .This application was developed as web based
both in training process of training document collections
application, that we could access it online using internet
or testing process using testing documents. Training result
will be matched with testing document. The experiment connection. We also combine NBC with Chi Square
result shows that NBC could be used to classify abstract stochastic test as feature selection method.
document based on its topics effectively. NBC also has a
simple algorithm, so it makes simple to be II. UNITS
implemented.Feature selection by using Chi Square also
could improve the accuracy of this classification system, Some equations used in this research, are shown in
even in this system evaluation the effect of Chi Square is this section.
not too significant because of the limited dataset used in
A. Chi Square Test (2)
training or learning process.
Chi-square testing (2) is a well-known discrete data
Index TermsText Classification, Nave Bayes, Chi hypothesis testing method from statistics,
Square, Feature Selection whichevaluates the correlation between two variables
and determines whether they are independent or
correlated [2]. The test for independence, when applied
I. INTRODUCTION to a population of subjects, determines whether they are
positively correlated or not. 2 value for each term t in a
N AIVE Bayes Classifier has been one of the core
frameworks in the information retrieval research
for many years. Recently Nave Bayes is emerged as
category c can be defined by equation (1) [1].
E3-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
E3-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Table 1. Text classification illustration [1] used Chi Square Test as feature selection method. This
Category Training Documents Testing Documents method was briefly explained at Units section.
d1 ... dj dj+1 ... dn
C. Learning and Classifying
C1 C11 C1j C1(j+1) C1n
Because we used Nave Bayes Classifier (NBC) as
... ... ... ... ... ... ...
classifier method and NBC is one of machine learning
Cm Cm1 ... Cmj Cm(j+1) .. Cmn
algorithm, so we need some dataset. First, NBC would
learn from dataset to produce some probabilistic models
Generally, text classification goes through three main of each words in each category. By using the
steps: pre-processing, text classification and evaluation. probabilistic models returned from learning process,
Pre-processing phase aims to make the text documents NBC will justify or classify some sentences for the most
suitable to train the classifier. Then, the text classifier is proper category.NBC equations was briefly given at
constructed and tuned using a text learning approach Units section above.
against from the training data set. Finally, the text
classifier gets evaluated by some evaluation measures D. Evaluation
i.e. recall, precision, etc. The following sections are To evaluate the classifier system performance, we
devoted to these three phases. used Precision and Recall that were explained also at
Units section above.
A. Dataset Gathering
Data used in this research are collections of paper IV. RESULT AND DISCUSSION
abstract of SENTIA. SENTIA is national seminar held
Dataset that was used for learning and testing process,
by Politeknik Negeri Malang on information technology
we got from collection of SENTIAs publication
including its application.
abstract.
We used 276 training data. These data was collected
These abstract (training data) classified into 11
from abstract collection of SENTIA 2011 and SENTIA
categories (Electronics and Control System, Informatics
2010. SENTIA is an annual seminar organized by
Politeknik Negeri Malang. For testig purpose we used and Computer, Electricity, Telecommunication,
148 testing data that were collected from SENTIA 2009. Bioengineering, Economic and Bussiness, Government,
Education, Chemical, Machine Engineering and Civil
B. Pre-processing Engineering). These training data will be used to train
Preprocessing is used to make the documents that will and make the classifier probabilistic model.
be processed are ready and suitable to be processed by
NBC. There are 3 types of preprocessing method used in Table 3. Comparison between Classification without Preprocessing
this research. Result and Classification with Preprocessing
Time
1. Stopword Removing (second)
Stopword is words that are not relevant to be Classification without
processed. It means that stopwords are not relevant or Preprocessing 207.8
unnecessary in classifying document to a category. For Classification with
example, yang, di, kepada, untuk etc.By removing Preprocessing 116.6
stopwords, we could reduce the dimension of words that
will be processed by classifier. It means, we could also Table 3 shows the comparison of time spent to do
reduce the time spent to do the classification or learning both of classification with preprocessing and
process. classification without preprocessing. Figure 1 gives the
graphical illustration of data shown in Table 3. It shows
2. Stemming
that preprocessing implemented in text classification
Stemming is process to which attempt to reduce a could reduce time spent to do classification process. By
word to its stem or root(basic) form. For example: implementing the preprocessing before classification
Menghitung hitung process, we could reduce the number of word or feature
Perhitungan hitung to be processed, so that time spent to classify text could
be reduced too.
In several natural language processing, stemming is not
relevant or is not necessary to be done. But in this text
classification research, stemming is relevant to be
implemented, because this teks classification process
just aware in basic form of the word, not into the word
form effected by its affix. For example, we just need to
process the basic form of hitung instead the verb
resulted by adding prefix me- to hitung (menghitung).
3. Chi Square Test as Feature Selection Method (2)
In order to reduce the word dimension, we applied
feature selection method. For this current research we Fig. 1. Comparison between Time resulted by Classification without
Preprocessing and Classification with Preprocessing
E3-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
To evaluate the system performance, we calculate make text classification system could perform
precision and recall. In this paper we just show precision optimally. By using more big number of training
and recall calculation result of Electronics (ELE) and documents, probability model resulted by Nave Bayes
Informatics (IT) category. We choose both of ELE and could be better than using training documents in small
IT because they have the a lot of dataset used in training number. By using the better probability model, the
process.Table 4 shows the value of precision and recall system could optimally classify the documents .
resulted in Electronics (ELE) category and Informatics
(IT) category.
E3-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
E3-5
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
I. INTRODUCTION
E4-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
With the first input is a plain text message m, the Figure 2.Modification of the Vigenere Table
second input is a number k which is a random number,
the third input is the numbers (A, p, g) which is a public The mathematical model of the Vigenere Cipher
key that is processed to produce output in the form encryption process which has been modified will bea
of C and C . Generally, the resulting cipher text is form of:
(C ,C ).
For the decryption process to be performed by the C = P + K mod 256 ............................... (8)
receiver, it requires three inputs which are then
processed into an output and is written in mathematical The mathematical model Vigenere Cipher decryption
form: process which has been modified will be a form of:
Withthe first inputis C , the second inputisC and Where C is a ciphertext, P is the plaintext, K is the key
thethirdinputisthe numbers (a, p) which is processedto
E4-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
and mod 256 states remainder (modulus) with 256. Vigenere Cipher and ElGamal, this algorithm also
Triple Vigenere Cipher has three processes in the processes each pixel component of the input image. The
process of encryption and decryption. Because it has selection of a prime number p is the number 257 with the
three processes in each process, the possible modes are: same process on the ElGamal algorithm. Processes in
a. EEE the combined algorithm can be described as follows:
(encryption(encryption(encryption(message)))) Mathematical form of create public key :
TVig(message,K1,K2,K3) = (P + K1 + K2 +
A=((g + 1)(TVig(a1,a2,a3,a4)+ 1) (mod 257) 1) ...... (14)
K3)(mod 256)
b. EED Mathematical form of encryption :
(decryption(encryption(encryption(message))))
TVig(message,K1,K2,K3) = (P + K1 + K2 - C1 = ((g + 1)k (mod 257) 1) ............................ (15)
K3)(mod 256) C2 = ((m + 1)(A + 1)k (mod 257) 1) .............. (16)
c. EDD Mathematical form of decryption:
(decryption(decryption(encryption(message))))
TVig(message,K1,K2,K3) = (P + K1 - K2 - Message= (((C1 + 1)(TVig(a1,a2,a3,a4) + 1))1 (C2 + 1)
K3)(mod 256) (mod 257) 1) ..................................................... (17)
d. EDE
(encryption(decryption(encryption (message)))) IV. DISCUSSION
TVig(message,K1,K2,K3) = (P + K1 - K2 + 4.1. ETV Create Public Key
K3)(mod 256) ETV create public key process in the application
e. DDD program is shown in Figure 3 and Figure 4.
(decryption(decryption(decryption(message))))
TVig(message,K1,K2,K3) = (P - K1 - K2 - K3)(mod
256)
f. DDE
(encryption(decryption(decryption(message))))
TVig(message,K1,K2,K3) = (P - K1 - K2 +
K3)(mod 256)
g. DEE
(encryption(encryption(decryption(message))))
TVig(message,K1,K2,K3) = (P - K1 + K2 +
K3)(mod 256)
h. DED
(decryption(encryption(decryptio(message))))
TVig(message,K1,K2,K3) = (P - K1 + K2 -
K3)(mod 256) Figure 3. ETV Create public key
Message= (((C1 + 1)(a + 1))1 (C2 + 1)(mod 257) 1) 4.2. ETV Encryption
..................................................................................................... (13) ETV encryption process in the application program
is shown in Figure 5 and Figure 6.
3.3. Combine Algorithm (ETV)
Cryptography to be built is a public-key
cryptography is the process includes create public key,
encryption and decryption. The entire process will use
the digital image as input variables. Similar to Triple
E4-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
V. CONCLUSION
This research proposed a new cryptographic
algorithm design is a combination of Triple Vigenere
CipherCryptographyand ElGamalCryptography. From
the design, implementation and testing of cryptographic
software obtained the following conclusion:
1. Triple Vigenere Cipher cryptography and
ElGamalcryptography can be combined using a
mathematical model of the Triple Vigenere Cipher
and ElGamalthat have beenmodified, which have
similar characteristics with ElGamal cryptography.
