Documente Academic
Documente Profesional
Documente Cultură
I am very happy to note that the University College of Engineering and Technology,
Acharya Nagarjuna University is organizing a International Conference on Navigational
Systems & Signal Processing Applications (NSSP-2013) during December 13th & 14th, 2013. I
understand that the above international conference will promote the areas of study, teaching
and research activities in field of technical education to the participants and budding
technocrats and they will prove their skills equipped with human and social values required to
provide integration of technical skills, social character, leadership, self confidence, creativity
and renovation for mankind and all round growth of an individual. I feel the importance of
troubleshooting in our day-to-day work and for preparing the future academic and
administrative action plan. The Research and Development plays a very crucial role in the
institutional developments. No one can deny the role of technology and It is needless to
emphasize that all round growth of the institution will be possible only when we focus our
attention on state of art teaching and research.
RECTORS MESSAGE
I am very happy to note that the University College of Engineering and Technology,
Acharya Nagarjuna University is organizing a International Conference on Navigational
Systems & Signal Processing Applications (NSSP-2013) during December 13th & 14th, 2013.
The conference is concerned with the latest in theoretical, mathematical and scientific
developments in Communications Engineering as well as applications of established techniques
to new domains. In the modern world, this type of conference is very relevant to meet the
emerging fields. I believe that this conference will provide an International Technical forum for
experts and researchers from both the academia and industry to meet and exchange new ideas
and present their findings of ongoing research in various disciplines of Electronics &
Communication Engineering. I am confident that the research papers of the international
conference provide an opportunity to rapidly disseminate research results that are of timely
interest to the engineering fraternity. I warmly welcome all the participants for the fruitful
technical sessions. Hearty congratulations to the organizers and best wishes for the success of
the International conference.
REGISTRARS MESSAGE
My Good wishes to all the participants and I wish the Seminar a grand success.
Prof. R. R. L. Kantam
26th Nov 2013
MESSAGE
The themes identified for discussions during the conference are very relevant to the
field of Navigation and Signal processing. I am also confident that this conference will prove to
be an excellent platform for the scientists, engineers and research scholars from various R&D
organizations, Academia and Industry to interact and exchange their concepts and ideas during
the event.
I am certain that the conference will be amply beneficial to the delegates and
participants since most of the emerging techniques and technologies in the field of Navigation
are proposed to be addressed adequately.
DEANS MESSAGE
FOREWORD
I greet and welcome the participants to our institution and to this conference with the
hope that, together we will be able to churn out new ideas and suggestions in the area of
Navigational Systems & Signal Processing Applications I do believe that the conference will
bear fruitful results and lay firm ground for future developments. I hope that each participant
will go back with a feeling of academic satisfaction by being actively involved in the discussions
and deliberations.
At first i would like to thank University authorities for their unstinting coordination and
cooperation. With regard to the conference, i would like to convey special thanks to the
Honorable Vice-Chancellor of this University, Prof. K. Viyyanna Rao, for this inspiration and
continuous encouragement to organize this International Conference Successfully. I thank
Rector, Prof. Y. P. Rama Subbaiah and Registrar, Prof. R. R. L. Kantam for their continuous
support to conduct the conference fruitful.
I also thank Prof.U.R.Rao, former ISRO chairman an outstanding Scientist with a vision to
be acting as a conference chair and for his guidance at each and every step, to bring the
conference on the stage. I would like to thank Prof. K. Giridhar, IIT, Madras, an Eminent
Academician to be with us as a Technical program Chair.
I would like to extend my heartfelt gratitude to Sri.G. Satheesh Reddy , Director RCI for
his immense support to conduct this conference in a grand way. I also thank Prof.U.B. Desai,
Director IITH ,Prof.Rameswar Rao ,Vice Chancellor of JNTUH. Sri.N. Prahalad Rao, Director ISRO,
Prof Raveendran Parmeswaran , University of malaya and Dr.V.Seshagiri Rao, Associate
Director, SHAR ISRO, Sri. P. Raghavendra Rao, Associate Director, CHES, Sr. G. J. Das, Former
Scientist ISRO, Sri. Jayaram Scientist F , ISTRACT, Dr. R.Sreehari Rao ,former chief controller,
DRDO for their continues support and acceptance to be the Guests of Honour for this
international conference.
I thank IEEE for readily agreeing to be the Technical sponsor of this International
Conference. I would like to extend my thanks Dr. M. Lakshmi Narayana, IEEE for Readily
agreeing to be the technical sponsor. I thank all the sponsors UGC, DRDO New Delhi, ISRO,
Effectronics PvtLtd, Vijaya Entuple technologies, AP State Council of Higher Education, DLD
Logistics,Banglore, Accord-SoftSystems etc. for their financial support to the Conference.
Finally, I thank the advisory, organizing and Technical committees for their commendable job in
realizing this goal.
A total of 300 paper proposals were received from all over the world, in the form of "full
papers". After extensive screening, 90 papers were selected for presentation at the Conference.
These were divided into 15 sessions and sessions would run parallel. We have also included
invited tasks from experts within the country and abroad. I hope that will certainly bring in a
standard and originality to the Conference.
Finally, I take this opportunity to thank all the authors and invited speakers profusely for
contributing and participating in the conference.
I welcome all the participants to imbibe the spirit of the originality and technical content
of the Conference.
I wish the Conference a grand success and I hope this will be a trend setter for future
Conferences.
Dr. P. Siddaiah
CONVENER
NSSP-2013
UNIVERSITY COLLEGE OF ENGINEERING & TECHNOLOGY
ACHARYA NAGARJUNA UNIVERSITY
NAGARJUNA NAGAR - 522 510, GUNTUR(DT), A.P., INDIA
PRINCIPALS MESSAGE
In this global scenario we are all aware that Indias secured future lies in making India a
complete knowledge nation. Keeping that in view, we are organizing an International
Conference on Navigational Systems & Signal Processing Applications (NSSP-2013) during
December 13th & 14th 2013. To strengthen India in the field of Science and Technology. The
International Conference also aims to bring in the unexplored areas of research in different
aspects of Navigational Systems & Signal Processing Applications in RADAR, Satellite imaging
and Medicine.
SESSION-II : ANTENNAS-I
S.No Paper Title and Author Page No
Design of Non-Uniform Circular Arrays for Side lobe Reduction Using
1. the Method of Real Coded Genetic Algorithm 39
Ms.M. Nirmala
Effect Of Dielectric Superstrate On The Characteristics Of Square
2. Patch Antenna 44
V. Saidulu, K. Srinivasa Rao, P.V.D. Somasekhar Rao
Analysis On Flexural Vibration Characteristics of Hemispherical Shell
3. 52
Shailesh kumar,Anil badiger, Kishore chandra das, Runa banerjee
Effect of dielectric superstrate on the characteristics of Rectangular
4. Patch antenna 57
V. Saidulu, K.Srinivasa Rao, P.V.D.Somasekhar Rao
Design of Printed Dipole antenna Integrated with Resonator and
5. Defected Ground Structure (DGS) for improving noise rejection 65
Takeshore Khumanthem, C.Sairam, S.D.Ahirwar and M.Santha
Design of Dual Frequency Circular Patch Antenna Using HFSS
6. 70
Akhila John. Davuluri, Dr.P.Siddaiah
SESSION-XIV: ANTENNAS
S.No Paper Title and Author Page No
Electronically Steerable Planer Phased Array Antenna using HFSS
1. 401
Vasujadevi M, Dr. P. Siddaiah, P.Sai Vinay Kumar
Analysis and Performance Comparison of DOA Algorithms for Continuous
Tracking And Quality Enhancement Of Desired Signal In Wireless Services
2. 406
With Smart Antenna Technology
M.Siva Ganga Prasad, KVR Chowdary, P.Siddaiah, P.Sai Vinay Kumar
Design and Analysis of C-Band Rectangular Microstrip Patch Antenna
3. 414
B. Sekharbabu , BV. Subba Rao , I. Kullayamma, Dr.R.V.S. Satyanarayana
The Dielectric Loss Tangent Effect on Radiation Characteristics of Co-
4. axially Feed Rectangular Patch Antenna: Design & Analysis 418
Ch. Raja, K. Praveen Kumar, Dr. M. Madhavi Latha, Dr.E.G.Rajan
AbstractThis paper presents the design of an array one bit of the multiplier. These partial products are shifted
multiplier, Wallace Tree multiplier and Booth in accordance to their bit orders and are then added
multiplier. Due to the advancements of technology in together [3]. The addition operation is performed with the
the area of signal processing, graphics, image and video Carry Propagate Adder. Number of adders required for the
processing etc. multipliers play an important role in purpose of multiplication is N-1 where N is the multiplier
these fields. Due to the numerous applications of length. The Basic multiplication process is shown in Fig. 1.
multipliers in different areas, the research is going in
this direction to achieve highly efficient, fast, low cost
multipliers. This paper presents the software design of
an array multiplier, Wallace Tree Multiplier and Booth
Multiplier with the help of flow charts. These
multipliers are implemented using Xilinx ISE Design
suite 14.2 and thus compared with each other based on
different criteria like number of look up tables used,
number of input output blocks used and others.
I. INTRODUCTION
1
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
multiplier is from LSB of the input to the MSB of the but the carry input in Full Adder becomes one of the input
product. The delay generated is 2*n. in CSA and the carry output from each bit becomes one of
the output along with the sum in CSA rather than being
moved to the next MSB [4]. It takes three numbers (i, j, k)
as input to add together and gives carry (c) and sum (s) as
output. In carry save adder, the The carry and sum is
brought to the last stage and are then added together by
ordinary addition. Carry Save Adders are used to obtain
partial products in integer or bit wise multiplication, whic
allows these adders to be used in those architectures where
a tree of carry save adders are needed and hence producing
fast partial product terms. Carry bits and the last partial
products are added by the normal adder to give the final
result of multiplication. For this last stage usually a carry
select adder or a carry look ahead adder is used for optimal
performance of the multiplier. By using carry save adder,
carry propagation in the adder is avoided due to which one
Figure 2. Basic Structure of Ripple Carry Array Multiplier gate delay is reduced because the gate delay of adder is
equal to the latency of one addition.
2) Carry Save Array Multiplier- A carry save array The multiplier implementation basically involves three
multiplier uses a carry save adder to add three or more steps as shown in Fig. 4. They are Partial-Product
binary numbers of n-bits. In ripple carry array multiplier generator, Wallace Tree structure and ordinary adder.
there was only one output but in carry save-array
multiplier there are two outputs as shown in Fig. 3. One is
partial sum bits and other is carry bits. These two bits are
then added together to produce the final result.
2
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
complement form. Let us illustrate the booth algorithm TABLE II. STEP 2 IN BOOTH ALGORITHM
with the following example that multiplies (3)10 x (-6)10.
1) Making the Booth table- From the two numbers pick the
number with the largest difference between a series of
consecutive numbers and make it a multiplier. i.e., 0011-
from 0 to 0 no change, 0 to 1 one change, 1 to 1 no
change, and so there is only one change in this one. 1010--
from 1 to 0 one change, 0 to 1 another change, 1 to 0 one
more change, so there are total three change on this one.
Hence, multiplication of 3 x (6), where 310 (0011)2 is the
multiplicand and (-6)10 (1010)2 is the multiplier.
Let X = 1010 (multiplier)
Let Y = 0011 (multiplicand)
Take the 2s complement of Y and call it Y.
-Y = 1101
Load the X value in the table. Load 0 for X-1 value it
should be the previous first least significant bit of X. Load
0 in U and V rows which will have the product of X and Y
at the end of operation. Make four rows for each cycle
because we are multiplying four bits numbers [8].
2) Booth Algorithm:-
Booth algorithm requires examination of the multiplier
bits, and shifting of the partial products [9]. Before the
process of shifting, the multiplicand may get added to the
partial product, get subtracted or left unchanged in
accordance to the following rules: Look at the first least The four cycles are finished, so the answer is obtained in
significant bits of the multiplier X,and the previous least the last rows of U and V which is: 111011102,
significant bits of the multiplier X -1.
V. IMPLEMENTATION OF MULTIPLIERS
0 0 Shift only
1 1 Shift only This section of the paper covers the implementation
0 1 Add Y to U and then shift of
1 0 Subtract Y from U and then shift or add (-Y) to multipliers with the help of flow chart.
U and shift
1) Array Multiplier
Take U & V together and shift arithmetic right shift which a) Method 1 to implement array multiplier :- The steps to
preserves the sign bit of 2s complement number such that carry out an array multiplication are written as below.
positive number remains positive and a negative number 1. Initiate P=00000000, A= 00001001, B= 1101 and n=4
remains negative. Shift X circular right shift because this 2. Check for if B (0) =1 or not. Here in this case B (0)=1 so
will prevent from the use two registers for the X value. P= P + A then P becomes 00001001 and move to next
step. If B (0) 1 then directly move to next step.
3
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
4
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
VI. RESULTS
5
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
REFERENCES
6
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract The main objective of this paper is inference system was presented to control a steam
controlling the speed of DC motor when the load is engine and boiler combination by linguistic rules.
varying. Various methods of speed control of speed Fuzzy logic is expressed by means of if-then rules
control of DC motor is available. This paper presents with the human language. In the design of a fuzzy
design and implementation of Fuzzy Logic in the speed logic controller, the mathematical model is not
control of DC motor. Fuzzy logic has found high necessary. Thus, the fuzzy logic controller owns good
applications as a speed control techniques because of its robustness. Fuzzy controller has been widely used in
ability to take into account vague and uncertainties . industry for its easy realization. However, the rules
this paper presents a MATLAB simulink model for and the membership functions of a fuzzy logic
speed control of DC motor using fuzzy logic. in this controller are constructed by expert experience or
paper TMS320F28335 DSP is used as the controller knowledge database.
necessary signal conditioning components are used to
ensure high processing speed and precision in the II. MODELING OF DC MOTOR
overall control system. The implemented system has a
A common actuator in control systems is the DC
motor. It directly provides rotary motion and, coupled
fast response with small overshoot and zero steady state
with wheels or drums and cables, can provide
error compared to conventional PI controller.
translational motion. The electric circuit of the
I. INTRODUCTION armature and the free-body diagram of the rotor are
shown in the following fig.1:
In spite of the development of power electronics
resources, the direct current machine became more
and more useful. Now a days their uses isnt limited
in the car applications (electrics vehicle), in
applications of weak power using battery system
(motor of toy) or for the electric traction in the multi-
machine systems too.
The speed of DC motor can be adjusted to a great
extent as to provide controllability easy and high
performance. The controllers of the speed that are
conceived for goal to control the speed of DC motor Fig. 1.Seperatly Exited DC Motor
to execute one variety of tasks, is of several For this example, we will assume that the input of
conventional and numeric controller types, the the system is the voltage source (V) applied to the
controllers can be: PID Controller, Fuzzy Logic motor's armature, while the output is the rotational
Controller; or the combination between them Fuzzy- speed of the shaft d(theta)/dt. The rotor and shaft are
Genetic Algorithm, Fuzzy-Neural Networks, Fuzzy- assumed to be rigid. We further assume a viscous
Ants Colony, Fuzzy-Swarm (Swarm). friction model, that is, the friction torque is
Fuzzy theory was first proposed and investigated proportional to shaft angular velocity.
by Prof. Zadeh in 1965. The Mamdani fuzzy
7
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
T s sJ . s B. s
J-moment of inertia 0.01 kg.m2
Kt -torque constant 0.052 Nm/A m m m
Kb-electromotive force constant 0.1 V/rad/s
B-viscous
T s
friction coefficient 0.004
N.m/rad/s
In general, the torque generated by a DC motor is s sJ B
m
m
7
proportional to the armature current and the strength
of the magnetic field. In this example we will assume
that the magnetic field is constant and, therefore, T s K .i s
m T a
8
that the motor torque is proportional to only the
armature current i by a constant factor Kt as shown in
the equation below. This is referred to as an E s k . s
b b m
9
armature-controlled motor. The angular acceleration is equal to 1 / J multiplied
Here, the differential equation of armature circuit is- by the sum of two terms (one positive, one negative).
di t t
Similarly, the derivative of current is equal to
E t R I t L 1
a 1/ L multiplied by the sum of three terms (one
a a a E a b positive, two negative).The final design should look
d t like the example shown in the figure below. The
following fig.2. shows the modeling of DC motor,
The Torque equation is- which is obtained from the above mathematical
analysis using MATLAB/Simulink.
d t
T t J
m
m
B. m t 2
d t
T t k i i
m f f a
3
Where , Kf is the constant. In armature-controlled D.C
motor the field current (if) is kept constant i.e.
T m
K T .i a 4 Fig. 2. Modeling of DC Motor
8
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
control system. Thus, it can be one of the best The fuzzy logic controller for speed controller for
available answers today for a broad class of DC motor was implemented in MATLAB/Simulink.
challenging controls problems. A fuzzy logic control Block diagram show the simulation model.
usually consists of the following:
Fuzzification: This process converts or transforms
the measured inputs called crisp values, into the fuzzy
linguistic values used by the fuzzy reasoning
mechanism.
Knowledge Base: A collection of the expert control
rules (knowledge) needed to achieve the control goal.
Fuzzy Reasoning Mechanism: This process will
perform fuzzy logic operations and result the control
action according to the fuzzy inputs.
Defuzzification unit: This process converts the result
of fuzzy reasoning mechanism into the required crisp
Fig. 5.Modeling of Fuzzy Logic Controller
value
The above fuzzy logic controller modeling has two
subsystems. One is DC motor, which shows fig.2.
And another is IGBT switch, shown in the following
fig.6.
9
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
E. Rule View
F. Surface View
10
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
V. CONCLUSION
ACKNOWLEDGMENT
Firstly, Author would like to pay Special thanks
to Department of Science & Technology (DST),
Ministry of Science and Technology, Government of
India, New Delhi, and thanks to Faculty and
Management of MIC College of Technology,
11
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Kanchikacharla, and also to the government of India, Devineni Venkataramana & Dr.HimaShekar MIC College
for encouragement. of Technology, Kanchikacherla ,A.P, India.
AUTHORS BIOGRAPHY
12
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract:-Data acquisition (DA) is the process of UAV Firstly, this paper introduced the structure and
sampling signals that measure real world physical working principle of system. Then the design of
conditions and converting the resulting samples into system's hardware and software programming were
digital numeric values that can be manipulated by a described, as well analyzed configuration of every
computer. The communication between them is
module and communication of serial ports Ethernet
standard asynchronous serial form by typical serial
control card, which have complicated cable wiring, connectivity is achieved with small Ethernet adapter
frequently interruption of CPU, which greatly reduces (serial to Ethernet) UART to Sensors that convert
the CPUs efficiency and impacts the systems real-time physical parameters to electrical signals[3]. Data
processing. To resolve these problems, a method of acquisition is the process of sampling signals that
multi-serials to Ethernet Gateway based on the field measure real world physical conditions and
programmable gate array (FPGA) with network converting the resulting samples into digital numeric
interface chip is implemented in this project. values that can be manipulated by a computer.
Finally, the timing simulation of two serial ports of
The Gateway will send data as Ethernet frame
UART was shown in Chip scope modules compared
format after receiving serial data, indirectly achieves
multi-serials communication, simplifies cabinet wiring with traditional methods, it possesses the characters
and improves CPUs efficiency. In this project we use of high-integration, low-power, and flexibility [5] ,
WIZ net W5100 Hardwired TCP/IP Embedded [6].
Ethernet Controller chip will be interfaced to FPGA to
provide Ethernet interface. Various digital signals will II. ARCHITECTURE DESIGN
be captured by FPGA and will be sent on a serial line.
The FPGA implements all the necessary logic to read
the data from these sensors and store the data in it. On
FPGA logic also will be implemented to read data from
multiple numbers of serial ports. The WIZ Net module
takes data from serial port and sends to PC in Ethernet
form. In PC application will be developed to read data
from Ethernet. The code will be developed in VHDL.
converter is a high speed serial interface that As shown in above figure Receiver section is a
interfaces easily to FPGAs. The A/D interface combination of a baud generator and receiver section
adapter (AD1_PMOD) is implemented within the module, each module has been explained below in
FPGA. Inside the FPGA, this adapter facilitates detail [3].
parallel data acquisition. Sampling is initiated at the
rising edge of a clock applied at the line sample. The Baud generation section is a clock divider circuit,
timing diagram of the communication protocol FPGA board clock runs at 50MHz, but UART
obtained with Modelsim is illustrated in Fig.2. transfers data at predefined standards that had to be
maintained, in present system is designed for a rate of
9600 bits/sec (i.e. 50x106 is scaled down for 9600). It
generates 9600 pluses per sec; this implies the speed
of UART is 9600 bits per sec [3]. Another clock is
with a 16 times faster than required, and it is given to
the receiver section, so that the data will not be
corrupted [4].
Fig. 2 PMOD connection diagrams
D. Receiver Section:
B. UART Block:
UART receiver handles reception of data from
The UART is a serial interface with a frame RS232 port. Main functions of receiver block are to
format of start bit of active low 0at beginning of convert the serial data to parallel data, and check the
frame and 8 bit of information with a stop bit of correctness of data from parity and store the
active high1 signal at the end The operation of received data. UART receiver state machine is shown
UART is controlled by Clock signal which is fed in Figure 5. The receiver is in IDLE state by default.
from external crystal [15]. When the serial data pin goes low, indicating the start
bit, the state machine enters DATA0 state. The data is
received; one bit at a time from LSB to MSB in states
DATA0 to DATA7. If parity is enabled, the state
machine checks the parity bit received against the
parity obtained from received data. If the data
received is fine, the data_Rx (data_Rx_done) bit is
set to 1 and the receiver goes back to IDLE state
again.
C. UART Receiver:
E. Enable Generator:
14
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
signal as a output signal. And this enable_ out signal format form, and that data is transmitted to the
will be given as a input signal to the transmitter destination system in the form of Ethernet packet
section as a enable input signal. This signal is used to .By Team Tera Pro software we observed the
synchronize the transmitter section, when the data is transmitted data in the output PC.[7]
to transfer is defined by this signal [18].
F. Transmitter Section:
Fig.8 Ethernet module
The transmitter block is responsible for the
transmission of serial data from UART. It takes 8-bit F. RS-232:
data from the receiver section in parallel and send
data in serial form. In Telecommunications RS232 (Recommended
Standard 232) is the traditional name for a series of
Data is inserted between start and stop bits. An standards for serial binary single-ended data and
optional parity bit also may be used for error control signals connecting between a DTE (Data
detection. State machine for transmitter is shown in Terminal Equipment) and a DCE (Data Circuit-
Figure 7. Transmitter stays in IDLE state unless terminating Equipment). It is commonly used in
transmit enable (tx_enable) is made as 1. The data computer serial ports. The standard defines the
transmission starts with tx_enable = 1. As mandated electrical characteristics and timing of signals, the
by the protocol, a 0 is transmitted to indicate start of meaning of signals, and the physical size and pin out
transmission or start bit. This is done in START state. of connectors.
Then data bits 0 to 7 are transmitted in states DATA0
to DATA7. If parity is enabled in configure ration
register, the data is attached with parity in PARITY
state. Then transmitter enters STOP state and sends
1 as stop bit. This indicates the completion of
transmission. Then the transmitter enters the IDLE
state and waits for next data transmission.[18]
Fig.9 Telecommunications RS232
15
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
16
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Waveform Description: Data is ready on the data _rx bus to transmit to the
next block transmitter section this will in the active
Transmitter section consists of clk , high(1) state. Else in active low state (0).
BDclk_Posedge, reset , data_in , data_ready as a
input and outputs are Send_data , tx . when rst equals
to 1, the data _in set to all zeros and present state set
to S0 (PS = S0).when clk and BDclk is active and on
raising edge and when data ready is one (1) the data
is transmitted serially from one state to next state.
When tx and send equals 1 ,s0 set to 1 ,s1 set to
zero as start bit and next eight states the data is
transmitted and the tenth state (s10) is set to 1 to
represents the end of data transmission. i.e stop
bit.s10 and s0 bit will one until the next data
transmission is identified.
Fig.15 Simulation Wave form of Top module
Waveform Description:
17
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Screen shot for UART Transmitter: system in real time. FPGAs flexible programming
features also allow further upgrade for system. This
design could be used in the domain of net control and
information management system. It has a certain
value for application. Low cost, as the multi card
solution can come in single FPGA card with small
modules around it, Low power in comparison with
multi card solution, Lot of scope for adding
additional functionality on FPGA.
ACKNOWLEDGMENT
REFERENCES
18
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Author Biography
19
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract WiFi (Wireless Fidelity) is a wireless MAC layer provides a CSMA/CA (Carrier Sense
technology based on a cellular architecture where system Multiple Access/Collision Avoidance). The optional
is subdivided into cells. WiFi is the wireless way to handle PCF based on the contention free service is suitable
networking. It is also known as 802.11 networking and for delay sensitive traffic. Due to technology
wireless networking. The big advantage of WiFi is its
advancement in the 21St Century, wireless
simplicity. You can connect computers anywhere in your
home or office without the need for wires. The computers Communication had been most popular choices of
connect to the network using radio signals, and computers communication. More and more people are turning to
can be up to 100 feet or so apart. wireless due to the convenience of mobility. An
802.11 LAN is based on a cellular architecture where
WiFi transmitter implements the Basic Service set (BSS) the system is subdivided in to cells, where each cell
and controlled by the base station called access point. [called Basic Service Set or BSS] is controlled by a
These access points are connected with distribution Base Station [called Access point, or in short
system. The transmitter is responsible for transmission AP].Even though that a wireless LAN may be formed
and generation of various entities like Preamble includes
by a single cell, with a Single Access Point[ can also
synchronization and Start frame delimiter (SFD), PLCP
Header used to decode the frame consists of PLCP_PDU work without an Access Point], most installations
Length word, PLCP Signaling Field and Header Error will be formed by several cells, where the Access
Check Field and Mac Data Frame. But compared with Points are connected through some kind of back
wired network, wireless communication is having low bone[ called Distribution System or DS],typically
speed, less range, less reliable, cost effective and less Ethernet, and in some cases wireless itself.
security. We can design and implement some protocol
standards to overcome the disadvantages of wireless II. IEEE 802.11 STANDARD
communication. In this project we are designed and
implemented MAC (Medium Access Control) layer
functions such as authentication, association, probe Due to technology advancement in the 21st
request and cyclic redundancy check as hardware to Century, wireless Communication had been most
increase speed and security of WiFi WLAN popular choice of communication. More and more
communication. This Design coding, Simulation, Logic people are turning to wireless due to the convenience
Synthesis and Implementation is done using Xilinx ISE. of mobility. An 802.11 LAN is based on a cellular
The designed hardware was tested by simulating in the
architecture where the system is subdivided in to
Model Sim 6d and was successfully implemented in the
SPARTAN3E FPGA board using Xilinx ISE as a cells, where each cell [called Basic Service Set or
synthesis tool BSS] is controlled by a Base Station [called Access
point, or in short AP]. Even though that a wireless
Keywords:- WiFi, IEEE 802.11, BSS, SFD, PLCP, MAC, LAN may be formed by a single cell, with a Single
WLAN, Xilinx ISE. Access Point [can also work without an Access
Point], most installations will be formed by several
cells, where the Access Points are connected through
I. INTRODUCTION some kind of back bone [called Distribution System
or DS], typically Ethernet, and in some cases wireless
In recent year, the popularity of WLAN (Wireless it self. The whole interconnected wireless LAN
Local Area Network) has generated much interest on including the different cells, their respective Access
improvement and performance analysis of the Points and the Distribution System, is seen to the
IEEE802.11 protocol. The IEEE 802.11 protocol upper layers of the OSI model, as a single 802
defines the MAC (Medium Access Control) and the network and is called in the standard as Extended
physical layer functions of WLANs. The WLAN Service Set [ESS]. The standard also defines the
MAC protocol employs two medium access methods concept of a portal, a portal is a device that
for packet transmission which are the DCF interconnects an 802.11 and another 802 LAN [3].
(Distributed Coordination Function) and the PCF However, all is not prefect in the WLAN world.
(Point Coordination Function) .We focus on the Offering nominal bit rates of 11Mbps [802.11b] and
20
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
54Mbps (802.11a and 802.11g) the effective MAC LAYER Frame Format:
throughputs are actually much lower-owing to packet
collisions, protocol overhead, and interference in the The basic format for all packets sent by the MAC
increasingly congested unlicensed bands at 2.4GHz layer. Some actual packets do not actually contain all
and 5GHz. Furthermore, operation in these bands of the fields.
entails a strict regulatory transmit power constraint,
thus limiting range and even bit rates beyond a
certain distance [5].
21
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
CLKGENERATION BLOCK: The clock generation So each and every frame transmission in transmitter
block is used to generate 1MHz clock frequency from of WLAN MAC is transferred only if the previous
50MHz that is available from the board. This frames response is received correctly. HEADER
generated clock is easy to use to the backoff block BLOCK: The header block defines the type of
and difs counter blocks which are sub multiples of frame that is transmitted, the contents of the frame
1MHz clock frequency. has
type, subtype values that define which frame body to
DIFS BLOCK: The DIFS counter block is operated transmit, and also the add1 recipient address of 6
by the main control machine which initiates the bytes
counter, the estimated period for the DIFS with and the transmitter address which is MAC address of
respect to PHY layer considering 802.11b is 50 micro transmitting node, duration of 2 bytes.
seconds, so the counter is made work with 50 as
counter under clk_1MHz clock frequency to obtain a PROBE BLOCK: The probe block has the contents
delay of 50 micro seconds. As soon as the block gets as
completed it intimates the control machine for further SSID 32 bytes and supported data rate and extended
back off operation. data rates of 8 bytes that are need to transmit in a
sequence.
BACKOFF BLOCK: This block is operated only
when the phy_idle signal that is provided by the AUTHENTICATION FRAME BLOCK: The
PHY layer becomes logic 0 before the DIFS gets authentication block transmits the frame body of the
completed. The start of backoff is initiated by control auth frame byte by byte. The content algorithm no,
machine, and initially the random value is taken as authentication transaction number status code and
30, that is defined by number of time slots. In this challenge text, that are send in sequence of bytes
case the slot time is taken as 20 micro seconds. through the block to MPDU block to PHY layer.
ASSOCIATION FRAME BLOCK: The association
CONTROL BLOCK: Back off state is an important frame
operation in wireless LAN MAC, that decides which block capability information, listen interval ,SSID of
node to access the channel, this is executed if the 32
phy_ilde becomes low within DIFS counter period bytes, supported rates are transmitted in a sequence of
else not, in our case the state transition has moved byte by byte to MPDU block.
DIFS to backoff state to test the states transition
operation. MPDU BLOCK: The MPDU block is a multiplexer
block that pass the bytes of information from each
The further states are for the transfer of the header frame block to the physical layer only one at a time,
frames and probe frames , auth frames association depending on the selection signals that are generated
frames and rts and data frames ,that are processed in from the control state machine.
sequence, by enabling the header block with
header_En so that the data_in byte of input that get 4.3 Finite State Machine
transfers framing it as the header frame through the
MPDU multiplexer block to the physical layer, very The below figure can represents the Functional
similar to the header block is the probe block is Diagram of Finite State Machine.
22
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Figure 4.3 Functional Diagram of Finite State Machine Figure 5.2 DIFS block diagram
IV. DESIGN IMPLEMENTATION After the end of DIFS and SIFS the outputs are given
as 1 to DIFS end and SIFS end.
Different Modules in the Project with Explanations:
5.1 CLOCK GENERATION BLOCK: 5.3 BACKOFF BLOCK:
23
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
24
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
multiplexer block. ASSO_END is the output port that the crc for bytes of information and the crc_stp is
gives the information of the completion of the used for stopping the crc calculation after the last
association frame to the control state block. byte is transmitted. The crc_out is 32bit of value that
CRC_STP_ASS is the output port that gives is shifted bytes in 4 times into the frame block to the
information for the CRC gen block to give the CRC MPDU block.
data to the ASSO_FRAME. Control state machine
has been implemented for the sequence of association 5.11 DATA TRANSMISSION BLOCK:
frame transmission. Each and every content is
specified will resemble a state that will transmit the
contents through the output 8 bit port. After the final
transmission of data AUTH_END is enabled that
gives the info to the control state machine.
25
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Figure 5.13 Frame Control block diagram Figure 6.2 RTL Schematic-1 of Top Module
6.1 Resources Utilized on FPGA Result Figure 6.4 RTL Schematic of input Module
26
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Figure 6.9 Simulation result for DIFS Figure 6.14 Simulation result for RTS
27
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
VII. CONCLUSION
ACKNOWLEDGMENT
Figure 6.16 Simulation result for Frame Enable We would like to thank the Organizations of
Acharya Nagarjuna University, Guntur, for support to
use R&D Laboratories.
REFERENCES
28
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
I. INTRODUCTION
DTMF signalling has many applications such as The transmitter of a DTMF signal simultaneously
telephone dialling, data entry, credit checking, and sends one frequency from the high-group and one
voice mail system control. A DTMF signal Consists frequency from the low-group. This pair of signals
of two superimposed sinusoidal waveforms with represents the digit or symbol shown at the
frequencies chosen from a set of eight standardized intersection of row and column in the Fig. 1.
29
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
For example, sending 1209Hz and 770Hz indicates including the extreme case of dial-tone being
that the "4" digit is being sent. The below figure sent by the central office at 0 dBm (which may
represents the basic DTMF Communication System. occur in extremely long loops). Above 600Hz,
any other signals detected by the receiver must
be at least -6 dB below the low-group frequency
signal strength for correct digit detection.
The values shown are those stated by AT&T in
Compatibility Bulletin 105. For compatibility with
ANSI T1.401-1988, the minimum inter-digit interval
Fig. 2: DTMF communication system shall be 45msec, the minimum pulse duration shall be
50msec, and the minimum duty cycle for ANSI-
At the transmitter, the maximum signal strength compliance shall be 100msec.
of a pair of tones must not exceed +1 dBm, and the
minimum strength is -10.5 dBm for the low - group Methods for Generating DTMF tones:
frequencies and
-8.5 dBm for the high-group frequencies. When the DTMF generation can be done by different
intentionally not sending DTMF tones (including the methods for analog and digital systems.
inter-digit interval), any leakage of these tones must Use special IC for generating DTMF tones.
not exceed -55 dBm. The frequencies generated by Modems and telephones use this method.
the transmitter must be nominally within 1.5% of the Generate DTMF tones using soundcard FM
stated values and the receiver must not accept signals synthesizer chip.
that deviate more than 3.5% from the stated Load sine wave sample to wavetable soundcard
values[8],[9]. memory. Play that sample using two instrument
channels at different frequencies.
The receiver is responsible for performing several Sample all DTMF tone combinations headed and
checks on the incoming signal before accepting the playback those samples as needed. 8 kHz at 8 bit
incoming signal as a DTMF digit: resolution is enough for that.
Generate the sample data which is played back
Energy from a low-group frequency and a high- using software.
group frequency must be detected. Our project is concentrating on the look-up table
Energy from all other low-group and all other method, because it is the most generic way to do
high-group frequencies must be absent or less the DTMF generation. You can use this method
than -55dBm. with every sound card which can play back
The energy from the single low-group and single samples and it is as well suitable for DSP
high-group frequency must persist for at least implementations also [5].
40msec*.
There must have been an inter-digit interval of at III. GENERATING DTMF
least 40msec* in which there is no energy
detected at any of the DTMF frequencies. The
minimum duty cycle (tone interval and inter-digit The another methods for the generation of the
interval) is 85msec*. DTMF tones will be
The receiver should receive the DTMF digits Polynomial approximation
with a signal strength of at least -25 dBm and no Look-up table
more than 0 dBm. Recursive oscillator
The energy strength of the high-group frequency
Dual tone generation can be done with 2 sine
must be -8 dB to +4 dB relative to the energy
wave sources connected in parallel. Different method
strength of the low-group frequency as measured
can be used for such implementation:
at the receiver. This uneven transmission level is
DTMF signal must meet certain duration and spacing
known as the "twist", and some receiving
requirements:
equipment may not correctly receive signals
where the "twist" is not implemented correctly. 10 Digits are sent per second.