Characteristics of the algorithm are a public key
cryptography, message expansion in cipherimage,
Figure 5. ETV encryption time of encryption process is not much different
from the ElGamal encryption, decryption processing
time and create public key processing time longer
than the ElGamalcryptography and Triple Vigenere
Cipher, combined cryptography has several modes
as well as the Triple Vigenere Cipher cryptography
is used in create public key and decryption, as well
the combined algorithm has four private key used in
the decryption process, thus increasing the security
of a digital image of the message can be obtained.
2. In the combined algorithm, the similarity of
messages and decrypt the result would be obtained if
the message in the form of bitmap digital images,
with 100% similarity in pixels and visually. When
using jpeg digital image, the result will be the same
Figure6.Result of ETV encryption visually, but the pixels are not 100% the same, due to
the compression techniques used in the jpeg image.
4.3. ETV Decryption 3. In the combined cryptography, entropy of the public
ETV decryption process in the application program key encryption will affect the results of the message.
is shown in Figure 7 and Figure 8. The greater the entropy of the public key, then the
encrypting result has a high randomness.
REFERENCES
[1] Adha, R.Vignere Cipher RotasiBerlapis.
http://www.informatika.org, October 2010.
[2] Afif, S.Kriptografi. http://javanusco.files.
wordpress.com, October 2010.
[3] Anonymous. PSEUDOCODE STANDARD.
http://users.csc.calpoly.edu, October 2010.
[4] Anonymous. Citra Digital. http://www.ittelkom
.ac.id, October 2010.
[5] Anonymous. PengantarKriptografi. http://
www.informatika.org, October 2010.
Figure 7. ETV decryption [6] Anonymous.
LandasanMatematikaUntukKriptografi.
http://haryanto.staff.gunadarma.ac.id, January 2011.
[7] Defls, H danKnebl, H. Introduction to
Cryptography - Principles and Applications.
Springer-Verlag Berlin Heidelberg.2007.
[8] Gonzalez, C dan Woods, E. Digital Image
Processing Second Edition. Addison-Wesley
Publishing Company, Inc.1993.
[9] Guiliang, Weiping, XiaoqiangdanMengmeng.
Digital Image Encryption Algorithm Based On
Pixels. Intelligent Computing and Intelligent
Systems (ICIS), IEEE International Conference on.
ISBN: 978-1-4244-6582-8. Page 769 772.
Xiamen, China.2010.
Figure 8.Result of ETV decryption
E4-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
E4-5
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
E5-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
x = ( x1 , x 2 ,...x N )
and two auxiliary vectors:
x + = (x1 , x 2 ,...x N 1 )
(1)
Fig. 1. RR interval (RRI) duration derived from an ECG
x = (x 2 , x3 ,...x N )
so that each point:
the autonomic nervous system, various substances
circulating in the blood and present in the heart tissues (x i
+
)
, xi , i = 1....N 1
[11]. The control of heart rate is modulated by both in the plot corresponds to two successive heart beats.
sympathetic and parasympathetic branches of Poincar plot provides sumary information as well as
autonomic nervous system as well as many other detailed beat-to-beat information on the behavior of the
autonomic reflexes. heart [16] . Poincar plot shown in Fig. 2 can be divided
Breathing is one of the important factor modulating into three regions. All points described by consecutive
heart rate [11]. It causes heart rate acceleration during cardiac beats of equal duration (RRn = RRn+1) are
inspiration and its deceleration during expiration. located on the identity line. The points above the identity
Another example of a separate system regulating the line correspond to all prolongations (RRn < RRn+1), and
heart rate are the changes in blood pressure modulated the points below this line represent all shortenings of the
by baroreflex. interval between two consecutive beats (RRn >RRn+1).
All these systems and reflexes are responsible for There are several descriptors in the Poincar plot. The
changing of the duration of RR interval from one beat to following 3 descriptors of the Poincar plot were used in
another and this phenomenon is called heart rate the study [14], as shown in Fig. 3.
variability (HRV) [12]. HRV is a strong and SD1 is the standard deviation calculated from
independent predictor of mortality following an acute
myocardial infarction [13]. The higher HRV the better
prognosis in survivors of myocardial infarction or
patients with heart failure.
A number of parameters are used in HRV analysis.
The Poincar plot of RR intervals is one of the recent
methods. The analysis of Poincar plot is an emerging
method of nonlinear dynamics applied in HRV analysis.
E5-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
x+ x
x1 = (5)
2
SD21 is the ratio of SD2 to SD1.
Fig. 4. RR interval time series (a) RR Interval Time Series for record
SD21 = SD2 / SD1 (6) 30 and (b) the 9000 s to 9300 s selected segment
E5-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Table I. Comparison of Poincar plot between Sudden Cardiac [2] Concise Dictionary of Modern Medicine. McGraw-Hill
Death Holter Database and MIT-BIH Normal Sinus Rhythm Companies, 2002.
Database [3] Lloyd-Jones D, Adams RJ, and Brown TM. Heart disease and
stroke statistics---2010 update. Circulation 2010;121:e46--215.
[4] Al-Khatib, SM, Sanders GD, Bigger JT, Buxton AE, Califf RM,
and Carlson M. Expert panel participating in a Duke's Center for
the Prevention of Sudden Cardiac Death conference. Preventing
tomorrow's sudden cardiac death today: part I: Current data on
risk stratification for sudden cardiac death. Am Heart
Journa,2007; 153: 941-950.
[5] Kuck KH, Cappato R, Siebels J, and Ruppel R. Randomized
comparison of antiarrhythmic drug therapy with implantable
defibrillators in patients resuscitated from cardiac arrest: the
Cardiac Arrest Study Hamburg (CASH). Circulation 2000; 102:
748754.
[6] Mirowski M, Reid PR, and Mower MM. Termination of
malignant ventricular arrhythmias with an implanted automatic
defibrillator in human beings. N Engl Journal Med 1980;
303:322-324.
[7] Zipes DP, Camm AJ, and Borggrefe M. ACC/AHA/ESC 2006
guidelines for management of patients with ventricular
arrhythmias and the prevention of sudden cardiac death: a report
of the American College of Cardiology/American Heart
Association Task Force and the European Society of Cardiology
Committee for Practice Guidelines (Writing Committee to
Develop Guidelines for Management of Patients With
Ventricular Arrhythmias and the Prevention of Sudden Cardiac
Death). Journal Am Coll Cardiol 2006;48:e247-e346.
[8] Zwanziger J, Hall WJ, Dick AW. The cost effectiveness of
implantable cardioverter-defibrillators: results from the
Multicenter Automatic Defibrillator Implantation Trial
(MADIT)-II Journal Am Coll Cardiol 2006;47:2310-2318.
[9] Mark DB, Nelson CL, dan Anstrom KJ. Cost-effectiveness of
defibrillator therapy or amiodarone in chronic stable heart
failure: results from the Sudden Cardiac Death in Heart Failure
Trial (SCD-HeFT) Circulation 2006;114:135-142.
[10] Kadish A & Mandeep Mehra. Heart Failure Devices.
Implantable Cardioverter-Defibrillators and Biventricular
Pacing Therapy. Circulation. 2005; 111: 3327-3335.
[11] R. Hainsworth, Physiology of the cardiac autonomic system in:
Clinical guide to cardiac autonomic tests edited by M. Malik,
Kluwer Academic Publishers, London, 1998, pp. 3-28.
[12] Heart rate variability. Standards o measurement, physiological
interpretation, and clinical use. Task Force of the Working
Groups on Arrhythmias and Computers in Cardiology of the
ESC and the North American Society of Pacing and
Electrophysiology (NASPE), European Heart Journal 93, 1996,
pp. 1043-1065.
[13] Bigger JT, Fleiss JL, Steinman RC, Rolnitzky LM, Kleiger RE,
Rottman JN. Frequency domain measures of heart Standards of
heart rate variability. European Heart Journal, Vol. 17, March
1996.
[14] P Guzik, Jaroslaw P, Tomasz K, Raphael S, Karel HW, Andrzej
Rec: record number WT, and Henryk W. Correlations between the Poincar Plot and
Conventional Heart Rate Variability Parameters Assessed during
Paced Breathing. Journal Physiol. Sci. Vol. 57, No. 1, Feb.
VII. CONCLUSION 2007. pp. 6371.
[15] J. Piskorski, P. Guzik. Filtering Poincar plots. Computational
Poincar Plots of RR intervals from a person who Methods in Science and Technology 11(1), 2005. pp. 39-48.
would have sudden cardiac death are characterized by [16] Toichi M, Sugiura T, Murai T, Sengoku A. A new method of
assessing cardiac autonomic function and its comparison with
the relatively large values of SD1 and SD2, and small spectral analysis and coefficient of variation of R-R interval.
value of SD21. Journal Auton Nerv Syst. 1997;62:79-84.
[17] Goldberger AL, Amaral LAN, Glass L, Hausdorff JM, Ivanov
PCh, Mark RG, Mietus JE, Moody GB, Peng C-K, Stanley HE.
REFERENCES PhysioBank, PhysioToolkit, and PhysioNet: Components of a
[1] A. A. Sovari. (2011, Dec 5). Sudden cardiac death. Medscape New Research Resource for Complex Physiologic Signals.