Nearly all modern DTMF decoders receive Sampling is done via a codec at 8Khz.
DTMF digits correctly despite twist errors. Each tone duration must be > 40msec and a
The receiver must correctly detect and decode spacing of 50ms Minimum between two
DTMF despite the presence of dial-tone, digits is required.
30
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
DTMF generator refers to any electronic device or other auxiliary telephonic devices that can decode
capable of generating dual tone multi-frequency signals produced by a DTMF generator. Telephones
(DTMF) signals used for dialling touch-tone equipped with DTMF-based touch-tone dialling allow
telephones. The DTMF dialling system operates by consumers to access their account balances at banks,
configuring a keypad in a three column by four row as well as retrieve other information from
grid representing the numbers 0 through 9 and other organizations that provide a telephonic interface
special characters such as the # and * keys. The touch based on the DTMF system.
pad grid is arrayed by assigning a specific high
frequency for each column and a designated low A computer equipped with a sound card and digital
frequency for each row [3]. When a user presses a audio software is capable of acting as a
digit on the keypad, a dual tone is generated that DTMF generator by producing the dual tone multi-
represents the intersection of the two distinct frequencies used on touch-tone telephone keypads
frequencies assigned for each column and row on the [3],[4].
grid. For example, when the "2" digit is pressed on
the keypad, a unique audio sinusoidal tone is emitted The software program interprets the phone digits
by mixing a pure high frequency of 1,336 Hz with a and then generates the appropriate audio files that
pure low frequency of 697 Hz. The DTMF signal conform to the standardized DTMF paired frequency
tone generated by the keypad is then transmitted to tones. The dialling process can be activated either by
the local office of the phone company where it is a mouse click or through keyboard commands. This
translated or decoded into the appropriate allows for rapid automated computer dialling of
corresponding digits which allow the designated outgoing telephone numbers and is used by many
phone number to be dialled. Therefore the tones in businesses [4].
keypad are generated by taking one of the frequency
from the row selector and the other frequency from IV. GOERTZEL ALGORITHM
the column selector. These two frequencies combine
and give the dual tone at output. The Goertzel algorithm is a digital signal
processing (DSP) technique for identifying frequency
The block diagram of DTMF tone generator will components of a signal, published by Dr. Gerald
be the first commercial application of Goertzel in 1958. While the general Fast Fourier
a DTMF generator was the introduction of the touch- transform (FFT) algorithm computes evenly across
tone pad dialling system, first offered by AT&T to its the bandwidth of the incoming signal, the Goertzel
customers in 1963. The touch-tone dialling system algorithm looks at specific, predetermined
was demonstrably faster and more convenient than frequencies, Some applications require only a few
the mechanically based rotary dial pulse system, DFT frequencies. One example is frequency-shift
which it supplanted. After its introduction, touch-tone keying (FSK) demodulation, in which typically two
became the standard dialling system for telephones in frequencies are used to transmit binary data; another
the United States and was eventually adopted example is DTMF, or touch-tone telephone dialing, in
worldwide [13]. which a detection circuit must constantly monitor the
line for two simultaneous frequencies indicating that
a telephone button is depressed [15].
Goertzel algorithm reduces the number of real-
valued multiplications by almost a factor of two
relative to direct computation via the DFT equation
[6].
For a length of N, the Goertzel's series is
31
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
VII. CONCLUSIONS
32
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
33
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract I. IMPLEMENTATION
This paper presents a method to describe all modulation A. Binary Phase Shift Keying (BPSK)
techniques (like: BPSK, ASK, FSK, QPSK, QAM) on Binary Phase Shift Keying (BPSK) [1], [4], [6], [7]
Field Programmable Gate Array (FPGA) development demonstrates better performance than ASK and FSK.
board which is widely available and inexpensive. To PSK can be expanded to M-array scheme, employing
develop the system blocks Simulink environment and
multiple phases and Amplitudes as different states.
system generator version 13.1 are used under MATLAB
version 7.11 (R2010B). To achieve simulation and Filtering can be employed to avoid spectral
synthesis of Spartan 3 FPGA tools from Xilinx ISE 13.1 spreading.
are used. Very High Speed integrated circuit hardware
description language (VHDL) is used for describing the
hardware in system understanding language. Digital to
Analog converter is used to interface both FPGA and
CRO which is used to visualize the analog output of the Fig. 1. BPSK modulation.
digitally modulated signal. If the received data is logic 1, then the modulated
KeywordsAmplitude Shift Keying (ASK),Binary Phase Shift signal has no phase shift, if the received data is logic
Keying (BPSK),Frequency Shift Keying (FSK), Quadrature 0, then the modulated signal has 1800 phase shift as
Phase Shift Keying (QPSK), Digital Communication, Field shown in figure 1[1], [6].
Programmable Gate Array (FPGA), Receivers, Transmitters.
1. Simulink block diagram for BPSK using system
I. INTRODUCTION generator
34
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
35
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
multiplexer is
1. QPSK Constellation Used, if the given data is shown in table 2[3].
2.
Carrier
Data transmitted Carrier phase S. Action BPSK ASK FSK QPSK
amplitude
o N
00 225 1.0
01 135o 1.0 1 Control bit 00 01 10 11
10 315o 1.0 2 Informatio 180,0 No change Carries 225,135,
11 45o 1.0 n phase amp f1,f2 315,45
,zero amp phase
Table 1. QPSK phases. 3 Parameter Phase Amplitude Frequency Quadrature
phase
135 o o
45
4 Carrier one one two Two
Table 2. Control Unit.
225 315o
o
Fig. 10. QPSK constellation.
To get the accurate output as shown in figure 12, the To get the accurate output as shown in figure 13, the
parameters for each block must be configured parameters for each block must be configured
precisely in the Simulink/system generator precisely in the Simulink/system generator
environment [2], [5]. environment [3], [5].
3. Scope result for Quadrature Phase Shift 2.scope result for control unit
Keying(QPSK)
36
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
B. hardware
The complete lab measurement setup used for
realizing all kind of modulators is illustrated in
figures 15, 16. Some of the resources use there:
Spartan 3 starter kit board [8], [10], Digital to Analog
converter (NIFCO7A) with PQ28 package, Function
generator, Regulated power supply from Aligent and 3. Simulated result for FSK
Cathode Ray Oscilloscope.
C. Complete setup
37
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
REFERENCES
38
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract - This paper is dealth with design problem other systems. In this Real Coded Genetic Algorithm
of non-uniform circular antenna arrays for maximal (RCGA) techniques is applied to design of non-
side lobe level reduction. This problem is modeled as uniform circular antenna arrays.The method of Real
a simple optimization problem. In this case, the Coded Genetic Algorithm (RCGA) is used to
antenna array[1] design problem consists of finding determine an optimum a set of weights that provide a
weights that provide a radiation pattern with maximal radiation pattern with maximal side lobe level
side lobe level reduction. Due to the great variety of reduction[4].
parameters involved, optimization techniques[11]
such as Real Coded Genetic Algorithm (RCGA)[5] II. ARRAY FACTOR ANALYSIS
are very appropriate tools to search for the best
antenna models. Genetic Algorithm techniques are a. Introduction to Real Coded Genetic
becoming widely used to solve electromagnetic Algorithm (RCGA)
problems due to their robustness, wide range of
applications and readiness in their implementation. The aim of antenna array geometry synthesis is to
The effectiveness of Real Coded Genetic Algorithm assess the physical setup of the array that produces
(RCGA) for the design of non-uniform circular arrays the radiation pattern that is nearest to the desired
is shown by means of experimental results. pattern. There are many problems in electrical
Experimental results reveal that design of non- engineering for which analytical solutions do not
uniform circular antenna array using the method of exist. In fact, most practical electromagnetic problems
Real Coded Genetic Algorithm (RCGA) provides a are nonlinear and non-differentiable. These problems
considerable side lobe level reduction with respect to require engineers and scientists to employ numerical
the uniform case. The purpose and contribution of methods. These methods provide no insight into the
this project is to present a model of problem that global optimum solution of the problem. Thus, the
includes design of non-uniform circular antenna numerical solvers resort to marching parameters up
arrays for side lobe level reduction using the method and down until a global optimum is obtained. These
of Real Coded Genetic Algorithm (RCGA). brute force methods are time consuming and
unintuitive. Real Coded Genetic Algorithm (RCGA)
I. INTRODUCTION stochastically evolve a population towards a solution
using the concepts of survival of the fittest. These
The design of circular antenna arrays[12] finds algorithms allow the global optimum, which is the
application in areas as mobile and wireless goal of all optimization techniques, to be attained far
communications systems. Generally speaking, the faster than the traditional brute-force techniques.
problem of designing antenna arrays is characterized
by different and conflicting requirements (beam b. Sequence of operation in Real Coded Genetic
width, side lobe level, directivity, noise sensitivity, Algorithm (RCGA)
robustness) to be satisfied. In this project a design 1. Generate an Initial Population
criterion is considered to evaluate the performance of Randomly generated quantized genes
circular array for minimum side lobe level[4]. (binary)
Each gene contains the genetic code of
Real Coded Genetic Algorithm (RCGA) techniques each individual
have been fairly successful at designing linear 2. Evaluate Fitness (or Objective) Function,
antenna arrays. However, array configurations in f(x)
which the elements are placed in a circular ring are of quality of individuals in the population
great interest. They have applications in radio
direction finding, air and space navigation, radar, and
39
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Use numerical solver (i.e. NEC, Wire) implemented instead of a common weighted roulette
to calculate properties such as wheel selection.
Directivity, Bandwidth, etc.. The function update population assigns ranks to
3. Select the best members of the population individuals in the population generated by the union
based on Fitness of parents and children. This is in order to hold the
There are many methods to select the best individuals in each generation. The explanation
best members of the procedures involved at each step of this
4. Crossover or mate the individuals in the algorithm in detail.
population
This occurs with a probability pcross
(.6 < pcross < .9 )
Choose two members, swap their
genetic codes to create two new
members.
5. Mutate the individuals
This occurs rarely with a probability
pmut (0 .001 < pmut < 0.01)
This allows changes in the population
to not only slow the convergence rate,
but to also allow unique members to
appear.
6. Evaluate Fitness again
7. If the population has converged to a Fig.1: Geometry and notations used for non-uniform
superior race, end the process, otherwise circular antenna array
loop back and continue
The individual representations as well as the
c. Problem Statement crossover and mutation operators are explained in the
following subsections.
Design of a Non Uniform circular array that provide a
III. ARRAY FACTOR
radiation pattern with maximal side lobe level[10]
reduction.Real Coded Genetic Algorithm is applied to
Consider a circular antenna array of N antenna
design of Non-Uniform circular antenna arrays.The
elements uniformly spaced with a distance d on a
method of Real Coded Genetic Algorithm is used to
circle of radius a in the xy plane as shown in below
determine an optimum a set of amplitude excitation
figure 1. If the N elements in the circular antenna
weights to provide a radiation pattern with maximal
array are taken to be isotropic sources, the radiation
side lobe level reduction. Initially Uniform circular
pattern of this array can be described by its array
radiation pattern is generated for number of elements
factor.
N=10 and the sidelobe level for that is calculated.
The array factor for the circular array in the xy plane
Design a Non-Uniform circular for N=10 and 20 in figure 1 is given by[2]
with maximum SLL for different set of amplitudes.
These different set of element amplitude
[ ( ) ]
excitations[3] are generated by using Real Coded ( , )=
Genetic Algorithm. The main purpose is to design a
low side lobe radiation pattern for non-uniform
circular antenna arrays. For this purpose, a Where
population-based stochastic procedure is used In = Excitation of nth element
denominated Real Coded Genetic Algorithm. I chose N= no.of elements in circular ring
this algorithm for its easiness of implementation. The
k= phase constant
procedure for used Real Coded Genetic Algorithm is
described as follows. a= where d =
The function generate initial population randomly and Then ka =
uniformly generates a set of individuals. The main
idea in classify individuals is to rank the individuals n =
according to their fitness values. A selection scheme Where n = 1,2,.N
combining fitness ranking and elitist selection is n = ka cos ( 0 )
40
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
I = [I1, I2,., IN], In represents the excitation of parameters[8], after a trial and error procedure, were
the nth element of the array, k=2/ is the phase set as follows:
constant, is the angle of incidence of a plane wave, maximum number of Generations G = 200
is the signal wavelength and 0 is the maximum Chromosome size c = 80
radiation angle. Gene size b = N
Now need to formulate the objective function we
Where N =10 and 20.
want to minimize. Let us introduce first a couple of
definitions for the radiation pattern[1]. Let 0 be the Crossover probability pc = 1.0 and
angle where global maximum is attained in = [-, Mutation probability pm = 0.1
]. Let msl be the angle where the maximum side lobe
is attained.Based on the above considerations the
objective function[2][7] can be written as
F 1= |AF(msl, I)| / | AF(0, I)|
Where
msl = The angle where the maximum side
lobe is attained
0 = The angle where global maximum is
attained
V. EXPERIMENTAL SETUPS
41
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
0.7
0.6
In the present work, a uniform circular array[9] is
Amplitude
0.5
0.1
array, the amplitude excitation coefficients are
0
optimized by using a Real Coded Genetic Algorithm. 1 2 3 4 5
Element number
6 7 8 9 10
Real Coded Genetic Algorithm is applied to a circular Fig.5:Amplitude distribution for N=10 using
array for different number of array elements for N=10 Real Coded Genetic Algorithm
and 20.
The amplitude excitation coefficients are synthesized The amplitude excitation coefficients and their
by using Real Coded Genetic Algorithm. By using respective Radiation patterns are presented in figure 5
the amplitude excitation coefficients, patterns are and figure 7. Computations are carried out for both
numerically computed. small and large arrays and the computed amplitude
excitation coefficients are presented in Table 1. The
1 reduced SL Level and Beamwidths are also presented
0.9 in the table.
0.8 0
0.7
-5
0.6
Amplitude
0.5
-10
Array factor (AF)in dB
0.4
0.3 -15
0.2
-20
0.1
0
1 2 3 4 5 6 7 8 9 10 -25
Element number
-30
-150 -140 -130 -120 -110 -100 -90 -80 -70 -60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150
theta(in degrees)
Fig.3: Amplitude Distribution for N=10
Fig.6: Radiation pattern of circular array for
N=10 using Real Coded Genetic Algorithm
0
-5
-10 0.9
Array factor (AF)in dB
0.8
-15 0.7
0.6
-20
0.5
Amplitude
0.4
-25
0.3
0.2
-30
-150 -140 -130 -120 -110 -100 -90 -80 -70 -60 -50 -40 -30 -20 -10 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150
theta(in degrees) 0.1
42
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
REFERENCES
0
[1]. G.S.N.Raju, Antennas and Wave Propagation, Pearson Edu,
-5
2005.
[2]. Marco A. Panduro, Aldo L. Mendez, Rene Dominguez,
-10 Gerardo Romero Design of non-uniform circular antenna
Array factor (AF)in dB
43
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract: A dielectric superstrate or radome is above a light weight, low profile, conformable to planar and
square patchmicrostrip antenna has a remarkable effect non-planar surfaces and easy to fabricate. The
on its characteristics such as gain, bandwidth, antenna is suitable for high speed vehicles,
beamwidth, radiation pattern and resonant frequency. aircrafts, space crafts and missiles because of Low
This paper experimentally investigates the effect of a
profile and conformal nature of characteristics [2].
dielectric superstrtes with and without on the
performance characteristics of coaxial probe fed square Microstrip antenna has inherent limitation of narrow
patchmicrostrip antenna with varying the various bandwidth. So, superstrate(radome) is used
thickness of dielectric superstrates on gain, bandwidth, on a microstrip antenna as a cover to protect the
beamwidth, and radiation pattern and resonance antenna from external environmental conditions like
frequency. The antenna designed frequency is temperature, pressure etc. When microstrip antenna
2.4GHz(ISM band) and fabricated on Arlondiclad 880 covered with a dielectric superstrate(radome) its
substrate, whose dielectric constant () is 2.2, loss properties like resonance frequency, gain,
tangent (tan ) is 0.0009 and the thickness (h) is 1.6mm bandwidth and beam width are changed which may
and the 1 same substrate can be used as dielectric
seriously degrading the antenna performance[1-4].
superstrat whose dielectric constant () 2.2 ,at
various thickness such as 0.2mm, 0.5mm, 0.8mm,
By choosing the thickness of the substrate
1.0mm, 1.3mm, 1.5mm, 2.2mm, 2.4 and 3.2mm on the superstrate layer, a very large gain can be achieved
patch antenna with same specification. Experimental [5-9]. Coaxial probe fed squaremicrostrip
measured result observed that when the dielectric antenna characterstics have been investigated using
superstrate is above the substrate, the antenna resonant High Frequency Structure Simulator (HFSS) softer
frequency is altered and shifted towards the lower side and measured experimentally. When microstrip
of frequency of operation from 2.4GHz to 2.33 GHz, antennas are covered with protective dielectric
while other parameters have slight variation in their superstrates, are subjected to icing conditions, or
values. The gain of single patch antenna without come into contact with plasma, the resonant
dielectric superstrate is 4.8 dB and the microstrip patch
with dielectric superstre is varied from 0.47dB to
frequency is altered and shifted to lower sides,
3.43dB based on the thickness of the superstrate. The causing detuning which may seriously degrading the
bandwidth of microstrip single patch without dielectric antenna performance. As the bandwidth and gain of
superstrate is 4% and the microstrip patch with microstrip antennas is inherently low, typically of
dielectric superstrte is 1.0% to 2.0% based upon the the order of 1-2% [1]. It is important to determine
thickness of the dielectric superstrtes. The beamwidth the effect of dielectric superstrate on the resonant
for single patch without 2 dielectric superstrte in E- frequency of the microstrip antennas and in order
Plane is 108.160 and in H-Plane is 90.20. The beamwdth introduce appropriate corrections in the design of
with dielectric superstre in E-Plane is 95.40to 105.330 microstrip antennas. In this paper experimentally
and in H-Plane is 74.860 to 83.610. The value of VSWR
is from 1.38 to 2.80 based upon the thickness of the
measured the effect of dielectric superstrates on the
dielectric superstrates,as dielctricsuperstrat thickness antenna parameters such as bandwidth, beamwidth,
increases, VSWR increases, return loss increases, gain gain, resonant frequency and radiation pattern with
and bandwidth decreases. varying variousthickness of dielectric superstrates
Keywords: Square patch microstrip antenna, dielectric etc.
superstrate, VSWR, Gain, Beam- width, Bandwidth,
resonant frequency etc. II. ANTENNA SPECIFICATIONS:
44
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
(tan )
A high loss tangent increases the dielectric
loss and therefore reduce the antenna
performance. The substrate and superstrate
parameters chosen for designing the patch antenna as
follows: The substrate material is ArlonDiclad 880
dielectric constant ( 1 ) is
is substrate and whose Fig 4. Square patch antenna with superstrate
2.2, loss tangent (tan )
is 0.0009 and thickness
(h) of the material is 1.6mm. The superstrate
materials as same as substrate, dielectric
constant is 2 = 2.2and whose substrate dimension
is
100mm100mm and same specification.
The low dielectric constant materials increase
efficiency, bandwidth and better for radiation. The
dielectric constant of the substrate and superstrateis
shown in table 1 and table 2 and also substrate
and superstrate material is shown in Fig 5. Structure of square patch with dimension.
Fig 8.
45
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
FRINGING EFFECTS:
The q factor is referred to as the fringing factor
The amount of fringing is a function of the (length reduction factor). As the substrate height
dimensions of the p atch and the h e i g h t of the increases, fringing also increases and leads to larger
s ub str a te . Since fo r microstrip antennas L/H >>1, separations between the radiating edges and lower
fringing is reduced; however, it must be taken into resonant frequencies.
account because it influences the resonant
frequency of the antenna. The effective dielectric DESIGN:
1 < <
constant has values in the range of most
applications where the dielectric constant of the
greater than the unity ( 1), the
substrate is much
value of will
be closer to the value of the actual
dielectric constant
of the substrate.
CONDUCTANCE:
46
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
or
Where the plus (+) sign is used for modes with odd
(ant
symmetric) resonant voltage distribution beneath the
patch and between the slots while the minus (-) sign V. MICROSTRIP ANTENNAS COVERED WITH A
is used for modes with even (symmetric) resonant DIELECTRIC SUPERSTRATES:
voltage
distribution. The mutual conductance is defined, in When microstrip line (antenna) is covered with the
terms of the far- zone fields, as dielectric superstrate or radome the characteristics of
antenna parameters change as a function of the
dielectric superstratelayer. The properties of a
Where 1 is the electric field radiated by slot 1, 2 is microstrip antenna covered with dielectric superstrate
the magnetic field radiated by slot 2 , 0 is the layer have been studied theoretical formulationusing
voltage across slot, and the integration is performed the transmission line and cavity model. The resonant
over a sphere of large radius. It can be shown that 12 frequency of a microstrip antenna covered with
can be calculated dielectric superstrate layer can be determined when
the effective dielectric constant of the structure is
47
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
48
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
49
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
50
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
and with superstrate is 2.80, the gain valve without gain 3.43dB and 0.2mm thicknes is best for higher
dielectric superstate is 4.8dB and with dielectric BW is 2%.
superstrate is varied according to thickness from
0.47dB to 3.43dB, the bandwidth value without REFERENCES:
dielectric superstrate is 4.0% and with dielectric
superstrate is 1.0%2.0%.We observed experimental 1. IE3D Manual, Zeland software Inc.,Fremount, USA, 1999
measured data in tables 3, 4 and 5 and also gain, 2. I J Bhal and P Bhartia, Microstrip antennas, Artech
radiation pattern in vertical and horizontal plane and house, 1980.
3. R.Shavit,Dielectric cover effect on Rectangular Microstrip
VSWR plot with and without superstrate is shown in Antennas array. IEEE Trans. Antennas propagat.,Vol 40,. PP.992-
Fig 9 to Fig 22.The discussion we conclude that in 995, Avg.1992.
the following points. 4. Inder ,Prakash and Stuchly, Design of Microstrip Antennas
1. Variation of VSWR with different dielectric covered with a Dielectric Layer. IEEE Trans. Antennas Propagate.
Vol.AP-30.No.2, Mar 1992.
superstrate (radome) thickness, as dielctricsuperstrate 5. O.M.Ramahi and Y.T.LO,Superstrate effect on the Resonant
thickness increases, VSWR increases. frequency of Microstrip Antennas, Microwave Opt.Technol. Lett.
2. Variation of Gain at different dielectric Vol.5, PP.254-257, June 1992.
superstrate(radome) thickness as dielectric superstrate 6. A.Bhattacharyya and T. Tralman, Effects of Dielectric
Superstrate on patch Antennas, Electron Lett., Vol.24,PP.356-
thickness increase, the gain decreases. 358, Mar 1998.
3. The antenna Beamwith in E-Plane increases from 7. Patil V.P, Kharade A.R Enhancement of directivity of RMSA
95.20 degree to 105.33dregree and the antenna Beam using multilayer structure, IJERD,79-84, 2012.
width in H-Plane increases from 74.86 degree to 8. Patil V.P, Kharadea.r Enhancement of Gain of RMSA using
multilayer structure, IOSRJECE,2278-34,2012.
83.61 degrees. 9. M..Younssi,A.Jaoujal Study of MSA with and without
4. The return loss first increases with increasing superstrate for Terahz frequency, ISSR Journal, 2013.
thickness of dielectric superstrates and then decrases. 10. L. Yousefi, H. Atta High gain patch antenna loaded with high
5. The bandwidth of the microstrip antennas also chr. Impedance superstrate, vol.10, 858-861, 2011.
11. S.D.Gupta, A. Singh, Design and analysis of multidielectric
increases with increasing thickness of dielectric layer MSA with varying superstrate layer chracterstics, IJAET,
superstrate for low dielectric constant materials, and vol.3, pp. 55-68, 2012.
decreases for high dielectric constant of the materials. 12. R.K Y adav, R.L.Yadave Effect on performance char. Of
rectangular patch antenna with varying height of dielectric cover,
IJPCSC, vol.2, no.1, ISSN: 0976-268X.
13. H.Attia, L.Yousefi and O.M.Ramahi Analytical model for
X. CONCLUSION: calculating the radiation fields of MSA with artificial
maganeticsuperstrates: Theory and experiment. IEEE Tranc.
The design of square patchmicrostrip antennas with Antennas and wave progation, vol.59,2011.
14. M..Younssi,A.Jaoujal Study of MSA with and without
dielectric superstrates have been presented. The superstrate for Terahz frequency, ISSR Journal, 2013
resonant frequency of a microstrip covered with
dielectric superstrates can be predicted accurately if
the effective dielectric constant of the structure is
known. The effective dielectric constant can be
calculated using transmission line and cavity model.
The experimentally results shows that the variation of
VSWR with different dielectric superstrate(radome)
thickness, as dielectric superstrate thickness
increases, VSWR increases. The variation of the
antenna gain at different dielectric superstrate
thickness as dielectric superstrate thickness increases,
the gain decreases. The bandwidth of the microstrip
antennas can also increases with increasing thickness
of the dielectric superstrate for low dielectric constant
materials, and decreases for high dielectric constant
of the substrate materials. Initially the return loss
increases with increasing thickness of dielectric
superstrates and then decreases. The antenna
beamwidth in E-Plane increases from 95.2 degree
to105.33 degree and beamwidth in H- Plane increase
from 95.20 degree to 105.33 degree and among
various thickness 2.2mm thicknes is best for higher
51
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract Mechanical gyros based on hemispherical right and left in the northern and southern
structure and working on the principle of Coriolis effect hemispheres, respectively. The effect was first
play a vital role in the navigation systems for very long observed by Gaspard-Gustave Coriolis, a French
range missions and deep space probes The precessional scientist who in 1835 described the forces that arise
phenomenon and flexural vibration characteristics of from the motion of an object in a rotating reference
hemispherical resonator shell for such a navigation frame [2].
application are investigated in this paper. Thin shell
theory and virtual displacement principle are employed Hemispherical Resonator Gyro (HRG), as a kind
to model hemispherical shell. The influence of shells of Coriolis Vibratory Gyros (CVG), is the only CVG
geometry parameters on its natural frequency and that meets the inertial grade performance. HRG is
precessional rate is analyzed. Depending on ABAQUS- solid state gyroscope whose sensing property is based
finite element analyzer, the influence of shells geometry on standing vibration precession. It has features of
parameters on each free frequency is analyzed for high accuracy, long life span, inherent high reliability,
optimal Q-factor. natural radiation hardness and small size.HRG is one
type of vibrating gyros that can work as an angle
KeywordsHemispherical Shell; flexural vibration; q- sensor as well as an angular rate sensor depending on
factor; resonant frequency.
its mode of operation. The two operation modes are
I. INTRODUCTION Whole Angle (WA) mode and Force to rebalance
mode (FTR). In WA mode, the standing wave is
Gyroscope is one of the most important sensors in allowed to precess freely with respect to the Coriolis
navigation applications. It is applicable to several forces due to external rotation. In FTR mode the
areas such as aircraft, satellite, and oil drilling- standing wave is fixed at a specific location by
exploration and so on. However, classical gyros are applying the rebalance force. The angular rate can be
composed of many mechanical moving parts which estimated by the force applied to fix standing wave. In
can cause the operation failure. In that sense, vibratory general, WA mode has relatively larger dynamic
gyroscope is relatively effective sensor since it has no range with relatively low accuracy while FTR mode
moving parts which mean that it has longer operation has shorter dynamics range with high measurement
time compared to classical mechanical gyros. accuracy [3].
Working principle of the vibratory gyros is based on
the Coriolis Effect. If the external rotation of the I. MODEL OF RESONATOR
platform occurs while the vibratory gyro is at its
resonance, this external rotation causes the precession
Dynamic Precession Analysis
of the gyros. By measuring the precession angle, one
can measure the amount of external rotation.
The Hemispherical Resonator Gyro, HRG, has its
notional roots back in the late 19th century when
G.H.Bryan had his seminal paper, On the beats in the
vibrations of a revolving cylinder or bell published in
the proceedings of the Cambridge Philosophical
Society in 1892 [1]. In this paper he mathematically
described the nature of the beats that are heard as a
result of rotating a vibration shell about its symmetric
(or cylindrical) axis. In a vibrating shell such as a
wineglass, rotation of the wineglass about its stem will
cause the nodes of vibration on the rim of shell to
move at precession rate slower than shell itself. The
physics of HRG is based on the forces arising from The basic dynamics of shell for the second
Coriolis acceleration. This is the same force that resonant mode, n=2, half period is shown in
drives direction of winds and ocean currents to the fig1.Assuming the shell rotates at an angular velocity
52
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
, analyzing a small element at a point A: its
amplitude and nullify the quadrature of vibration
without creating drift. The undesirable generated drift
transport velocity is V1 and Coriolis acceleration is depends on two characteristics:
Wk 2 V1 . For an element m, if it receives a The modulus of the forces needed to
maintain the vibration and nullify the
Coriolis acceleration Wk , it can generate an inertia quadrature which are defined by the physical
characteristics of the resonator (damping and
force Fk with same magnitude but in opposite anisofrequency) and directly linked with the
direction. For other elements the elastic deformations precision of manufacturing process.
are similar. The Coriolis acceleration and inertia force The errors generated by electronics used to
are shown in fig.1. Due to the inertia force caused by
estimate the position of vibration and to
, shadow region is compressed, while blank area is apply the forces to the resonator.
extended [4]. The result is precessional phenomenon: The displacement vector of an arbitary point on
the vibration wave is not standing anymore, but a the shell is given by:
negative relative angular velocity with . ua v b wc
(1)
Co-ordinates and Equations of Vibration Mode:
A model for a thin axis-symmetrical Where:
hemispherical shell is shown in fig 2. It has two Y in
u, v,w are the magnitude of the displacement
orthogonal modes of vibration that are coupled by the generatrix, ring, and radial directions,
Coriolis forces. The basic equation of the system is respectively[8].
X
the following:
fx a , b , c are the unit vectors in corresponding
x 2. x . x . x x2 . x 2. .. y directions.
m
By detailed analysis of inertia force F, it can be
fy derived that: when resonator rotates at an angular
y 2. y . y . y . y 2. .. x
2
y
m velocity of , x yz , if 2 2 ,
/ t 2 , the influence of original stress can
In this equation, x and y stands for the be neglected, the force caused by angular acceleration
position of the vibration in each of the two modal doesnt affect the vibration mode.
direction, x and y are the pulsation of each mode, The vibration of a rotating hemispherical resonator
x and y are the damping of each mode, f x and which is assumed to be isotropic can be formulated as:
u ( , , t ) u ( ) cos n ( ) cos t
f y are the forces applied to each mode, m is the
v ( , , t ) v ( ) sin n ( ) cos t
modal mass, is the rotation rate and is a w ( , , t ) w ( ) cos n ( ) cos t
coefficient called Bryan Coefficient. (2)
The solution of such a system is an ellipse of
amplitude a, angle , phase and quadrature b. Where:
is the precession angle,
t
0
Pdt , P K x
K is precession coefficient.
n is number of ring waves
is resonator shell frequency.
In above equation we can see that Coriolis
forces are homogeneous to a damping force making The kinetic energy of the shell can be expressed as:
the angle of the vibration to rotate. This is the / 2 2
1 2
K r h ( u v 2 w 2 ) sin d d
2
physical phenomenon that allow to measure
rotational rates. 2 0 0
53
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
54
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
/2
If the stem radius R2decreases, each order of
I sin 3 tan 2 n d ;
2
frequencies decreases. On the other hand if
0 the stem radius increases each order of
/2
(2 cos )
frequencies increases, but the distance
J 2
2 sin 2 sin tan 2 n d between neighbours are shortened.
0
2
(13)
IV.FINITE ELEMENT ANALYSIS
Modal Analysis
ABAQUS-Finite Element Analyzer is utilized for
analysis of Hemispherical Shell as per the geometry
parameters of the resonator are shown in fig.3.
CONCLUSION
Fig.4 Illustration of shells modes shapes
The resonator achieves different vibrating shapes This paper investigates the vibrational and
on a series of frequncies. As the number of the ring precessional mode of the hemispherical shell.
waves n increases, the frequncies increases too, it will Utilizing the virtual displacement principle the
be difficult to excite the resonator. As the number of kinetic equation of hemispherical shell is analyzed.
ring waves increases, vibrating shape become The influence of shells geometry parameters on its
complex, detection of signal will be more complex. natural frequency and precession rate are analyzed as
analytic solution of model. Utilizing ABAQUS-Finite
Geometry parameters influence on frequencies Element Analyzer, the influence of shells geometry
By changing the geometry parameters, several finite parameters on each free frequency is investigated [5].
element models are studied and analyzed utilizing The analysis in this paper may be useful in the
Abaqus development of Hemispherical Shell for optimal Q-
When the radius of shell R1 gets smaller, factor and in the analysis of source of error in
each series of frequencies become higher. performance evaluation.
The shell thickness has notable influence on ACKNOWLEDGMENT
each vibration mode. As the shell grows
thicker the n=2 mode frequencies increasebut
the distance between neighbours are The authors are grateful to the Research Centre
shortened. Imarat of Defence Research and Development
55
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
REFERENCES
56
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
57
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
58
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
59
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
over a sphere of large radius. It can be shown that V. MICROSTRIP ANTENNAS COVERED
12 can be calculated WITH A DIELECTRIC SUPERSTRATES:
Where,
= Effective dielectric constant with dielectric
superstrte
0 =Effective dielectric constant without dielectric
superstrate
RESONANT FREQUENCY: = Change in dielectric constant due to dielectric
superstrate
For the rectangular patch antenna is shown in Fig1, =Fractional change in resonance frequency
the resonant frequency f, can be accurately =Resonce frequency
60
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
61
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
62
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
63
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
In order to verify the experimental results for the thickness increases, the gain decreases. The
rectangular microstrip antenna with and without bandwidth of the microstrip antennas can also
dielectric superstrates. The antenna designed increases with increasing thickness of the dielectric
frequency 2.4GHz fabricated on Arlondiclad 880 superstrate for low dielectric constant materials, and
substrate and the substrate dimension is 100 decreases for high dielectric constant of the substrate
100 , whose dielectric constant(= 2.2), loss materials. Initially the return loss increases with
tangent(tan )is 0.0009, increasing thickness of dielectric superstrates and
thickness(h) is 1.6mm. The dielectric superstrate is then decreases. The antenna beamwidth in E-Plane
same as the substrate whose dielectric constant is 2.2, increases from 84.20 degree to 9.94 degree and
loss tangent is 0.0009; thickness is 1.6mm.We beamwidth in H-Plane is increases from 67.91 degree
observed the resonating frequency is shifting to lower to 77.63 degree.The amongvariouethickness at
side 2.40 GHz to 2.33GHz, while other parameters 1.3mm thicknes will give higher gain 6.0dBand
have slight variation in their values. The gain of higher BW 5%.
single patch antenna without dielectric superstrate is
7.3 dB and the microstrip patch with dielectric XI. REFERENCES:
superstre is 3.3dB to 6.0dB. The bandwidth of
microstrip single patch without dielectric superstrate 1. IE3D Manual, Zeland software Inc.,Fremount, USA,1999
2. I J Bhal and P Bhartia, Microstrip antennas, Artech house,
is 2% and the microstrip patch with dielectric 1980.
superstrte is 2. % to 5.0% based upon the thickness of 3. R.Shavit,Dielectric cover effect on Rectangular Microstrip
the dielectric superstrtes. The beamwidth for single Antennas array. IEEE Trans. Antennas propagat.,Vol 40,. PP.992-
patch without dielectricsuperstrte in E-Plane is 88.360 995,Avg.1992.