Reference: Drugs, Diseases, and Procedures. Available: Circulation 101(23):e215-e220 [Circulation Electronic Pages;
http://emedicine.medscape.com/article/151907-overview#show http://circ.ahajournals.org/cgi/content/full/101/23/e215]; 2000
all. (June 13).
[18]
E5-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
E6-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31,
31, Brawijaya University, Malang, Indonesia
Once a cancerous lung tumor grows, it may or may features which are extracted from texture analysis.
not shed cancer cells. These cells can be carried away in These features contain the information about the image
blood or float away in the natural fluid, called lymph, such as homogeneity,ty, contrast, the complexity of the
that surrounds lung tissue. Lymph flows through tubes image, and etc. They are used in many applications such
called lymphatic vessels that drain into collecting as biological applications and image retrieval.
stations called lymph nodes, the tiny, bean-shaped
bean
organs that help fight infection. Lymph nodes are
located in the lungs, the center of the chest, and
elsewhere in the body. The natural flow of lymph out of
the lungs is toward the center of the chest, which
explains why lung cancer often spreads there. When a
cancer cell leaves its site of origin and moves into a
lymph node or to a faraway part of the body through the
bloodstream,
am, it is called metastasis [4].
The stage of lung cancer is determined by the
location and size of the initial lung tumor and whether it
has spread to lymph nodes or more distant sites. The
typee of lung cancer (NSCLC versus small cell) and stage Figure 2 Direction of calculation in Haralick texture features
of the disease determine what type of treatment is
needed. Based on figure 2.3, Haralick texture features can be
extracted from eight directions. Four equations are used
B. Gray Level Co-occurrence
occurrence Matrix :
Texture analysis is an essential issue in computer
vision and image processing, such as in remote sensing,
content based image retrieval.
retrieval The Gray-Level $%
+, -+ , ,
Co-occurrence
occurrence Matrix (GLCM) is one of the statistic
method that can be used as texture analysis based on the ,
&'
'
$%
extraction of a gray-scale
scale image. Coocurance Matrix is
+. -+ , ,
extracted by considering the relationship between
betw two
neighborhood pixels. Based on the hypothesis that in a ,
texture configuration reccurence is occurs, the first pixel &'
'
$% $%
is known as a reference and the second is known as a
+,/. --+ , ,
neighbor pixel [5].
,
GLCM can be measured as follows : &' &0
2, 3,, . , 289 ",
(3)
1) Create the GLCM symmetrical
2) Calculate the probability of each combination, the 3 ,
, , , # , , , (
probability is calculated :
$% $%
+,4. --+ , ,
,
&' &0
, 2, 3,, . , 289 ",
(4)
, 0, )| , | |,
, ,
(1)
, , , , (
!"
14 derived features from Haralick
aralick are written as follows:
If the co-occurrence
occurrence matrix is symmetric then p(i,j) p( = 2) Contrast
(p(i,j) + p(i,j)T ) = 2 that T indicates the transpose matrix Direction of contrast calculation is to measure the
and will be 0, 45, 90 and 135. intensity or gray-level
level variations of reference pixel and
its neighbor.
3) Texture features calculations
In order to estimate the coocurence in texture
analysis using GLCM, Haralick proposed 14 statistical
E6-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
, = |
$% 4' $% $%
5: - :
<- - + ,
9) Entropy
;&0 &' &'
By extract the entropy features of a grayscale
|
(6) image, value of image which is needed for image
compressions.
$% $%
5I --+ , GH + ,
3) Correlation
In co-occurene matrix, gray lavel values linear , , (13)
dependency of a grayscale image is calculated in order &' &'
to obtain the correlation features. Which is represent the
relation between a pixel with pixels neighbor in eight If high entropy obtained, it is mean that a grayscale
+ , ,
directions. image has high contrast from one pixel with it is
$% $%
&' &' ?, ?. neighbour and cannot be compressed as a low contrast
5>
@ @
as a result low entropy
, .
(7)
OPQ OPQ1
12) Information Measures of Correlation 1
5':
4) Sum of Squares (Variance)
KR OP, OQ
Sum of Squares is statistic equation for extract the
(16)
variance of image gray tone in a grayscale image.
$% $%
5A -- ? :+ , ,
S'>
(8) 13) Information Measures of Correlation 2
&' &'
OPQ --+ , GH + ,
grayscale image. Homogeneity returns the measures of
the closeness of the distribution of the GLCM elements , , (18)
to the GLCM diagonal. &' &'
$% $%
1
5B -- + , ,
13
HX and HY are entropies of px and py
: (9)
&' &'
6) Sum of Average $% $%
OPQ1 --+ , GH +, +.
Sum of average calculate the average or mean of a
grayscale image. ,
:$%
(19)
&' &'
7) Sum of Variance - - +, +. GH +, +.
:$%
&' &'
5E - 5F : +,/. (11)
&:
14) Maximal Correlation Coefficients
8) Sum of Entropy 5'A XT YNMG KLHNZ[ N HN JK \N G5 ]
:$% (21)
5F - +,/. GH+,/. (12)
&:
+ , + , ,
Where
,
] , -
In equation of sum of entropy, in some case the
probability is equal to zero. Causing the log(0) cannot be +, +. ^
^
(22)
defined, therefore to solve this problem, it is
recommended to use log(p+) is an fluctuative small
positive constant.
E6-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
, G[NLh ZN
Image histogram in MATLAB choose the grayscale
transformation T to minimize :
b; b0 ;
object except In. If Rn is the segmented image, then :
Where c0 is the cumulative histogram, c1 is the
(35)
cumulative sum of histogram for all intensities k. The
b < 0
equation is based on constrain that T should be
5 b l o
0,
b , G[NLh ZN
monotonic and c1(Ta)) always less than c0(a) by more
than half the distance between initial histogram. In
MATLAB algorithm, image histogram equalization of
chest image is written as : H = histeq(I);
E6-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
E6-5
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31,
31, Brawijaya University, Malang, Indonesia
xu = Output Target
Table 3. Haralick feature extractions of chest X-Ray
X image (3) where
2 :
Num
Sum Sum Diff. xwu 1= Output Training
Variance Entropy Av. n = Number of training data Target
1 1529.50234 0.26245 2.03841 0
2 2319.44702 0.31598 1.66829 Output
-1
3 2119.04656 0.34456 2.91122 1 3 5 7 9 11 13 15 17 19 21 23
4 1222.52621 0.21021 1.41418 Figure 6 Testing Results
5 1160.28484 0.21258 1.62962
1 is represent that there are malignant cell present
6 1051.05972 0.18830 1.28160 in the lung from processed Chest X-Ray
X image, while 0
7 4409.11707 0.49555 2.31462 represent the normal lung. If the output results are less
than 0.5 then it classified as normal lung, in the other
8 1754.03879 0.26790 1.65172 hand, if the results are > 0.5 it classified as lung cancer
9 1170.49981 0.19786 1.23188 image. From the testing graphic plot in figure 6, the
validation results can be seen in table 4
10 1844.86317 0.27771 1.70144
Table 4. Results of Validation
Extracted features from Chest X-ray
X image are set
as training and testing data which contain of the desired Classified Misc. Accuracy
input and output data pairs. Membership function 23 2 92%
parameters can be estimated in training phase of ANFIS.
This project uses the ANFIS Editor GUI menu bar to
load a FIS S training initialization, and then save the V. CONCLUSIONS AND FUTURE WORKS
trained FIS. To open the ANFIS GUI, following syntax A new method of lung cancer detection system has
is typed in MATLAB command window : been developed by using Haralick Features of Gray
Level Cooccurence Matrix as feature extraction method
anfisedit
to get the ANFIS input data from initial Chest X-Ray
X
image for cancer detection purpose in imaging test.
Before extracted, the image are processed through
1.30 several stages, i.e. (1) Resizing, (2) Grayscaling, (3)
Histogram equalization for contrast enhancement, (4)
Lung segmentation based on their region of interest, and
0.80 (5) Image matrix reduction by specific reduction
r factor.
Target
There are nine features (Energy, Correlation,
Output Entropy, Inverse Diff. Moment, Sum Average, Sum
0.30
Variance, Sum Entropy, Inf. measure of correlation 1,
and Inf. measure of correlation 2) appropriate as data
-0.20 1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 input in ANFIS as information of malignant
ma cell present
in the lung Based on testing results,
results ANFIS parameters
Figure 5 Training Results with 9 inputs features can predict lung cancer in imaging
test with 92% accuracy with 2 missclasified from 25
In order to obtain the training performance numerically, images
Root Mean Square Error (RMSE) value of each training
is calculated by using the following equation
vu&' xu xwu :
t
X REFERENCES
RMSE
n [1] Le Kim. Automated Detection of Early Lung Cancer and
Tuberculosis Based on X Ray Image Analysis.
Analysis International
= 0.135026 Conference on signal, speech, and Image Processing WSEAS.
2006. 110
25 data are used in testing phase for validation. to load [2] American Society of Clinical Oncology. Guide to Lung Cancer.