4. Inder ,Prakash and Stuchly, Design of Microstrip Antennas
and in H-Plane is 90.20 . The beamwdth with covered with a Dielectric Layer. IEEE Trans. Antennas Propagate.
dielectric superstre in E-Plane is 84.200 to 9.940 and Vol.AP-30.No.2,Mar 1992.
in H-Plane is 67.910 to 77.630. The value of VSWR is 5. O.M.Ramahi and Y.T.LO, Superstrate effect on the Resonant
from 1.6935 to 2.2373 based upon the thickness of frequency of Microstrip Antennas, Microwave Opt.Technol. Lett.
Vol.5, PP.254257,June 1992.
the dielectric superstrates. The variation of VSWR 6. A.Bhattacharyya and T. Tralman, Effects of Dielectric
with different dielectric superstrate thickness, as Superstrate on patch Antennas, Electron Lett., Vol.24,PP.356-
dielectric superstrate thickness increases, VSWR 358, Mar 1998.
increases. The variation of gain at different dielectric
superstrate thickness, as dielectric superstrate
thickness increases, gain decreses. The beamwidth in
E-Plane decreases from 90.94 degree to 90.50 degree
and in H-Plane increases from 70.71 degree to 71.80
degree. The bandwidth of microstrip antenna is
increases with increasing thickness of dielectric
superstrates for low dielectric constant and decreases
for high dielectric constant.The experimental
measured results is shown in table 4, 5 and 6 and
corresponding measured VSWR and radiation pattern
plot for various thickness is shown in Fig 8 to Fig 19.
X. CONCLUSION:
64
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract:- In this paper design of printed dipole outside its band. Lower spectra to ISM band is the
antenna integrated with ring resonator and DGS is GSM-1800 and above it is the UWB band. To further
presented along with measured results. The ring improve the fidelity of data transfer rate, an antenna
resonator bandpass filter is coupled with DGS to can be designed to rejects the above unwanted bands.
provide wide band rejection characteristics. This
Recently, a numbers of research papers and literature
antenna shows high signal rejection outside the required
frequency band of 2.4GHz to 2.5 GHz (~ ISM band), on antenna with band rejection characteristics are
thereby improving the noise rejection in the unwanted published. To achieve band rejection, they either used
band. The measured VSWR of the antenna is less than different shaped slots or filters [1][2][3].In this paper
1.5:1 over the band. The antenna shows good radiation a printed dipole antenna in ISM band (2.4-2.5 GHz)
efficiency and omni-directionality. The antenna is is designed. This antenna is coupled with ring
simulated using CST microwave studio and then resonator and DGS to achieve wide band signal
practically realized on RT duriod substrate. Thus, the rejection outside the desired frequency band without
proposed antenna finds applications in trans-receive degrading its overall radiation efficiency and omni-
modes in todays wireless arena where high signal
directionality.
selectivity is in demand.
Keywords: Printed dipole, ISM band, ring resonator, II. DESIGN APPROACH
DGS, noise rejection.
A printed dipole antenna with center frequency
I. INTRODUCTION 2.45 GHz was designed. The length of the dipole is
taken less than /2 in order to cancel the inductive
Printed antennas are emerging as preferred choice reactance. The antenna is integrated with printed
among the different antenna designs for various balun as shown in figure 1[4]. The overall dimension
applications. Owing to their low profile, low cost, of the antenna is length x width = 108mm x 66mm
ease of fabrication and compatibility with the respectively. The balun and ring resonator are printed
microwave integrated circuit technology, these on front side of the substrate and the DGS and printed
antennas are in high demand from the system dipole are printed on back side. The substrate is RT
designers point of view. They can be implementing duriod with r = 2.2 and thickness = 30 mil. The ring
on the same PCB as the circuitry with little or no bandpass filter with wide band rejection
additional cost. For a wireless communication system characteristics as proposed by Chul Soo Kim and et.al
which requires an omni-directional radiation patterns, [5] is integrated with the printed antenna as shown in
as in a broadcasting system or a communication figure 1.
system, a printed monopole or dipole is a good
choice.
The conventional printed dipole antennas are
narrow band in nature with impedance bandwidth of
hardly 5-10% of its centre frequency. However, they
show wide radiation pattern and gain bandwidth.
Sometimes, it is required to have narrow gain
bandwidth antenna in order to minimize the
interferences from the unwanted signals. The antenna
for high data transfer rate in ISM band will be
required to have high band rejection characteristics
65
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Most of the conventional BPF's are designed with B. Ring resonator with DGS
parallel coupled lines [6][7]. However, these filters
have a limitation in realization due to, Figure 4 shows the ring resonator coupled with
i) Large size because of /2 resonator. DGS as proposed by Chul Soo Kim and et.al [5]. The
ii) Very narrow gap between the coupled lines is same is integrated with the proposed printed dipole
required to antenna to achieved wide band signal rejection
get strong coupling. outside ISM band. From the figure 5 & 6 it is seen
On the other hand, the ring resonator can give that both the simulated and measured VSWR is less
compact size, narrow bandwidth and low radiation than 1.3:1 over the band. Figure 7 shows the insertion
loss [8]. But ring resonator has a passband at every loss of the filter and is less than 1 dB over the band.
harmonic frequency. These pass bands can be These results show high band rejection outside the
suppressed by using DGS along with resonator. By required band. Overall, there is good agreement
properly designing the DGS, the coupling value between the simulated and measured results.
controlled and good impedance matching over the
desired band is achieved.
66
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
D. Current distribution
67
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
The antenna far field radiation patterns are Figure 12(b): Measured Azimuth plane patterns of Printed dipole
measured inside a rectangular Anechoic chamber antenna with ring resonator and DGS
using PNA based antenna measurement system.
Figure 12(a) shows the measured elevation patterns
of the proposed antenna at 2GHz, 2.45 GHz and 3
GHz. From the figure it is seen that there is
reasonable gain variation within desired and
undesired frequency bands. This envisages the noise
rejection capability of the proposed antenna
68
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
IV. CONCLUSIONS
ACKNOWLEDGEMENT
REFERENCES
69
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract This paper presents a circular patch The circular microstrip antenna is a metal disk of
microstrip antenna operate in dual band (1.28 GHz
radius a and has a driving point location at r which
and 2.67GHz). The proposed Circular patch antenna
will be in light weight, flexible, slim and compact unit makes an angle with the x axis. The thickness of the
compare with current antenna used in dual band. The substrate is h, where h<<o , Which has a relative
paper also presents the detail steps of designing the dielectric constant of r [2].The resonant frequency fnm,
circular patch antenna and the simulated results. for each TM mode of a circular microstrip antenna is
HFSS software is used to compute the return loss,3D given by
polarplot,3D rectangular plot,3D gain (LHCP), ,3D
gain (RHCP) and radiation pattern of the antenna.
(1.4)
(1.5)
Figure 1.1 Circular microstrip antenna geometry where a is the physical radius of the antenna.
70
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
(1.7)
C. 3D Rectangular Plot
In the typical design procedure of the Microstrip
antenna, the desired resonant frequency, thickness
and dielectric constant of the substrate are selected
initially. In this design of Circular Microstrip Patch
antenna, FR4 dielectric material ( r=4.4) with
dielectric loss tangent of 0.02 is selected as the
substrate with 1.6 mm height. Then, a patch antenna
that operates at the specified operating frequency. f0 =
1.28GHz and 2.67 GHz can be designed by using the
above said equations. TM11 mode is selected to design
the radius of circular patch. The units of h must be
in mm and fr in GHz. Figure 3.3. 3D Rectangular Plot
71
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
. 4.CONCLUSION
ACKNOWLEDGMENTS
72
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
About Authors:
Dr. P Siddaiah obtained
B.Tech degree in Electronics
and communication Engineering
from JNTUA college of
engineering in 1988. He
received his M.Tech degree
from SV University Tirupathi.
He did his Ph.D program in
JNTU Hyderabad. He is the Chief Investigator for
several outstanding Projects sponsored by Defense
Organizations, AICTE, UGC & ISRO. He is currently
working as Professor & DEAN, Department of ECE in
University College of Engineering and Technology,
Acharya Nagarjuna University, Guntur, India. He has
taught a wide variety of courses for UG & PG students
and guided several projects. Several members
successfully completed their Ph.D under his guidance.
Several members pursuing their Ph.D degree. He has
published several papers in National & International
Journals & Conferences. He is the life member of
FIETE, IE & MISTE.
73
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
INTRODUCTION:
74
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
RADIO OCCULTATION
75
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
inverted to give information on the density, pressure and one of the filters to make such a second order
and temperature of the atmosphere. A typical data PLL is
processing chain for Radio Occultation is shown in
fig. 4.
76
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
the LEO satellite and the occulting GPS satellite. It in Doppler calculated from the model as compared to
estimates the Doppler using the GPS ephemeris and the Doppler seen by the spacecraft. Data to track
the LEO spacecrafts position and velocity vectors. using open loop tracking is generatedin matlab and
noise has been added to this data.Open loop tracking
The Doppler equation can be written as
simulations have been carriedout on this data and the
phase of output signal hasbeen calculated. This is
shown in fig. 8.
REFERENCES
77
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract-
The modernization of the GPS led to availability of tracking loops for tracking code and carrier
additional GPS signals with more features. L2C, the signals and further demodulation of Navigation
second civilian GPS signal is going to become data of GPS satellites.
available in full constellation. L2C signals have longer Different methods for acquiring CM code like parallel
ranging codes which leads to larger search space and code search acquisition, parallel code search
increase in computational complexity. L2C signals make acquisition with reduced points and parallel
use of two PRN sequences CM (Civil Moderate) and CL
frequency search acquisition methods are used and
(Civil Long) of 20ms and 1.5sec duration. In this paper
compared.
two local replica codes, NRZ and RZ CODES are presented.
The code phase and carrier Doppler are estimated based on
the parallel code space search and parallel frequency II. L2C SIGNAL STRUCTURE
search space acquisition.
The L2C signal (1227 MHz) is composed of two
Keywords- GPS acquisition, Time domain ranging codes, namely L2 CM (Civil Moderate) and
correlation, FFT, IFFT, down sampling. L2 CL (Long).The L2 CM code is 20 ms long and
contains 10230 chips, while the L2 CL code has a
I. INTRODUCTION period of 1.5 s, containing 767250 chips. The CM code
is modulo-2 added to data, and the resultant sequence
The current civil GPS signal consists of a single of chips is time multiplexed with the CL code on
frequency transmission L1 (C/A). The Global a chip-by- chip basis. The individual CM and CL
Positioning System (GPS) modernization program codes are clocked at 511.5 kHz, while the
deploys new Block IIR-M satellites, equipped to composite L2C code frequency is 1.023 MHz. Code
initiate the transmission of the new civilian signal
boundaries of CM and CL are aligned, and each
L2 civil (L2C). One of the most commonly
CL period contains exactly 75 CM periods.
referenced limitations with GPS is the vulnerability
of the L1 signal to Radio Frequency Interference
(RFI) either intentional or unintentional. With its
novel code structure and compact data format, the L2C
signal can offer advantages like indoor positioning,
ionosphere error elimination, and improved
tracking performance.
and reset with a specified initial states as described is multiplied with the Fourier transform of the PRN
for specific satellites. The CM codes are reset after code. The result of the multiplication is transformed
10230 chips, 20 ms long and CL code is 1.5s long into the time domain by an inverse Fourier transform.
and reset after 767250 chips. The absolute value of the output of the inverse Fourier
transform represents the correlation between the input
and the PRN code.
III. LOCAL L2C CODES
79
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
frequency (100 KHz in case of receiver on a LEO (pre-correlation filtering and down sampling), so that
satellite) to get higher (finer) frequency resolution. incoming samples in 1 chip are averaged to one sample
Finally the Fourier transform is taken and correlated value. This leads to high computational efficiency and a
output is viewed. This is repeated until incoming signal small correlation loss. Simulation for the correlation
code phase and the locally generated code are in same values with locally generated RZ and NRZ CM codes
phase. This algorithm reduces the number of FFT with Doppler frequency of 0 Hz to 50 have been carried
computations as compared to the other algorithms. out.
V. SIMULATION RESULTS
80
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
set at same frequency and by varying the code phase VI. CONCLUSION
correlation values are plotted.
The L2C code is a Time Multiplexing of two separately
generated codes (L2 CM and L2 CL). CL code is not
used for acquisition based on the NRZ and RZ local
codes. The acquisition based on circular correlation
FFT, parallel frequency space search and parallel code
space search acquisition are performed for various
Doppler carrier and code phase and the FFT
computations and acquisition time are analyzed. The
results show that the parallel frequency space search
acquisition provides less FFT computation and is very
efficient.
REFERENCES
Fig.11. Comparison of correlation values between NRZ [1] A Software Defined GPS and Galileo Receivers A Single
and RZ local codes for various Doppler frequencies Frequency Approach - Kai Borre, Dennis M. Akos,Nicolaj
Bertelsen,Peter Rinder,Sren Holdt Jenson.
[2] GPS directorate and system engineering and integration interface
specification IS-GPS-200F -21 SEP-2011.
[3] A SANA U. QAISAR, ANDREW G. DEMPSTER, Senior
Member IEEE University of New South Wales Australia Assessment
of the GPS L2C Code Structure for Efficient Signal Acquisition
IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC
SYSTEMS VOL. 48, NO. 2 APRIL 2012.
[4] TUNG HAI TA,Hanoi University of Science and
Technology,Vietnam, SANA U. QAISAR, ANDREW G.
DEMPSTER, Senior Member, IEEE, University of New South Wales,
Australia, FABIO DOVIS, Politecnico di Torino, Italy ,Partial
Differential Postcorrelation Processing for GPS L2C Signal
Acquisition, IEEE TRANSACTIONS ON AEROSPACE AND
ELECTRONIC SYSTEMS VOL. 48, NO. 2 APRIL 2012.
[5] S.Mariappan, Swarna Ravindra Babu, Sasi Bhusana
Rao,Acquisition and tracking strategies of Modernised GPS L2C
Fig.12. correlation values with varying the code phase in steps of 5 signal,IEEE- International Conference on Recent Trends in
samples (Doppler is zero) Information Technology, ICRTIT 2011.
[6] Yanhong Kou, Xingyun Zhou, Beihang University, China,Yu
Morton, Miami University,Lei Zhang, Beijing Institute of
Technology/Miami University ,Processing GPS L2C Signals under
Ionospheric Scintillations.
81
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract- The Global Positioning System is undergoing the relative motion of the satellite with respect to the
continual modernization, providing continuous user, causing a Doppler effect. The Doppler
improvements to users worldwide. The development of frequency shift depends upon the velocity of the
the next generation of satellites, called GPS III. An GPS satellites and user velocity. This can be 5 kHz
integral part of the GPS III capabilities being for a stationary receiver on Earth and as high as 50
developed is a new civil signal, called L1C, which KHz for a receiver on Low Earth Orbiting (LEO)
will be transmitted on the L1 carrier frequency in
addition to current signals. The development of L1C
Satellites. The code phase denotes the point in the
signal represents a new stage in international GNSS. current data block where the L1C code begins. If a
The L1C signal is being designed for interoperability data block of 10ms is examined, the data includes
between GPS and Galileos Open Service Signal and an entire L1C code and thus one beginning of
also Japans Quazi-Zenith Satellite System (QZSS). The a L1C code. These measurement values are
L1C signal has many advanced features and employs used by tracking loops for tracking code and
the BOC modulation scheme. This paper describes the carrier signals and further demodulation of
features of the L1C signal and different methods for the Navigation data of GPS satellites.
acquisition of the L1C signal. The methods used for Different methods for acquiring like Fishman & al.,
unambiguous acquisition of the BOC signal include
BPSK- like methods (Fishman & al. method and Martin
Martin & al., and Sub Carrier Phase Cancellation
& al. method) and the Sub Carrier Phase Method are used and compared.
Cancellation (SCPC) method.
Keywords- GPS acquisition, Time domain correlation, II. L I C SIGNAL STRUCTURE
sub carrier phase cancellation, single side band, binary
phase shift keying. The L1C (1575.42 MHz) signal consists of two main
components, one denoted L1CP to represent a pilot
I. INTRODUCTION signal, without any data message and the data
message L1CD. The pilot and data components are
The Global Positioning System is undergoing transmitted using ranging codes. The pilot
continual modernization, providing on-going component is further modulated by an overlay code
improvements for users worldwide. The next which is unique to each satellite. The ranging codes
generation of satellites, called GPS III includes a new used to spread the pilot and data components are
civil signal, called L1C, which will be transmitted on also separate and unique to each satellite. Length
the L1 carrier frequency in addition to current 10230 sequences are used to spread both the pilot
signals. The L1C signal will be available with the and data components of L1C, with a repetition
first Block III launch, scheduled for 2014. L1C signal period of 10 ms at the 1.023MHz chip rate.
has advanced features like
Pilot component is sent along with the data. The pilot Different spreading sequences are used for each
component is useful for accurate acquisition. The satellite and for each pilot and data component. The
power split is unequal with 75% in the pilot spreading codes corresponding to the ith satellite are
component and 25% in the data component. The L1C denoted as L1CPi(t) and L1CDi(t). In addition, the
signal employs advanced Weil based spreading codes. pilot component is also modulated by an overlay
Advanced forward error control coding: the FEC code of duration 18s at 100 bps with length 1800.
coding provides an improvement in data The overlay code is unique to each satellite. The
demodulation threshold of more than 7 dB compared overlay code corresponding to PRN signal number
with no FEC, such as the C/A code signal. Data i is denoted as L1COi(t). The spreading codes
symbol interleaving to combat fading. are based on Weil sequences which in turn are
The purpose of acquisition is to identify the visible derived from the length 10223 Legendre sequence.
satellites to the user. The two important parameters The modulation used for L1C signal is binary offset
estimated are the Carrier Doppler and the Code carrier (BOC). A BOC signal comprises of the data
Phase. As the carrier and code signals are affected by message, a PRN spreading code, square sub carrier
82
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
and a sinusoidal carrier. The BOC modulated signal signals. This paper describes the acquisition of data
is generated by the multiplication of these component signal which is BOC(1,1). As BOC(1,1)
components in the time domain. Both the spreading signal auto correlation function (ACF) has multiple
code and the sub carrier are binary-valued. Thus, the peaks, which cautions the user to acquire the main
BOC modulated signal may be viewed as a BPSK peak rather to side peaks for effective initialization
signal multiplied with a square sub carrier. of tracking loops. Several techniques can be used to
Multiplication by the sub carrier causes the spectrum eliminate the ambiguity of the ACF envelope. Two
of the BPSK signal to split into two parts with main- such methods which can be used to acquire BOC
lobes on either side of the carrier frequency. For this signal include BPSK-like technique and the Sub
reason, the BOC modulation is also referred to as Carrier Phase Cancellation (SCPC) technique.
split spectrum modulation. A. Fishman & al. BPSK Method
The BOC signal has two parameters: the sub carrier The main idea behind these BPSK-like methods is
frequency fsc the chip rate of the spreading code that the BOC-modulated signal can be obtained as
fc. The sub carrier frequency and the chip rate are the sum of two BPSK-modulated signals, located at
chosen to as multiples of a reference frequency, positive and negative sub-carrier frequencies. The
such that: effect of sub-carrier modulation can be removed
fsc = f0 and fc = f0, Where, the reference by using a pair of single- sideband correlators. We
frequency f0=1.023MHz. may have a single-side band (SSB) receiver, where
either the negative or the positive of the sidebands
The BOC signal is thus represented as BOC (, ).
correlators is used, or a dual-side band (DSB)
The modulation for the L1C data component is
receiver, when both bands are combined non-
strictly BOC(1,1). The L1C pilot component
coherently. Due to filtering and correlation losses,
uses a time multiplexed combination of
the BPSK-like methods bring some degradation in
BOC(6,1) and BOC(1,1) known as TMBOC. L1C
the signal level. A SSB BPSK-like method induces
signal from one satellite is
at least 3 dB degradation in SNR. If DSB processing
is used, this loss can be partially compensated,
excepting anyway the squaring losses in non-
coherent integration. Compared to coherent
processing of both sidebands, non-coherent
processing loses about 0.5 dB of SNR.
Where
the total signal power is denoted as C(Watts) which
includes any antenna gain and receiver
implementation losses, dP (t) and dD(t) are the series
of pilot overlay code and data bits, cP (t) and cD(t)
are the spreading codes for the pilot and data
components, gP(t) and gD(t) are the periodic
repetition (repeating every spreading code chip) of
the spreading symbols (also known as the
subcarrier) for the data and pilot components, and
fd are the unknown delay and Doppler frequency Figure 1: Principle of Fishman method
of the signal, the signal is at an intermediate In this method, the receiver selects only the main
frequency fIF (Hertz), the unknown phase term is lobes of the BOC-modulated received signal and the
(radians), which is the phase difference reference code. The main lobe of one of the
between the received signal and the locally sidebands (upper or lower) of BOC-modulated
generated signal used for down conversion, received signal is selected (via filtering) in each
And n(t) is additive white Gaussian noise with channel as shown in Figure1. Both sidebands are
power spectral density N0/2. translated to the baseband by multiplication with
complex exponential. In each channel, the resulting
III. ACQUISITION METHODOLOGIES signal is correlated with the reference code, having
the tentative delay and the tentative Doppler
Acquisition identifies the visible satellites to the user frequency fD. Hence, the DSB Fishman & al.
and determines two properties of the signal: code approach needs two sideband selection filters. The
phase and carrier Doppler. Either the pilot or data outputs of the two channels are combined non-
component can be used for acquisition GNSS coherently. This method has rather high complexity
83
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
due to the fact that two complex filters are needed search and transition because there is only one main
for DSB processing. Instead of the DSB statistic, energy peak. The advantage of the technique is an
only one of the channels may be retained for increased simplicity of the implementation, and
lower complexity. This gives the SSB statistic. above all, the least effect of the filtering.
However, as mentioned earlier, in this case the
SNR degrades by at least 3dB. The correlation
function between the received signal and reference
code, on each sideband, will be unambiguous and
will resemble the ACF of a BPSK- modulated
signal. However, the shape of resulting ACF is not
exactly the one of a BPSK modulated signal,
since there are information losses due to selection of
main lobes.
B. Martin & al.BPSK Method Figure 3: Correlating the filtered signal with the local
Here, a single centered filter is used to select the BPSK Signal
main lobes and the secondary lobes between
them. For this reason, this method has lower
implementation complexity as compared to the C. Sub Carrier Phase Cancellation (SCPC)
Fishman & al. method. This method uses only one method
centered filter with a bandwidth including the two
principal lobes of the spectrum and the secondary The Sub Carrier Phase Cancellation technique
lobes between the two principal lobes. (SCPC), allows ones to make a non ambiguous
Two correlation channels are generated: right acquisition of a BOC signal. Figure4 present the
channel where the filtered signal situated at fIF+fsc general principle. The idea here is to get rid of the
sub carrier signal as it is done for the carrier signal.
and correlated with a BPSK signal and left channel
In addition to the local in phase and quadrature
where the filtered signal situated at fIF-fsc and
carrier signals, an in phase and a quadrature local
correlated with a BPSK signal. The principle of this sub carrier signals have to be generated. Thus, two
method is illustrated in Figure2. The correlation correlation channels are generated here. On one
outputs of the two channels are combined. channel, the received filtered signal is correlated
with the local BOC signal in sub carrier phase, and
on the other one the received filtered signal is
correlated with the local BOC signal in sub
carrier quadrature. When these two correlation
channels are combined, an ACF similar to the BPSK
one is obtained.
Due to unsymmetrical spectrum, considering only The acquisition methods are carried out in MATLAB
one lateral main lobe, the correlation function is simulation software. BOC(1,1) signal is generated by
complex. The consequence is a slight coupling modulating the PRN code onto the square wave
between the carrier phase and code delay measures. sub carrier. In this case the code rate fc = fref =
However this feature is not critical for the energy 1.023 Mcps and the sub carrier frequency is fsc = fref
84
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
85
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
V. CONCLUSION
Of the three methods, the peak is highest for the
SCPC method. From the above simulations it is
seen that with respect to the correlation function and
the energy search, the SCPC outperforms the other
two methods considerably.
REFERENCES
86
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract- The responses of the ionosphere TEC due tracking performance (Das Gupta et al., 2004, Dubey
to the Geomagnetic storm have been investigated by et al., 2006, Skone, S., 2000)(3),(4)&(5). Since every
using the Global Positioning System (GPS) data geo magnetic storm has unique character and
during 29 June, 2013 at low latitude station, K L forecasting the response of ionospheric parameters
University, Vaddeswaram,(Geographic latitude like total electron content (TEC) is one of the major
16.440N, Geographic longitude 80.620E),India. The issues in space weather studies. The geomagnetic
variations of VTEC during this storm are analyzed, storms drastically affect the TEC of GPS signals,
significant deviations in VTEC during the main especially of low latitude region such as India and
phase of the storms are observed. During June Brazil etc,. In this context, the response of the
storm,TEC depeltions are observed. The depletion in ionosphere over low latitudes needs to
VTEC is due to the thermospheric composition investigated.Therefore there is a necessity of
changes; however disturbance dynamo electric field incorporate a provision in the ionospheric models for
also plays an important role. The GPS VTEC values detecting the ionospheric anomalies and magnetic
compared with VTEC derived from IRI-2007 & IRI storms (Ho et al., 1996) (6).
2012 models. It is observed that IRI models could
not predict these variations during storm days. It The severity of geomagnetic storm is usually
has been found that the development of more experiences its geomagnetic storm can be explained
accurate TEC models would be needed. So, it is in terms of Dst (Disturbance storm time) index as
necessary to modify the IRI models to get better well as Kp index. The Dst index is a measure of geo
accuracy. magnetic activity used to assess the severity of
magnetic storms. It is based on average value of
Key words: GPS TEC, IRI TEC, low-latitude horizontal component of earths magnetic field
Indian station, geomagnetic storm. measured hourly at four near geo magnetic
observatories. Use of Dst as index of storm strength
I. Introduction because the strength of the Earths surface magnetic
field at low latitudes is inversely proportional to the
The sun is the major source that controls the near energy content of the ring current, which increases
earth surface and drives its weather and climate. during geomagnetic storms. Generally, geomagnetic
Coronal mass ejections from the sun interact with storms can be classified according to different Dst
solar wind and the interplanetary magnetic field index levels: weak; -50nT Dst -30nT; moderate; -
(IMF) during their propagation disrupts the solar 100nT Dst -50nT and intense Dst -100nT. The
wind. Fast CMEs eventually drive shocks on the Kp index is computed on a three hour basis and
earths magnetosphere and trigger geo magnetic represents the overall level of planetary geo magnetic
storms. The magnetospheric storm is a global electro field disturbance. It is derived from ground based
dynamic disturbance in the magnetosphere- magnetic field measurements ranges from 0-9. A
ionosphere. Recently, the Global positioning system typical quiet day will have Kp values of 0-2.
(GPS) has been broadly used on scientific research According to NOAA scale, storms are classified into
because of its global and on time characteristics. strong (Kp = 7), Severe (Kp=8) and extreme (Kp=9).
During geomagnetic storms, the ionospheric From Dst and Kp indices, it is observed that the
disturbances cause data loss in satellite strong strom is occurred on 29th June, 2013.
communication links and loss of signal for GPS
satellite (Das Gupta et al., 2004, Chen Wu et al., In this paper the variations of VTEC measured by
2008)(1), (2). This can also reduce the availability of monitoring GPS L1 and L2 signals across low
satellites for GPS navigation, which may affect the latitude station, Vaddeswaram, India is investigated.
positional accuracy, navigational accuracy and GPS Measured VTEC were then compared with the TEC
87
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
predicted by the IRI-2007 and IRI-2012 models to model for all phases of the solar cycle. More
assess its predictability and observation. information about IRI can be found on the IRI
homepage at http://IRI.gsfc.nasa.gov/. The
II. GPS TEC data corresponding IRI TEC predictions are calculated
from IRI 2007 and IRI 2012 models. To predict
The GPS TEC data used in this study were recorded the IRI TEC from IRI 2007 and IRI 2012 we can
by the GPS receiver (Model: GPS 6 Station) which is access the
installed at KL University, siteathttp://ccmn.gsfc.nasa.gov/model/web/models/iri
VaddeswaramduringJanuary to May, 2013. The slant _vtmo.php/which has more recently become available
TEC (or STEC) from a satellite to a receiver can be as an interactive web interface accessible from the
obtained from the difference between the pseudo IRI home page.
ranges (P1 & P2) or the difference between the
carrier phases (L1 and L2) of the two frequencies IV. TEC estimation methodology
(Blewitt, 1990) (7) i.e.,
The dual frequency GPS data provides both code and
( ) carrier phase measurements on the two GPS
= ( ) (
+ +
) frequencies and navigation data parameters due to all
the visible satellites. In the estimation of TEC and
Or instrumental biases, several data processing steps
including extraction of desired data, smoothing,
2( 1 2) computation of satellite position, receiver position,
= + +
( 1 2)2 ( 1 1 2 2) slant factor and mean longitude of sun is to be carried
out. These data processing steps are implemented in
MATLAB code.
Where, k relate to the ionosphere refraction, is
80.62m3/s2; 1 & 2 are the wavelengths V. Results and discussions
corresponding to f1 &f2 respectively; r& r are the
differential code biases corresponding to the pseudo Major Geomagnetic storms originated from solar
ranges (P1 & P2) and the Carrier phases L1 & L2 bursts can cause strong disturbances in the geo space
respectively; s& s are the inter-frequency biases environment which ultimately affects the
corresponding to the pseudo ranges (P1 & P2) and the performance of GPS system. During geomagnetic
Carrier phases (L1 & L2) respectively. storm period, the ionospheric disturbances cause loss
VTEC in electron/m2 can be calculated as of signal for GPS satellite and data loss in satellite
(Ma and Maruyama, 2003) (8)as communication links. Usually these storms associated
VTEC = STEC *(cos X), with increased electron densities in the lower
Where the zenith angle X is expressed as X = arc sin ionosphere. GPS technology requires better space
(RE Cos /RE + h). weather support to compensate the ionospheric
Where is the elevation angle of the satellite, R E is induced errors. Hence there is a necessity to
the mean radius of the earth and h is the height of understand and model the effects of ionospheric delay
ionospheric layer, which is assumed to be 350 km. on radio system during the storm period.According to
previous studies, middle and low latitude ionosphere
II. IRI model can be affected with an Equatorial anomaly expanded
III. poleward and become pronounced during an storm
The IRI model is empirical model of the ionosphere events (Tanaka 1986, Abdu 1997)(14),(15). Migoya
by a working group that is jointly sponsored by the et al (2009)(16) reported that large values of TEC at
committee on Space Research (COSPAR) and the low latitudes during geo magnetic storms in
International union of Radio Science (URSI) (Bilitza correspondence with well developed crest of TEC.
and Rawer, 1993)(9). The model provides spatial and Das Gupta et al., (2007)(17) reported that the
temporal TEC representations. depletion in TEC are quiet frequent even under
magnetically quiet condition in the equatorial
The IRI model is continually upgraded as new data ionosphere particularly during equinoctial months of
and new modeling approaches become available and solar maximum period. Shagimuratov et al.,reported
this process results in several major milestone that the response of TEC to geo magnetic storm
editions of IRI (Rawer et al., 1978a,b, 1981; Bilitza& consists of effects of both enhancement and depletion
Rawer, 1996)(10),(11),(12), & (13) that progress (positive and negative disturbances). Dashora et
from a set of tables for typical conditions to a global al.,(2009) (18) studied the TEC response to the geo
88
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
magnetic storm of 15th May 2005 at Udaipur a station The performance of VTEC obtained from GPS
near the crest of equatorial ionization anomaly and derived TEC are compared with IRI model data in
Hyderabad a low latitude station using GPSTEC. order to determine their accuracy when storms takes
They found that the large enhancement (or place. During the storm day the maximum TEC
decrement) in VTEC is attributed to prompt obtained from IRI-2007 model is around 75 TECU.
penetration of electric fields and Travelling IRI model over estimates the GPS TEC value during
Ionospheric Disturbances. Bagia mala et al., (2009) the storm period and IRI model fials to predict
(19) studied the 24th August 2005 geo magnetic storm variability in the TEC value during geomagnetic
for Rajkot station using GPSTEC and found that the storm period.
enhancement to VTEC is due to prompt penetration
electric fields during storm day and thermospheric Conclusions
composition changes and disturbance dynamo electric
fields on the recovery day. In this paper, Analysis of TEC depletions during
29 June, 2013 Storm adverse space weather conditions is investigated.
The main feature of 29th June 2013 storm under
Fig.1 represents the variation of GPSTEC, and TEC investigation was a negative VTEC effect. The VTEC
derived from IRI-2012 and IRI-2007 models with Dst depression amounted up to 4.68 TECU. GPS TEC
index during storm period (26-07-2013 to 30-07- data results reveals that IRI models are
2013). The sun spot number during this period is overestimatedwhen compared with the GPS TEC
52.5. During pre storm day i.e., on 28th June, 2013 at measurements made with a ground-based GPS
00:00LT VTEC reaches to 2.33TECU, at 12:00LT receiver during storm days observation period.
VTEC attains a value 23.83 TECU with Dst index - Between these two models IRI-2012 model shows
13nT and at 22:00LT Dst index reaches to -72nT and better results than IRI-2007 model. Based on these
VTEC reaches to 8.3TECU during initial phase of the observations, it is necessary to modify the IRI models
storm. During storm day i.e., 29th June, 2013 at to get better accuracy.
00:00LT Dst index reaches to -96nT and VTEC
attains 2.472 TECU with Dst index 7nT. At 08:00LT Acknowledgement
Dst index reaches to -98nT and VTEC reaches to
4.68TECU during Main phase of the storm. After The above work has been carried out under the
10:00LT Dst index slowly reaches to -87nT and TEC project entitled Development of Ionospheric
reaches to 18.95 TECU and at 20:00LT Dst index Forecasting models for Satellite based Navigation
attains -60nT and VTEC reaches to 4.5 TECU during systems over the low latitude stations sponsored by
recovery phase. From fig.1, it is observed that there is Department of Science and Technology, New Delhi,
depletion in TEC during this storm day. India, vide sanction letter No: SR/FTP/ETA-
0029/2012, dated:08.05.12.
References
89
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
90
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
AbstractReceiver architecture for recovery of Delta Hence, when the measurements are differenced the
Differential One-way Ranging (DDOR) signals has errors that are common to both the measurements get
been described. Characteristics of Cascaded Integrator eliminated resulting in higher accuracies as compared
Comb (CIC) filters have been analyzed. Extraction of Delta to plain Differential One- way Ranging (DOR).
DOR tones from the spacecraft downlink signal has been
demonstrated using Cascaded Integrator Comb (CIC) based
Digital Down Converter (DDC). The accuracies are good enough for satellite orbit
determination and navigation.
Index TermsDelta Differential One-way Ranging The concept of DDOR is graphically represented in
(DDOR), Cascaded Integrator Comb (CIC) filters, Digital Fig. 1. The line through the earth joining the two
Down Converter (DDC), I-Q signal processing. ground stations participating in the DDOR
measurements is known as the baseline. If B is the
length of the baseline and is the angle subtended
I. INTRODUCTION
by the spacecraft with respect to the baseline, then the
differential delay experienced by the signal in reaching
DELTA-DOR is a Very Long Baseline the two ground stations is given by the formula:
Interferometry (VLBI) technique that can be used in = B Cos() / c, where c is the velocity of light in
conjunction w i t h Doppler and ranging data to free space. Once is obtained from measurements, the
improve spacecraft navigation by more efficiently value of can be obtained from the above formula.
determining spacecraft angular position in the plane of
sky. An application of VLBI is spacecraft navigation in
space missions where delay measurements of a
spacecraft radio signal are compared against similar
delay measurements of angularly nearby quasar radio
signals. In the case where the spacecraft measurements
are obtained from the phases of tones emitted from the
spacecraft, first detected separately at each station, and
then differenced, this application of VLBI is known
as Delta Differential One-Way Ranging (DDOR or
DOR). The data produced during a DDOR
measurement session are complementary to Doppler and
ranging data.
91
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
II. RECEIVER ARCHITECTURE FOR DDOR discrete-time domain is carried out as sample- to-sample
multiplication. Heterodyning results in the signal
The block schematic of a typical ground segment for centered at IF being down converted to baseband
DDOR operations is shown in Fig.2. frequency. The heterodyning process also yields an up-
converted signal at double the IF frequency. This signal
is not of interest and will be filtered. The output of filter
contains signal of interest at baseband. This signal,
being a narrow band signal centered at zero frequency,
need not be sampled at the rate at which the original IF
signal was sampled. Hence, decimation can be
carried out so that subsequent discrete-time signal
processing can be carried out at a much lower sampling
rate. When I-Q processing is carried out, the minimum
sampling rate required for each of the two streams after
decimation can be as low as the bandwidth of the
original signal.
92
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
centered around zero frequency. Hence, the sampling Fig.5 Block diagram of a 3 stage decimating CIC Filter
rate at this stage will be much higher than the Nyquist
criterion that needs to be satisfied. Reducing the The transfer function of a CIC filter at fs is given by the
sampling rate (Decimation) can lead to ease of discrete- equation:
time signal processing in the subsequent stages. The
ratio of IF low-pass sampling frequency and bandwidth
of the desired signal being very high, the process of
decimation can involve large rate change factors. Large
The above equation shows that even though a CIC has
rate changes require high order narrow band filters
integrators in it, which by themselves have an infinite
requiring fast multipliers. This, in several cases, can be
impulse response, a CIC filter is equivalent to N FIR
the bottleneck of the discrete-time system.
filters, each having a rectangular impulse response. As
Hogenauer devised an efficient way of performing
all the coefficients of these FIR filters are unity, and
decimation and interpolation using a flexible, ultiplier-
hence symmetric, CIC filter exhibits linear phase
free, linear phase filter suitable for hardware
response and constant group delay. For CIC decimators,
implementation that can handle arbitrary and large rate
the gain G at the output of the final comb section is
changes [1]. These are known as Cascaded Integrator
given by the equation:
Comb filters or CIC filters for short. CIC filters
G = (RM)N
performing decimation operation are only considered in
When two's complement arithmetic is used, the number
this paper as the requirement for interpolation does not
of bits required for the last comb due to bit growth can
arise for the application being considered. CIC filters
be calculated from the above equation. If B is the
consist of N cascaded stages of Integrators followed
number of input bits, then the number of output bits,
by Decimator and N cascaded stages of Comb
Bout, is given by:
sections as shown in Fig. 5 [2]. An integrator is a
single-pole IIR filter with a unity feedback coefficient
and is described by the equation: Bout bits are needed for each integrator and comb stage.
y[n] = y[n -1] + x[n] The input needs to be sign extended to B bits, but
A comb filter running at the high sampling rate, f, for a LSB's can either be truncated or rounded at later stages.
rate change of R is an odd symmetric FIR filter Frequency response of CIC filter for different
described by the equation: decimation factors and different number of stages are
y[n] = x[n] -x[n - RM] shown in Figs. 6 11.
In this equation, M is a design parameter called the
differential delay. M can be any positive integer, but it
is usually limited to 1 or 2.
A CIC filter can be built by cascading N integrator
sections together with N comb sections. However, this
filter can be made simpler by introducing a rate changer
before the Comb sections. The discrete-time output in
such a case would be given by the following equation at
the slower sampling rate:
fs2 = fs1/R
y[n] = x[n] - x[n - M]
By implementing the filter in this fashion the second
half of the filter following the decimator is operated at a Fig.6 Magnitude and Phase response of CIC filter
lower frequency. Also, the number of delay elements (R=8, N=1, M=1)
required for realizing the Comb section is reduced.
More importantly, the integrator and comb structures
are made independent of the rate change factor R.
Hence a CIC filter can be designed to have a
programmable rate change factor while retaining the
same filtering structure.
A CIC decimator consists of N Cascaded Integrator
stages clocked at f, followed by a Decimator of factor
R, followed by N Cascaded Comb stages running at fs2
= fs1/R as shown in Fig. 5.
93
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
94
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Fig.12 Spectrum of PM Signal containing fm, 2fm, Fig.16 Spectrum of Comb Output. 3fm component is
components centered around IF=3.125 MHz more than 30 dB down as compared to fm component.
V. RESULTS OF HARDWARE
IMPLEMENTATION OF CIC FILTER BASED
DDC
CIC filter based DDC was implemented in hardware
using Xilinx Virtex-5 130T device. Parameters chosen
for practical demonstration were same as those used for
MATLABsimulation. Fig. 18 represents the spectrum
of the PM signal centered at 3.125 MHz. Sampling
frequency of ADC is 12.5MHz. Spectrum of the signal
obtained by heterodyning the PM signal with the NCO
Fig.14 Spectrum of Integrator Output generating 3.125 MHz sinusoidal signal is shown in
Fig. 19. The output of the Integrator (1 stage)when the
heterodyned signal is fed is shown in Fig. 20. Fig.
21shows the output of Decimator for decimation factor
R of 8.The decimated output when fed to the Comb
section (1 stage)results in signal having the spectrum as
shown in Fig. 22. It can be seen from this figure that the
PM modulating tone of195.3125 KHz (f) has been
recovered at this stage. All otherfrequency components
are of much lower power and are insignificant.
Fig. 23 shows the recovered PM modulating tone
fordifferent test conditions. Sampling frequency of
Fig.15 Spectrum of Decimator Output ADC was chosen to be 50 MHz. Sinusoid of frequency
95
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
VI. CONCLUSIONS
96
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
AbstractOne of the most important ground elements steering towards UTC, time transferbetween IRNWT
of a GNSS system is the Precise Timing Facility (PTF). and a standard metrological lab is carried out.
The timescale generated by the Precise Timing Facility Generally GNSS All-in-view (AV) technique is used
(PTF) is usually derived from an ensemble of Active for the time transfer for high accuracy. The offset of
Hydrogen Maser Clocks and High performance Cesium
the metrological lab with respect to UTCis obtained
Clocks.The time generated at the PTF needs to be
synchronized and steered towards Co-ordinated from Circular-T, a monthly bulletin published by the
Universal Time (UTC) for catering to various user Bureau of International Weigths and Measures
requirements and for traceability of GNSS time. For a (BIPM). Accordingly the relative offset of the GNSS
GNSS PTF, traceability to UTC is achieved by PTF with respect to UTC is worked out as follows
performing time transfers with national or international 6 offset values corresponding to UTC-
metrological time labs. For GNSS applications, the UTC(Met Lab) at 5 days intervals is taken from
offset of the PTF with respect to UTC or any other Circular-T provided by BIPM for the preceding
standard referenceneeds to be made available to the month. There values usually correspond to
user through satellite broadcast. Generally , the offset of
Modified Julian Dates (MJDs) ending with 4
the PTF with the national metrological lab and/or
UTC is provided in terms of bias (a0), drift (a1) and and 9.
drift rate (a2). This paper deals with the study carried The GNSS AV exercise provides the difference
out for mathematically characterizingaPTF between PTF and Met lab. This is derived as a
timescalewith respect to UTC and other metrological five day average spread 2.5 days on either side
standards. The applicability of a quadratic model for of the day with MJD ending either 4 or 9.
relative characterization of two steered timescales is The offset corresponding to UTC-UTC(PTF) is
discussed along with the results achieved. The optimum then be computed as a combination of the
number of days/data samples required to minimize the above two and can be described as follows:
prediction error in the offset value has also been
recommended. This work was carried out for Indian
Regional Navigation Satellite System(IRNSS). UTC-UTC (PTF) =
[UTC-UTC (Met Lab)] (Derived from Circular-T)
KeywordsPrecise Timing Facility, Traceability, Bias, [UTC (PTF)-UTC (Met Lab)] (Derived from
Drift, Drift Rate and Steering AV)
The offset of the PTF with respect to metrological
I. INTRODUCTION and UTC needs to expressedin terms of bias, drift and
drift rate and broadcast to the user through the
The timescale generated at the PTF is an ensemble of navigation data. This helps the GNSS user in
Active Hydrogen Maser Clocks and Cesium Clocks achieving traceability of GNSS time to international
which is being generated continuously to be used as a and metrological standards.The following sections
reference time scale for the GNSS. To achieve describe the details of the analysis carried out for
traceability towards UTC, the PTF is continuously characterszingthe GNSS PTF with respect to
steered to the reference timescale i.e. UTC. For metrological lab and UTC in terms of bias, drift and
97
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
drift rate. The results of the same have been provided obtained for the end user. Hence a quadratic model is
in subsequent sections. used for the relative characterizing.
98
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Fig. 7 A 7th order curve fit to UTC-UTC (IRNWT) Fig. 9 Quadratic fit applied of 10 samples
99
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
100
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
CONCLUSION
101
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
I.INTRODUCTION TO COMPRESSIVE
SENSING
102
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
III.RECONSTRUCTION ALGORITHMS
3.1 L1 Mini mization Algorithm
103
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
104
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
105
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
This algorithm generalizes the iterative hard 3.3.2 MFR wi th Least Squares
thresholding (IHT) algorithm. The algorithm consists
of two parts, an update and a thresholding step. The MFR algorithm can be modified by stealing a
(a) Update: An update of the measurements is leaf from the orthogonal matching pursuit class of
performed as algorithms. Here the frame reconstruction algorithm
is used to provide column selection, by choosing the
columns of that correspond to the entries of x that
where y is the vector of measurements, is the have largest magnitude and then solve the least
measurement matrix and r is a control parameter squares problem on this set of columns. (k) Hence the
which is referred to as the step-length. idea here is to keep running the frame step until the
(b) Thresholding: The second part of the algorithm set of columns we select changes, and then perform
is the thresholding procedure where we generate the least squares. Convergence of this new algorithm is
next signicantly quicker than the plain MFR a lgorith m.
solution
106
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
REFERENCES
[1] Cands, E.J.J. Romberg, and T. Tao, Robust uncertainty
Figure 4.1: Simulated results for several algorithms principles: exact signal reconstruction from highly incomplete
using matrix sizes with n = 400 and m = 50, 100, 150, frequency information. Information Theory, IEEE Transactions
and 200. on
52(2), 489509, 2006 Feb
[2] Cands, E.J.J. Romberg, and T. Tao, Stable signal recovery
from incomplete and inaccurate measurements. Communications
on Pure and Applied Mathematics 59(8), 12071223, 2006.
[3] Cands, E. J. and T. Tao, Decoding by linear programming.
Information Theory, IEEE Transactions on 51(12), 42034215,
2005 Dec.
[4] [2] D. Donoho, Comp ressed sensing, IEEE Transact ions on
Information Theory, vol. 52, no. 4, pp. 12891306, Apr.2006.
[5] R.G. Baraniuk, M . Davenp ort , R. DeVore, and M .B. Wakin,
A simple proof of the restricted isometry principle for random
matrices , Constructive Approximation, 2007.
[6] Cands.E. and J. Romberg. Sparsity and incoherence in
compressive sampling. Inverse Problems 23(3), 969985, 2007.
[7] Joel A. Tropp , Anna C. Gilbert , Signal Recovery From
Random Measurements Via Orthogonal Mat ching Pursuit . IEEE
transactions on information theory, vol.53, no.12, december 2007.
[8] D. Needell And J. A. Tropp , Cosamp : Iterative Signal
Recovery From Incomplete And Inaccurate
Samples,arXiv:0803.2392v2 [math.NA] 17 Apr 2008.
[9] Graeme Pop e, Modied Frame Reconst ruct ion Algorit hm
for
Comp ressive Sensin g, arXiv :0906.1079v1 [cs.IT ] 5 Jun 2009.
[10] R. G. Baraniuk, V. Cevher, M . F. Duarte, and C. Hegde,
Model-based comp ressive sensing, 200 . Prep rint . Availab le at
http://dsp.rice.edu/cs.
[11] Mallat, S. G. and Z. Zhang , Matching pursuits with
timefrequency dictionaries. Signal Processing, IEEE Transactions
on
41(12), 33973415, 1993, Dec.
107
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Micro Electro Mechanical Systems (MEMS) Bulk micro machined accelerometers developed by
technology enables fabrication of miniature, low LEOS, ISRO are based on capacitance detection
power and high performance sensors. The scheme. The schematic of the MEMS accelerometer
significance of this sensor is the fact that by sensor structure is as shown in Fig.1.
integrating the output signal, an accelerometer can
also provide a measure of velocity and position. The
in-house fabricated capacitive MEMS based
accelerometer consists of two fixed electrodes and a
seismic mass acting as a movable electrode. The three
electrodes together form a differential capacitor
assembly. A change in acceleration causes the
seismic mass to move and causes a differential
change in capacitance. It is proposed to use this
accelerometer as seismometer for measurement of
lunar seismic signals by integration of the Figure.1 Typical capacitive MEMS accelerometer
acceleration output. The sensitivity of the in-house sensor structure
fabricated sensor is around 6-8pF/g. The sensor is
proposed to be packaged separately and interfaced The accelerometer consists of a spring-mass
with the discrete readout electronics in printed circuit system also called as seismic mass acting as a
board form. Detection of lunar seismic signals movable electrode in a differential capacitor
involve measurement of very low magnitude and low assembly. Seismic mass is inter-digitized with fixed
frequency acceleration signals [1] leading to electrodes on both sides of a capacitor assembly.
measurement of low capacitances of the order of 1 These capacitors also called as the nominal
ppm of the nominal capacitance (few atto farads (10- capacitance are equal where there is no acceleration.
18
F) change in a nominal capacitance of 10pF).
Conventional seismometers are generally bulky and A small change in the relative position of these
have a very large mass resulting in an instrument electrodes due to an external acceleration, changes
noise well below the seismic noise. MEMS based the distance between the proof mass electrodes and
sensors due their low weight have relatively larger the fixed electrodes which can be transduced as a
self noise and also require integration to give an change in capacitance as given in Equation (1)
108
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
109
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
electrode is connected to the capacitance to voltage tapped from sensor fixed electrodes. The output from
converter. The capacitance to voltage converter is each of the Capacitance to voltage converter is
configured to give an output proportional to the proportional to the individual capacitor values as
change in capacitance C as given in the equation (3) given in the equation (4) and not the difference as in
the previous scheme and requires further processing
to obtain an output proportional to the difference in
capacitance value.
(3)
Where Vexc is the excitation voltage, C is the
change in the sensor capacitane value and Cf is the
feedback capacitor. (4)
MEMS accelerometer sensor being differential in Where Vexc is the excitation voltage, C0 is the
nature, the excitation signal can be applied to top and sensor dead capacitance value and Cf is the feedback
bottom fixed electrodes of the sensor and the sensor capacitor.
movable electrode can be connected to the
capacitance to voltage converter. Fig. 3 shows the Cf
capacitance to voltage converter with sensor
connected in differential configuration.
-
+
Vexc Cf Vex
c
Cf
-
+ -
+
Vexc
Figure.4 Capacitance to Voltage converter with
sensor in single ended configuration
Figure.3 Capacitance to Voltage converter with
sensor in differential configuration The main advantage of this scheme is that this
scheme requires a single excitation source and being
For fixed excitation voltage and sensor sensitivity, a differential scheme any common mode noise or
gain of capacitance to voltage converter is limited by offset can be cancelled in further differential
the feedback capacitor used. Smaller the feedback processing. The main drawback of this scheme is that
capacitor higher is the gain. But, the change in sensor capacitance to voltage converters sensitivity is
capacitance is of the order of few aF and hence the limited as it is not possible to increase the gain of the
feedback capacitance has to be comparable in value circuit by increasing the excitation voltage or
to the change in sensor capacitance to have an reducing the Cf as this can cause saturation of the
appreciable output from the capacitance to voltage front end capacitance to voltage converter.
converter. Operational amplifiers internal
capacitance, PCB stray capacitance and availability The other option is to reduce the sensor dead
of small capacitance values in discrete form pose a capacitance value. But achieving a higher sensitivity
limit on Cf and hence the achievable gain. sensor with a low dead capacitor value is a
challenging task from sensor fabrication point of
Capacitance to Voltage converter with sensor in view. Another problem associated with this circuit is
single ended configuration: that the imbalance in the sensor capacitance due to
fabrication tolerances can not be balanced by adding
In this method, the capacitance to voltage converter balancing capacitors in parallel.
is configured to give an output proportional to the Buffer
sensor nominal capacitance. In this mode of
operation, two separate Capacitance to voltage Another simple configuration for converting the
converters are used for sensing of top and bottom change in sensor capacitance values to equivalent
sensor capacitors as shown in Fig 4. Excitation is fed voltage is by providing an Operational Amplifier
to the movable electrode of the sensor and output is buffer at the front end as shown in Fig 5.
110
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Vex
c
-
+ R
Vexc 1
+
-
R
+ Vex
- 2 R
c
3
Voltage amplifier
VI. CONCLUSION
111
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Acknowledgement
REFERENCES
112
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract This papers focus is primarily on good solution. WSN provides advantage in cost, size,
providing an overview of automated polyhouse. This complexity, power etc. Environmental monitoring
survey is mainly concentrated on the study of using wireless sensortechnology has become
automated polyhouse with regard to the sensors that
more and more important in todays world. WSN
are needed to be used for effective monitoring of
various climatic parameters that exist in a polyhouse carved a niche for itself among the galaxy of
environment, which in turn produce accurate results technologies and has become the mostly used one.
while saving energy and to develop a low cost One of its applications lies in agriculture also. WSN
automated polyhouse for commercial use. This review is used for monitoring the above said parameters that
also gives a projection on different wireless exist in the polyhouse via the sensors.
technologies that can be deployed for developing
an automated polyhouse and emphasize on
system integration and controlling various parameters.
The communication between the sensors unit and
control unit can be achieved using different wireless
sensor network technologies. The sensors unit collects
the data related to the environment in the polyhouse
using different sensors located in different parts of the
polyhouse and communicates with the control unit via
WSN protocols.
113
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
B. Humidity Sensors
114
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
V. CONCLUSION
115
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
intensity, soil moisture and humidity should be [7] Fungal Diseases, http://www.ag.arizona.edu/.
[8] George. N.Agrios, Plant Pathology, Elsevier Science, January
monitored and controlled to prevent the polyhouse
2005, pages: 256-257.
plants from various diseases. The data about the [9] Texas instruments IC Temperature sensor,
environment in the polyhouse can be periodically http://uk.farnell.com/texas-instruments/tmp102aidrlt/ic-
collected from various sensors that are located in the temperaturesensor/dp/1652477
[10] Honeywell sensing hch-1000 series product sheet,
polyhouse and can be controlled by taking required
http://www.digikey.com/product-
control action by the control unit. The future scope of search/en/sensorstransducers/moisture-sensors-humidity/1966708
this survey lies in the [11] LI-COR LI-192 Underwater PAR Sensor,
implementation. http://www.fondriest.com/li-cor-li-192-underwater-par-sensor.htm
[12] Hanson, B., and G. Dickey. "When measuring soil water
content, field practices affect neutron moisture meter accuracy."
REFERENCES California Agriculture 47, no. 6 (1993): 29-31.
[13] Sengunthar Gayatri R, Ekata Mehul, A SURVEY ON
[1] Quan, Vu Minh, Gourab Sen Gupta, and Subhas GREENHOUSE AUTOMATION SYSTEMS, In International
Mukhopadhyay. Journal of Engineering & Science Research (IJESR), Feb 2013,
"Review of sensors for greenhouse climate monitoring." In Sensors Vol.
Applications Symposium (SAS), 2011 IEEE, pp. 112-118. IEEE, 3, no. 5, pp: 2338-2343.
2011.
[2]Polyhouseconstructionmodel,
http://agribusinessblog.wordpress.com/tag/polyhouse/
[3] C. Arun, K. Lakshmi Sudha. Agricultural Management using
Wireless Sensor Networks - A Survey. In 2nd International
Conference on Environment Science and Biotechnology, 2012
ICESB,
pp. 76-80. DOI: 10.7763/IPCBEE. 2012. V48.
[4] Park, Dae-Heon, Beom-Jin Kang, Kyung-Ryong Cho, Chang-
Sun
Shin, Sung-Eon Cho, Jang-Woo Park, and Won-Mo Yang. "A
study on greenhouse automatic control system based on wireless
sensor network." Wireless Personal Communications 56, no. 1
(2011): 117130.
[5] Kang, BeomJin, DaeHeon Park, KyungRyung Cho, ChangSun
Shin, SungEon Cho, and JangWoo Park. "A study on the
greenhouse auto control system based on wireless sensor network."
In Security Technology, 2008. SECTECH'08. International
Conference on, pp. 41-44. IEEE, 2008.
[6] Agro meteorology-Temperature, Relative Humidity,
http://www.agritech.tnau.ac.in/.
116
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
K.Murali
Asst.Professor in ECE
Vijaya Institute of Technology for Women
Enikepadu, Vijayawada
kalipindimurali@gmail.com
Abstract-Spectrum sensing and localization are two reasoning capability to automatically disable its radio
important tasks in electronic warfare, signal frequency functions.
intelligence and cognitive radios. In cognitive radios,
unlicensed users implement spectrum sensing[1] and 2. Problem Formulation
localization not to interfere with licensed users.
Cooperation in spectrum sensing and localization In this section, we first present the general model for
improves the signal detection and position estimation channel sensing, and then review the energy detection
performance under fading, shadowing or noisy scheme, radiometry and cyclo stationary feature
conditions. detection and analyze the relation between the
This paper implements spectrum sensing and probability of detection, probability of miss detection
localization tasks using a radio frequency sensor [2]and probability of false alarm and three-bit hard
network and analyze the performance of this combination scheme is superior to the traditional hard
implementation through simulation. combination schemes in terms of false alarm
reduction.
Key words: spectrum sensing, cognitive radio,
sensor network, EIRP, localization scheme. 3. System Model and Background
The terms "Cognitive Radio" and "Dynamic The goal of spectrum sensing is to determine if a
Spectrum Access" arise from recent technological licensed band is not currently being used by its
and regulatory trends in the area of wireless primary owner. This in turn may be formulated as a
communication. A "radio" is any device that can binary hypothesis testing problem
receive and transmit radio frequency signals, Two conditions are possible under hypothesis test
everything from cell phones to WiFi-enabled 1. When the primary user is not active, the received
computers. Recent advances in radio technology now signal at the secondary user can be represented as
allow radios to be "agile" in the sense that they can y n u n 1
reconfigure themselves to dynamically receive and
transmit on a wide variety of channels using a variety Where y n is the signal received by the
of communication modes.
secondary user and u n is noise.
A cognitive radio is a radio that has some ability to
represent and reason about goals, policies, features of 2. When the primary user is active, the received
the external world, aspects of its internal state, etc., signal is given by
that relate to radio communication. For example a
y n h n s n u n 2
cognitive cell-phone, upon boarding an aircraft, could
use its policy knowledge, context awareness, and
117
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Where s n the signal is transmitted by the primary (2) Probability of false alarm, f , defines at the
users and received by secondary users over a channel hypothesis 0 , the probability of the sensing
h n . When the channel is non-fading, h n is algorithm claiming the presence of the primary
signal. That is, if we decide 1 , but is true, it is
constant. It is assumed that noise samples u n are
called a false alarm error. In Fig. 2 (b), the PDFs
independently and identically distributed (i.i.d.) with exceeding the threshold under the hypothesis 0 are
zero mean and variance u n u based on
2 2
defined as f .
the received signal [4]
The goal of spectrum sensing is to make a decision, As a preliminary, we set up the model for signal
i.e. to choose between 0 and 1 , based on the detection. Consider a time continuous signal t .
received signal [5]. We wish to express the time-continuous signal in a
discrete vector representation over a finite time
The probability density function (PDF) under each interval. In general, the signal can be expressed by a
hypothesis is shown in Fig. 1(a) and in Fig. 1(b), basis expansion
where the threshold value for each hypothesis is
denoted as. t xii t 3
i 1
x1 , x1 ,......x
118
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Under 1 ,
y
1
C 0, n2 s2 samples. The energy detection should be carried out
over all logical channels defined by the CR network.
Assuming that the channel is time-invariant during
Therefore log-likelihood ratio is
the sensing process, the energy detection on the given
1 y
2
channel is performed by accumulating the energy of
y exp 2 2
n s
samples and comparing it with the predefined
n2 s2
log 1 log 7 threshold, to decide whether signal is present or not
y y
2
1 [7].
0 exp 2
2
n n
5. Minimizing Sensing Time for Detection
By removing all constraints that are independent of
the received vector y, we obtain the optimal Neyman- In order to minimize the sensing time an algorithm
Pearson test has been proposed, whose flow chart is shown in
1 2 Figure 3. According to this algorithm there are three
e y yi e 8
1
2
possible states for the output of each detection
technique i.e. Low L, Medium M and High H. If
i 0 0
its output is H; it indicates the presence of primary
In this case the energy detector is also known as user, if its output is L; it means that primary user is
radiometer [6] not present. If the output is M then detection
4. Local Spectrum Sensing technique is not sure about the presence or absence of
The performance of a given spectrum sensing scheme primary user[8] The received signal is first passed
is fundamentally limited by the radio propagation through Matched Filter, as it takes least time for
channel. Typically, the effects of a radio channel can sensing among all three mentioned techniques. If
be divided into three main parts: path loss, small- output is H or L then its fine that we concluded
scale fading, and large-scale fading (shadowing). about presence or absence of primary user.
Path loss effects are incorporated in the received SNR
at a cognitive radio terminal. Small-scale fading
causes rapid, random variations in the signal strength
at the CR receiver and is modeled by Rayleigh fading
in this paper. Shadowing is the slow variation of
received signal power as the cognitive radio moves in
and out of the shadow of large structures like
mountains, buildings, and so forth. Shadowing is
often modeled as a lognormal distributed random
process that varies around a local mean given by the
path loss and with the standard deviation dB which
depends on the environment
119
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
(9)
The number of samples required to optimally detect
the incoming signal is O(1/SNR2)
120
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
signals and hence more sophisticated algorithms overlapping features in power spectral density are
should be employed for signal detection. non overlapping features in cyclic spectrum. Hence
different types of modulated signals that have
6.2 Cyclo stationary Feature Detection identical power spectral density can have different
cyclic spectrum. Because of all these properties cyclo
An alternative method for the detection of primary stationary feature detector can perform better than
signals is Cyclo stationary Feature Detection in which energy detector in discriminating against noise.
modulated signals are coupled with sine wave However it is computationally complex and requires
carriers, pulse trains, repeated spreading, hopping significantly large observation time. For more
sequences, or cyclic prefixes. This results in built-in efficient detection, the enhanced feature detection
periodicity. These modulated signals are scheme combined with cyclic spectrum analysis and
characterized as cyclo stationary because their mean pattern recognition based on neural networks is
and autocorrelation exhibit periodicity. proposed.
121
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
localization scheme[11].
-2
Pm
10
In this network configuration, it is implicitly assumed Fig 10: Perfect energy detection under
that the radiation patterns of the transmitter antennas AWGN channel
are azimuthally omni-directional. In this scheme at 1
x 10
-11
PD
on any node. -1.5
-2
8. Results : -2.5
-3
The proposed sensor network based cooperative 0 10 20 30 40 50
SNRreq (dB)
60 70 80 90 100
3
In particular, redundant exhaustive sensing on empty
bands is avoided with MRSS, and less overhead in 2.5
PD
0.9
0.8
0.6
0.7
0.6
0.4
0.5
PD
0.4
0.2
0.3
0.2
0
0 10 20 30 40 50 60 70 80 90 100 0.1
No of Users --->
0
-10 -8 -6 -4 -2 0 2 4 6 8 10
Fig 9:Signal Received in Cogntive Radio SNRreq (dB)
Environment
Figure:13 Wide Band Radiometry
122
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
1
networks. An RF sensor network can be deployed
0.9
over an area of interest to detect the signals in the air,
to determine the frequencies of these signals, and to
0.8
estimate the positions and effective isotropic radiated
0.7
powers (EIRPs) of the transmitters emitting these
0.6
signals. Maximizing the signal detection performance
0.5
PD
References:
123
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract - This paper presents the design of various parameters can be monitored and stored in
water consumption monitoring system for gated the Windows PC. A current popular and
communities using wireless sensor network. The widely used technology, Zigbee based wireless
application of flow-sensor for a water usage
monitoring system is composed of a number of
sensor network is used in this work as it is able to
flow-sensor nodes with Zigbee network that can be provide flexibility, low cost implementation and
deployed for continuous monitoring purpose. The reliability. Zigbee is a communication standard for
water consumption monitoring system will monitor the use in the wireless sensor network defined by the
flow of water, detects abnormal flow if any and Zigbee Alliance that adopting the IEEE
warns the occupant by alarming. Therefore, the use 802.15.4 standard for its reliable communication.
of the monitoring system will prevent the wastage of
The fig.1 shows the block diagram of overall
water and the cost of destruction to the property
that can be caused when the flow of water is not system with sensor unit at the individual and
properly controlled. In this paper, the fundamental monitoring units at base station.
design of the system is proposed and tested practically
using Zigbee technology[1]. The developed system is
cost-effective and allows easy customization. Results of
water consumption measurement system are also
presented.
Keywords - Flow-Sensor, Water Consumption Monitoring
System, Zigbee, Gated Community
I. INTRODUCTION
During the past few years, gated communities have
The sensor unit at the pump is optional to monitor the
increased rapidly in the Indian municipality and
incoming water level. Each individual Sensor and
metropolitan cities. Water is used by the people in
Monitoring units are connected with EEPROM
the gated communities for a variety of domestic uses
memory to store timely day wise results. The
such as washing, bathing and flushing toilets etc.
monitoring unit collects the data from individual
Usage of water by the gated communities in a proper
sensor units and stores all the information in the data
way is important in order to reduce the wastage of
base of the PC.
water. However, it is a challenging task for the
community to continuously monitor the consumption
II. HARDWARE DESIGN
of water and any leakage in the water pipes.
Therefore to control the wastage of water, a A. Sensor Unit
continuous monitoring system[3] is needed. For this
reason, it is important to have a monitoring system The Fig.2 shows the block diagram of the sensor
with the characteristics of autonomous and unit with a LCD display, Key pad, Real Time Clock,
reliability. The use of such monitoring system EEPROM memory and a Zigbee module connected
will reduce the reliance on man power and reducing with a programmable P89C51 microcontroller. A
the cost. This work focuses on the use of multiple sensor unit[4] is basically consists of flow-sensor
water flow- sensors working together as a device to used to detect the predetermined parameters like
check the consumption, leakage, level and flow of consumption, leakage, level and flow of water. All
water in the individual of the gated communities. The the sensor units use battery for their operation. The
flow-sensor unit in the individual of the gated P89C51 microcontroller in the sensor node is
community is able to continuously monitor these programmed such a way that the pulses sensed by the
parameters from the water pipes. Then, the sensed sensor are counted using the counter. Then, the same
data from the sensor unit will be transmitted information is converted to number of liters and is
wirelessly to the base monitoring unit where sent to the monitoring unit wirelessly by Zigbee
124
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
communication. The Fig.3 shows a sensor hardware monitoring unit[6] consists of Zigbee receiver
unit connected with a SF800 flow-sensor and a module programmed such a way that it receives the
Xbee module. data sent from the sensor units which are called end
devices wirelessly. For every 30 minutes, the
monitoring unit stores all the collected data from
various sensor units in the EEPROM memory. The
duration is maintained with reference to the Real
Time Clock connected to the unit. If there is any
abnormal flow through the water pipe, then it is
detected by the sensor unit and the same is informed
to the monitoring unit.
125
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
signal reception from the sensor unit and sending the III. RESULTS
collected data to the PC through RS232 which can
be seen on the monitor. The monitoring unit and the The tested network is connected in star topology with
sensor u n i t s a r e p o w e r e d w i t h a three sensor units and with a monitoring unit. Each
battery. The Z i g b e e compliant wireless senor node is placed at 12 meters apart from the
devices are able to transmit 10-75 meters, depending monitoring unit. The monitoring unit is connected to
on the RF environment and the power output the PC through RS232 to show the sensed data on
consumption required for a given application which the system monitor in a readable format. The
will operate in the unlicensed RF range at 2.4GHz following Figure.6 shows the consumption of water
Global. The XBee module operating data rate is in liters from different sensor units on the screen of
250kbps at 2.4GHz. The Zigbee network [2] can be Windows PC.
configured to operate in a variety of different ways From the Fig.6 it can be easily observed that flat
to suit the application and environment with no.201 consumes more water than the other flats.
supported topologies like star, cluster tree and Furthermore, flat no.102 consumes least amount of
mesh. The use of high power Wireless Sensor water. The Fig.7 also shows water consumption at
Network is suitable for activities in industries three senor nodes 101,102 and 201 on their
involving large area monitoring. individual LCDs respectively. Hence, the results
shows that the proposed system working properly.
D. Working of the System:
The following algorithms indicate the functioning of
the sensor and monitoring units. The proposed sensor
unit will send the data to the monitoring unit for
every 30 minutes. This timing can be adjusted
with different duration depending on the need.
The RTC at every unit will be set with common
clocking. The sensor node at the water tank is
optional which can be used to identify the inflow
of the water in to the water tank, So that we can find
the water consumption with respect to the water in
the tank.
126
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
REFERENCES
[1] Zulhani Rasin and Mohd Rizal Abdullah Water Quality
Monitoring System Using Zigbee Based Wireless Sensor
Network International Journal of Engineering & Technology
IJET Vol: 9 No: 10.
[2] Gopala Krishna Moorthy .K, Dr.C.Yaashuwanth,
Venkatesh.K AWireless Remote Monitoring Of Agriculture
Using Zigbee (IJEIT) Volume 2, Issue 8, February 2013.
[3] K.Vairamani, K. Arun Venkatesh and N.
Mathivanan.Design and Development of Zigbee Based
Instantaneous Flat-plate Collector Efficiency Measurement
System MEASUREMENT SCIENCE REVIEW, Volume 11,
No. 2, 2011.
[4] Junguo ZHANG, Wenbin LI, Ning HAN, Jiangming KAN
Forest Fire detection system based on a Zigbee wireless sensor
Network Front. For China 2008, 3(3): 369374, DOI
10.1007/s11461-008- 0054-3.
[5] Peng Jiang, Hongbo Xia , Zhiye He and Zheming Wang
Design of a Water Environment Monitoring System Based on
Wireless Sensor Networks Sensors 2009, 9, 6411-6434;
doi:10.3390/s90806411.
[6] C. Arun and K. Lakshmi Sudha Agricultural
Management using Wireless Sensor Networks - A Survey,
Singapore, 2nd International Conference on Environment Science
and Biotechnology DOI: 10.7763 /IPCBEE. 2012. V48. 15.
127
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract Cognitive radio is a promising technology to ii)Spectrum Decision: Users can be allocate a
utilize the spectrum efficiently as radio frequency channel, based on spectrum availability. This
spectrum is precious and scarce. The main intension of allocation not only depends on availability of
cognitive radios is whenever licensed users are not in
use opportunistic access of spectrum is carried out.
spectrum but also based on internal policies.
However in practical scenario some applications like iii)Spectrum Sharing: If there are multiple CR users
medical, military have more importance than the others. trying to access the spectrum, there should be
Based on this paper proposes a cognitive radio network,
which contains primary users and three types of coordination to prevent collision.
prioritized cognitive users. Arrival times and conference iv)Spectrum Mobility: CR users are regarded as
times of all the users are generated by using exponential
random distribution. Based on these priorities channel
visitors to the spectrum. Hence, if the specific portion
allocation is done. Blocking probability for the cognitive of the spectrum in use is required by a primary user,
users is calculated. This work is carried out with the the communication must be continued in another
help of MATLAB. vacant portion of the spectrum and it has to change its
properties according to the available channel
I. INTRODUCTION characteristics.
The life cycle of cognitive radio is shown in figure1.