Alexandria. Conquer Cancer Foundation. 2011: 2.
the testing data, testing button is selected and data
da is [3] Floche. Backgroundd information Non-small
Non Lung Cancer.[pdf]
uploaded from MATLAB workspace. Testing phase or (URL:http://www.roche.co.id/fmfiles/re7229001/Indonesian/m
model validation is required to test the performance of edia/background.library/oncology/lc/Lung.Cancer.Background
model in data fitting process. er.pdf accesed on July 30, 2011)
[4] Anonymous. Kanker Paru Pedoman Diagnosis dan
Penatalaksanaan di Indonesia.
ndonesia. Perhimpunan Dokter Paru
Indonesia. 2003
[5] Haralick, Robert M. Textural Features for Image Classification.
Classification
IEEE Transactions Onn Systems, Man, And Cybernetics. 1973,
3(6) : 612-619
E6-6
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Abstract The purpose of this study is to compare mouth stands [1]. In this case, it Is thought to be an
performance of the EDR methods for sleep apnea obstruction at the URT of patient. In order to prevent the
classification. EDR methods used are two algorithm, R blockage, an intense activity in the chest and abdomen is
Wave Detection and RR Interval Correction, and using
observed. Central sleep apnea (CSA) is the state in the
Naive-Bayes Classification. Data obtained from the
absence of both respiratory effort and air flow together.
polysomnography recordings related to the 7 different
patients are used for evaluations. Data of each subject taken
Central apneas grow by the corruption of the central
at a specific time frame for 15 minutes. timescales are chosen regulation of respiration. Mixed sleep apnea is the state
based on annotation data with apnea detection. Results of starting with central sleep apnea and continuing the
both algorithm shows have generated signal similarity model absence of oral and nasal air flow when the respiratory
(wave form),and the classification results for the RR- effort begins. How the respiratory effort after the central
intervals correction method is slightly better than the R- sleep apnea starts is still a unresolved research topic [2].
wave detection method, for R Wave Detection, correctly The device used for measuring and recording
classified 0.5333 and kappa statistic 0.0395, for R-R Interval
physiological signals during sleep is called as
correction , correctly classfied 0.5333 and kappa statistic
polysomnograph and the signals retrieved from the device
0.0676
are called as polysomnography (PSG). By the use of PSG,
it is possible to observe the physiological changes in
Index terms Apnea Classification, ECG, EDR, R-R humans during sleep. Various physiological signals of
Interval. the subjects are recorded simultaneously by the PSG
device, which has an embedded multi-channel data
acquisition system. The recording process made as analog
I. INTRODUCTION recordings in the 90s has left its place to digital recorders
after the development of digital systems. Thus, the
The most important one of the sleep repiration prevention of errors caused by the hardware chaos of
disorders is sleep apnea. Sleep apnea is a common analog systems is provided [3]. By the use of these
disorder in which there is a pause in breathing or shallow devices, Electroencephalogram (EEG), Electrocardiogram
breaths during sleep. There are three types of sleep apnea, (ECG), Electromyogram (EMG), Electrooculogram
central,obtructive and mixed apnea. Obstructive sleep (EOG), breathing, Pulseplethysmograph (PPG) and
apnea (OSA) has the highest prevalence. Together with various desired or necessary signals of subjects in sleep
the absence of respiratory effort in the lungs, the absence are recorded. In this way, the subjects statuses are
of air flow inside the mouth and nose is defined as central determined during the night sleep and their diagnosis and
sleep apnea. Despite the respiratory effort, the lack of air treatment outcomes can be delineated. The classification
flow in the nose and mouth is obstructive sleep apnea. of sleep apnea is also realized by the investigation of
The situation starting with central sleep apnea and these physiological signals obtained from the PSG device.
continuing as obstructive sleep apnea is defined as mixed The following physiological activities in your body
sleep apnea. Mixed apnea subjects can be treated by the are recorded during polysomnography [11].
methods applied to the subjects with obstructive sleep
apnea. Obstructive sleep apnea is the most common sleep Brain activity is measured with EEG
apnea syndrome. Obstructive sleep apnea is the state of (electroencephalogram). Disruptions in sleep stages
may suggest narcolepsy or REM sleep behavior
absence of oral and nasal air flow despite the respiratory
disorder.
effort. Although the diaphragm and intercostal muscle
activity continued, exchange of air through the nose and
E7-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Eye movements are recorded with a EOG (electro- normally visible in 50-75% in the ECG. Fig. 1 shows a
oculargram). This helps to determine the presence of typical example of an ECG signal. The duration of a
sleep stages particularly REM sleep stage. heartbeat is the time interval from one R wave to next R
Muscle activity such as face twitches, teeth grinding wave, also known as RR-Interval [5]
and leg movements, is measured by EMG
In 2004, Chazal et al, suggested an obstructive sleep
(electromyogram). This helps to determine if the REM
sleep is present during sleep. Frequent leg movements apnea detection using ECG signal. Features used in these
may indicate a periodic limb disorder such as restless studies are the statistical measurement of variables
leg syndrome which is a sleep disorder because these derived from RR-intervals and ECG-derived respiratory
movements often disrupt your sleep. signal (EDR) [6].
Airflow in and out of the lungs while you are asleep is
measured with a nasal airflow sensor.
Snoring activity is measured with the help of snoring
microphone. Very loud snoring is indicative of sleep
apnea
The percentage of oxygen in your blood is measured
(oxymetry) by a bandage like oxymeter probe or
sleeve that fits painlessly on one of your fingers. Low
oxygen levels may indicate a sleep apnea.
male,aged 32 to 56 (mean age 43), with weight ranging Data of each subject taken at a specific time frame for 15
from 89 to 152 kg (mean weight 119 kg). There were two minutes. timescales are chosen based on annotation data
subjects that divided into 2 record, subject with code with apnea detection. Spesific time frame of each subject
slp01 and slp02, so total records are 18. are shown in table3.
E7-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Naive-Bayes Classification
A Naive-Bayes classifier is a probabilistic classifier based
on Bayes theorem with strong (naive) independence
assumptions. Naive Bayes classifier assumes that the
presence (or absence) of a particular feature of a class is
unrelated to the presence (or absence) of any other
feature. This classification method relies on transforming
the discrete decision labels output by the individual
matchers into continuous probability values.
CA
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
OA
OA
OA
MA
Fig 4. Respiratory rate detection using the size of R wave
Fig 6. EDR Signals from RRinterval and Rwave with subjects code
slp03
2.0
Rwave
1.5 RRin
1.0
0.5
0.0
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
MA
OA
OA
OA
OA
OA
OA
OA
OA
OA
OA
Fig 7. EDR Signals from RRinterval and Rwave with subjects code
slp16
2.00
1.50
1.00
0.50 Rwave
RRin
0.00
N NOAOAN N NOAOAOAN N N N N NOAOAN NOANOANOANOANOAN
Fig 8. EDR Signals from RRinterval and Rwave with subjects code
slp32
E7-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
1.50 calculation, and the x-axis shows the type of sleep apnea
that occurs.
Process with the naive-Bayes classification, performed
1.00
on each of the EDR calculation algorithm, R Wave
Rwave Detection and R-R Interval corection are, for R Wave
0.50 Detection, correctly classified 0.5333 and kappa statistic
RRin
0.0395, for R-R Interval correction , correctly classfied
0.00 0.5333 and kappa statistic 0.0676
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
N
OA
OA
CA
OA
OA
OA
OA
OA
OA
OA
OA
OA
OA
Fig 9 EDR Signals from RRinterval and Rwave with subjects code IV. CONCLUSION
slp45
Based on the results of the experiment, the respiratory
1.20 signal can be obtained from the ECG signal processing.
1.00 method used is to use the RR-interval correction and R-
0.80 wave detection. Results of both shows have generated
RRin signal similarity model (wave form).
0.60 The signal generated from the two methods are used
0.40 Rwave for sleep apnea classification using Naive-Bayes
0.20 Classification. The classification results for the RR-
intervals correction method is slightly better than the R-
0.00 wave detection method, with a kappa value of 0.0676.
N
N
N
N
N
N
N
N
CA
CA
CA
CA
CA
CA
CA
CA
CA
CA
OA
OA
OA
OA
OA
OA
OA
OA
CA
Keywords CellProfiler, Central Vein, Congested, CellProfiler is freely available modular image analysis
Pipelines, Plasmodium berghei software that is capable of handling hundreds of
thousands of images. The software contains
I. INTRODUCTION already-developed methods for many cell types and
Malaria is the most serious and widespread assays and is also an open-source, flexible platform for
parasitic disease of humans. It affects at least 200 to 300 the sharing, testing, and development of new methods by
million people every year and causes an estimated 3 image analysis experts. CellProfiler uses the concept of a
million deaths per year. Malaria is initiated by 'pipeline' of individual modules. Each module processes
Plasmodium sporozoites, which are inoculated by the images in some manner, and the modules are placed
mosquitoes. The disease is characterized by a range of in sequential order to create a pipeline: usually image
clinical features from asymptomatic infection to a fatal processing, then object identification, then
disease [1]. There are four species of Plasmodium that measurement. Most modules are automatic, but
infect man and result in four kinds of malaria fever: P. CellProfiler also allows interactive modules (for
falciparum, P. vivax, P. ovale, and P. malariae [2]. example, the user clicks to outline a region of interest in
Malarial involvement of liver is now a known entity each image). Modules are mixed and matched for a
with it is specific histopathological lesions. specific project and each module's settings are adjusted
Histopathologic lesions often described in the liver of appropriately. Upon starting the analysis, each image (or
rodent with malaria is congested central vein with group of images if multiple wavelengths are available)
neutrophils and eusinophils within the lumen (Fig. 1) travels through the pipeline and is processed by each
[2,3]. Detection of congested central vein commonly module in order [5].