Nowadays, people are very much interested in
Sensing all the radio frequency bands, taking a
wireless devices and technology like Wireless
decision, sharing of spectrum and transmission of the
Routers, Mobile Phones, MODEMs, and GPS....etc.
signal is the main aim of cognitive radio.
Increase in wireless devices demand for radio
frequency spectrum, which is limited. So, efficient
utilization of the spectrum is required. FCCs
Spectrum Policy Task Force which has determined
that, in some locations or at some times of a day,
about 70 % of the allocated spectrum may not be in
use [1].Measurement [2] reveals that spectrum
utilization is often heavy in unlicensed bands while
low in TV bands or medium in some cellular bands.
These observations of actual spectrum usage made
the researchers to think of opportunistic spectrum
access. Cognitive radio is one of the solutions to the
above mentioned problem. Cognitive Radio offers a
solution by utilizing the spectrum holes that represent
the potential opportunities for non-interfering use of
spectrum which requires four major steps[3].
i)Spectrum Sensing: Unused portion of spectrum can Figure1: Cognitive Radio Life Cycle
be allocated to a cognitive radio user. Therefore
available spectrum bands should be monitored by the Sensing the entire spectrum is to be done for every
CR user.CR user needs to capture their information instant of time every channel is to be sensed and
and detect the spectrum holes. Various spectrum finding out which channels are vacant depending on
sensing techniques which, in general, could be user characteristics. Once spectrum holes are
classified as (1)energy-based sensing,(2) identified, cognitive radio users opportunistically
cyclostationary feature-based sensing,(3)matched utilize these holes for communication without
filter-based sensing and (4)other sensing techniques. causing any interference to primary users. But in this
128
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
scenario decision making plays a major role for interference as possible. Ultra Wideband (UWB)
different applications so prioritizing secondary users technologies are using this kind of spectrum sharing.
and allocating the channels to secondary users [6]
depending upon the priority. The rest of this paper Underlay Spectrum Sharing: The spectrum is shared
is organized as follows. Section II deals with in one of the following three ways:
available spectrum sharing techniques, section III i)Opportunistic: -The spectrum is shared among the
explains our proposed work, section IV contains the SUs whenever the PUs are not using it. [6]
simulation results, and sectionV describes ii)Cooperative: -The spectrum is allocated after
conclusion and future scope. negotiation with the PUs. [6]
iii)Mixed: - The spectrum is accessed in either a
II SPECTRUM SHARING cooperative or opportunistic way. [6]
Basically there are 2 types of spectrum sharing
The sharing of spectrum among cognitive users techniques. One is based on game theory and the
is divided into three main classes. This other is on genetic algorithm. Game theory based
classification is based on the assumptions of the cognitive channel assignment theory has originally
architecture, the way in which the spectrum is been proposed by the mathematician John von
allocated and spectrum access techniques. [4] Neumann to study human behavior [7]. It is an
First Classification: The first classification which is interdisciplinary research area where mathematics,
based on the architecture is illustrated as follows: social and behavioral sciences are brought together.
Centralized spectrum sharing: In this technique there Game Theory is one of the tools that more and more
exists a central entity that controls the allocation of researchers in this field try to apply to radio resource
spectrum and channels. All the users in the management. Significant work has been done by
cognitive network forward their measurements James Neel in applying Game Theory to the radio
towards the central entity that will make decisions resource management of cognitive radio systems
about the spectrum allocation. [4] [8],[9],[10],[11],[12].
Distributed Spectrum Sharing: Distributed spectrum Second one is a genetic algorithm which is an
sharing technique is implemented in the case where optimization technique that uses a number of bio-
there cannot be defined any infrastructure of the inspired evolutionary concepts like inheritance,
network. In this scheme each individual user selection, mutation and crossover. A random solution
makes decisions about the spectrum and channel is usually generated at the beginning of optimization.
allocation. [4] Then at each generation, the fitness of the solution is
evaluated by a predefined fitness function, and the
Second Classification: The second classification that solution will be modified accordingly. The algorithm
is based on the spectrum access behavior is is terminated when a satisfactory fitness level is
described as follows achieved or a maximum number of generations are
Cooperative spectrum sharing: The measurements of reached. Applications of a genetic algorithm can also
interference made by the individual cognitive users be applied to distributed optimization problems. A
are shared among all users. This information genetic algorithm based frequency allocation
collected from all the users is used for the approach for distributed cognitive radio networks is
allocation of the spectrum. Almost all of the proposed by Si Chen and Alexander M. Wyglinski in
centralized spectrums sharing techniques are [13]. A fitness function is developed to intelligently
cooperative. [4] Security is the major issue in this allocate frequency bands for subcarriers in a Non-
technique. [5] Contiguous Orthogonal frequency Division
Non-cooperative Spectrum Sharing: In non- Multiplexing (NC-OFDM) system.
cooperative spectrum sharing the cognitive users are
making decisions based on local observations. They III. PROPOSED WORK
do not cooperate with each other a n d ar e m a k i n g
d eci sio n s i n d e p e n d e n t l y f r o m others. The Compared to the above mentioned techniques, our
ad vantage o f this approach is that control traffic proposed work is simple and efficient. Considering
is mitigated to a large extent and the channel number of channels is equal to the number of primary
allocation behavior is also simplified. users, as primary users have own their individual
license. Keeping in mind that some applications are
Third Classification: Finally, the third technique is more important than the others secondary users are
based on how the spectrum is accessed: classified into three priorities secondary users1 (su1),
Overlay spectrum sharing: In this type of sharing the secondary users2 (su2) and secondary users3 (su3)
SUs are accessing the spectrum at the same time respectively. Figure 2 shows how the channel is
the PUs are accessing it, but with as little occupied by primary and secondary users
129
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Non-blocking state: A secondary user is considered to Now search for medium priority secondary users
be in this state if it is completely serviced without arrival and conference time if any of the medium
being interrupted by a primary user on that channel. priority secondary users arrival time and conference
time are present within the time interval T and also
Dropping state: When the primary user of a channel
find if any of the primary user or high priority
returns, the secondary user utilizing that channel
secondary user is using that channel at that instant if
should vacate. If there are no more free channels
primary user or high priority secondary user is not
available then it is semi-serviced and its call is
using the channel allocate the channel for medium
dropped.
priority secondary user if primary user or high
Forced termination state or Transition state: This is priority secondary user is using the channel then
the state during which the secondary user is shifting block the medium priority secondary user.
its channel due to the return of the licensed user into
the previous channel. In this case there are free Searching for low priority secondary users arrival
channels to shift to and so the secondary user and conference time if any of the low priority
performs a spectrum hand-off. secondary users arrival time and conference time are
present within the time interval T and also find if any
of the primary user or high priority secondary user
or the medium priority secondary user is using that
channel at that instant. if primary user or high
priority secondary user or the medium priority
secondary user is not using the channel allocate the
channel for low priority secondary user if primary
user or high priority secondary user or the medium
priority secondary user is using the channel then
block the low priority user. MATLAB software
which is a computationally intensive environment is
used for simulate of the proposed work.
Figure 2: Random access in five channels.
ALGORITHM
130
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
R EFERENCES
[1] Spectrum Policy Task Force report, Technical report 02-135,
Federal Communications commission, Nov. 2002.
Figure4: Blocking probability for 3 channels [2] M. McHenry, Report on spectrum occupancy measurements,
http://www.sharedspectrum.com/?section=nsf summary.
131
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
132
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract : Now a days image processing plays a key produce synthesis in FPGA, HDL simulation and
role in many areas like defence, medical and radar development tools. The tool retains the hierarchy of
signal processing etc. Image processing algorithms Simulink when it is converted into VHDL.
can be implemented using PC or dedicated H/W. But
it takes large computational time. The speed of The shape of objects in images is described many
implementation can be reduced using FPGA which parameters like the edges, depth discontinuity,
supports parallelism. In this paper Image filtering surface orientation discontinuity, reflectance
operations are explained using Xilinx System discontinuity, illumination discontinuity, and noise
Generator(XSG) and SIMULINK. The filtering is level in the images.
performed using Sobel and Median filters on CT and
MRI images. The results are validated using PSNR II. XILINX SYSTEM GENERATOR
and MSE.
Keywords: Digital Image processing, Matlab, Xilinx System Generator (XSG) is an integrated
Xilinx System Generator, FPGA. design Environment (IDE) for FPGAs within the ISE
13.4 development suite, which uses Simulink, as a
I. INTRODUCTION development environment and is presented in the
form of model based design. It has an integrated
The handling of digital images has become a subject design flow, to move directly to the Bit stream file (*.
of widespread interest in different areas such as legal, bit) from Simulink design environment which is
medical, technological applications and many others. necessary for programming the FPGA.One of the
There are many areas where the image processing most important features of XSG is that it possesses
helps to analyze, infer and make decisions. The main abstraction arithmetic that is working with
objective of image processing is to improve the representation in fixed point with a precision
quality of the images for human interpretation or the arbitrary, including quantization and overflow. XSG
perception of the machines independently. FPGAs are can only perform simulations as a fixed point double
increasingly used in modern imaging applications precision type.
like image filtering, medical imaging, image
compression, and wireless communication. The XSG automatically generates VHDL/Verilog code
various applications where image filtering operations and a draft of the ISE model being developed. It also
applied are noise removal, enhancing edges and generates hierarchical VHDL/Verilog synthesis, floor
contours, blurring and so on. plan and mapping hardware. In addition to this it also
generates a user constraint file (UCF), simulation and
The main aim of this paper is to implement image testbench and test
filtering algorithms in H/W like Xilinx Virtex6. The vectors.
need to process the image in real time using dedicated
H/W or PC is time consuming, leads to this XSG was created primarily to deal with complex
implementation in hardware level, which offers Digital Signal Processing (DSP) applications, but it
parallelism, and thus significantly reduces the also deals with implementation of many images
processing time. It uses XGS, a tool with graphical processing application. The blocks in XSG operate
interface under the Matlab Simulink, based blocks with Boolean values or arbitrary values in fixed point
which makes it very easy to handle with respect to type, for a better approach in hardware
other software for hardware description. implementation. In contrast Simulink works with
numbers of double-precision floating point. The
In addition to offering computation speed this tools connection between XSG blocks and Simulink blocks
works at graphical simulation level. This paper are the gateway blocks. The Fig.1 shows the broad
presents architecture development of image flow design of XSG.As previously mentioned, XSG
processing application in Simulink which is a is configured to program the FPGA. The synthesis
bookstore called Blocks Xilinx. They are mapped and implementation of the program are done
architectures, entities, ports and attributes, whose subsequently.
script file is to
133
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
134
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
For post-processing it uses a Buffer block which Fig.9 Salt & Pepper and Guassian noise
converts scalar samples to frame output at lower
sampling rate, followed by a 1D to 2D (matrix)
format signal block, finally a sink is used to display
the output image back in the monitor, utilizing the
Simulink block sets.
135
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Conclusion
REFERENCES
136
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract - In recent days, the Morphological image segmentation [12-17] approach has found wide
segmentation is an important problem and widely in applications in various types of images which are not
use in different fields of computer vision and image only confined in the optical spectrum but also in
processing. To extend the concept of segmentation, other frequency ranges like infrared, microwave and
an ordering of the data is required. In recent days, millimeter wave based imaging, viz radar imaging,
there are several methods in image segmentation. The satellite imaging. The main advantage of the
focus of this research paper is a morphological image watershed algorithm is that it can guarantee closed
segmentation tool called marker controlled watershed contours, and there is a good match between the
segmentation with Laplacian of gaussian spatial high boundaries of regions in the result and those that
pass filter. The efficiency of the results demonstrates appear in the original image. Though watershed
that combining of these two strategies has been very algorithm a very good method for image
helpful for image segmentation and for computer segmentation, but this technique suffers from over
vision. segmentation problem [8-10]. Applying of watershed
segmentation algorithm directly on the gradient
images produce high over segmentation problem. The
Keywords:Image segmentation, , Laplacian of
problems caused by these factors can be serious
Gaussian, markers, watershed algorithm. enough to render the result useless.
137
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
out with the Laplacian and Gaussian (LOG) filtering The mathematical formulation of watershed
approach. An approach with modified LoG spatial algorithm is sated below.
filter having 9x9 mask have been used for generating
gradient images. The LOG filtering mask although is 3. MARKER CONTROLLED WATERSHED
a 5x5 matrix as proposed by Marr-Hildreth [11], the SEGMENTATION
mask can be extended dimensionally on the basis of
the need of the image processing situation. The
Though watershed algorithm a very good method
proposed gradient kernel is capable of removing the
for image segmentation, but this technique suffers
noise much better than the standard LoG spatial filter
from over segmentation problem. An approach used
and marker controlled watershed segmentation is
to control over segmentation is based on the concept
used to controlled over segmentation. Comparative
of controlled marker. The concept of markers is a
analysis also has been done in this research paper.
good approach to control over segmentation. The
The Gradient images obtained using conventional
markers are connected component of an image.
LoG filter are compared with Gradient images of
There are internal markers and external markers
proposed or modified Log spatial filter having 9x9
where i n t e r n a l markers a r e a s s o c i a t e d
mask. The final segmented results are also compared.
w i t h o b j e c t of interest and external markers are
The results obtained show the good performance of
associated with the background. This is a very useful
this approach.
technique for improvisation of watershed transform
to overcome the over segmentation problem. In this
This paper is divided into a choice of sections. In
technique, the set of the catchment basins of the
section 2 introduces the watershed algorithm. Section
grayscale image function f with values in [l ,l ]
3 describes the marker controlled watershed
is equal to the set Y obtained after the following
segmentation. Section 4 gives a brief description on
Laplacian of Gaussian spatial filter. Section 5 recursion: Y = T (f) , where T is the
presents the proposed or modified Laplacian of threshold set at level l
Gaussian mask. The proposed approach has been
discussed in section 6. Experimental results are = , ( ) , 1
discussed in section 7 and we nish this paper with (2)
some concluding remarks with section 8.
Where, MIN1 is the union of all regional minima at
2. WATERSHED ALGORITHM altitude l
And ( )= ( )
(3)
Where, =
{ | [1, ]{ }: ( , )< ( , )}
(4)
138
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
4. LAPLACIAN OF GAUSSIAN SPATIAL FILTER which constitute the fuzzy set, there will be no gross
deviation from real situation and the sum of the
The Laplacian of Gaussian filter (LOG) plays a very weights inside the mask 0.47 which is sufficient to
important role in image segmentation. It is a help enhance a contour pixel.
convolution filter that is used to detect edges of
different objects. This filter first applies a Gaussian
blur, then applies the Laplacian filter and finally
checks for zero crossings, i.e. when the resulting
value goes from negative to positive or from negative
to positive.
The main objective of this filter is to highlight edges
different objects. As a input the LOG operator takes a
single gray level image and produces another binary
image as output. A 5x5 mask LOG filter has been
shown in Figure 3.
7. PROPOSED APPROACH
Watershed segmentation is an effective method for
image segmentation. The flowchart of the present
algorithm is stated below. This approach is based on
the concept of LoG spatial filter with markers. In first
Figure 2: A conventional 5x5 LOG surround filter. step of present algorithm three real life color images
has been chosen and accordingly converted into a
gray scale image in second step. The gray scale
5. PROPOSED OR MODIFIED LAPLACIAN OF image also called as black and white image only
GAUSSIAN FILTER contains the intensity information of an image
varying from black at the weakest intensity to white
A 9x9 mask has been proposed in the present work. at the strongest. In third step the gradient image is
There are two annular regions in this mask, i.e., the computed using proposed or modified LoG spatial
inner being excitatory and the outer being inhibitory filter having 9x9 mask in present study. The gradient
in nature which surrounds the central pixel where the magnitude image has high pixel values along object
positive weights (excitatory) are assigned for inner edges and low pixel values everywhere else.
one and negative weights (inhibitory) are assigned for
outer one. The influence of a point in the surround on Read the color image
the cell respond depends on its distance from the
centre and the nature of variation approximately Convert the image into gray
follows a Gaussian profile and also the excitatory scale
influence is weaker than the inhibitory influence [18],
[1].
Obtain gradient image using
After large number of trials using masks of larger Proposed 7x7 LoG mask
dimensions, the size of the optimal mask was
obtained and found to have 9x9 in dimensions. There
is no loss of generality if the weights of individual Enumerate internal markers
pixel in the masks are changed slightly, keeping in
the mind basic requirement of being isotropic and Enumerate external
ability to emulate HVS. The mask has been shown in markers
Figure 4. Where, by sampling a Gaussian profile of
standards deviation 1.4 the weights in the outer
inhibitory region of the optimal mask were found. If Apply watershed algorithm
anybody slightly alters the pixel values in the mask Figure 4: Flow diagram of the proposed approach.
139
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
140
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Figure 8: Gradient image of fruits with conventional LoG Figure 12: Gradient image of flower with proposed 9x9
mask. LoG mask.
Figure 11: Gradient image of fruits with proposed 9x9 LoG Figure 15: Marker controlled watershed segmented image
mask. of flower with conventional Log mask.
141
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
8. CONCLUSION
Figure 17: Marker controlled watershed segmented image In this research paper deals with the problem of
of fruits with proposed 9x9 Log mask. morphological image segmentation. segmentation,
for which the marker controlled watershed algorithm
has been one of the most widely accepted solutions.
An effective methodology for digital color image
segmentation has been publicized in this paper with
LoG spatial filter and marker controlled watershed
algorithm. The proposed methodology has been
successfully tested on three real life digital color
images. The results demonstrates that the
arrangement of modified LoG spatial filter which is
proposed in this paper with marker controlled
watershed algorithm has been very effective for
image segmentation and for computer vision. The
experimental results and statistical measurements
Figure 18: Marker controlled watershed segmented image
confirm the efficiency of the proposed methodology.
of flower with proposed 9x9 Log mask.
DEDICATION
REFERENCES
Figure 19: Marker controlled watershed segmented image
1. Glasbey, C.A. and G.W. Horgan, Image analysis for
of stones with proposed 9x9 Log mask. the biological science, Statistics in Practice, Series
Editor Vic Barnett., John Wiley and Sons (1994).
142
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
143
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract - In recent days many digital image Prewitt, Roberts etc. In this research paper, the
segmentation technique have been found in published gradient image is generated using morphological
or online literature, among them morphological approach. The image smoothing [13-16] technique is
image segmentation methods have found wide also introduced with mathematical morphology.
applications. In this research paper an approach of This paper is divided into various sections. Section 2
image segmentation using mathematical morphology introduces the morphological gradient. In section 3
is propose . The proposed approach has been applied morphological smoothing is presented. A brief
to images with almost similar objects and the description on Watershed algorithm and its
segmentation results are found to be very efficient mathematical formulations are given in section 4.
and encouraging. Section 5 presents the proposed scheme or approach.
The experimental results are discussed in section 5
and finally conclusion is discussed in section 7.
Keywords- Image segmentation, morphological
gradient, morphological smoothing, watershed 2. MORPHOLOGICAL GRADIENT
algorithm.
Morphological gradient can be generated using
1. INTRODUCTION dilation and erosion. Dilation gives the original set
plus an extra boundary, the size and shape of the
Mathematical morphology [1-3] is a nonlinear branch boundary depends on the shape and size of the
of the signal processing field and concerns the structuring element. Erosion gives the points for
application of set theory concepts to image analysis. which the structuring element is contained in the
Morphology refers to the study of shapes and original set. The outer boundary of the original shape
structures from a general scientific perspective. Many is removed by erosion. The morphological gradient is
morphological techniques have been proposed in the generated by subtracting an eroded image from its
published or online literature [4-10] for the dilated version. The morphological gradient
segmentation of digital images. Morphological image
highlights sharp gray-level transitions in the input
segmentation methods, particularly those using a
image.
watershed algorithm [7-10], have found wide
= ( )
applications.
( ) (1)
In this research paper an effective approach based on
mathematical morphology is presented. In 3. MORPHOLOGICAL SMOOTHING
conventional approaches the watershed algorithm is
applied on the gradient images [10-13] for image One way to achieve smoothing is to perform a
segmentation purpose. The gradient images are morphological opening followed by a closing.
usually generated using spatial high pass filters [12- Opening smoothes the contour by removing thin
13] like, Sobel, Laplacian of Gaussian (LoG), bridges and eliminating thin protrusions.
144
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
= ( ) [ ] = ( 1) ( 2) . . . . . . ( )
(2) (7)
Closing also smoothes the contour, but by enforcing [ + 1] = ( 1) ( 2) . . . . . . ( )
bridges and closing small holes. (8)
If the algorithm keeps on increasing flooding level
= ( ) then Cn(Mi) and T[n] will either remain constant or
(3) increase. Algorithm initializes [ + 1] =
The boundary of opening with a circular structuring [ + 1], and then precedes recursively by
element corresponds to rolling a ball on the inside of assuming that at step n C [n - 1] has been
the set. The boundary of closing corresponds to constructed.
rolling a ball on the outside of the set. Let, G is a set of connected components in T[n] and
for each connected component g G[n], there
4. WATERSHED TRANSFORM possibilities will arise.
1. g C[n - 1] is empty.
In mathematical morphology watershed transform is 2. g C[n - 1] contains one connected component of
the technique which is commonly used in image C[n - 1].
segmentation. A gray level image can be interpreted 3. g C[n - 1] contains more than one connected
as the topographic image. It can be assumed as a component of C[n - 1].
landscape where the gray intensities of higher
magnitude corresponds local maxima, whereas those
5. PROPOSED APPROACH
of lesser value are local minima or catchment basins.
When a landscape or topographic relief is flooded
The proposed image segmentation approach here is
with water, water will be filled up at these starting
able to segment the digital images with minimum
local minima and at points where water coming from
drawbacks of over segmentation. The steps of the
different basins would meet and dams will be built.
proposed approach are shown in figure 2. In first
When the water level reaches the highest peak in the
step, a color image is chosen for the processing
landscape the process is stopped. As a result, the
purpose and converted into gray scale image in the
landscape is partitioned into regions or basins
second step.
separated by dams, called watershed ridge lines or
simply watersheds. The mathematical formulation of
Gray scale images, also very popularly known as
watershed algorithm is sated below.
black-and-white images are mainly combination of
two colors, black and white. It carries the intensity
Assume, Mi where i= 1 to n be the set of coordinates
information where, black have the low or weakest
points in the regional minima (catchment basins), of
intensity and white have the high or strongest
the image P(x,y) and C(Mi) be the coordinates points
intensity. In third step the gradient image generated
of catchment basins associated with the regional
using morphological approach with dilation and
minima Mi
erosion. The gray scale erosion is performed using
= {( , ) | ( , ) < }
flat structural elements. A square element is applied
(4)
for the erosion purpose in this proposed approach.
Where,
The morphological gradient is generated by
T[n] = set of points in P(x,y) which are lying below
subtracting an eroded image from its dilated version.
the plane p(x,y) = n
min, max = minimum or maximum gray level value.
Watershed algorithm produces over segmentation if it
n = stage of flooding varies from min + 1 to max + 1
directly applied on gradient images. So in fourth step,
Let Cn(M1) be the set of points in the catchment basin
morphological smoothing operation is done on the
associated with M1 that are flooded at stage n.
gradient image. To achieve smoothing, an opening
( 1) = { ( 1), [ ]} operation is performed followed by a closing.
(5) Opening smoothes the contour by removing thin
Where, bridges and eliminating thin protrusions. Closing also
1, ( , ) ( ) ( , ) [ ]
( )= smoothes the contour, but by enforcing bridges and
0, closing small holes.
(6)
C[n] is the union of flooded catchment basin portions
In next step watershed transform is done and the final
at the stage n.
segmented map is obtained in the sixth and final step.
Where,
145
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Read the color image analysis. The statistical measurements are also
included and shoewn in table 1. They are analyzed in
respective of entropy, peak signal to noise ratio
Convert the image into (PSNR) and mean square error (MSE).
gray scale Experimental results presented in this paper are
obtained by using MATLAB.
Obtain gradient image
using morphological
approach
Morphological smoothing
146
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
(b) (c)
Figure 5: Gradient images after morphological smoothing
operation, (a) Bird, (b) Elephant, (c) zebra.
(c)
(a)
(b)
(b)
(c)
147
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Figure 6: Segmented images obtained by applying image segmentation techniques, mostly those using a
watershed algorithm directly on gradient images, (a) Bird, watershed transform, have found in various
(b) Elephant, (c) zebra. applications of image processing. The proposed
approach that is presented in this paper is fully
developed using mathematical morphology and has
been applied to images with almost similar objects
and the segmentation results are found to be very
efficient and encouraging.
DEDICATION
REFERENCES
148
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
I. INTRODUCTION
149
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
150
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
SM-DCT
III. CONCLUSIONS
151
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
that it is an unreliable measure, especially for 11. D. Van der Weken, M. Nachtegael and E. E. Kerre, A new
similarity measure for image processing, Journal of
correlation across distortion levels. The performance
Computational Methods in Sciences and Engineering, Vol.3,No. 2,
of SM- DCT has a perfect match with the quality pp. 209-222, 2003.
perceived by human observers. The Proposed 12. A. Beghdadi and B. Pesquet-Popescu, A new image distortion
measure is also superior to state-of-the-art metrics Q measure based on wavelet decomposition, 7th International
Symposium on Signal Processing and Its Applications, Paris,
and MSSIM. In future work, we will use more color
France, July 1-4, 2003.
images, and extend the measure for evaluating the 13. A. M. Rohaly, J. Libert, P. Corriveau, and A. Webster
quality of watermarked images and video sequences. (editors),Final Report from the Video Quality Experts Group
REFERENCES on the Validation of Objective Models of Video Quality
Assessment,March 2000.
1. A. C. Bovik and S. Liu, DCT-domain blind measurement of
blocking artifacts in DCT-coded images, Proceedings of
International Conference on Acoustics, Speech, and Signal
Processing, Salt Lake City, UT, May 7-11, 2001.
2. Z. Wang, A. C. Bovik and B. L. Evans, Blind measurement of
blocking artifacts in images, Proceedings of IEEE 2000
International Conferencing on Image Processing, Vancouver, BC,
Canada, September 10-13, 2000.
3. Z. Wang, H. R. Sheikh and A. C. Bovik, No- reference
perceptual quality assessment of JPEG compressed images,
Proceedings of IEEE 2002 International Conferencing on
Image Processing, Rochester, NY, September 22-25, 2002.
4. L. Meesters and J.-B. Martens, A single-ended blockiness
measure for JPEG-coded images, Signal Processing, Vol. 82, pp.
369-387, 2002.
5. P. Marziliano, F. Dufaux, S. Winkler and T. Ebrahimi, A
noreference perceptual blur metric, IEEE 2002 International,
Conference on Image Processing, Rochester, NY, September
22- 25, 2002.
6. E.-P. Ong, W. Lin, Lu, Z. Yang, S. Yao, F. Pan, L. Jiang and
F.,Moschetti, A no-reference quality metric for measuring image
blur, 7th International Symposium on Signal Processing and Its
Applications, Paris, France, July 1-4, 2003.
7. M. Carnec, P. Le Callet and D. Barba, An image quality
assessment method based on perception of structural information,
2003 International Conference on Image Processing, Barcelona,
Spain, September 14-17, 2003.
8. Z. Wang and E. P. Simoncelli, Reduced- reference image
quality assessment using a wavelet- domain natural image
statistic model, Human Vision and Electronic Imaging X
Conference, San Jose, CA, January 17-20, 2005, Proceedings of
SPIE, Vol. 5666.
9. Z. Wang and A. Bovik, A universal image quality index,
IEEE Signal Processing Letters, Vol. 9, No. 3, pp. 81-84, March
2002. 10. Z. Wang, A. C. Bovik, H. R. Sheikh and E. P.
Simoncelli, Image quality assessment: from error measurement to
Structural similarity, IEEE Transactions on Image
Processing,Vol. 13, No. 4, April 2004.
152
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
AbstractImpedance cardiography is a non-invasive ensemble averaging but this suppresses the beat-to-
technique for monitoring stroke volume, based on beat variations that occur during respiration (Permutt
sensing variation in the thoracic impedance, z(t), due to and wise, 1986; Wang et al., 1991) and cardiac
blood flow.In this paper, we present two additions to the arrhythmias. Wang et al. (1991) have shown that the
LMS algorithm in order to improve the beat by beat r-r interval changes during different phases of the
estimation of fiducial points (B and X) and the respiratory cycle. if the r-z interval (time from the r-
maximum value of the derivative in impedance
wave of the e c g to the peak of dz/dt) changes
cardiography signal: addition of a reference input and
scanning each beat more than once. On data synthesized proportionally, then the peak of dz/dt will occur at a
from real ICG which exhibits beat by beat changes and different time with respect to the peak of the r-wave,
baseline variations, we used statistical measures to show hence, ensemble averaging will smear the peak of
that the modified algorithms give accurate beat by beat dz/dto
estimation conversely to the original version of LMS.
When accurate determination of cardiac output is
crucial, impedance cardiography may be used in
conjunction with a standard technique to establish a
baseline reference, thereby permitting further analysis.
The ease and precision of this technique warrants its
more widespread use in the assessment of pacemaker
patients.
I. INTRODUCTION
technique is computationally cumbersome and slow. was used to eliminate the breathing component
because it allowed averaging over exactly two
The purpose of our research was to develop a complete breathing cycles and ten cardiac cycles
digital filter that can remove respiratory and locked in phase with the respiration cycle. For
movement artefact signals without attenuating the normals at rest and during exercise, the ratio of
cardiac signal or causing phase distortion. heartbeats to respiratory cycles is between 4 : 1 and 6
In Impedance Cardiography, there is partial : 1 (hildebrandtand Dauman, 1957; Almasi,1972).
overlap between z(t) and respiratory artifact spectra Asking a subject to breathe at a fixed ratio of 5 " 1
and statistical properties of both vary. hence non- heartbeats per respiratory cycle does not cause any
adaptive digital filter is not effective in removing significant stress or discomfort. A tone generator
respiratory artifact from z(t). spectral subtraction triggered by the ECG was turned on for two beats,
method [21] which involves estimating noise signalling inspiration, and off for three beats,
magnitude spectrum and subtracting it from the signalling expiration. Before designing the digital
contaminated spectrum, cannot be applicable here filter the power spectra of the dZ/dt signals were
because respiratory artifact spectrum cannot be examined. The first spike in the power spectrum
dynamically estimated using average or quantile corresponds to the breathing frequency and the higher
based estimation. However, an adaptive filter [22], spikes correspond to the cardiogenic components.
[23] can be employed to dynamically change its Examining the spectrum shows that a filter with a
transfer function to remove respiratory artifact. very narrow transition band is required. A higher-
Respiratory artifact is the variation in the sensed order filter (sixth order) was chosen because it would
thoracic impedance, caused primarily by change in be able to filter out breathing artefacts without
the dimension of thoracic cage during inhale and causing significant attenuation of the cardiac signal.
exhale phases of respiration. Hence air flow during A Butterworth filter was chosen because of its fiat
respiration is directly related to the respiratory frequency response in the pass band and the stop
artefact and can be used to provide a reference for band. A recursive filter was used instead of a finite
adaptive cancellation of respiratory artifact from the impulse response (FIR) filter to avoid distortions that
z(t) signal. the respiratory signal acquired from a can be introduced due to windowing. Another
respiratory sensor, simultaneously with z(t) signal, is advantage of an infinite impulse response (IIR) filter
taken as reference input, contaminated z(t) being the is the computational economy.
primary input. the objective is to adaptively remove The IIR filter designed had a nonlinear phase shift
the respiratory artefact from the z(t) signal without which could cause distortions in the waveform.
having to control the respiration. Advantage of this Back filter-ing was used to eliminate the distortions
method is that it does not require any control of that could be caused by the phase shift. In back
respiration. Respiratory signal can be acquired during filtering the data are passed through the filter twice,
exercise. Hence beat-by-beat stroke volume first in a normal manner, forward in time, and then
calculation is possible, even if respiratory artifact has in a reverse manner, backward in time, which
large variation. Also, beat-by-beat variation in stroke cancels the phase shift caused by the forward filter.
volume and Tlvet may provide additional diagnostic
information about cardiovascular functioning. It is to A Signal Processing
be noted that the technique suppresses additive
respiratory artifact. The stroke volume itself gets We acquired four signals, each sampled at
modulated by the respiratory cycle. This modulation 1000 Hz and processed in 1-min segments:
should remain unaffected by the proposed technique. respirometry, spirometry, dZ/dt, and Z. Because the
Z signal is coupled inductively to the output of the
II. METHOD Minnesota impedance device, this signal is subject to
low frequency filtering and distortion. Consequently,
The standard reference Z and dZ/dt signals used we derived a measure of Z ( Zd) by integrating the
to compare with the different filtered outputs signals dZ/dt signal (to reconstruct Z undistorted by this
was obtained by ensemble averaging ten complete inductive filtering). The Zd displayed a closer
cardiac cycles over two complete respiratory cycles. correspondence to direct spirometric measures of
Voluntary cardio respiratory synchronisation respiration than did the Z output of the Minnesota
(VCRS), developed by Almasi and Schmitt (1970), device and was used for the primary analyses.
was used to achieve a lock between the respiratory After decimating each signal to 250 Hz, we applied a
and the cardiac signal and time stationarily. With this point by-point integration to the dZ/dt signal and to
technique the breathing cycles are synchronised with the spirometry signal. All four signals respirometry
the cardiac cycles in a predetermined ratio. VCRS Z, Zd, and spirometry were low pass filtered (0.5
154
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Hz), using an interpolated finite impulse response filter to shift the coefficients to match for the
filter, to remove cardiac signals and high frequency difference in the delay. Adaptive filter output r(n) is
noise. subtracted from contaminated input z(n) so that the
All signals were then linearly detrended and total output error e(n) is progressively minimized.
zero-meaned, and a power spectrum Fast Fourier Fig. 2.Fig. 2. Adaptive respiratory artifact canceller
Transform (FFT) was calculated for each minute to For adaptation, we use the least mean square (LMS)
decompose the signals in the frequency domain. This algorithm, also known as stochastic gradient
method provided descriptive characteristics of the algorithm [22], [23]. The LMS algorithm is used to
respiration, and served as the basis for weighting of control the digital FIR Filter.
transfer functions as described below. In addition, a
Hanning window was applied to each of the four C. Signal Analysis
signals, and transfer function analyses were used to
compare Zd, Z, and respirometry (outputs) to the The recorded thoracic impedance signal was low
criterion spirometry signal (input). Transfer function pass filtered to attenuate high frequency noise. A 2-
analyses were run on each 60-s epoch, and were point differentiation was used to obtain ICG
based on seven 15-s overlapping windows to waveform. The resulting ICG waveform exhibited
minimize problems of nonstationarity and to smooth strong presence of respiratory artifact, particularly in
the resulting functions (Oppenheim & Schafer, 1989). the post-exercise recordings. The adaptive
These analyses yielded estimates of the parameters of cancellation scheme of Fig. 2 was implemented for
coherence (similarly of waveform morphology), gain different combination of filter tap length and delay in
(amplitude ratio) and phase relations of the signals. the primary signal path. Based on maximum eigen
The transfer functions provide parameter estimates value of autocorrelation matrix of the input signal
over the entire range of frequencies within the z(n), = 0.0044 was used for adaptation. Most
analytical window (0.050.50 Hz), whereas the satisfactory results were obtained for tap length M
distribution of respiratory power was more restricted. 200 and sample delay 70. The filter coefficients
Because respiratory frequencies are not uniformly were found to get stabilized in approximately 500
distributed across the frequency spectrum, the samples.
transfer functions were weighted using the output Because peak dZ/dt is an important parameter in
power spectrum. This method of weighting provided the calculation of stroke volume, the percentage error
more meaningful summary statistics of these was cal-culated for this parameter. The peak height of
parameters, appropriately scaled by the actual the unfiltered ensemble averaged waveform was
frequency distribution of power (for general rationale taken as the standard signal. The same data were then
for the weighting procedure, see Porges et al., 1980). passed through the filter with cutoff frequencies
The value of each point of the transfer function was selected as a fraction of the heart rate. The percentage
multiplied by the relative amplitude of the power error was then calculated between the peaks of the
spectrum of the output signal at that frequency standard waveforms and the filtered data with various
(expressed as a proportion of total cut-off frequencies.
power). This calculation yielded an overall average
estimate of the transfer function parameter, The percentage peak dZ/dt error plotted against
appropriately weighted by the relative power of the different cutoff frequencies. The percentage peak
specific respiratory frequency contributions to that dZ/dt error for all conditions was less than 2 percent
estimate. for a cut off frequency of less than 50 per cent of the
heart rate.
B. Adaptive Cancellation Scheme
A schematic of adaptive respiratory artifact
canceller is shown in Fig. 2. Signal z(n) is the
thoracic impedance signal contaminated by
respiratory artifact from the thoracic impedance
sensor. Reference signal r(n) is obtained from the
respiratory sensor. The respiratory sensor used here
senses the airflow at nostrils, and hence the reference
signal has a delay with respect to the respiratory
artifact because of the movement of thorax cage. To
compensate for this, a delay of nd samples is
introduced in signal path of z(n). Number of taps in
the FIR filter should be large enough for the adaptive
155
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
156
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Authors Biography
157
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
158
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
159
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
160
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
161
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
162
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
163
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
164
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
ACKNOWLEDGEMENTS
165
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract- This paper describes the State-of-the art composite signal, one Interplex Signal is added.
modulation scheme designed and developed for the The modulator system generates the BOC (5,2)
RNSS (Regional Navigation Satellite System). It spreaded signals for RS service and then modulates
elaborates theory, design, implementation and these signals along with the OS signal.
important results of the modulator. The modulator unit There are three baseband navigation signals as
generates the composite modulated signal with following:
constant amplitude vectors which uses the BPSK
(Binary Phase Shift Keying) modulation and BOC
(Binary Offset Carrier) modulation to provide OS 1. OS Data Signal (sos(t))
(Open Services) for civilian purposes and RS 2. RS BOC Pilot Signal (srs_p(t))
(Restricted Services) for military applications 3. RS BOC Data Signal (srs_d(t)) The sub-carrier is
respectively using Interplex technique. Here sinusoidal defined as :
BOC (5, 2) modulation scheme is chosen for RS signal.
scx(t,) = sgn[sin(2fsc,xt + )] (1)
The first constraint that any new signal has to fulfill is it
should be efficiently share the allocated spectral bands where fsc = Subcarrier frequency;
with the already existing signals, without the = Subcarrier phase
degradation on the other signals to become too high. So
the new spreading modulation technique (BOC The IRNSS RS data and pilot BOC signals are
Modulation) is designed to meet the frequency sinBOC. Hence the subcarrier phase =0. The
compatibility which is vital for the coexistence of
complex envelope of composite signal with Interplex
different GNSS (Global Navigation Satellite System).
BOC spreading modulations are one way to accomplish signal I(t) is:
this. s(t) = [2 (sos(t) + srs_p(t)) + j(2. srs_d(t) I(t))]
Keywords- RNSS, BPSK, BOC, OS, RS, CASM, (2)
Interplex. The Interplex signal is generated from the navigation
signals to realize the constant envelope composite
I. INTRODUCTION signal.
166
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
167
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Fig.4 (a) Simulated Spectrum with higher measuring Fig.6 (a) Measured Spectrum of in-house space-
frequency qualified modulator
Fig.4 (b) Simulated Constellation Plot Fig.6 (b) Measured Constellation of in-house
The above described modulator system was developed space-qualified modulator
developed and tested in-house as shown in fig.5. Measured Results o f d e s ig n e d mo d ulato r are
Measured spectrum and constellation of designed shown in table 2 along with the specifications.
hardware are shown in the fig.6.
V. CONCLUSION
Fig.5 In-house developed space qualified Modulator
with test setup The BOC/Interplex modulator has successfully been
developed and the results of ground testing are quite
satisfactory, meeting the simulated ones. This
modulation scheme using code spreading as BOC is
efficient for bandwidth utilization where the same
168
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
REFERENCES
169
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
170
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
frequency and time which can be obtained from the State equation:
satellite ephemeris. These measurements of pseudo- xk = Axk-1 + Buk + wk-1... (ii)
ranges are noisy and not accurate. The accuracy of Output measurement equation:
final position and velocity vectors depend upon the zk = Hxk + vk... (iii)
accuracy of the pseudo-ranges. The errors introduced
Where x n is the state of a discrete time controlled
in the pseudo-ranges can be assumed to be random
and Gaussian in nature. process to be estimated and z m is the
The observable in GPS orbit determination is the measurement. A relates the state at the previous
time step (K-1) to the statesat the current step (K), in
pseudo-ranges () which means the apparent
the absence of either a driving function or process
distance between tracked jth GPS satellite to user
noise. B relates the optional control input u l(if
LEO spacecraft. This pseudo-range (j) can be
available) to the state x. H in the measurement
obtained by
equation relates the state to the measurement z k. The
= x x + y y + z z random variables wk-1 and vkrepresent the process and
measurement noise respectively. They are assumed to
+ ( ) be independent, white, and with normal probability
where (xj , yj , zj ) are the coordinates of jth GPS distributions
satellite, (xu , yu , zu ) are the coordinates of user p(w) ~ N(0,Q)
LEO spacecraft, tu is an offset of receiver clock from p(v) ~ N(0,R)
the system time, and c is a speed of light (c = Where Q and R are process noise covariance and
299,792,458m/s). The receiver estimates the pseudo- measurement noise covariance respectively and these
range. The GPS satellite position can be calculated by might change at every time step.
using the demodulated navigation data received from Each of these quantities is vectors and therefore
GPS satellites. The value of coordinates xj,yj and zj contains more than one element. The vector x
can be computed from the ephemeris of that contains all of the information about the presentstate
particular satellite from the demodulated navigation of the system, but we cannot measure x directly.
data. Four equations are needed to solve for four Instead we measure y, which is a function of x that is
variables xu,yu, zuand tu. Thus a minimum of four corrupted by the noise z. We can use y to help us
visible GPS satellites are required to solve for LEO obtain an estimate of x, but we cannot necessarily
spacecraft position vectors. take the information from y at face value because it is
corrupted by noise.
III. KALMAN FILTER The state estimation in kalman filter is a linear
combination of priori state estimate and the weighted
Kalman filter is a data processing algorithm difference of the measurementZkobtained and the
that estimates the states of a linear system from predicted measurementH x K -.This difference(Zk - H x
indirect, inaccurate and uncertain observations. It is a -
K )is called as residual. The weighting factor is called
set of mathematical equations that implement a as a gain or the blending factor and can be obtained
predictor-corrector type estimator that is optimal. by
This means that it minimizes the mean square error of Kk = P K -HT (H P K - HT +R)-1
the estimated parameters. Kalman filter is recursive
in nature, so the new measurements are processed as This value of gain is such that it minimizes the
soon as they are available. This makes it capable of posterior error covariance P K. The Kalman filter can
performing real time filtering of the noisy system be thought of as a feedback control system. It
variables corrupted by guassian white noise. The estimates the process state at some time and then
primary objective of Kalman Filter is to estimate the obtains feedback in the form of noisy measurements.
states of a discrete-time controlled process that is As such, the equations for the Kalman filter fall into
governed by the linear stochastic difference equation. two groups:
The extended Kalman filter (EKF) is a nonlinear 1.) Time update equations predict the state and
version of the Kalman filter and is widely used for error covariance estimates at current time
position estimation in GPS receivers. The EKF is using state model to obtain the priori estimates
formulated with the assumption that the measurement for the next time step. The time update
noise is white. equations are responsible for projecting
forward in time the current state and error
MATHEMATICAL FORMULATION OF KALMAN FILTER covariance estimates to obtain the priori
estimates for the next time step.
A linear process can be described by the following Time Update Equations
two difference equations:
xk = Axk-1 + Buk ... (iv)
171
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
172
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
where x is the state vector, v is the additive The algorithm is implemented on the hardware based
measurement noise; andH the pseudo-range on Xilinx FPGA and ARM7TDMI processor. It is a
measurement matrix and is obtained by linearizing part of the navigation solution computing the state
the pseudo-range equation with respect to the filter vectors every one second. Code to implement the
states algorithm was optimized tomeet all the timing
The linearized measurement matrix is given by constraintsand compute the filtered solution along
with the conventional method once in a second.
( ) ( ) ( )
= 1 0 0
Algorithm is tested by obtaining the raw
Measurement noise matrix can be given by measurements from the GPS simulator simulating the
orbit scenarios. Simulations were carried out on
0 0 dataset generated from high dynamics GPS simulator
0 0 that simulates the GPS constellation. It also simulates
= the highly dynamic orbits of the LEO spacecraft.
Pseudo-ranges were obtained from the GPS
0 0 0 simulator, along with the ephemeris of GPS satellites.
N is the number of satellites being tracked and is This high dynamics signal was given to the GPS
the measurement noise variance for the range receiver which processes it and provides the state
measurements ofith satellite. vectors. State vectors were obtained from both the
iterative method as well as from the Kalman filter
algorithm. State vectors were compared and
significant improvement in 3-dimensional accuracy is
observed.
V. IMPLEMENATION
15
The steps for implementation of Kalman Iterative
Filtered
filter are shown in fig3. 10
5
Error(m)
-5
-10
0
between the projected range j and the measured
ranges cj
-5
= +
-10
The residual Z can be used to estimate the states at
current time step. The predicted error covariance
-15
estimates are given by (v). The Kalman Gain Kkcan 0 1000 2000 3000 4000 5000 6000 7000 8000 9000
Samples
be computed using (vi) and error covariance P K
using (viii). Fig. 5: Error in y position
173
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
-60
-80
-100
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
Samples
V. CONCLUSON
REFERENCES
174
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract IRNSS is an Indian Regional Navigation a) The space segment consists of three satellites
Satellite System, which relies on constellation of in GEO orbit and four satellites in GSO orbit
regional navigation satellites to provide navigation inclined at 29 to the equatorial plane. IRNSS
services to various users in India and a region extending
satellites have two types of payloads, navigation
about 1500 km around India with a user position
accuracy of less than 20 meters. This paper addresses and ranging payload. the navigation payload
how the IRNSS measurements have been simulated to will transmit the ranging codes being generated
test and validate the navigation software in the absence onboard with the navigation data uplinked from
of IRNSS satellites one way range measurements. This the ground stations.
simulation software will simulate pseudo range
measurements in meters and carrier phase b) The ground segment is responsible for
measurements in cycles. maintenance and operation of IRNSS constellation
which comprises of seventeen one way ranging
IRNSS one way range measurements on L5 stations (IRIMS) spread over Indian landmass and
(1176.450MHz) and S (2492.028MHz) frequencies for around as shown in Figure 2, four two way ranging
all possible line of sights between seven IRNSS satellites stations, nine laser ranging stations, one centre for
(i.e., IRNSS 1A to 1G) and identified IRIMS (IRNSS health monitoring, analysis and control of IRNSS
Range and Integrity Monitoring Station) which are satellites, one precise time generation centre for
distributed in and around Indian land mass. The
IRNSS system time maintenance, and one centre for
simulated range measurements includes all the possible
errors such as free space delay, atmospheric delays, IRNSS navigation parameters generation by
clock offsets, multipath, phase center offsets, noises etc navigation software.
as close to the real measurements in RINEX (Receiver
Independent Exchange Format). And the paper c) The IRNSS user segment comprises of single and
discusses motivation and importance of this work along dual frequency navigation operating on L5/S and L5
with different error models included in the simulated and S frequencies respectively.
range measurements.
Keywords - MEASUREMENTS SIMULATION, SIGNAL IRNSS one way range measurements are the
DELAY MODELS primary source for the generation of navigation
parameters. In the absence of real time IRNSS range
measurements there is a need of a simulator which
I.INTRODUCTION
simulates the IRNSS one way range measurements on
IRNSS envisages establishment of regional both L5 and S frequencies, this simulator is
navigational satellite system using a combination of developed and used to generate the range
3GEO (Geo Stationary Earth Orbit) and 4GSO (Geo measurements on both L5 and S frequencies.
Synchronous Orbit) Satellites as shown in Figure 1.
IRNSS will provide basically two types of services:
Standard Positioning Service (SPS) and Restricted
Service (RS). Both of these services will be provided
at two different frequencies one in the L5 band and
the other in S-band.
Thus, user receiver can be operated in single and /or
dual frequency operation mode. IRNSS is expected to
provide position accuracy better than 20 meters over
India and a region extending about 1500 km around
India. IRNSS consists of three segments namely the
space segment, the ground segment and the user
segment.
Figure 1: IRNSS Satellites constellation
175
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
176
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
177
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
satellite and HASSAN station a clock offset with respect to IRNSS time resulting in
random behavior of clock. These satellite and receiver
clocks offset errors along with randomly occurring
phase jumps can be simulated using following two
models and the corresponding graphs are shown in
figures 10 and 11
1) Polynomial method:
Clock Error (t) can be modeled using
t = af0 + af1 (t - toc) + af2 ( t toc)2 +
Where,
Figure 8: Variation of Ionospheric Delay for S Signal using af0 = Clock bias, af1 = Clock drift
MLDF grid Model with IRNSS time for LOS between IRNSS 1A
satellite and HASSAN station af2 = Frequency drift or aging,
toc = Clock data reference time
2) Tropospheric models
t = Current time epoch, =clock noise
Troposphere stretches to about 16km above the
equator and about 9km above poles from earth 2) Realistic clock Errors
surface. The general empirical models of the IGS precise clocks data available in public
troposphere depend on the station height, elevation domain is used to generate clock offsets with a
angle, pressure, temperature and relative humidity in provision to add phase jumps and simulated
signal line of sight. The simulator would be able to noise.
choose any of the following tropospheric delay
models and it has provision to take real time
meteorological data or computed meteorological data
using mathematical models as input.
178
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Hardware delays
179
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
V. CONCLUSION
180
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
ACKNOWLEDGMENT
The authors would like to thank Shri A.S.
Ganeshan, Program Director, Satellite Navigation
Program and Sri S.C. Rathnakara, Division Head,
Navigation Software Division for many useful
conversations regarding this project. The authors also
wish to acknowledge all the members of Space
Navigation Group, ISAC for their motivation and
help.
REFERENCES
181
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract This paper describes the novel scheme for code on 5.115MHz sub carrier. The major
design and development of Actel FPGA based functions of NSGU are to receive uplink NAV
Navigation Signal Generation Unit (NSGU) for the data, store it in onboard memory, extract data
navigation payload, whose key function involves storing, depending upon time of applicability and validity
formatting and processing of bulk uplinked Navigation period, add timing information and format to
(NAV) data to prepare base band NAV signal for construct NAV subframes, process these subframes
onboard modulators. There are two critical prerequisite to prepare baseband signals for onboard modulators
from NSGU which will help in achieving position
of L5 and S band respectively. NAV data
accuracy < 20m on ground. One of the requirements is
to get Timing Delay bias in Nano seconds for which a Processing in NSGU includes Forward Error
novel method is adopted for design of NSGU using Correction (FEC) encoding, interleaving of encoded
master clock of 10.23MHz to achieve Timing Delay bias symbols, encryption and spreading with onboard
in nano seconds. Second requirement is setting of generated codes for SPS and RS services and
onboard time with ground reference time in nano selection of different data patterns at the output of
seconds which is most crucial necessity for position NSGU. This paper is further outlined as follows.
accuracy. An innovative scheme is developed in which Section II discusses the basic architecture of NSGU.
time synchronization between onboard and ground can Section III discusses the FPGA design for NSGU for
be achieved in nano seconds. This paper describes the
processing of NAV subframe to achieve Timing
design methodology of NSGU in Actel FPGA for
realization of Timing Delay bias in nanoseconds and Delay bias in nanoseconds. Section IV illustrates the
procedure for time setting to achieve time generation of onboard time and method of time
synchronization in nano seconds. synchronization between onboard time with ground
Keywords: NSGU, SPS, RS, NAV, Timing Delay Bias, reference time. Section V concludes the paper.
TOWC, WN, FPGA.
II. BASIC ARCHITECTURE OF NSGU
I. INTRODUCTION
As shown in Fig.1 NSGU consists of three major
Satellite Navigation has revolutionized the navigation functional units. SUMMIT DXE device for receiving
world, opening new opportunities in an increasing uplink NAV data on MIL-1553B interface to
number of sectors that requires high precision TCTM, Micro controller as main controller and for
positioning and timing information [2][4]. Generally formatting of NAV data subframes which uses
any Navigation satellite system targets basically two external RAM and PROM for storage/archival of
types of services named as Standard Positioning NAV data and program memory for software. Actel
Services (SPS) and Restricted services (RS). In FPGA which is the target hardware for complete
Navigation payload the Atomic Frequency Standard processing of formatted NAV data, timing generation
(AFS) acts as a basic clock source for the payload and setting, Encryption of RS code and spreading of
configuration which in turn provides master clock processed NAV subframe with SPS and RS codes.
source of 10.23MHz for NSGU and other The uplinked NAV data received from onboard
subsystems of payload. The NAV data is up TCTM through MIL 1553B protocol are stored in
linked periodically from ground and received by SRAM. Micro controller reads NAV data from
NSGU through onboard Telecommand-Telemetry SRAM and formats into subframe as per signal
(TCTM) unit. This NAV data is stored in NSGU, structure document. NAV subframes are then
formatted with onboard time and processed to transferred to FPGA memory. The complete
prepare as base band modulating signal for L5 and S processing of NAV subframe is carried out in FPGA
band onboard modulators. Each of L5 and S band of NSGU and processed frames are provided as base
provides SPS service with BPSK modulation with band signals to L5 and S band modulators. Fig. 2
1.023Mbps Spreading code and RS service with shows Flight Model (FM) of NSGU which consists
BOC ( 5, 2 ) modulation o n 2.046Mbps spreading of the main and redundant units.
182
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
183
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
purpose [3]. In Navigation payload this onboard uploaded from ground, so that the time taken by
time is presented by a 27 bits counter with LSB telecommand to reach onboard and other processing
17 bits as Time of week counter (TOWC) and MSB delays doesnt affect the scheme, because difference
10 bits as week number (WN). TOWC and WN are of onboard time and Ground Reference time always
generated in FPGA of NSGU. remain within clock drift limit. After this the
onboard and ground time are set with accuracy of
A. TOWC Generation nanoseconds.
S-SPS epoch@1ms is used for generation of
TOWC in FPGA of NSGU. A novel scheme
consisting of counters [1] is implemented for TOWC
generation, which helps in achieving time
synchronization between onboard and ground in
nano seconds
184
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
185
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
II. CASE STUDY AND RESULT ANALYSIS Figure 4: Matched eye template from the database.
Figure 4 show that the resultant matched template
After processing and analyzing the iris signal,
from the database. The Hamming distance
The template of the each database signal have been distribution and matching condition between the
plotted in Matlab as shown in corresponding figures. templates are shown below. The result is plotted in
The results are plotted in terms of scales (Hamming terms of scales (Hamming distance) on the y-axis
distance) on the y-axis and the number of images in and the number of images in the database (sample
the database (sample images) on x-axis. The images) on x-axis.
matched result at that point is represented by a blue
color bar with the threshold value. Red bar indicates In figure5, the blue bar at position25 indicates that
that unmatched condition. In this exposition two matching condition below the threshold value. Since
cases are considered under test. database contains the different persons data,
For every Image in the dataset the following values tested image is matched with the best compatible
have been analyzed: image. The image is being used for test is matched
1.Iris radius and its center coordinates. with processed left eye images in the database
2.Filter parameters. corresponds to the same person, the Hamming
distance between tested template and matched
186
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
template is taken as the optimum threshold horizontal black line, it will be differed for each
value for the same category. In the above example individual test. Hence test of above signal is done
the optimum threshold value is set programmatically successfully and the corresponding observations will
about 0.4, it will be differed for each individual test. be discussed in section 4.
Figure 6: Test right sample template. Where d' is decidability, mean of the intra-class
After processing and analyzing the iris signal, the distribution s and the mean of the inter-class
template of the database signal has been plotted distribution d and also the standard deviation of
as shown in above figure. In figure6 the image is the intra-class and inter-class distributions 2, and
being tested S1094R03.jpg, name reveals that third 2d respectively. The results of experiment, various
right eye sample image of a 94th person in the filter parameters to encode iris templates are
database. observed below table1.
Figure 7 show that the resultant Hamming distance
distribution and not-match condition between the
templates are shown below, all the bar lines
exceeds the threshold value line 0.4, hence there is
no match at any case, the result is plotted in terms of
scales (Hamming distance) on the y-axis and the
number of images in the database (sample images) on
x-axis.
187
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
As the number of shifts increases, the mean of the This exposition has presented an Inter-class iris
inter-class distribution will converge to an optimum recognition system, which was tested using set of
value below the statistical value 0.5. It can be seen in images as databases of greyscale eye images from
figure 9.
different. Firstly, an efficient iris processing method
is processed. In the processed method the iris
segmentation, normalization, feature extractions are
being done. Then the encoding technique is applied
to extract the unique feature of iris image, and
188
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
REFERENCES
[1] http://www.csse.uwa.edu.au/~masekl01/
[2] Recognition of Human Iris Patterns for Biometric
Identification, Libor Masek The University of Western Australia,
2003
[3] http://en.wikipedia.org/wiki/Iris_(anatomy)
[4] Canny, John, "A Computational Approach to Edge Detection,"
IEEE Transactions on Pattern Analysis and Machine
Intelligence,Vol. PAMI-8, No. 6, 1986, pp. 679-698.
[5] "A Human Identification Technique Using Images of the Iris
and Wavelet Transform" W. W. Boles and B. Boashash
[6] "Iris Recognition: An Emerging Biometric Technology"
RICHARD P. WILDES, MEMBER, IEEE
[7] Optimized Daugmans Algorithm for Iris Localization Dr.
Mohamed A. Hebaishy National Authority for Remote
Sensing and Space Science Gozif Titp St., Elnozha Elgididah.
Egypt (11769), Cairo.
[8] Recognition of Human Iris patterns for biometric
identification, Libor Masek, Dr.Peter Kovesi, The University of
western Australis.
[9] Exploit modeling for categorization of electrocardiogram
signals through wavelet transform. prof.P.V.Ramaraju,
V.Malleswara Rao, Mr CH. Ganesh. 6th international multi
conference on intelligent system, IISN-2012, March 16-18,NIT
Klawad,Haryana,India.
[10] ECG signal processing and classification via heart rate
fortitude in the vicinity of continuous wavelet
transforms.prof.P.V.Rama raju, V.Malleswara rao and Mr
CH.Ganesh, IEEE conference, International journal ACNCN-
2012,AU college of engineering, Visakhapatnam.
189
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract: Accuracy and repeatability of IRNSS (Indian Earth Orbit) orbit and four satellites in GSO
Regional Navigation Satellite System) is ultimately (Geosynchronous orbit) orbit. All the IRNSS
determined by the inherent stability of the on-board satellites are intended carry 3 RAFS: one primary
clocks. In IRNSS, the users position is determined by and two redundant (hot redundant and cold
measuring the time of travel of the signals broadcasted redundant).
from satellites. Hence, any error on the measurement
Accuracy and repeatability of IRNSS will be
of time leads to an error on the users position.
Therefore the on-board clock behaviour has to be
ultimately determined by the inherent stability of the
continuously monitored and any malfunctioning has to on-board clocks. The variation in the clock
be noticed immediately to improve the capability of the frequency or phase with time directly affects the
timing system for accurate positioning of the user. accuracy of the navigation solution. The Overlapped
Each IRNSS satellite carries three Rubidium Atomic Allan variance (OAVAR) is a standard statistical
Frequency Standards (RAFS), one primary and tool to estimate clock stability which is a
two redundant clocks (Hot redundant & Cold fundamental quantity used to assess the performance
redundant). Allan variance is commonly used to of a clock and its impact on the performance of the
estimate the stability of IRNSS satellite clock. The
system that uses the clock [1,2]. In navigation
classical Allan variance cannot capture the non-
stationarity in the random signal and it will give the systems, a stable clock is predictable and yields a
average stationary behavior over the whole low user range error, resulting in better positioning.
period. The stability of IRNSS satellite clock can A real time computation of Allan variance has been
change with time due to several factors, such as implemented to study instantaneous clock stability
temperature, humidity, radiations, aging and sudden and thus detect the possible non-stationary behaviour
breakdowns. Hence, it is important to represent the and related clock anomalies.
time-varying stability of the clock. In the present paper the initial results of the IRNSS-
This paper provides the In-Orbit clock performance 1A clock performance are shown using conventional
of IRNSS RAFS using both conventional Overlapping
Overlapping Allan Variance (OAVAR) and the real
Allan Variance (OAVAR) and the newly proposed Real
time Allan Variance (RAVAR). RAVAR has been time Allan variance (RAVAR) computation.
implemented to study instantaneous clock stability and
thus detect the possible non- stationary behaviour and II. SATELLITE CLOCK BEHAVIOR
related clock anomalies such as data gaps, outliers,
phase jumps and frequency jumps. IRNSS payload A thorough understanding of the behaviour of the
records the phase difference between primary and satellite clocks is required in order to use the current
secondary RAFS and transfers it to ground station system optimally and to plan future improvements
through telemetry(TM). Along with the stability the in upcoming navigation satellites. Deterministic and
temperature of RAFS and attitude variations of the
stochastic perturbations cause the variation of
IRNSS-1A are also monitored.
clocks frequency stability and accuracy. The main
Keywords: Satellite clock, Allan variance, Atomic clock, deterministic behaviour in atomic clocks is the drift
Clock characterization.
that is a systematic variation of the frequency
generated by the clock. It may be due to intrinsic
changes in the clock itself, to the ageing or to
I. INTRODUCTION environmental factors. Generally, the satellite clock
behaviour is mainly analysed in terms of:
Indian Space Research Organization (ISRO) is
implementing the Indian Regional Navigation 1. Clock deterministic behaviour: Bias, Drift, and
Satellite System (IRNSS) to provide navigation Drift- Rate.
services to various users over the intended service 2. Clock stochastic behaviour: Frequency
area. It is an independent navigation system of India Stability.
designed to provide position accuracy better than Frequency stability indicates the degree to which an
20m for dual frequency users over India and the oscillating signal produces the same frequency for a
region extending about 1500 Km around India. The specified interval of time. Depending on this
space segment of IRNSS will consist of seven information, one can judge whether the clock is
satellites: three satellites in GEO (Geo-stationary behaving as per the specification or not. If it is not
190
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
meeting the specification, switching over of primary III. LIMITATIONS OF OVERLAPPED ALLAN
clock is done or satellite clock is physically adjusted. VARIANCE
The OAVAR is the widely used statistics to from a set of M frequency measurements for averaging
characterize the frequency stability of atomic clocks, time t=mt0, where m is the averaging factor and t0 is
but it has its limitations that it cannot capture anomalies the basic measurement interval, by the expression
and the nonstationarity in the random signal. It simply
indicates the average stationary behavior of the clock
for the given period. But the OAVAR fails to provide
the stability of the clocks under the following cases. Now in real time for each next coming data point, the
1. Data Gaps: During the IOT since various satellite first data point will be removed from the existing data
subsystems are tested, continuous availability of so that always same length of data will exist. The
telemetry data from onboard phase meter was not proposed algorithm follows as:
guaranteed, thus leading to data gaps. a) Outlier screening and removal of bad data.
2. Clock Anomalies: The anomalies such as phase b) Computation of frequency accuracy
jump, frequency jump and outliers cannot be detected c) Phase jump detection and computation of its
by OAVAR since they are averaged out in the process. magnitude.
d) Frequency jump detection and computation of its
The proposed algorithm (RAVAR) with real time clock magnitude.
stability technique is designed to provide the clock e) Computing OAVAR corresponding to first data
stability measure even with data having the above point
discussed problems. In order to show the effectiveness f) Computing OAVAR corresponding to last data
of proposed algorithm in detecting time-varying clock point
behaviour that are not often visible from the OAVAR, a g) Computing final OAVAR for currently existing
simulation of all clock related anomalies was done and data length by removing and adding the effect of first
the results are discussed in the later section. and last data points OAVAR respectively.
h) Take next coming data point and remove first data
IV. PROPOSED ALGORITHM FOR REAL-TIME point so that the data length is always same.
CLOCK For each new epoch all the above steps will be repeated.
CHARACTERIZATION So the output will give the real time stability of the
clock behaviour and also the real time information such
Phase meter will give the offsets between the on-board as time and magnitude about the clock anomalies.
primary and hot redundant clock. The proposed
algorithm is designed to work for both clock phase data V. RESULTS AND DISCUSSIONS
and frequency data.
To show the effectiveness of the proposed clock
stability technique in case of clock anomalies, a clock
phase data simulation has been done with anomalies
such as outliers, clock jumps and data gaps. The results
of the study are shown in the below section.
Simulated data:
The simulated frequency data with phase jumps and its
stability plots are as shown in figure 2 and figure 3. The
phase jumps can be observed from figure 3 as
jumps/spikes in the surface plot.
191
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
192
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Fig.10 OAVAR for RAFS1 RAFS3 motivation, help and their constructive comments on
the real time characterisation of on-board clocks.
REFERENCES
ACKNOWLEDGMENT
193
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract Avisual cryptography scheme (VCS) is a kind (i.e., sharing secret image), have been found, for
of secret sharing scheme which allows the encoding of a example, authentication and identification,
secret image into shares distributed to participants. The watermarking and transmitting passwords etc.
beauty of such a scheme is that a set of qualified The associated secret sharing problem and its
participants is able to recover the secret image without physical properties such as contrast, pixel expansion,
any cryptographic knowledge and computation devices. and color were extensively studied by researchers
An extended visual cryptography scheme (EVCS) is a worldwide. For example, showed constructions of
kind of VCS which consists of meaningful shares threshold VCS with perfect reconstruction of the
(compared to the random shares of traditional VCS). In black pixels.
this paper, we propose a construction of EVCS which is
realized by embedding random shares into meaningful
covering shares, and we call it the embedded EVCS.
Experimental results compare some of the well-known
EVCSs proposed in recent years systematically, and show
that the proposed embedded EVCS has competitive visual
quality compared with many of the well-known EVCSs in
the literature. In addition, it has many specific advantages
against these well-known EVCSs, respectively.
Furthermore, Eisenet al. proposed a construction of
I. INTRODUCTION threshold VCS for specified whiteness levels of the
recovered pixels. The term of extended visual
The basic principle of the visual cryptography cryptography scheme (EVCS) was first introduced by
scheme (VCS) was first introduced by Naor and Naoret al. in, where a simple example of (2,2)-EVCS
Shamir. VCS is a kind of secret sharing scheme that was presented. In this paper, when we refer to a
focuses on sharing secret images. The idea of the corresponding VCS of an EVCS, we mean a
visual cryptography model proposed in is to split a traditional VCS that have the same access structure
secret image into two random shares (printed on with the EVCS.
transparencies) which separately reveals no
information about the secret image other than the size Generally, an EVCS takes a secret image and original
of the secret image. The secret image can be share images as inputs, and outputs shares that satisfy
reconstructed by stacking the two shares. The the following three conditions: 1) any qualified subset
underlying operation of this scheme is logical of shares can recover the secret image; 2) any
operation OR. In this paper, we call a VCS with forbidden subset of shares cannot obtain any
random shares the traditional VCS or simply the information of the secret image other than the size of
VCS. In general, a traditional VCS takes a secret the secret image; 3) all the shares are meaningful
images.
image as input, and outputs shares that satisfy two
conditions:
II. IMPLEMENTAION AND METHODOLOGY
1) any qualified subset of shares can recover the
secret image; 2) any forbidden subset of shares We implement the Visual Cryptography. We used
cannot obtain any information of the secret image LZW Data Compression algorithm. The LZW data
other than the size of the secret image. An example of compression algorithm is applied for the gray scale
traditional (2,2)-VCS can be found in Fig. 1, where, image here. As a pre-processing step, a dictionary is
generally speaking, a VCS means any out of shares prepared for the gray scale image. In this dictionary,
could recover the secret image. In the scheme of Fig.
1, shares (a) and (b) are distributed to two the string replaces characters with single quotes.
participants secretly, and each participant cannot get Calculations are done using dynamic Huffman
any information about the secret image, but after coding. In compression of greyscale image select the
stacking shares (a) and (b). information pixels. Then generate halftone shares
the secret image can be observed visually by the using error diffusion method. At last filter process is
participants. VCS has many special applications, for applied for the output gray scale images. Filtersare
example, transmitting military orders to soldiers who used to improve the quality of reconstructed image to
may have no cryptographic knowledge or minimize the noises for sharpening the input secret
computation devices in the battle field. Many other
applications of VCS, other than its original objective image.
194
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
ALGORITHM: Decoding:
Input: The c x d dithering matrix D and a pixel with The decoding algorithm works by reading a value
gray-level g in input image I. from the encoded input and outputting the
Output: The halftoned pattern at the position of the corresponding string from the initialized dictionary.
pixel At the same time it obtains the next value from the
input, and adds to the dictionary the concatenation of
For i=0 to c-1 do the string just output and the first character of the
For j=0 to d-1 to do string obtained by decoding the next input value. The
If g<=Dij then print a black pixel at position (i,j); decoder then proceeds to the next input value (which
Else print a white pixel at position (i,j); was already read in as the "next value" in the
previous pass) and repeats the process until there is
no more input, at which point the final input value is
decoded without any more additions to the dictionary.
In this way the decoder builds up a dictionary which
is identical to that used by the encoder, and uses it to
decode subsequent input values. Thus the full
dictionary does not need be sent with the encoded
data; just the initial dictionary containing the single-
character strings is sufficient (and is typically defined
beforehand within the encoder and decoder rather
than being explicitly sent with the encoded data.).
Creating Transparencies:
195
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
cases: Secret sharing scheme is a method of sharing known schemes, such as the fact that it can deal with
secret information among a group of participants. In a gray-scale input images, has smaller pixel expansion,
secret sharing scheme, each participant gets a piece of is always unconditionally secure, does not require
secret information, called a share. When the allowed complementary share images, one participant only
coalitions of the participants pool their shares, they needs to carry one share, and can be applied for
can recover the shared secret; on the other hand, any general access structure. Furthermore, our
other subsets, namely non-allowed coalitions, cannot construction is flexible in the sense that there exist
recover the secret image by pooling their shares. In two trade-offs between the share pixel expansion and
the last decade, various secret sharing schemes were the visual quality of the shares and between the secret
proposed, but most of them need a lot of image pixel expansion and the visual quality of the
computations to decode the shared secret information. shares.
The basic 2 out of 2 visual cryptography model
consist of secret message encoded into two
transparencies, one transparency representing the REFERENCES
cipher text and the other acting as a secret key. Both
transparencies appear to be random dot when
inspected individually and provide no information [1] G. R. Blakley, Safeguarding cryptographic keys, in Proc.
National Computer Conf., 1979, vol. 48, pp. 313317.
about the original clear text. However, by carefully [2] T. H. Chen and D. S. Tsai, Owner-customer right
aligning the transparencies, the original secret protection mechanism using a watermarking scheme and a
message is reproduced. The actual decoding is watermarking protocol, Pattern Recognit., vol. 39, pp.
15301541, 2006
accomplished by the human visual system. The [3] M. Naor and A. Shamir, Visual cryptography, in Proc.
original is encrypted into 2 transparencies you need EUROCRYPT 94, Berlin, Germany, 1995, vol. 950, pp.
both transparencies to decode the message. 112, Springer-Verlag, LNCS.