done manually under microscope. This process has a Here the used of CellProfiler an open access cell
possibility to do automatically using image analysis image analysis software for automated detection
software. With the availability of digital photography, congested central vein liver histology of mice infected
the congested central vein detection process can be done with Plasmodium berghei is reported. The results are
E8-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
3.5
0.5
2.5. Automated detection congested central vein 0
0 1 2 3 4 5 6 7 8
CellProfiler Results
An open access cell image analysis software
CellProfiler 2.0 r10997 that developed by Broad Fig 3. Scatter plots comparing congested central vein
Institute was used for an automated detection congested defined by manually and with CellProfiler.
central vein. CellProfiler (CP) runs on Microsoft
Windows XP SP 2 32-bit platform. Processor type used
3.2. Statistical analysis compare between automated
inside the computer is AMD Athlon(tm) 64 X2 Dual
and manual results
Core 5000+ with memory (RAM) is 1.87 GB. A
pipelines was developed to doing automatic detection Statistical analysis using Wilcoxon Rank test show
congested central vein (Fig. 2). that there are no significant different between manual
counting and automated counting (P = 1), because the
p-value is bigger than 0.05 it is mean H0 hypothesis is not
rejected.
E8-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
IV. DISCUSSION
Convert color image
Our pipelines consist several steps to detected the to grayscale color
congested central vein in liver histology. First we rescale
the image to speed up the time required for detected
congested central vein in the image, and then we convert
the images to grayscale images. Grayscale images show
bright color on objects and a dark color in background.
Because the central vein area is cover by a bright color Detection Central Vein
Using Threshold methods
then it can be easy to identify using the thresholding
methods in Identify Primary Objects module.
Unfortunately because sometimes the material inside
Morphology image
central vein is attached to the border line of the central consist dilate, fill
vein, then the central vein area that detected by holes and erode
thresholding method became very tight. image
In order to refine the central vein area results we apply
several images morph processing to get better central Invert images so material inside
vein detection. Based on our experiment dilate the the central vein can be detected
images, fill holes after dilate the images and last erode using threshold methods
the images can be used for getting a better detection
central vein area. Interestingly CellProfiler provided all
image morph processing in Morph module. To define Detected material inside the
which one is the congested central liver we must detect is central vein using threshold
there any materials inside the central vein. Because methods
thresholding method is detected the bright area and the
materials are in gray color, first we must invert the
images so the material inside the central vein became
bright and can be detected by thresholding methods. In Detection Central Vein that have
our pipelines, the congested central vein detection is a material in the lumen
based on parent-child relationship between the central
vein and the materials inside the central vein. The central
vein (parent) should define as the congested central vein
Fig 4. Flowchart automatic detection of congested
if it has at least one or more material (child) in it. It will
central vein defined by CellProfiler.
not define as congested central vein if the material inside
the central vein is not overlapping with the central vein.
Details of all steps of our pipelines are shown in Fig 4. REFERENCES
A significant variation between manually and [1] SIO, S.W., SUN, W., KUMAR, S., BIN, W.Z., TAN, S.S., ONG,
S.H., KIKUCHI, H., OSHIMA, Y., and TAN, K.S.
automatically detection of congested central vein MalariaCount: an image analysis-based program for the accurate
observed in one image. This happened because there are determination of parasitemia. J Microbiol Methods 2007,
large sinusoids areas and because the inside the sinusoid 68:11-18.
[2] HISAEDA, H., YASUTOMO, K., and HIMENO, K. Malaria:
areas there are many Kupffer cells then the pipelines also
immune evasion by parasite. Int. J. Biochem. Cell Biol. 37,
detected as a congested central vein. A large sinusoid 700-706.
area is also commonly histological lesions because [3] KHAN, Z.M., NG, C., and VANDERBERG, J.P. Early Hepatic
infection of Plasmodium, this phenomenon usually calls Stages of Plasmodium berghei: Release of Circumsporozoite
Protein and Host Cellular Inflammatory Response. Infection and
a sinusoid dilatation. Baheti et al research show that Immunity, 1992, 60(1), p. 264-270.
seventy five percent histological lesion in liver cause by [4] SILVA, A.P.C., RODRIGUES, S.C.O., MERLO, F.A.,
Plasmodium is sinusoid dilatation [7]. We are now in the PAIXO, T.A., AND SANTOS, R.L. Acute and chronic
histopathologic changes in wild type or TLR-2-/-, TLR-4-/-,
process of developing a better pipelines than can be used TLR-6-/-, TLR-9-/-, CD14-/-, and MyD-88-/- mice experimentally
for detected the congested central vein and can infected with Plasmodium chabaudi. Braz J Vet Pathol, 2011,
differentiate between the congested central vein and 4(1), 5-12.
sinusoid dilatation. [5] Carpenter, A.E., Jones, T.R., Lamprecht, M.R., Clarke, C.,
Kang, I.H., Friman, O., Guertin,D.A., Chang, J.H., Lindquist,
R.A., Moffat, J., Golland, P., and Sabatini, D.M. CellProfiler:
V. CONCLUSION image analysis software for identifying and quantifying cell
We have developed pipelines for CellProfiler phenotypes. Genome Biology, 2006, 7:R100.
[6] IVANOVSKA, T., SCHENK, A., HOMEYER, A., DENG, M.,
software that can be used to detect the congested central DAHMEN, U., DIRSCH, O., HAHN, H.K., AND LINSEN, L. A
vein in liver histology section of mice infected with fast and robust hepatocyte quantification algorithm including
Plasmodium berghei. Overall, our pipelines worked very vein processing. BMC Bioinformatics 2010, 11:124
[7] BAHETI, R., LADDHA, P., and GEHLOT, R.S. Liver
well to detection of congested central vein. Involvement in Falciparum Malaria A Histo-pathological
Analysis. JIACM CM, 2003; 4(1): 34-8
E8-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
E9-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Chromium Browser
Webmin (DNS Server)
ZoneMinder Linux CCTV Server
Java Android script programming
B. HARDWARE
Hardware that used to develop the system consist of :
Smartphone Android
o 2.3.3 Android Version (GingerBread
OS)
o 1 GHz Processor
Fig. 1. Application Architecture using Android Smartphone o 2.6.32.9 Kernel Version
IP Camera
o VGA Camera Lens
o MJEPG Video Encoding Standard
Figure 1 describes how the application running
o 270 Remote Pan, 120 Tilt Control
through smartphone android. Administrator can
o Wifi IEEE 802.11 b/g
monitored the situation from smartphone android which
Wireless ADSL2 + Gateway
connect to WLAN PT. Pertamina RU IV Cilacap and
o Modem Linksys WAG54G2
directly access hse.telemonitoring.com /
o Wifi IEEE 802.11 b/g
192.168.100.10.
Notebook
In other way, smartphone android connect via
o Intel Core 2 Duo
WLAN PT. Pertamina RU IV Cilacap through the
o T5450 Processor 1.66 GHz, 667 MHz
server for automatic storage, download video data and
FSB, 2MB L2
o Cache
o RAM 1,5 GB DDR3
C. Files that Used in the System
There are 3 packages contained in this application.
Where in the package there are 8 java class (*.java).
Table 1 indicates file that used in this system
No File Name Remark
1 Package akseshttp Package contains classes that
manage basis64 and streaming
Basis64.java File that used to password
Fig. 2. Application Architecture using Personal Computer (PC) encryption for security
GambarBerubahListene Contains change image
r.java detection method and error
notification.
monitored multiple camera at the same time. PemisahStream.java File that used to separating the
Figure 2 describes how the application running image from http header
through Personal Computer (PC). PC connect via StreamKamera.java Implementation from
GambarBerubahListener.java
WLAN PT. Pertamina RU IV Cilacap then access interface and used to connect
telemonitoring.com through the server for automatic to the camera and take pictures
storage, control the camera and monitored multiple repeatedly
camera at the same time.
2 Package Package contains application
hse.telemonitoring menu
III. IMPLEMENTATION IpCameraActivity.java File that used to control the
camera
This section explained the implementation of the LoginActivity.java File that used to controlled
Telemonitoring Application in Health Safety and menu
Environment at PT. Pertamina Refinery Unit IV
3 Package Video Package to take byte data
Cilacap using Android Smartphone. IImgData.java Contains set bytes and get
bytes method.
ImgData.java Implementation from
A. SOFTWARE IImgData.java interface and
used to wrap the image data
Software that used to develop the system consist of :
Ubuntu 11.04 Natty Narwhal
Apache Webserver
Eclipse IDE Helios + Android SDK and ADT Application Display from Android Smartphone
Plugin
E9-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
IV. CONCLUSION
The conclusion of this research is that has been built
the Telemonitoring Application in Health Safety and
Environment at PT. Pertamina Refinery Unit IV
Cilacap using Android Smartphone. The system has
been succesfully tested and have capability running in
2 ways. Firstly, from Android Smartphone (through
Fig 3. User Interface Login
smartphone web browser or telemonitoring
application). Secondly, from web browser which uses
ZoneMinder as a server.