[4] C. Blundo, A. De Bonis, and A. De Santis, Improved
schemes for visual cryptography, Designs, Codes and
Un-hiding Image from Transparency: Cryptography, vol. 24, pp. 255278, 2001.
[5] A. Shamir, How to share a secret, Commun.ACM, vol.
22, no. 11, pp. 612613, 1979.
The simplest form of visual cryptography separates [6] M. Naor and B. Pinkas, Visual authentication and
an image into two layers so that either layer by itself identification, in Proc. CRYPTO97, 1997, vol. 1294, pp.
322336, Springer-Verlag LNCS.
conveys no information, but when the layers are [7] N. K. Prakash and S. Govindaraju, Visual secret sharing
combined the image is revealed. One layer can be schemes for color images using halftoning, in Proc. Int.
printed on a transparency, and the other layer Conf. Computational Intelligence and Multimedia
Applications (ICCIMA 2007), 2007, vol. 3, pp. 174178.
displayed on a monitor. When the transparency is [8] P. Tuyls, T. Kevenaar, G. J. Schrijen, T. Staring, and M.
placed on top of the monitor and aligned correctly, Van Dijk, Security displays enabling secure
the image is revealed. For each image pixel, one of communications, in Proc. First Int. Conf. Pervasive
the two encoding options is randomly selected with Computing, Boppard Germany, Springer-Verlag Berlin
LNCS, 2004, vol. 2802, pp. 271284.
equal probability. Then, the appropriate colorings of [9] G. Ateniese, C. Blundo, A. De Santis, and D. R. Stinson,
the transparency and screen squares are determined Visual cryptography for general access structures, Inf.
based on the color of the pixel in the image. Computat., vol. 129, pp. 86106, 1996.
III. CONCLUSION
196
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
197
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Recoding of binary numbers was first hinted by COMPRESSOR FOR PARTIAL PRODUCT REDUCTION
Booth four decades ago. MacSorley proposed a The Wallace tree method is used in high speed
modification of Booths algorithm a decade after. The designs in order to produce two rows of partial
modified Booths algorithm (radix-4 recoding) starts products that can be added in the last stage. The
by appending a zero to the right of X0 (multiplier critical path and number of adders reduced when
LSB). Triplets are taken beginning at position of X0 compared to the conventional parallel adders. Using
and continuing to the MSB with one bit overlapping the carry save adder, three partial product terms can
between adjacent triplets. If the number of bits in X be added at a time to form the carry and sum. The
(excluding X0) is odd, the sign (MSB) is extended one Sum signal is used by the full adder of next level. The
position to ensure that the last triplet contains 3 bits. carry signal is used by the adder involved in the
In every step we will get a signed digit that will generation of the next output bit, with a resulting an
multiply the multiplicand to generate a partial product overall delay proportional to logarithm of rows.
entering the Wallace reduction tree. The meaning of A multiplier consists of various stages of full
each triplet can be seen in Table 1. adders, each higher stage adds up to total delay of the
Table 1 Radix-4 Recoding system. In the first and second stage of Wallace tree
Triplet Value Signed digit value structure, the partial products do not depend upon any
000 0 other values other than the inputs obtained from the
AND array. However for the immediate higher stages,
001 +1 the final value depends upon the carry out of previous
010 +1 stage. This operation is repeated for further stages.
011 +2 Hence the major cause for the delay is propagation of
100 -2 the carry from previous stage to next stage.
101 -1
110 -1 In the proposed structure compressors are used in
place of full adders and final carry propagate stage is
111 0
replaced by a SQRT CSLA with a BEC circuit. The
first stage consists of a full adder. In the second stage
This recoding scheme applied to a parallel two full adders have been grouped and implemented
multiplier halves the number of partial products so the using a 4:2 compressor. The number of
multiplication time and the hardware requirements interconnections is taken care since they playa
decrease. This gain is possible at the expense of important role in the flow of carry from one stage to
somewhat more complex operations in every step. next stage in the tree. A 4: 2 compressor is shown in
Radix-8 recoding applies the same algorithm as the figure2.
radix-4, but now we take quartets of bits instead of
triplets. Each quartet is codified as a signed-digit
using the table 2.
Table 2 Radix-8 Recoding
Quartet Value Signed digit value
0000 0
0001 +1
0010 +1
0011 +2
0100 +2
0101 +3
0110 +3
0111 +4
1000 -4
1001 -3 Figure 2. 4:2 Compressor
1010 -3
1011 -2
1100 -2 In this compressor the outputs generated at each
stage are efficiently used by replacing the XOR blocks
1101 -1 with multiplexer blocks. The select bits to the
1110 -1 multiplexers are available much ahead of the inputs so
that the critical delay path is minimized.
198
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
199
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Slice Logic Utilization Used Available Utilization [1] S.F hsiao, M R jiang and J S yeh,Design of High speed low
power 3:2 counter and 4:2 compressor for fast
Number of Slice multipliers,Electron.lett,vol.34, no.4,Pp 341-343
Registers 520 28,800 1% [2] K.prasad and K.K parhi,Low Power 4:2 and 5:2
compressor, in proc of 35th asilomar Conf on
Signals,systems and computers, Vol1,2001,pp129-133
Number of Slice LUTs 739 28,800 2% [3] Lakshmanan, masuri Othaman and Mohammad Alauddin
Mohd,Ali, High Performance parallel multiplier using
Number of fully used Wallace-Booth Algorithm, Semi conductor Elecronics,
LUT-FF pairs 394 753 52% IEEE International Conference, Pp:433-436, Dec 2002.
[4] C S wallace,A suggestion for a fast multiplier, Electronic
Number of bonded computers, IEEE Transactions, vol13,pages 14-17,Feb 1964
IOBs 32 480 6% [5] T.G Noll,Carry save architectures for high speed digital
signal processing, Journal of VLSI Signal Processing, 1991,
Pp 121-140.
Table 6 Comparision of Multipliers [6] O.J Bedri, Carry Select Adder,IRE
Radix -8 Booth Trans.Electron.Comput..Pp340-344,1962
Radix -8 Booth
Logic Utilization Algorithm with [7] Steve Kilts, Advanced FPGA Design :
Algorithm with
( XC3S5000 fg1156 - 4:2 compressors
4:2 compressors Architecture,Implementation, and optimization, Wiley-IEEE
4) and CSLA with
and CLA
BEC
press, 2007.
Number of Slice
LUTs 855 739
Number of Slice
Registers 980 520
Number of bonded
IOBs 44 32
VII. CONCLUSION
In this paper implementation and analysis of low
power area efficient Radix 8 multiplier with 4:2
compressor and SQRT CSLA using BEC is proposed.
The Proposed multiplier is compared with Radix 8
booth multiplier with 4:2 compressor and CLA. Both
designs are implemented on Virtex 5XC5VLX50T
FPGA board. The proposed multiplier shows more
reduction in power and area as compared to other
multiplier. The proposed design is to be used to design
a FIR filter and is to be implemented on a FPGA. In
the next stage physical design of FIR filter is realized
using Cadence Encounter Tool towards optimizing
area, power and delay.
200
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract Stereo vision is required in a field of 3D which determines the range of depth that can be
applications, like robotic navigation. This paper describes measured (the horopter). A difference between
a real-time stereo vision system that is required to support locations of the same object on the two images (the
high-level object based tasks in a tele-operated shift) is called disparity. The depth information (Z axis)
environment. Stereo vision is computationally expensive,
is derived from the disparity, found between two
due to having to find corresponding pixels. Correlation is
a fast, standard way to solve the correspondence problem. images as shown in Fig1.
This paper analyses the behaviour of correlation based
stereo to find ways to improve its quality while
maintaining its real-time suitability. The paper describes
a correlation based stereo matching algorithm and a DSP
based implementation. Results are given on real stereo
images. A comparison is made in terms of times they will
take by taking two different stereo image pairs with
windows of different dimensions by using black fin
processor.
Keywords: Stereo Matching, Computer Vision, 3D Scene
Reconstruction, Image Processing, DSP, Real Time, Robotic
Navigation.
I. INTRODUCTION
201
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
202
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
compared to the reference. The following pseudo code window does not overlap the depth discontinuity to the
describes the matching (stereo correspondence) process. same extent. Generally, the choice of the correlation
The support region is relatively small window defined window size is a trade off between increasing reliability
around the correlated pixel on the left (reference) in areas with constant depth and decreasing errors in
image. In this manner the correlation process takes into areas where depth changes.
account the neighbors of the pixel, not only the pixel
itself. The search window, defined on the right image, Optimization
is relatively large window. Its horizontal size is The correlation is performed using the SAD (Sum of
determined by the baseline between two cameras Absolute Differences) approach. The approach may be
(which affects the horopter). The vertical size may be 1 expressed mathematically by the following equation:
pixel for ideal stereo camera, or bigger for the practical
case. For each pixel inside the search window, a
correlation window (around it) is defined. The size of
the correlation window must be the same as the support
region size on the left (reference) image. Than, each M and N are corresponding to the vertical and
correlation window (inside the search window) is horizontal sizes of the correlation windows. The
correlated with the support region. The whole process measure of similarity is inversed, i.e. smaller SAD
must be performed for each pixel on the left (reference) value means better similarity. When support region is
image. two small, the correlation becomes ineffective, and the
quantity of mismatch errors increases. When using a
large support region, a lot of processing must be done
to accomplish the correlation task. To optimize the
correlation process, proposed the following
For each pixel in the left frame do
improvement.
begin
Set support region around the pixel (left frame) Data Reuse algorithm
Set search window in right frame
For each pixel in search window (right frame) When processing next pixel on the same line on the
begin reference image, a new support region (on the left
Set correlation window around the pixel image) and a new search window (on the right image)
Correlate support region with correlation window must be set. Since both windows are moving
end synchronously one pixel to the same direction
Find the best match (compared to previous point), only one
Calculate disparity single column of data has been changing inside them.
end All other data is still there, but moved to different
locations inside the windows. This fact is used to
Another problem with the correlation matching is that perform a fast correlation of the next search window, as
the selection of window size. Correlation works by only single column must be processed and added to the
using a usually fixed, rectangular window around the previous data. In this manner only for the first left pixel
pixel of interest in the first image. The window is on the left reference image the search window must be
correlated with a second window, which is moved over processed fully, while the rest of the pixels on the same
all possible positions in the second image. The possible line may be processed very fast using Data Reuse
positions are defined by the minimal allowed distance algorithm.
between the camera and an object, which gives the
Matching constraints
maximum disparity. The position where correlation has
the highest value determines the pixel in the second
image that corresponds to the pixel of interest. Bigger Stereo matching process is a very difficult search
correlation windows increase the reliability by procedure. In order to minimum false matches, some
averaging over a bigger area, thus reducing the effect of matching constraints must be imposed. Below is a list
noise. Consequently, this part of the window of the commonly used constraints.
introduces an error in the calculation. Furthermore, it
should be noted that the left part of the correlation Similarity: For the intensity-based approach, the
window in the left image is at least partly occluded in matching pixels must have similar intensity values (i.e.
the right image. The size of the occluded part depends differ lower than a specified threshold) or the matching
on the disparity difference and the size of the windows must be highly correlated. For the feature-
correlation window. The use of a smaller correlation based approach, the matching features must have
window reduces the problem, because a smaller similar attribute values.
203
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
204
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
VI.CONCLUSIONS
The above results show that the system can distinguish
between different objects in the observed scene and can
determine the relative distances to the objects. The
hardware implementation enables fast processing times
and acceptable quality, so the system may be used for a
real time applications, like robotic navigation.
REFERENCES
205
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
206
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
They also correspond to the FIRs impulseresponse. For have three rows of full adders to add width to the
LTI systems it is sometimes more convenient to express layout. This resulted in a more rectangular and flat
in the z-domain with Y(z)=F(z)X(z) circuit, but allowed us to reduce drastically the distance
the signal had to travel between chips, and decrease the
where F(z) is the FIRs transfer function defined in the overall complexity of connections within the circuit.
zdomain by Dissecting the Timing:
When we take a closer look at the timing contribution
of
i).the partial-product generation,
ii) the reduction tree, and
iii) the final adder to the total delay for MB and BW
multipliers, we see that the delay that is gained by
having a smaller reduction tree for the MB
implementation is canceled by the large delay of the
MB recoding circuitry. Table I shows that the partial-
Folded Bit-Plane Architecture product generation of the MB multiplier is about two
Using a new assignment of folding sets, which is times as slow as the generation in the BW multiplier.
applied on the transformed DFG from [4], we obtain This comes as no surprise, since for BW there is only
folded Bit-Plane architecture in general form (Fig. 1). one 2-input AND -gate in the critical path, while for the
Input Data Entering Module (IDEM), denoted with MB multiplier the critical path goes through the
dashed lines in Fig. 1, provides input data for the folded encoder followed by the decoder. Taking gate fanout
architecture in accordance with the allocation table into consideration, the primary inputs of a BW
(Table 1). Sections S0, S1, , Sk-1 in Fig. 1. are multiplier need to drive only N minimum-sized 2-input
Processing Elements (PE) of the folded architecture. A ND gates, while for the MB multiplier the x primary
Each section is devoted to computations from the inputs first have to driveN / 2 encoders, which in turn
corresponding folding set. Sections are implemented as drive N decoders. The delay difference of the reduction
rows of basic cells, where the basic cell is comprised of tree is not as large as above.
AND gate and full adder.
Baugh-wooley multiplier: The Baugh Wooley RESULT:
multiplier is an interesting implementation of a
multiplier. It produces a relatively standard cell shape SIMULATION:
for easier manufacturing, while maintaining good
driving caharacteristics. Each full adder within the
multiplier performs a similar number of computations,
with the diagonal ripple carry propagating the input
signal a constant and similar number of times. Baugh-
Wooley Multiplier is used for both unsigned and signed
number multiplication. Signed Number operands which
are represented in 2s complemented form. Partial
Products are adjusted such th at negative sign move to
last step, which in turn maximize the regularity of the
multiplication array. Baugh-Wooley Multiplier operates
on signed operands with 2s complement representation
to make sure that the signs of all partial products are
positive. An issue encountered during the design and
layout of the multiplier involved the layout of the
required full adders. A single full adder circuit SYNTHESIS: RTL SCHEMATIC:
naturally lays out in a very wide (or tall) chip, which
creates problems when working toward smallest form
factor and efficiency of cost. We attempted a number of
creative solutions to work around this, including
stacking these chips to create a more square form. This,
however, produced long connections as the signal
wrapped around from the output of one full adder to the
input of another and introduced both losses and
capacitance. It was determined that a better solution
was to implement the multiplier in a single, linear
direction through four stages of full adders, but only
207
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
ADVANTAGES: [14] E. Borch, E. Tune, S. Manne, and J. Emer, Loose loops sink
chips, in Proc. of HPCA-8, February 2002, pp. 270281.
[15] J. A. Butts and G. S. Sohi, Use-based register caching with
Power optimizing FIR filters consumes less power and decoupled indexing, in Proceedings of the 31st annual international
occupies less area. High throughput FIR filters used symposiumon Computer architecture, 2004, pp. 302313.
for all industrila applications.Mulyi hopping is main [16] R. E. Kessler, The alpha 21264 microprocessor, IEEE Micro,
cretireia in now a days. So, this designed FIR filter vol. 19, no. 2, pp. 2436, March-April 1999.
includes all these type of propoerties. [17] A. Capitanio, N. Dutt, and A. Nicolau, Partitioned register files
for vliws: a preliminary analysis of tradeoffs, in Proc. of Micro-25,
1992, pp. 292300.
APPLICATIONS: [18] K. I. Farkas, P. Chow, N. P. Jouppi, and Z. Vranesic, The
multicluster architecture: reducing cycle time through partitioning, in
Proc. Micro- 30, 1997, pp. 149159.
Bio medical needs for finding response of all solutions
[19] R. Canal, J.-M. Parcerisa, and A. Gonzalez, Dynamic cluster
with diferent chemical mixtures. Tele communications assignment mechanisms, in Proc. of HPCA-06, 2000, pp. 132142.
uses these FIR filters for finding impulse response of [20] A. Gonzalez, J. Gonzalez, and M. Valero, Virtual-physical
demodulated waves. Military applications involves registers, in Proc. of HPCA-4, 1998, pp. 175184.
triggering effects with different dimentions of bullets. [21] T. Monreal, A. Gonzalez, M. Valero, J. Gonzalez, and V. Vinals,
Delaying physical register allocation through virtual-physical
registers, in Proc. of Micro-32, 1999, pp. 186192.
CONCLUSION:
[22] S. Jourdan, R. Ronen, M. Bekerman, B. Shomar, and A. Yoaz,
A novel renaming scheme to exploit value temporal locality through
An efficient and enhanced FIR filter which deals with physical register reuse and unification, in Proc. Micro-31, 1998, pp.
all type of IEEE testing paramets like area, power and 216225
efficiency is implemented in this project. Baugh-wooley
multiplier is used for increasing efficiency. Proposed
algorithms in this project are more effective and
efficient when compare with all other technologies. All
proposed algorithms in this project are supports all type
of applications and yields high throughputs with more
security.
REFERENCES:
208
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Dr. D SRINIVASARAO 3
Prof & Head,
Department of ECE,
J N T U H, HYDERABAD, dsraoece@gmail.com
Abstract - Character Recognition has been an connectivity. The security systems, one different
active area of research in the past and due to its approach of dealing authentication is to use
diverse applications it continues to be a BIOMETRICS - the science of identifying or
challenging research topic. Methods based on verifying the identity of a person based on
multi layer perceptron (MLP) neural network physiological or behavioural characteristics.
provide attractive possibilities for solving pattern Physiological characteristics include fingerprints, iris,
classification problems. Several algorithms have hand geometry and facial image. The behavioural
been proposed for training the multi layer characteristics are actions carried out by a person in a
perceptron (MLP) neural network. We explore characteristic way and include recognition of
these techniques to design an optimal handwritten signature, machine printed characters, handwriting,
number recognition system. The main approach and voice [1].The character recognition is required
for handwritten number recognition is based on for different applications like postal address system,
segmentation. After segmentation the problem signature verification system and recognition of
gets reduced to the recognition of simple isolated characters from filled applications [2].
digits. We here adopt segmentation based
handwritten digit recognition where neural Character recognition is classified into two
networks are used to identify individual digits. In categories, off-line character recognition and on-line
this paper multi layer perceptron (MLP) neural character recognition. In off-line character
network using back propagation training with recognition the system accepts image as input from
tunnelling algorithm is proposed for handwritten the scanner [3]. It is more difficult than on-line
digit recognition application. The efficiency of the character recognition because of unavailability of
proposed method is tested on the handwritten contextual information and lack of prior knowledge
digits of different fonts and found that it is of the character. Noise will also exist in the images
successful in recognizing the digits. acquired in off-line character recognition. Machine
Printed character recognition comes under this
Index Terms - ANN, MLP network, Back category.
Propagation Training with tunneling , Zoning method On-line character recognition system accepts the
moment of pen from the hardware such as graphic
1. INTRODUCTION tablet, light pen and lot of information is available
during the input process such as current position,
The digital revolution has drastically changed our moments direction, start points, stop points and
perspective of the concept of communication and stroke orders [4-5].
209
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
For digitizing the handwritten document first the causal chain. With unsupervised learning it is
source material must be scanned using a scanner. possible to learn larger and more complex models
Older character recognition systems match these than with supervised learning. This is because in
images against stored bitmaps based on specific supervised learning one is trying to find the
fonts. The hit-or-miss results of such pattern- connection between two sets of observations.
recognition systems evaluate the performance of the Backpropagation method is the basis for training a
system accuracy. The current character recognition supervised neural network. The output is a real value
systems add the multiple algorithms of neural which lies between 0 and 1 based on the sigmoid
network technology. function [8].
In unsupervised learning, all the observations are Considering the MLP Neural Network with one
assumed to be caused by latent variables, i.e. the hidden layer as shown in Fig.2 consisting number of
observations are assumed to be at the end of the input nodes I, the number of hidden nodes J and the
210
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
number of output nodes K . The I-dimensional input z When the MLP Neural network is used in
is passed directly to a hidden layer. Suppose there are classification, the hidden layer performs clustering
J neurons in the hidden layer, each of the J neurons in while the output layer performs classification. The
the hidden layer applies unipolar sigmoid activation hidden units apply Non linear sigmoid activation
function defined by using (1) function for the input patterns. The output layer
would linearly combine all the outputs of the hidden
( )= ( . )
layer. Each output node would then give an output
value, which represents the probability that the input
pattern falls under that class. Back-propagation can
(1)
be
applied to networks with any number of layers. In
this paper a feed-forward neural network with a
single layer of hidden units is used with a sigmoid
activation function for the units.
211
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
implements the BP learning. The second phase 3) Error signal vectors and of both layers are
implements dynamic tunneling in the weight space computed. Dimension of Vector is (K X 1)
avoiding the local trap and thereby generates the and Dimension of is (J X 1). The error
point of next descent. Alternately, repeating the two signal terms of output layer is given by using
phases forms a training procedure that leads to the
global minimum. This algorithm is computationally =( )(1 ) , (6)
efficient.
= 1,2, .
The first phase in the Back Propagation tunneling The error signal terms of hidden layer is given by
training algorithm is Back-Propagation training using (7)
Algorithm. A two layer feed-forward network did not
present a solution to the problem of how to adjust the = 1 ,
weights from input to hidden units The solution is
that the errors for the units of the hidden layer are
= 1,2,
determined by back-propagating the errors of the
(7)
units of the output layer. The method is often called
the back-propagation learning rule. Back-propagation
4) The output layer weights are adjusted as
can also be considered as a generalization of the delta
= + , (8)
rule for non-linear activation functions layer
networks [13]. =
The error measure E is defined as the total quadratic 1,2,
error for pattern p at the output units. When a = 1,2, .
learning pattern is clamped, the activation values are
propagated to the output units, and the actual network 5) The hidden layer weights are adjusted by using
output is compared with the desired output values, we (9)
usually end up with an error in each of the output = + ,
units [14]. = 1,2,
(9)
Let I be the Number of input nodes, J be of Number = 1,2, .
of hidden nodes and K be the Number of output 6) Repeat the steps 1) to 5) for all the feature
nodes for the MLP neural network. Consider V be the vectors
weight vector for hidden layer and W be the weight
vector for output layer. The size of W matrix is K X 7) The training cycle is repeated for 1000 epochs.
J. The size of V matrix is JXI
5. TUNNELING ALGORITHM
A. The steps for the training cycle The Second phase in the Back Propagation tunneling
training algorithm is dynamic tunneling. A random
1) By applying the feature vectors one by one to the point W in the weight space is selected. For a new
input layer the output of hidden layer is point + , where is a perturbation related to W,
computed as if ( + ) ( ), the BP is applied until a local
= = 1,2, minimum is found; otherwise, the tunneling
(3) technique takes place until a point at a lower basin is
found as shown in the Fig 3. The algorithm
The function f1(.) is unipolar Sigmoid Function automatically enters the two phases alternately and
defined by using (1). Output of output layer is weights are modified according to their respective
computed by using (4) update rules. The tunneling is implemented by
solving the deferential Equation by using (10)
= ( ) = 1,2,
(4) ()
() ( ) /
= ( )
The function f2(.) is Generalized Sigmoid Function (10)
defined by using (2)
( )
Where is the learning rate, is the local
2) The error value is computed by using (5) () () ( ) ()
minima of , = +
= ( ) + , = 1,2, . ,
(11)
(5)
212
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
The equation is integrated for a fixed amount of time, This sequence is repeated until no additional image
with a small time-step . After every , E(W) is is found
()
computed with the new value of keeping the A. Thinning
emaining components of W unchanged. Thinning is the process of transforming a digital
pattern from one form to another with less thickness
Tunneling comes to a halt when E(W) E(W*), and while maintaining connectivity of the original
initiates the next gradient descent. If this condition of pattern[17]. It is used in several applications , one of
descent is not satisfied, then this process is repeated which is character recognition in the field of
( ) document analysis. In this contest the scanned image
with all the components of until the above
is converted into a binary image (I). It has a finite set
condition of descent is reached. If the above of points in the Euclidean plane where each point is
()
condition is not satisfied for all , then the last mapped into 0 or 1. The set of pixels in I that are
local minimum is the global minimum [8-14]. assigned the value 1 is called the foreground or the
pattern and denoted by P where as the background
6. PRE PROCESSING pixels are denoted by Pc. A good thinning algorithem
is one that ensures some basic desirable properties
The hand written digits document was first such as: maintaining connectivity of resulting
preprocessed before applying to the MLP skeletons, insensitivity to noise. Parallel
network.Preprocessing of hand written documents implementation of thinning algorithems are
consists three major steps [15]. First the document significantly faster than sequential ones. After
image was segmented and then the segmented digit is thinning process, the features of the image are
applied to thining alogorithm. obtained by feature extraction method.
B. Feature Extraction Using Zoning Method
A. Segmentation To train the MLP network with any image requires its
features.In this paper to extract the features Zoning
Character segmentaion is an opreation that seeks to method is used . In zoing method, the scanned image
decompose an image of a sequence of character in to is divided into NxM zones. From each zone, features
sub image of individual symbols . it is one of the are extracted to form feature vector [16]. The goal of
decision processes in a system for optical character zoning is to obtain the local characteristics instead of
recognition .Its decision, that a pattern isolated from global characteristics. Each sub-zone consists number
the image is that of a character or some of foreground pixels. A feature is considered as an
identificatiable unit), can be right or wrong . it is average of foreground pixel density.
wrong sufficently often to make a major
The scanned image is divided into 4-by-4 zones asa
shown in the Fig. 4. In addition, in each sub zone, the
foreground pixel values are taken and by averaging
all those pixel values, we can obtain the feature of
that sub zone. Such that the features of remaining sub
zones are obtained. For each digits 16 features are
obtained which is placed as feature vector of 1 x 16.
213
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
70
Thinned image was obtained for each segmented
% of correct classification
60
hand written digit and 16 features are extracted using
Zoning method.The features so obtained is kmown as 50
feature vector.The input training data set is a martrix
40
of dimensions 250 x 16 obtained by the combination
of feature vectors of total 250 hand written digits. 30
Fig. 6. The training algorithm initialized the weight Fig. 8 Percentage of Correct Classification Versus
matrix W to zero. The no. of hidden nodes
214
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
The Graph in the Fig. 9 shows the Mean square Error is found that this method requires less time for
versus number of Iterations for training the MLP training the MLP neural network and has good
neural network. success rate in recognizing the handwritten digits.
C. CONCLUSION
215
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Table 3: Target vector for hand written digits for 0 to [10] Tom Mitchell, Artificial Neural Networks, in Machine
Learning, 1st ed. McGraw- Hill ,1997 pp.95-111
9
[11] O.Batsaikhan and Y.P. Singh, Mongolian Character
Recognition using Multilayer Perceptron, Proce edings of the
9th International
[12] Kambiz rahbar, Saeid rahati quchani, Muhammad rahbar
recognition persian handwritten digits using templates and
back-propagation network with adaptive learnings rate
proceedings of the 5th international conference on signal
processing, istanbul, turkey, may 27-29, 2006 (pp46-51)
[13] D.Y. Lee, Handwritten digit recognition using K nearest
neighbour, radial basis function and backpropogation neural
networks, Neural Computation, Vol. 3, pp. 440-449.
[14] A.M numan-al-mobin, mobarakol islam, kaustubhdhar, tajul
Islam, md. rezwan,m. Hossain backpropagation with vector
chaotic learning rate
[15] Jaekyu Ha & Robert M. Haralick Ihsin, T. Phillips,
For supervised learning at each instant of time when Document Page Decomposition by the Bounding-Box
the input is applied, the desired response of the Projection Technique, IEEE Transactions on Systems, Man,
network known as target vector is provided by the and Cybernetics, Vol.18, No.1, January 1995, pp.1118-1122.
[16] Oivind Due Trier, Anil K.Jain and Torfinn Text, Feature
teacher. The distance between the actual and desired Extraction methods for Character recognitionA Survey,
response serves as an error measure and is used to Pattern Recognition, Vol. 29, No. 4, 1996, pp.641662.
modify weights so that the error decreases. The [17] P. S. P. Wang and Y. Y. Zhang, A fast and flexible thinning
target vectors for one font is mentioned for digits 0 to algorithm, IEEE Transactions on Computers, Vol.38, No.5,
May 1989.
9 in Table 3. By using these target vector the training
of MLP neural network will terminate. for the
designed MLP network there are 10 output nodes and
16 input nodes ,to which the feature vector of 1 x 16
is applied ,for example for digit 0 ,the output of the
first node is 1 and the output of the remaining nodes
are 0.The target vector is 1 x 10 vector.
References
216
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract: This paper describes t he various anal og which is given by n L/C where n refractive index ;
signal processing schemes used for detecting the L length of the fiber ; C velocity of light.
rate inform at ion from square wave bias modulated
Interferometric Fiber Optic Gyro(IFOG).
Development of Single axis fiber optics gyro carried
out in LEOS in open loop and closed loop schem es
are described here. The design consideration,
tradeoffs are highlighted for the front end
electronics o f an all - digital closed loop approach.
Achieved results are also presented .
1. INTRODUCTION
217
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
nonreciprocal phase shift within the interferometer to open loop system, even though the SLED
counteract the Sagnac phase shift and thereby temperature was controlled to 25C + 0.5C using
maintaining the detected built-in thermoelectric cooler.
output to zero.
218
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
In this scheme, instantaneous peak to peak noise is 30 as shown in fig.10. PINFET detector signal is
deg/hr and thermal sensitivity is also high. Upon converted to
analysis heat dissipating elements like SLED & DC by integrators and sample & hold for positive and
Processing electronics are integrated in a single box. negative bias modulation time and the analog
This problem is solved in the next version by demodulation is done by differential amplifier.
isolating the heat dissipating elements from the Amplified rate information is filtered by 4th order
sensor head and compact mechanical design for 4axes Bessel filter and converted to digital signal by ADC.
FOG version. Achieved results are shown in table-3. ARW plot,
Stability plots are shown in fig.11 & 12 respectively.
Dynamic range and resolution of this FOG is
improved, but instantaneous noise peak to peak is 20
deg/hr and drift is 0.2 deg/hr. Though heat dissipating
elements inside the SHA are isolated bias drift in
thermo vacuum cycling ( 0 to 50C) is of the order of
8 deg/hr/C, which is shown in fig.13 along with
SHA temperature..
B.Version-2 Table-3
In Sensor Head Assembly (SHA) non dissipating Closed loop scheme version-2 results
components like fiber and MIOC alone assembled
and all the heat dissipating elements are shifted to
Processing Electronics Module (PEM) as shown in
fig.7. High power SLED source is set to 1mW and
the detector power is 100 W. Both the source and
detector are changed to get better signal to noise
ratio. Coupler is also replaced with Circulator (CIR)
to avoid the light reflecting back to SLED. Bobbin
size is reduced to 110xh20mm and the assembled
single axis sensor is shown in fig.8 and 4-Axes SHA
mounting on a tetrahedral assembly is shown in fig.9
219
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Commercially available off the shelf PINFET 1) Detector selection: Pin diode junction capacitance
detectors are having very high bandwidth and output decides the detector bandwidth and hence the noise.
saturation voltage is 2V only. PIN diode to be chosen carefully and layout
guidelines to be strictly followed not to increase the
capacitance.
220
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
FPGA. Such digital demodulation is intrinsically free Fig.18 Allan Variance Plot of version-3
of any source of electronic long-term drift and analog
input offset of the A/D converter is cancelled out by
the subtraction of A & B samples as shown in fig.16.
With the above scheme achieved performance of the
FOG is shown in table-4.
Table-4
Closed loop scheme version-3 results
4. CONCLUSION
Signal processing schemes for detecting the rate
information for IFOG has taken a decade. For each
scheme layouts and PCBs are made and testing in
live environmental conditions are carried out.
Selection of precision components and up screening
of opto components for space use is also attempted.
Detecting the signal from noise is quite challenging
and winding of fiber also carried out in-house using
Fig.16 Buffer output of version-3 semi automatic winding machine. Signal integrity
issues are solved and 4axes FOG in tetrahedral
Inertial grade FOG performance is achieved by the configuration is in progress.
improved analog processing scheme and modulation
depth is also increased to 0.65instead of 0.5to ACKNOWLEDGEMENT
reduce the RIN noise. Stability plot and Allan
variance plot is shown in fig.17 & 18 respectively. The authors are thankful to Dr. G. Nagendra Rao,
Fig.19 shows the earth rate plot for the Bangalore Dy.Director, LEOS and Shri J.A.Kamalakar Director
latitude (14.59deg/hr) with an offset of 1.5deg/hr. LEOS for their guidance, constant encouragement
and support in the development of interferrometric
fiber optic gyroscope. Authors are also wish to
express their gratitude to Dr.T.K.Alex, Ex Director,
LEOS for his support and guidance in developing the
gyro.
REFERENCES
221
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
[7]. W.Marshall Leach, Jr., Senior Member, IEEE, Fundamentals of Low Noise Analog Circuit Design
222
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Step 3: 0.01
Vi 0 [Vi10 , Vi 02 , Vi 30 , Vi 04 , Vi 50 ] -0.01
i=1, 2 ,.....5
-0.02
0 0
V ij V j -0.03
Each elements of is selected a random digits for
example, between -0.04
x -0.05
0 2 4 6 8 10 12 14
[0, 0.1 i ,max ] n
Step 4: Fig.(1)
Set iteration count K=1.
Step 5:
Calculate Error value by using eq (5).
min[1 ,2,3,4,5 ]
And corresponding particle is the gbest is the particle
which leads to .
Step 6: Update velocities of each particle using
following relation
Pbestik
Where is the best previous position of the i th
particles, is particle which leads to
Vi k W *Vi k 1 c1* r1( pbestik 1 U ik 1 )
223
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
frequency response
1.2
0.8
0.6
0.4
gain in db
0.2
-0.2
-0.4
-0.6 Fig.(5)
0 0.2 0.4 0.6 0.8 1 1.2 1.4
error plot
normalized frequency 24
Fig.(2). 23.5
error plot
25.2
23
25.1
22
Mean square error
24.9
21.5
24.8
21
24.7
20.5
24.6 0 10 20 30 40 50 60 70 80 90 100
number of iterations
24.5
0 10 20 30 40 50 60
number of iterations
70 80 90 100 Fig.(6)
Mean Square error obtained for same specifications.
Fig.(3) N=16(TYPE-4)
TYPE MSE()
3 24.6
4 20.8
Table.(1)
5. CONCLUSIONS
The information mechanism in PSO is significantly
different. So the whole population moves like a one
group towards an optimal area. In PSO, only gbest
gives out the information to others. PSO converges to
the best solution quickly and gives optimal filter
coefficients. The amplitude response of TYPE-4 is
smoother than the TYPE-3 and mean square error is
Fig.(4) small in case of TYPE-4. So TYPE-4 filter is better
suited for design of optimal digital FIR low pass
differentiator than TYPE-3.
REFERENCES
224
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract A parallel and pipelined Fast Fourier reduce the computational burden.