REFERENCES
[1] _________, 2011, Apache HTTP Server.
http://en.wikipedia.org/wiki/Apache_HTTP_Server, (accessed
at Nov 20, 2011).
[2] _________, 2011, FFmpeg. http://ffmpeg.org/index.html,
Fig 4. User Interface Camera (accessed at Nov 20, 2011).
[3] Charibaldi, Novrido. (2010). Solusi Pemrograman Java
Application Display from Android Smartphone Web (Dilengkapi Contoh Soal dan Penyelesaian). Pyramida,
Yogyakarta.
Browser (telemonitoring.com/zm/skin?=mobile) [4] Eclipse Foundation., 2011, Eclipse.
http://www.eclipse.org/org/, (accessed at Nov 20, 2011).
[5] Fowler, Martin. 2005. UML Distilled 3th Ed, Yogyakarta:
Andi.
[6] Geoffrey, S., 2011, BIND.
http://en.wikipedia.org/wiki/BIND, (accessed at Des 12,
2011).
[7] Gombang, 2011, Wi-Fi.
Fig 5. User Interface Application via Android Smartphone Web http://id.wikipedia.org/wiki/Wi-Fi, (accessed at Sept 14,
Browser / Menu IPCAM MOBILE VIEW 2011).
[8] Graemel, 2011, IP camera.
Application Display from PC Web Browser http://en.wikipedia.org/wiki/IP_camera, (accessed at Sept 14,
2011).
(telemonitoring.com) [9] Kadir, Abdul. 2002. Pengenalan Sistem Informasi
Yogyakarta: Andi.
[10] Macks, D., 2011, Server (computing).
http://en.wikipedia.org/wiki/Server_ %28computing%29,
(accessed at Nov 20, 2011).
[11] Nicolas Gramlich, Andbook : Android Programming,
Download 10 Oktober 2009, http://andbook.anddev.org/
[12] Nugroho, Adi. 2005. Analisis dan Perancangan Sistem
Informasi Dengan Metodologi Berorientasi Objek.
Fig 6. User Interface Application via PC Web Browser
Informatika. Bandung.
[13] OHA, Android.
Implementation telemonitoring.com aimed to help http://www.openhandsetalliance.com/android_overview.html,
administrator to controlled and stored data. There are (accessed at Sept 19, 2011).
[14] Samulo, A., 2011, Intranet.
three menus, IPCAM VIEW(hse.telemonitoring.com), http://id.wikipedia.org/wiki/Intranet, (accessed at Nov 2,
IPCAM MONITOR ZONEMINDER 2011).
(telemonitoring.com/zm/skin?=classic), IPCAM [15] Scott, C., 2011, Basic Access Authentication.
http://en.wikipedia.org/wiki/Basic_access_authentication,
MOBILE VIEW (telemonitoring. com/zm/skin?=
(accessed at Nov 10, 2011).
mobile). IPCAM VIEW used to controlled the camera. [16] Setiawan, A., 2007, Perancangan Dan Implementasi Sistem
IPCAM MONITOR ZONEMINDER used to automatic Monitoring Jarak Jauh Berbasis Protokol AX.25 Dengan
storage. IPCAM MOBILE VIEW used to store data in Menggunakan Mikrokontroler.
smartphone. http://digilib.ittelkom.ac.id/index.php?option=com_repository
&Itemid=34&task=detail& nim=111020129, (accessed at Oct
Application Display from ZoneMinder CCTV Server 13, 2011.
[17] Stachura, Max E. (2010). Telehomecare and Remote
(telemonitoring.com/zm/skin?=classic) Monitoring : An Outcomes Overview. Georgia : Advamed
[18] Triornis LTD, 2011, ZoneMinder.
http://www.zoneminder.com, (accessed at Nov 20, 2011.
[19] Webmin, 2006-2011, Webmin.
http://www.webmin.com/intro.html , (accessed at Dec 12,
2011).
E9-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
E10-1
The 6th Electrical Power, Electronics, Communications, and Informatics
Informat International Seminar 2012
May 30-31,
31, Brawijaya University, Malang, Indonesia
E10-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Table 1. Database tables and fields information such as address and type or category of
NO TABEL FIELD puskesmas, health workers held, and the essential public
catagory_name health tools and equipment essential maternal and child
1 Sdm_catagory id_category_sdm health.
id_group_sdm
group_name
2 group_sdm
id_group_sdm
Status_name
3 status_sdm
id_status
7. Software Development
Software development is a coding process based
database and software design that has been created in
a programming language.
8. Testing and Repair Figure 6. Search menu and tracking.
Software testing performed on every menu that is
created by using the blackbox methods. Users can utilize the tracking menu to look for a route
These applications use the Google map, while for the from where he was to the location of the targeted health
server and database using Xampp and AJAX are used centers, as shown in figure 6.
for web-based programming to create interactive
applications in GIS.
E10-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
III.2 SIGAPP KES for Goverment Next on the menu to choose the level of statistical
The second major function of SIGAPP KES are for data to be seen whether the provincial, city, up to the
government. This function is used to look at the health district level. We can also see the data of each clinic.
center human resources information data, medical Figure 13 shows the doctor HR data statistics from the
devices, the type and category of health centers ( provincial level (western Java), Tasikmalaya city to
puskesmas ) from provincial to district level. Data are kawalu district.
presented in graphical form and GIS, this is to facilitate
the government in monitoring the development of health
centers as well as policy analysis function to take on the
latest conditions. Figure 10. display shows the login to
the government. There are two levels of logging that is a
government or puskesmas staff for the data update
function.
E10-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
E10-5
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
SAGPP KES. Administrator managed user accounts and IV. CONCLUSIONS AND RECOMMENDATIONS
government health centers authorized to view and
IV.1 Conclusions
change data. Figure 21. Display shows the admin login
menu SIGAPP KES application is made to help RIFASKES
program to be more optimal utilization in results. This
application is expected to benefit the community and
government. The application can update the data more
quickly, so that both the government and the public can
obtain puskesmas information better. With this
application the Government may monitor and take a
strategic policy to develop and improve the quality of
health services and health center facilities ( puskesmas ).
IV.2 Recommendations
This application is still far from perfect, so we expect
to continued this riset for some improvement. For future
Figure 21. Admin login menu display.
development, we can used map server and our own map
for easy processing of geography. Addition of data in
On the menu there is a list of admin users that are
which if the critical parameters such as drugs, health
currently active, there is a menu that user ca search by
information, disease information. There should also be
name, code and name of the clinic group. Figure 22.
developed GIS hospitals, laboratories, pharmacies. One
shows the admin menu.
of the roadmap of this research is to make Indonesia GIS
Health System.
REFERENCES
[1] Balitbang Ministry of Health of the Republic Indonesia,
preliminary results of RIFASKES 2011.
[2] Act ( UU ) No. 36 of 2009 on Health.
[3] Kaswidianti Wilis, Budi Santosa, Rifky Satya, "Geographic
Information System Health facilities in the town of Magelang
web-based", National Seminar on Informatics 2008 UPN
yogyakartaUU No. 36 of 2009 on Health.
[4] Software applications Healthmapper, WHO.
[5] Women Research Institute, Availability and Utilization of
Figure 22. Administrator menu display Health Services For Maternal, in 2008
[6] Women Research Institute, Utilization of Reproductive Health
Care for Poor Women, in 2007
Group code indicates that the user root as the main [7] RI health minister's decision 1457/menkes/sk/x/2003 number of
admin, while the user admin as admin for puskesmas minimum standards of health care in the district or city.
[8] Fathul Wahid, Information Systems Research Methodology: an
and Government code indicates that users of the overview, Media Informatics vol 2 no 1, June 2004.69 to 81.
government.
Istikmal ST. MT. was born on 11 november 1979 in Kebumen.
Graduate degree in STT Telkom and continuing education for S2 at
ITB. currently a lecturer in Telecommunications Engineering
Program, Faculty of Electro and Communications, Telkom Institute of
Technology, Bandung, Indonesia.
E10-6
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
E11-1
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
f f
fx = (p + q) fy = ( p + q ) (3)
x y C. Fitted surface differences as a conventional method
to compute the E matrix.
with the directional first-order derivatives of image A FSD method is FQS approach combined with HD
function are fx and fy , respectively. approach [6]. This method in the state of art, namely the
FQS-HD.
Harris and Stephens applied the eigenvalues to the
matrix E, which consists of enough local information in In the FQS approach computation, a point set V can
conjunction with the neighborhood structure. In the fit a quadratic surface[16]. Derivatives calculation
computational process, to prevent the highly eigenvalue perform to fit a quadratic surface to the set of
calculation, they designed to entrust with each pixel in transformed points. Using least square approach to
the approximation function image the subsequent value: discover a paraboloid of the form:
S (x , y ) =
a 2 c
x + bxy + y 2 + dx + ey + f (5)
h ( p ) = det (E ) k (tr (E ))
2
(4) 2 2
E11-2
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
(
x2 + y2 )
X = nv 2 e 2
. f x (x , y ) dx dy The following Figure 2 depicts any point the meshes
2 2
(8)
R T(P), a manifold can be represented including index k
(
x2 + y2 ) of the triangle enclosing it and coefficients of the convex
Y = nv 2 e 2
. f y ( x , y ) dx dy
2 2
(9)
R
(
x2 + y2 ) (
x = u 1 p t1 + u 2 p t 2 + 1 u 1 u 2 p t 3 ; ) x i [0 ,1] (18)
Z = nv 2 e 2
. f x (x , y ). f y ( x , y ) dx dy
2
(10) k k k
nv =
1 (11) ( )
Vector u = u 1 , u 2 is called Barycentric coordinates.
2
X = 2 a 2 + 2b 2 + d 2 (12)
Y = 2b 2 + 2 c 2 + e 2 (13)
Z = 2 ab + 2 bc + de (14)
Fig. 2. Barycentric coordinates and manifold triangular meshes
Last, to calculate the matrix E associated with the
points :
III. FITTED QUADRATIC SURFACE COMBINED WITH
X Z
E = (15) HILBERT SCANNING DISTANCE (FQS-HSD) AS A
Z Y PROPOSED METHOD TO COMPUTE THE E MATRIX
Given two finite points sets P {p1, pI} and Q
Because the proposed method use Barycentric {q1, qJ} such that each point p P and q Q, has
Coordinates [17-19], then if the object tessellation is integer coordinates. Firstly, the Hilbert scanning is used
uniform, i.e., almost all triangles in the manifold to convert them to new sets S {s1, sI} and T{t1,
triangular meshes have the same size, this computation tJ} in the 1-D sequence, respectively. Then the directed
can use a constant number of rings to all points, or use HSD from P to Q is computed by
the points contained in a ball of radius r and centered in
points vertices.
( )
The HD approach can compute the E matrix using a
d HSD (P , Q ) =
1 I
weighted average of the derivative 3D manifold
I
i =1 s i t j (17)
triangular meshes data for differents points. It can
perform the distance points between two points sets P
and Q using the difference fitted quadratic surface, where || . || is the Euclidean norm distance in the 1-D
space and function is defined as:
d HD (P , Q ) = max sup inf p q , sup inf p q (16) x (x )
p P qQ q Q p P (x ) = (18)
(x > )
D. Barycentric Coordinates and Manifold Triangular where, is called the threshold elimination function and
Meshes is a the threshold prefined.
The definition of a 3D data structure for boundary Then the directed HSD Q to P hhsd(Q,P) is obtained
representation, such triangular meshes, involves the similarly and HSD is defined by
topological entities coding, with the linked up geometric
information and of a suitable subsets of topological
relationships between such entities. A data structure is H HSD (P , Q ) = max (h HSD (P , Q ), h HSD (Q , P )) (19)
made up of vertices, edges and faces. The faces can be
represented as NF x 3 matrix of indices. The meshes
E11-3
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Fig. 3. 2-D Hilbert scanning (above) and 3-D Hilbert scanning (below)
E11-4
The 6th Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
running time in the experiment for computing 250 Harris Processing, volume 8, no. 7, pp. 964973, July 1999.
operator values. The noise which is used in the (references)
experiment is 1 to 7 and the has mean Gaussian noises [2] Harris, C., and Stephens, M., A combined corner and edge
detection, In: Proceeding of The Fourth Alvey Vision
from 0 to 100. Conference, pp. 147151, 1988.
[3] A.S. Mian, M. Bennamoun and R. Owens, A novel
representation and feature matching algorithm for automatic
pairwise registration of range image, International Journal of
Computer Vision. Copyright Springer Verlag. vol. 66, no. 1, pp.
19-40, 2006.
[4] A.S. Mian, M. Bennamoun and R. Owens, 3D model-based
object recognition and segmentation in cluttered scenes, IEEE
Transaction in Pattern Analysis and Machine Intelligence, vol.
28, no. 10, pp. 15841601, 2006.
[5] B. Taati, M. Bondy, P. Jasbedzki, and M. Greenspan, Variable
dimensional local shape descriptors for object recognition in
range data, In: Proceedings of International Conference on
Computer Vision 2007 3D Representation for Recognition
(3DRR), October 2007.
[6] Glomb, P., Detection of interest points on 3D data : extending
the Harris operator, In: Computer Recognition System 3.
Advances in Soft Computing, vol. 57, pp. 103-111. Springer,
Fig. 6. Comparison the proposed method (EHO-FQS-HSD) with Berlin, 2009.
method in the state of art, namely the Extension of Harris [7] Sipiran, I. and Bustos, B., Harris 3D: a robust extension of the
operator-Hausdorff distance (EHO-FQS-HD) for computing 250 Harris operator for interest point detection on 3D meshes, The
Harris operator values Visual Computer vol. 27, no. 11, pp. 963-976, 2011.
[8] Schmid, C., Mohr, R. and Bauckhage, Evaluation of interest
V. CONCLUSIONS point detectors, International Journal of Computer Vision.
Copyright Springer Verlag. vol. 32, no. 2, pp. 151-172, 2000.
In this paper, the interest points detection analysis on 3D [9] Vincent, E., and Laganiere, R., Detecting and matching feature
triangular meshes using extension of Harris operator points, Journal of Visual communication and Image
representation. volume 16, pp. 38-54, 2005.
combined with fitted quadratic surface (FQS) approach
[10] Moreels, P. and Perona, P., Evaluation of features detectors and
and Hilbert scanning distance (EHO-FQS-HD) is descriptors based on 3D objects, International Journal of
proposed. Computer Vision. Copyright Springer Verlag. volume 73, no. 3,
Our major contribution is the Hilbert scanning pp. 263-284, 2007.
distance application on interest points detection [11] G. Rote, Computing the minimum Hausdorff distance between
two points sets on a line under translation, Inf. Process. Lett.,
algorithm. volume 38, pp. 123-127, 1991.
We demonstrated the speed and accuracy of this [12] D.P. Huttenlocher, G.A. Klanderman, and W.J. Rucklidge,
algorithm in the presence of noise. That is to say, for Comparing images using the Hausdorff distance, IEEE
computing 250 Harris operator values the proposed fast Transaction in Pattern Analysis and Machine Intelligence, vol.
EHO-HSD is 5 10 times faster. Comparison with the 15, no. 9, pp. 850863, 1993.
[13] Garland, M., and Heckbert, P.S., Surface simplification using
conventional algorithm EHO-FQS-HD [6], our proposed quadric error metrics, In: SIGGRAPH, pp. 209216, 1997.
algorithm of EHO-FQS-HSD is superior. [14] (2010, Jan.) Queens Range Image and 3-D Model Database.
The future studies will aim at extending the [Online]. Available:
application fields, not just in matching objects, but also http://rcvlab.ece.queensu.ca/qridb/QR3D/DatabasePagexyz.htm
in 3D registration and recognition objects. l, 2009
[15] (2010, Jan.) Database for 3D object recognition in cluttered
scenes. [Online]. Available: http://www.csse.uwa.edu.au/
ACKNOWLEDGMENT ajmal/recognition.html
We would like to acknowledge: A.S. Mian dataset [16] Z.L. Cheng and X.P. Zhang, Estimating differential quantities
from the University of Western Australia and B. Taati from point cloud based on a linear fitting of normal vectors,
Journal of Science in China Series F: Information Sciences, vol.
Queens Range Image for providing 3D model range 52, no. 3, pp. 431444, 2009.
data. This research is sponsored and supported by [17] T. Tibyani and S. Kamata, Registering 3D objects triangular
Indonesian Government Scholarship (Beasiswa Luar meshes using an interest point detection on Barycentric
Negeri DIKTI-Kementerian Pendidikan dan coordinates, In: IEEE/OSA/IAPR International Conference on
Kebudayaan Republik Indonesia) Informatics, Electronics and Vision 2012 (ICIEV12), Dhaka,
Bangladesh, 18-19 May 2012.
[18] Bottema, O. On the area of a triangle in Barycentric
REFERENCES coordinates, Crux. Math. 8, 228-231, 1982.
[19] Coxeter, H. S. M. Barycentric Coordinates. in Introduction to
[1] S. Kamata, R.O. Eason, and Y. Bandou, A new algorithm for Geometry, 2nd ed, New York: Wiley, pp. 216-221, 1969.
N-dimenstional Hilbert scanning, IEEE Transaction on Image
[20]
E11-5
th
The 6 Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
Abstract Biometric is a personal information and is systems. It is more and more difficult to remember so
supposed to be unique. Biometric information can be lengthy passwords. Otherwise, by keeping the same
obtained generally from finger print or palmprint, iris, password for all systems, their passwords are prone to be
voice, or face by using such instrument and methods with
vulnerable. Moreover, some systems may even prevent
appropriate algorithms to read their raw information.
Each of biometric sources typically results in biometric people from keying the incorrect passwords in a certain
information which are different from one person to times, such as accessing a banks ATM.
another. By combining multiple biometric information Biometric is individual information and is considered
with considering security aspects, it is convincingly able unique. It is unlikely that two people have the same
to generate stronger key combinations. The key identical biometric, even for twin people. Biometric
generation is conducted by combining two or more
information can be taken from such sources as finger
biometric information with an algorithm of Longest
Common Sequences. The complexity of the number of print, palm print, iris, voice, and face. Generating key not
new combinations resulting from various combination of only includes digitizing from a biometric source, but it
two or more biometric sources is O(n2). By applying to also minimizes errors [1][2]. A single biometric has
some extent of the hamming distances between the various key lengths [1][6][7] depending on the methods.
database keys and the combination keys, the Never the less, in fact, some biometric information could
combination keys are acceptable and applicable in the use
for cryptography system.
meet a theft problem. They could be duplicated easily. A
solution to this problem and still keeping people with
Index Terms biometric, cryptography, key many keys is by using combination of biometric data.
generations, longest common sequences, combination of People are still able to have many keys without the need
multiple biometric sources. of memorizing various keys they might use. In order to
reduce, if not eliminate, this duplication problem by un
I. INTRODUCTION authorized people, combinations of biometric data are
applied to produce different but still secure unique keys.
I N recent years, the widely use of digital data and Combining data from multiple sources operates methods
data communication enabling people to deliver or algorithms and it results in various keys. By doing such
digital data among them dramatically increase. combination, it may lead to reducing percentage errors as
Obviously, digital data traversing communication media well.
are unsaved and easily eavesdropped. It is also supposed The methods in combining biometric keys to produce
to that private and personal digital data are frequently stronger unique keys are provided, such as concatenating
misused by non authenticated people. For this reasons, keys, adding keys and using Longest Common
people require an equipment to save their data or an Subsequences of keys. The first two methods will not
authentication tools of which their data are surely their reduce the errors nor keeping the length the same. The
own data. Hence, it is a need of cryptography which has last method can be considered good, as by using longest
capability of providing these requirements. common sequences bits which are the same in the
Meanwhile, people are likely to use their different sequence are preserved, whereas the not same bits in the
passwords for accessing different systems. In general, the sequence are omitted.
passwords are relatively simple and short in order to There can be error bits existing in original biometric
easily remember the passwords. For example, a user keys, which but are kept as low as possible. These error
employs one password (or key or account) to log into a bits are possibly measured as input for the combination
computer, and employs another password he/she has to process. The other thing is that there is a need to keep the
access building system. Thus, it would be difficult for key length not more than the shortest original keys, as
them to remember many passwords to corresponding quite short keys are easily broken by cryptanalysts using
E12-1
th
The 6 Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
such brute force algorithm [3]. [2]. Reducing error, in particular background noise as
If new keys resulted from this combination is to be proposed by [1] [2] was based on Hadamard transform
applied in encryption and decryption which operates the [10] (also called as Hadamard Code) which encodes
exactly same keys without any tolerances, it may lead to a (k+1) bits into 2k bits. Meanwhile, reducing burst error as
problem. It is because the combination key must be used by [1][2] was based on Reed Solomon, whose
exactly the same as the key in data base. Nevertheless, feature is that two information can be added together with
if the keys are for digital signature and allowing to some the length of n without reducing the distance [11]. Fig 2
extent of hamming distance, it would be greatly shows the process of biometric binaries into biometric
acceptable and applicable properly. information with applying both mechanisms.
The main focus of this research is to compare methods
to combine various keys with different length. It also
includes how the algorithm is built with number of
generated biometric keys into new key combinations.
However, it should be kept in mind that newly combined
keys must be long enough in order to avoiding the keys Fig 2. Biometric Information Process.
from being breakable.
However, based on some research works as in
[1][2][6][7] and [9], biometric information have been
II. PREVIOUS WORK obtained with various bit lengths. These bit length
comparisons are shown in following table.
Biometric sources are taken from part of human body.
These sources include face [1], iris [7], fingerprint [6] or Table 1. Comparison of Biometric Information Sources
palm print, and voice [9]. Examples of sources are shown Biometric Source Bit Lengths FAR1 FRR2
in Fig1. Face [1] 240 bits - 28%
Iris [2][7] 140 bits 0,005% 0,235%
Fingerprint [6] 73 bits - 1%
Voice [9] 46 bits - 20%
E12-2
th
The 6 Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
biometric template [8]. This entropy, based on Shannons III.1 Longest Common Subsequence (LCS)
entropy [12] which quantifies with regard to expected Longest Common Subsequence is to find common sub
value, measures the uncertainty regarding a random sequences from (usually two) groups of sequences. Given
sample. A biometric information source as a key sequence in X and in Y [4], and will result in LCS5. The
(comprising bits) is read many times to minimize the rule is illustrated as follows.
uncertainty of the key.
To sum up, most biometric sources, except voice,
provide appropriate key length which is suitable for
modern cryptography system. With FRR is less than 30%,
the biometric key is still acceptable.
This algorithm will show that for starting, cells in 0th
row and cells in 0th column are with value of 0. Value of
III. ALGORITHMS other cells depends on whether there is a commonness of
a character between two groups of sequences.
III.1 Strings (Bits) Matching The algorithm for LCS is composed of two main
methods. The first method, namely LCSLength(), is to
There are some algorithms or methods about how two
memo the common sub sequence characters from string X
groups of strings (or groups of bits) are to match and join
and string Y. It is precisely to mark the common character
together. The algorithms include concatenation [4],
in both sub sequences with the incremented value (in
adding two value of strings (bits) and string matching in
integer). This method is as follows.
particular using longest common subsequences [4]. The
concatenation of two group of string S1 and S2 with function LCSLength(X[1..m], Y[1..n])
length of L1 and L2 respectively, will result in a new C = array(0..m, 0..n)
string of S1 adjacently followed by S2 with the total for i := 0..m
C[i,0] = 0
length of an addition of L1 and L2. The second approach
for j := 0..n
is based on addition value, such as in ASCII4, of each C[0,j] = 0
character (or bit) of first group with value of each for i := 1..m
character (or bit) of the other group at corresponding for j := 1..n
if X[i] = Y[j]
position. This will result in another string (represented by C[i,j] := C[i-1,j-1] + 1
the resulted value) with at most length of either (L1 + 1) else:
or (L2 + 1). The last method is to match sub sequences of C[i,j] := max(C[i,j-1], C[i-1,j])
first group into sub sequence of second group, or vice return C[m,n]
5
Longest Common Subsequences is of dynamic programming.
4
ASCII stands for American Standard Code for Information Dynamic programming is not actually related to programming, but is
Interchange, which is a character encoding based on the English related to creating a table whose cells are filled with value according to
alphabet. some rules[4].
E12-3
th
The 6 Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
IV.1. Combinations
In this research, multiple biometric information as keys
are combined in such a way to generate other keys. The
combination is merely based on mathematic formula; that
is combination. Combination formula is
n!
C kn =
k!(n k )!
where :
n : number of a set of n biometric information
k : number of k biometric information of a set of n.
E12-4
th
The 6 Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
E12-5
th
The 6 Electrical Power, Electronics, Communications, and Informatics International Seminar 2012
May 30-31, Brawijaya University, Malang, Indonesia
REFERENCES [8] Jovan Dj. Golic and Madalina Baltatu, Entropy Analysis and
New Constructions of Biometric Key Generation Systems, IEEE
[1] B. Chen and V. Chandran, Biometric Based Cryptographic Key Transactions On Information Theory, Vol. 54, No. 5,, May 2008.
Generation from Faces, IEEE 2007 [9] Fabian Monrose, et al., Cryptographic Key Generation from
[2] Hao, F., Anderson, R., Daugman, J., Combining Cryptography Voice, In Proceedings of the 2001 IEEE Symposium on Security
With Biometrics Effectively, Technical Report UCAM-CL-TR- and Privacy, May 2001.
640, University of Cambridge, 2005. [10] M. Kunt, "On Computation of the Hadamard Transform and the R
[3] W. Stalling, Cryptography and Network Security: Principles and Transform in Ordered Form," IEEE Transactions on Computers,
Practices, 4th ed., Printice-Hall inc, 2006. vol. 24, no. 11, pp. 1120-1121, Nov., 1975
[4] S. Dasgupta, C. H. Papadimitriou, and U. V. Vazirani, [11] Reed, I. S. and Solomon, G., Polynomial Codes Over Certain
Algorithms, July 18, 2006. Finite Fields, SIAM Journal of Applied Math., vol. 8, 1960, pp.
[5] B. Bose, T.R.N. Rao, "Theory of Unidirectional Error 300-304.
Correcting/Detecting Codes," IEEE Transactions on Computers, [12] I.J. Taneja, Generalized Information Measures and Their
vol. 31, no. 6, pp. 521-530, Jun., 1982 Applications , 2001,
[6] Ulrike Korte, et. Al, A cryptographic biometric authentication http://www.mtm.ufsc.br/~taneja/book/node1.html
system based on genetic fingerprints, 2007, [13] U. Uludag, S. Pankanti, S. Prabhakar and A. K. Jain, Biometric
http://www.secunet.de/fileadmin/Downloads/Englisch/Sonstiges/s cryptosystems: issues and challenges, Proceedings of the IEEE,
ecunet_Fachartikel_BioKey.pdf. Vol. 92, No. 6, pp. 948960, 2004
[7] Padma Polash Paul, and Md. Maruf Monwar, Human Iris
Recognition for Biometric Identification, IEEE, 2007.
E12-6
GRATITUDE
WE WOULD LIKE TO THANKS TO ALL THAT DESCRIBED BELOW WHO HAVE GIVEN US INVALUABLE SUPPORTS
AND PARTICIPANTIONS
AND OTHERS....