Transform (FFT) processor for use in the Orthogonal In the communication world of today more and more
Frequency division Multiplexer (OFDM) . Unlike being OFDM systems are brought on-line with an ever
stored in the traditional ROM, the twiddle factors in our increasing number of standards, applications, and
pipelined FFT processor can be accessed directly. A services, all with different requirements on the
novel simple address mapping scheme and the physical layer of the transceiver. The physical layer
modified radix 4 FFT also proposed. Finally, the
pipelined 64-point FFT processor can be completely
is most often, due to speed and power constraints,
implemented within 23.750ns. implemented as an ASIC and thus locked to one
Index Terms OFDM, radix- 4, FPGA. specific case. Hence, the number of required
implementations will grow fast. Due to this fact,
1. INTRODUCTION there will be an interest from the industrys point of
view to search for a flexible architecture that can be
Active radar systems are commonly used in military configured to function with several standards,
and commercial applications for the detection and applications, or services. A flexible architecture will
tracking of objects through a medium such as air. lead to reusage of design and implementation and
Most active radar systems work by transmitting a thus reduced cost. This project addresses both OFDM
signal pulse through the medium and receiving the in theory and the implementation aspects of a
reflected energy wave field as it scatters off objects flexible hardware solution for digital baseband
in the medium. By processing the received wave OFDM. This project a flexible FFT processor and a
field, an active radar system can determine an flexible OFDM transmitter are presented. The FFT
object's distance, velocity, and other features. A processor is a central part of an OFDM transceiver,
passive radar system can consist solely of an array of and has been fabricated both as a standalone chip and
receiving antennas as opposed to an active radar as part of an OFDM transmitter chip. The transmitter
system with a co-located transmitter and receiver. chip includes in addition to the FFT, a signal
Without a transmitter the passive radar system mapper and a cyclic prefix inserter. It is shown
relies on other sources of electromagnetic waves, that high flexibility is obtained in the transmitter
such as AM or FM radio waves, TV broadcast, with a reasonable amount of extra hardware.
nearby radar, cell towers, or a wideband waveform to Comprehensive measurements on power consumption
illuminate objects in the medium. Various signal are presented for both chips.
processing algorithms can be applied to the data from A scheme to reduce hardware and delay in an OFDM
the antenna array. One algorithm of interest is the transceiver is proposed in this project. The scheme
Direction of Arrival (DOA) of the incoming scattered shows that the data buffers in an OFDM transceiver,
waveform and therefore the object. A computationally due to a cyclic prefix and a bit reversing pipelined
efficient version of the DFT is known as the Fast FFT processor, can be significantly reduced with a
Fourier Transform (FFT) and is commonly used to cyclic suffix and a bidirectional FFT processor.
225
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
FFT Architecture
The FFT and IFFT Equation 2.1 and 2.2 has the
property that,
226
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
building block [23]. There are two types of butterfly butterflies but the number of complex multipliers are
operations, decimation in time (DIT) and decimation reduced as if it were a radix-4 [24]. Figure 3 shows
in frequency (DIF), both are shown in Figure 2. The an N = 64-point radix-22 DIF FFT. Each stage in a
difference between DIT and DIF lies in the position radix-22 FFT consists of two butterflies,
of the twiddle factor multiplication, which is either one trivial multiplier, i.e. multiplication with -j, and
performed before or after the subtraction and one
addition. Since the FFT is based on divide and complex multiplier that multiply data with the
conquer, it will be most efficient if the input sequence twiddle factor.
is of length N = rp, where N is called point, r is called
radix, and p is a positive integer. To compute an N- 3. FPGA IMPLEMENTATION
point FFT, p stages of butterflies are connected. An
example of a hardware mapped N = 64-point radix-2 The Field Programmable Gate Array is majorly used
DIF FFT is shown in Figure 3. Observe that the input for
data, x(n), occurs in consecutive order, whereas the generation ASIC ICs to the computations. They offer
output data, X(n), is rearranged. The order of the more speed in execution process. SO, for generation
rearranged data is called bit reversed order [23] . It is ASIC ICs FPGAs are majorly used. The 64 FFT
possible to reconfigure the FFT in Figure 3 so that the with radix 4 is simulated and synthesized as well as
input occurs in bit reversed order and the output implemented on the FPGA of below configuration.
occurs in natural order. The reconfiguration can be
thought of as vertically moving the horizontal data
paths in Figure 3, while the cross connections remain
fixed to respective path, until the output is in natural
order. Also notice that, if all arrows in Figure 3 is
turned around a DIT FFT is performed instead of a
DIF FFT. 4. RESULTS:
227
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
228
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
[3] ETSI EN 300 744, Digital video broadcasting (DVB);framing [25] K. Nazi and G. Hasson, Industrys First RTL Power
structure, channel coding, and modulation for digital terrestrial Optimization Feature Significantly Improves Power Compilers
television, 2001. Quality of Results, 1998,
[4] European IST Project, Power Aware Communications for www.synopsys.com/news/pubs/rsvp/spr98/rsvp spr98 6.html.
Wireless OptiMised personal Area Networks (PACWOMAN), [26] W. Li and L. Wanhammar, A Pipelined FFT Processor, in
http://www.imec.be/pacwoman/. IEEE Workshop on Signal Processing Systems, 1999, pp. 654
[5] S. B. Weinstein and P. M. Ebert, Data transmission by 662.
frequency division multiplexing using the discrete Fourier [27] S. Johansson, S. He, and P. Nilsson, Wordlength
transform, IEEE Transactions on Communications, vol. 19, pp. Optimizationof a Pipelined FFT Processor, in Proc. of 42nd
628634, Oct. 1971. Midwest Symposium on Circuits and Systems, Las Cruces, NM,
[6] A .Peled and A. Ruiz, Frequency domain data transmission USA, Aug. 8-11 1999.
using reduced computational complexity algorithms, in Int. Conf. [28] Hongjiang He and Hui Guo, The Realization of FFT
Acoustic, Speech, Signal Processing, Denver, CO, 1980, pp. 964 Algorithmbased on FPGA Co-processor, Second International
967. Symposiumon Intelligent Information Technology Application,
[7] ETSI TS 101 475, Broadband Radio Access Networks vol. 3, pp.239-243, Dec. 2008.
(BRAN); HIPERLAN Type 2 Physical (PHY) layer, v1.1.1,
2000, http://portal.etsi.org/bran/.
[8] L. J. Cimini, Analysis and Simulation of a Digital Mobile
Channel Using Orthogonal Frequency Division Multiplexing
IEEE Transactions on Communications, vol. 33, pp. 665675,
July 1985.
[9] IEEE std 802.11a, High-speed Physical Layer in 5 GHz
Band, 1999, http://ieee802.org/.
[10] ] IEEE std 802.11g, High-speed Physical Layer in 2.4 GHz
Band, 2003, http://ieee802.org/.
[11] J. Bingham, Multicarrier Modulation for Data Transmission:
An Idea Whose Time Has Come, IEEE Communications
Magazine, vol. 8, pp. 514, May 1990
[12] O. Edfors, M. Sandell, J. van de Beek, D. Landstrm and F.
Sjberg, An introduction to orthogonal frequency-division
multiplexing, TULEA 1996:16, Div. of Signal Processing, Lule,
Tech. Rep., a University of Technology, Lule 1996.
[13] J. W. Cooley and J. W. Tukey, An Algorithm for Machine
Calculation of Complex Fourier Series, Math. Comput., vol.19,
pp. 297301, Apr. 1965.
[14] R. Grunheid, E. Bolinth, and H. Rohling, A blockwise
loading algorithm for the adaptive modulation technique in OFDM
systems, in Proc. of Vehicular Technology Conference, VTC
2001 Fall, Atlantic City, NJ, USA, Oct. 7-11 2001, pp. 948951.
[15] J. G. Proakis, Digital Communications. McGraw-Hill, 2001.
[16] M. Russel and G. Stuber, Interchannel interference
[17] N. Petersson, Peak and power reduction in
multicarriersystems, 2002, licentiate thesis, Lund University,
Sweden.
[18] S. Johansson, ASIC Implementation of an OFDM
[19] R. Morrison, L. J. Cimini, and S. K. Wilson, On the Use of a
Cyclic Extension in OFDM, in Proc. of Vehicular Technology
Conference, VTC 2001 Fall, vol. 2, Atlantic City, NJ, USA, Oct.7-
11 2001, pp. 664668
[20] K. Parhi, VLSI Digital Signal Processing Systems. New York,
NY, USA: John Wiley & Sons, 1999.
[21] K. Parhi, VLSI Digital Signal Processing Systems. New York,
NY, USA: John Wiley & Sons, 1999.
[22] Transmeta Corporation, Transmeta Long Run Power
Management, http://www.transmeta.com
[23] J. G. Proakis and D. G. Manolakis, Digital Signal Processing.
Prentice-Hall, 1996.
[24] S. He, Concurrent VLSI Architecture for DFT Computing
and Algorithms for Multi-output Logic Decomposition, Ph.D.
dissertation, Lund University, 1995.
229
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract-Binary Multipliers plays an important role by the multiplier on a FPGA (Field Programmable Gate
in digital circuits. There are many methods for Array) increases. Hence various algorithms have been
generating a Simple binary multiplication and some developed.
of them are like Ripple carry array multipliers, Row
adder tree multipliers, Partial product LUT in order to achieve lesser complexity in computation
multipliers, Wallace trees, Booth recoding etc.,. Our involving minimum calculation steps, which in turn can
project mainly concentrates on 8x8 Wallace tree reduce delay, power and area constraints of multipliers.
multiplier. It uses a famous Wallace tree structure
which is an implementation of an adder tree The Wallace tree has three computation steps:
designed for minimum propagation delay. Rather
than completely adding the partial products in pairs 1. Generation of Partial products multiplying each
like the ripple adder tree does, the Wallace tree bit of one binary input with every bit of the other
sums up all the bits of the same weights in a merged binary input. If each input has n-bits the result of
tree. Usually full adders are used, so that 3 equally this step will give us n2 number of binary bits
weighted bits are combined to produce two bits: one called Partial products distributed in n-rows and
(the carry) with weight of n+1 and the other (the 2n-columns. This step is very same as what we do
sum) with weight n. Each layer of the tree to multiply two numbers by hand.
therefore reduces the number of vectors by a factor 2. Reduction of partial products the partial products
of 3:2. are to be added according to their place values (or
weights) using half adders and full adders until
A conventional adder is used to combine these to only two rows of partial products are left.
obtain the final product. The benefits of the Wallace 3. Last stage addition remaining two rows will be
tree is that there are only O(logn) reduction layers, added using a conventional adder to get final result
and each layer has O(1) propagation delay. As of multiplication.
making the partial products is O(1) and the final
addition is O(logn), the multiplication is only
O(logn), not much slower than addition (however,
much more expensive in the gate count). Naively
adding partial products with regular adders would
require O(log2n) time. Our project is to develop 8 x 8
Wallace tree multiplier using VHDL and will be
simulated with the help of XILINX simulator and
verified on Spartan-3E FPGA circuit board.
INTRODUCTION
230
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
If b0 = 0:
If b0 =1:
FINAL RESULT BITS
Then the first row (a7 a6 a5 a4 a3 a2 a1)*b0
will be equal to
(a7 a6 a5 a4 a3 a2 a1)*1 = (a7 a6 a5 a4 a3 a2 a1) = A
231
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Representation of signals:
As described in an algorithm, we have to use
different variables to indicate the partial products in
different levels of reduction stages. For example, we
used a, b in first stage, then P0, P1 in the next and Scheme of representation of signals
later S, C, M, N etc. in the figures of chapter 2. These To satisfy the above four requirements, we adopt the
variables are of our choice and we must make sure above representation scheme.
that PPs in different levels of reduction do not have
the same representation. It means that designer has to Ri reduction stage number i ; Eg: R1,
make a note of which variables he is using in what R2, R4 etc.
stage of reduction, clearly and without confusion. Sj sum bit of column j+1
232
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Delay(ns)
20 13.924
If there are 4 sum bits in the column k takes the
values of 1, 2, 3 and 4. CSHM
10
Always the sum bits are taken first and the carry bits
are taken next to sums. Reverse will also give the
0 Booth
same answer, but to avoid confusion sums are given 8*8 Multiplier Multiplier
first priority in any column. Let us consider a column
having 4 sums and 3 carries. Let all the bits belong to
column number 6 (j=5) of 3rd reduction stage. It will
be represented as follows.
Comparison of Multipliers in terms of Area:
R3S5_1
R3S5_2 200 146 169180
180
WTM
Area(LUT)
R3S5_3
R3S5_4
R3C5_1 CSHM
0
R3C5_2 8*8 Multiplier
R3C5_3
A column of signals of 3rd reduction stage, 5th column
Designing, Synthesis and Results of WTM for the
Spartan 3E family FPGA chip. Comparison of Multipliers in terms of PDP:
0 CSHM
8*8 Multiplier
3 8 13.924
233
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Features of WTM:
234
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract - The aim of this paper is to investigate the for a large area, it is sufficient to use fixed-point
speech signal enhancement using Adaptive filters. arithmetic for the adaptive filter to work well. The
Background noise elimination is very important in algorithm used to update the filter coefficient is
many applications like hearing aids, forensic the Least Mean Square (LMS) algorithm which is
applications, military, cellular environments etc. The
known for its simplification, low computational
adaptive filters has advantages like low processing
delay and better tracking of the trajectory of non- complexity, and better performance in different
stationary signals. If the evaluation of background running environments. In terms of the number of
noise is more slowly than the speech, i.e., noise signal iterations required for convergence, LMS
is more stationary than the speech signal, we can algorithm is seen to perform very well when
easily estimate the noise during the pauses in speech. compared to other algorithms used for
Otherwise it is more difficult to estimate the noise implementing adaptive filters.
which results in degradation of speech. In order to
enhance the quality a n d intelligibility of speech, here II. SPEECH ENHANCEMENT
we proposed a design based on LMS algorithm to
remove n o i s e from speech signal using VHDL based
on commercially available FPGAs. In this design the Speech enhancement is concerned with improving
LMS algorithm used as a noise canceller and the some perceptual aspect of speech that has been
reference sig nal was adaptively f i l t e r e d a n d d e g r a d e d b y additive noise. In most applications
subtracted from primary signal to obtain the the aim of speech enhancement is to improve the
estimated speech signal. quality and intelligibility of degraded speech. The
improvement in quality is highly desirable as it can
Keywords: LMS, Noise canceller, FPGA, VHDL listener fatigue particularly in situations where the
and Speech Enhancement. listener is exposed to high levels of noise for long
period of time. The need to enhance speech signals
I. INTRODUCTION
arises in many situations in which the speech signal
To model the relationship between two signals in
originates from a noise location or is affected b y
real time in an iterative manner, An Adaptive filter
noise location or is affected b y noise over a
is used which is a computational device. Adaptive
communication channel.
filters are often realized either as a set of program
Speech enhancement algorithms can therefore be
instructions running on an arithmetical processing
used to improve the quality of speech at the
device such as a DSP chip or as set of logic
receiving end. The solution to the general problem
operations implemented in a field programmable
of speech enhancement depends largely on the
gate array (FPGA) or in a custom VLSI integrated
application at hand, the characteristics of noise
circuits.
source or interference, the relationship of the noise
Adaptive filters, as a part of digital signal system,
to the clean signal.
have been widely used in communication industry,
as well as one of the applications such as adaptive Critical to the design of speech enhancement
noise cancellation. However, implementation of algorithms is knowledge of the range of speech
adaptive filter takes a great deal and becomes a and noise intensity levels in real-world scenarios.
very important role in digital system design. From that we estimate the range of signal-to-noise
Adaptive filters are usually implemented in DSP r atio [SNR] levels encountered in realistic
processor because they are capable of performing environment. This is important since speech
fast floating-point arithmetic operations. But now a enhancement algorithms need to be effective in
day in signal processing, FPGA becomes an suppressing the noise and improving speech quality
important competitor by providing a lot of facilities within that range of SNR level.
to designers in the market. In addition, FPGA is a
form of programmable logic that can offer III. ADAPTIVE FILTERS
flexibility for repetitive reconfiguration. Because
of slices organized as array of rows and columns Adaptive filters learn the statistics of their operating
in FPGA, a great deal of parallelism can be environment and continually adjust their parameters
explored. Although it is not efficient to use accordingly. This chapter presents the theory
floating-point arithmetic in FPGA due to its need of the algorithms needed to train the filters. In
235
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
practice, signals of interest often become error signal, forcing the adaptive filter output signal
contaminated by noise or other signals occup ying to approximate the reference signal in a statistical
the same band of frequency. When the signal of sense.
interest and the noise reside in separate
frequency bands, conventional linear filters are IV. LMS ALGORITHM
able to extract the desired signal . However, when
there is spectral overlap between the signal and The LMS algorithm is a stochastic gradient
noise, or the signal or interfering signals algorithm, which uses a fixed step-size parameter to
statistics change with time, fixed coefficient filters control the updates to the tap weights of a
are inappropriate. transversal filter, (Widrow and Hoff 1960). The
The goal of any filter is to extract useful algorithm aims to minimize t h e mean-square error,
information from noisy data. Whereas a normal the error being the difference in y(n) and yest(n). The
fixed filter is designed in advance with knowledge dependence of the mean-square error on the
of the statistics of both the signal and the unknown tap weights may be viewed as a
unwanted noise, the adaptive filter continuously multidimensional paraboloid referred to as the error
adjusts to a changing environment through the surface .The surface has a uniquely defined
use of recursive algorithms. This is useful when minimum point defining the tap weights for the
either the statistics of the signals are not known optimum Wiener solution. However, in the non
beforehand o f change with time. stationary environment t h i s error surface is
The adaptive filter has advantages like lower continuously changing, thus the LMS algorithm
processing delay and better tracking of the trajectory needs to be able to track the bottom of the surface.
of non-stationary signals [2]. These are essential The LMS algorithm aims to minimize a cost
characteristics in applications such as noise function,
estimation, echo cancellation, adaptive delay V (w(n)), at each time step n, by a suitable
estimation and channel equalization, where low choice of the weight vector w(n). The strategy is
delay and fast racking of time-varying to update the parameter estimate proportional t o
environments and time-varying processes are the instantaneous g r a d i e n t value, dV (w(n))/dw(n),
important objectives. The e x i s t e n c e of a so that:
reference signal which is hidden in the fixed-filter
approximation step, defines the performance w(n + 1) = w(n) dV (w(n))/dw(n) (2)
criterion. The general adaptive-filter configuration where is a small positive step size and the
is shown in fig 1. minus sign ensures that the parameter estimates
descend the error surface. The cost function, V
(w(n)), which minimizes the mean-square error,
results in the following recursive parameter
update equation
w(n + 1) = w(n) x(n) (y(n) yest (n)) (3)
The recursive relation for updating t h e tap
weight vector Equation (1) may be rewritten as
w(n + 1) = w(n) x(n)(y(n) xT(n)w(n)) (4)
which can be represented as filter output ,
estimation error and tap weight adaptation.
The basic configuration of an adaptive filter,
operating in the discrete-time domain k, is
illustrated in Figure 1. In such a scheme, the input
signal is denoted by x(n), the reference signal d(n)
represents the desired output signal (that usually The LMS algorithm requires only 2N + 1
includes some noise component), y(n) is the multiplications
output of the adaptive filter, and the error signal is and 2N additions per iteration for an N tap weight
defined as vector. Therefore it has a relatively simple
e(n) = d(n)y(n) (1) structure and the hardware is directly proportional
The error signal is used by the adaptation to the number of weights.
algorithm to update the adaptive filter coefficient The L M S a l g o r i t h m is b y far the most widely
vector w(n) according to some performance used algorithm in adaptive filtering for several
criterion. In general, the whole adaptation reasons. The main features that attracted the use of
process aims at minimizing some metric of the the LMS algorithm are low computational
236
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
237
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
B. Chipscope Results
ChipScope provides you with a convenient software
based interface for controlling the integrated logic REFERENCES
analyzer, including setting in the triggering options
[1] S. Haykin, Adaptive Filter Theory, Prentice-Hall, third
and viewing the waveforms. Time required for
edition, 2002.
simulating co mplex design for all possible test [2] Roger Woods, FPGA-based Implementation of Signal
cases becomes prohibitively large and simulation Processing Systems John McAllister, Gaye Lightbody, Ying Yi.
approach fails. For rapid testing, such designs can be [3] U. Meyer Baese, Digital Signal Processing with Field
Programmable Gate Arrays, Springer, 2001.
loaded on to the target FPGAs and tested by
[4] Xilinx Inc., Block Adaptive Filter, Application Note XAPP
applying test inputs and directly observing their 055, Xilinx, San Jose, CA, January 9, 1997.
outputs. The ChipScope Pro tools integrate key [5] www.xilinx.com
logic analyzer and other test and measurement [6] P. Waldeck and N. Bergmann, Evaluating software and
hardware implementations of signal-processing tasks in an FPGA,
h a r d wa r e components with the target design
in Proc. IEEE International Conference on Field-Programmable
inside the FPGA. Computer based software tool Technology, Brisbane, Australia, Dec. 2004, pp. 299-302.
communicate with these hardware components and [7] Ravi Bolimera, T.Kranthi Kumar, M.Paul Sandeep, Speech
provide a designer robust logic analyser solution. Enhancement using Normalized Kernel Affine Projection
Algorithm, Proc. of the Intl. Conf. on Advances in Computer
Waveforms are showed in fig. 4.
Science and Electronics Engineering CSEE 2013, ISBN: 978-
981-07-5461-7
doi:10.3850/ 978-981-07-5461-7_42.
[8] A. Elhossini, S. Areibi and R. Dony, An FPGA
Implementation of the LMS Adaptive Filter for Audio Processing,
in Proc. IEEE International Conference on Reconfigurable
Computing and FPGAs, Sept. 2006, pp. 1-8.
[9] www.digilentinc.com/
C. Timing Summary
Speed Grade: -2
Minimum period: 23.452ns
(Maximum Frequency: 42.641MHz)
Minimum input arrival time before clock:
15.361ns
Maximum output required time after clock:
2.826ns
Maximum combinational path delay: 3.237ns
VIII. CONCLUSION
238
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract: With the adiabatic switching approach, complexity. Adiabatic logic offers away to reuse the
the circuit energies are conserved rather than energy stored in the load capacitors rather than the
dissipated as heat. Depending on the application traditional way of discharging the load capacitors to
and the system requirements, this approach can the ground and wasting this energy.
be used to reduce the power dissipation of the This demonstrates the low power dissipation of
digital systems. This work focuses on the design of Adiabatic Logic by presenting the results of
a new full adder circuit, using adiabatic logic designs designing various design/ cell units employing
such as ECRL and PFAL are implemented in Micro- Adiabatic Logic circuit techniques. A family of full-
wind & DSCH. The efficiency of the circuits is explored custom conventional CMOS Logic and an Adiabatic
and compared for different adiabatic logics. With the Logic units for example, an inverter, a two input
help of adiabatic logic, the energy savings of upto NAND gate, a two-input NOR gate, a two-input
76% to 90% can be reached. Circuit XOR gate, a two-to-one multiplexer and a one-bit
simulations show that the adiabatic design units Full-Adder were designed in Mentor Graphics IC
can save energy by a factor of 10 at 50MHz and Design Architect using standard TSMC0.35 m
about 2 at 250MHz, as compared to the logically technology, laid out in Micro wind IC Station.
equivalent conventional CMOS implementation. All the circuit simulations has been done using
Index Terms-: Adiabatic, ECRL Adder, PFAL Adder, Full various schematics of the structures and post-layout
adder. simulations are also being done after they all have
been laid-out by considering all the basic design
I. INTRODUCTION rules and by running the LVS program. Finally, the
analysis of the average dynamic power dissipation
The main objective of this design is to provide new with respect to the frequency and the load
low power solutions for Very Large Scale Integration capacitance was done to show the amount of power
(VLSI) designers. Especially, this work focuses on dissipated by the two logic families.
the reduction of the power dissipation, which is
showing an ever-increasing growth with the scaling II. MOTIVATION
down of the technologies. Various techniques at the
different levels of the design process have been In the past few decades ago, the electronics industry
implemented to reduce the power dissipation at the has been experiencing an unprecedented spurt i n
circuit, architectural and system level. growth, thanks to the use of integrated circuits in
Furthermore, the number of gates per chip area is computing, telecommunications and consumer
constantly increasing, while the gate switching electronics. We have come a long way from the
energy does not decrease at the same rate, so the single transistor era in 1958 to the present day
power dissipation rises and heat removal becomes ULSI (Ultra Large Scale Integration) systems with
more difficult and expensive. Then, to limit the more than 50 million transistors in a single chip.
power dissipation, alternative solutions at each level The ever-growing number of transistors integrated on
of abstraction are propose. The dynamic power a chip and the increasing transistor switching speed
requirement of CMOS circuits is rapidly becoming a in recent decades has enabled great performance
major concern in the design of personal information improvement in computer systems by several orders
systems and large computers. of magnitude. Unfortunately, such phenomenal
In this work, a new CMOS logic family called performance improvements have been accompanied
ADIABATIC LOGIC, based on the adiabatic by an increase in power and energy dissipation of the
switching principle is presented. The term adiabatic systems. Higher power and energy dissipation in
comes from thermodynamics, used to describe a high performance systems require more expensive
process in which there is no exchange of heat with packaging and cooling technologies, increase cost,
the environment. The adiabatic logic structure and decrease system reliability. None the less, the
dramatically reduces the power dissipation. The level of on- chip integration and clock frequency will
adiabatic switching technique can achieve very continue to grow with increasing performance
low power dissipation, but at the expense of circuit demands, and the power and energy dissipation of
239
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
high-performance systems will be a critical design the conventional CMOS gates and adiabatic counter
constraint. For example, high-end microprocessors parts. To convert a conventional CMOS logic gate
in 2010 are predicted to employ billions of into an adiabatic gate, the pull-up and the pull-down
transistors at clock rates over 30GHz to achieve networks must be replaced with complementary
TIPS (Tera Instructions per seconds) performance transmission-gate(T-gate) networks. The T- gate
[1]. With this rate, high-end microprocessors power network implementing the pull-up function is used to
dissipation is projected to reach thousands of Watts. drive the true output of the adiabatic gate, while the
This thesis investigates one of the major sources of T-gate network implementing the pull-down
the power/energy dissipation and proposes and function drives the complementary output node.
evaluates the techniques to reduce the dissipation. Note that all the given inputs should also be
Digital CMOS integrated circuits have been the available in the complementary form. Both the
driving force behind VLSI for high performance networks in the adiabatic logic circuit are used to
computing and other applications, related to science charge-up as well as charge-down the output
and technology. The demand for digital CMOS capacitance, which ensures that the energy stored at
integrated circuits will continue to increase in the the output node can be retrieved by the power supply,
near future, due to its important salient features at the end of each cycle. To allow adiabatic
like low power, reliable performance and operation, the DC voltage source of the original
improvements in the processing technology. circuit must be replaced by a pulsed- power supply
The word ADIABATIC comes from a Greek word with the ramped voltage output.
that is used to describe thermodynamic processes
that exchange no energy with the environment and III. ADIABATIC LOGIC TYPES:
therefore, no energy loss in the form of dissipated
heat. In real-life computing, such ideal process Practical adiabatic families can be classified as either
cannot be achieved because of the presence of PARTIALLY ADIABATIC or FULLY
dissipative elements like resistances in a circuit. ADIABATIC [12]. In a PARTIALLY ADIABATIC
However, one can achieve very low energy CIRCUIT, some charge is allowed to be transferred
dissipation by slowing down the speed of to the ground, while in a FULLY
operation and only switching transistors under ADIABATICCIRCUIT, all the charge on the load
certain conditions. The signal energies stored in the capacitance is recovered by the power supply. Fully
circuit capacitances are recycled instead, of being adiabatic circuits face a lot of problems with respect
dissipated as heat. The adiabatic logic is also known to the operating speed and the inputs power clock
as ENERGY RECOVERY CMOS. synchronization.
It should be noted that the fully adiabatic operation
of the circuit is an ideal condition which may only be ECRL-Efficient Charge Recovery Logic:
approached asymptotically as the switching process Efficient ChargeRecovery Logic (ECRL) proposed
is slowed down. In most practical cases, the energy by Moon and Jeong [8], shown in Figure, uses cross-
dissipation associated with a charge transfer event is coupled PMOS transistors. It has the structure similar
usually composed of an adiabatic component and a to Cascode Voltage Switch Logic (CVSL) with
non-adiabatic component. Therefore, reducing all differential signaling. It consists of two cross-
the energy loss to zero may not possible, coupled transistors M1 and M2 and two NMOS
regardless of the switching speed. With the adiabatic transistors in the circuit.
switching approach, the circuit energies are An AC power supply pwr is used for ECRL gates, so
conserved rather than dissipated as heat. Depending a store covers and reuses the supplied energy. Both
out and/out are generated so that the power clock
on the application and the system requirements, this
generator can always drive a constant load
approach can sometimes be used to reduce the power
capacitance independent of the input signal. A more
dissipation of the digital systems.Here, the load
detailed description of ECRL can be found in .Full
capacitance is charged by a constant-current source
output swing is obtained because of the cross-
(instead of the constant-voltage source as in the
coupled PMOS transistors in both pre charge and
conventional CMOS circuits).
recoverphases. But due to the threshold voltage of
Here, R is the resistance of the PMOS network. A
the PMOS transistors, the circuits suffer from the
constant charging current corresponds to a linear
non-adiabatic loss both in the pre charge and recover
voltage ramp. Assume, the capacitor voltage VC is
phases. That is, to say, ECRL always pumps charge
zero initially. on the output with a full swing. However, as the
In the following, we will examine simple circuit voltage on the supply clock approaches to ck.So the
configurations which can be used for adiabatic recovery path to the supply clock is disconnected,
switching. shows a general circuit topology for thus, resulting in incomplete recovery. Vtp is the
240
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
threshold voltage of PMOS transistor. The amount of PMOSFETs as in ECRL logic, and that the
loss is given as functional blocks are in parallel with the
transmission PMOSFETs. Thus the equivalent
resistance is smaller when the capacitance needs to
Thus, from Equation , it can be inferred that the non- be charged. The energy dissipation by the CMOS
adiabatic energy loss is dependent on the load. Logic family and Adiabatic PFAL Logic family can
beSeen.
IV. POSITIVE FEEDBACK ADIABATIC
LOGIC
Figure2: The Basic Structure of the Adiabatic PFAL Figure5: ECRL SUM Circuit
Logic.
241
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
VI. CONCLUSION
242
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
243
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
( )= + ( )= ( ) ( )
(15)
(3)
Then the reconstructed signal in equation (12) can
( )= + be represented as
(4) ( )= ( ) ( )+ ( ) ( )) ( )
Z-Transform of interpolated signal is
(16)
( ) = ( ) = ( ( ) + ( ))
(5)
244
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
( ) = ( ( ) ( ))
(17)
IV. PERFORMANCE ANALYSIS OF
( ) = ( ( ) ( ))
PROPOSED SYSTEM
(18)
The distortion transfer function can be A QMF bank in which the analysis and synthesis
represented in frequency domain by putting = filters are FIR is said to be an FIR QMF bank.We
))
know that the FIR filters are having linear phase
( )= ( ) ( (
condition and symmetric conditions. If ( ) has
(19) linear phase, then T(z) is also linear phase, thereby
eliminating phase distortion. Let ( )=
Where ( ) represents amplitude distortion
( ) with ( ) is real.The linear phase
arg ( ( ) ) Represents phase distortion.
constraint requires ( ) = ( ) since
If a QMF bank is free from aliasing,
amplitude distortion and phase distortion is said to be ( ) has to be low pass filter, the only possibility is
a Perfect reconstruction filter [6]-[7] i.e equivalent to ( ) = ( ).
( )= . Where c=constant. From the equation With this choice ( )= / ( )
(15)
where ( ) is a real for all
( )= ( ) Using above equation we get
(20) ( )=
From the time domain representation of perfect ( )
( ) (1) ( ) (24)
reconstruction filter is given by
If N is even then the above expression reduces to
( )= ( ) zero at = , resulting in severe amplitude
(21)
The reconstructed signal y(n) is merely a scaled distortion. So we have to chose N to be odd so that
and delayed version of x(n). ( )= ( ) +
( )
III. DESIGN OF PROPOSED LINEAR PHASE ( ) (25)
FIR QMF BANK But ( )= ( ( )
) so the above
We know that the transfer functions of low
equation can be written as
pass and high pass analysis filters of a two channel
linear phase FIR QMF bank be H0(z) and H1(z) ( )= ( ) + ( )
respectively, and those of the synthesis filters of the
filter bank be F0(z) and F1(z) respectively. (26)
If this condition is not satisfied, then it is
Here H0(z) is a transfer function of the prototype amplitude distortion and termed as reconstruction
filter and impulse response of the prototype filter is error or peak reconstruction error (PRE) is given as
h0(n) . Let the poly phase components of H0(z) be
= 20 log ( ) +
E0(z2) and E1(z2) then poly phase type-I
representation of system[8] is given below. ( ) (27)
245
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
If the prototype filter is assumed to be ideal in the Figure 4(a) shows the normalized frequency
pass band and stop band, then there is a vsMagnitude plot for a LPF of analysis section. A
reconstruction error in the transition region, which range of -100 to 50 dB values are considered for
can be minimized for improving the performance Magnitude. The plot reflects the performance of the
ofthe filter. This can be achieved by the linear phase system, in terms of Magnitudeversus normalized
analysis filters having power complementary. frequency illustrated. Figure 4(b) shows the
normalized frequency vs Magnitude plot for a HPF of
V.SIMULATION RESULTS AND analysissection.A range of -100 to 200 dB values are
DISCUSSION considered for
gain in db
gain in db
0 1 -50 -200
Magnitude
Magnitude
Magnitude
gain in db
/ /
-50 -200
Figure 5(a) shows the normalized frequency vs
-100 -400 Magnitude plot for a LPF of synthesis section. A
0 0.5 1 0 0.5 1
normalised frequency normalised frequency
range of -100 to 50 dB values are considered for
(c)amplitude response of both LPF &HPF (d)amplitude distortion function Magnitude. The plot reflects the performance of the
200 1.005
system, in terms of Magnitude versus normalized
frequency illustrated. Figure 5(b) shows the
0 1
Magnitude
Magnitude
246
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
to 200 dB values are considered for Magnitude. The Guntur, India.We would like to express our special
plot reflects the performance of the system, in terms thanks of gratitude to the dean of Acharya Nagarjuna
of Magnitude versus normalized frequency University college of engineering & technology
illustrated.Figure 5(d) shows the normalized Prof.P.Siddaiah for his valueble suggestions.
frequency vs Magnitude plot for amplitude distortion
of LPF&HPF of synthesissection. A range of -100 to REFERENCES
200 dB values are considered for Magnitude. The plot
reflects the performance of the system, in terms of [1]. Vaidyanathan, "Multirate Systems and Filter Banks", Prentice
Magnitude versus normalized frequency illustrated. Hall, 1990
[2] Y.-P. Lin and P. P. Vaidyanathan, Linear phase cosine
Figure 6 shows the normalized frequency vs modulated maximally decimated filter banks with perfect
Magnitude plot for a analysis section. A range of -1 reconstruction, in Proc. Int. Symp. Circuits Systems
(ISCAS), May 30Jun. 2, 1994, vol. 2, pp. 1720.
to 1.5 dB values is considered for Magnitude. The
[3] T. Q. Nguyen, Near-perfect-reconstruction pseudo-QMF
plot reflects the performance of the system, in terms banks, IEEE Trans. Signal Process., vol. 42, no. 1, pp. 65
of Peak Reconstruction error 76, Jan. 1994.
peak reconstruction error [4] C.-W.Kok, W.-C. Siu, and Y.-M. Law, Peak constrained least-
1.5
squares QMF banks, Signal Process., vol. 88, pp. 23632371,
2008.
1
[5] J. D. Johnston, A filter family designed for use in quadrature
mirror filter banks, in Proc. Int. Conf. Acoustics, Speech,
Signal Processing (ICASSP), Apr. 1980, vol. 5, pp. 291294.
0.5
Magnitude
VI. CONCLUSION
This correspondence proposes a modified method
for the design of a two channel linear phase
symmetric FIR QMF bank, so that a weighted sum of
the maximum amplitude distortion of the analysis
filter bank and synthesis filter bank are minimized.
The simulation results show the improved
performance with regard to the amplitude distortion
and complexity and also decreases the peak
reconstruction error. It is also observed that the
proposed method is efficiently implemented by using
poly phase type-I representation. If order of the filter
is N then we requires a total multiplications per
unit time (MPU) and additions per unit time
(APU).
ACKNOWLEDGMENTS
247
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
E x(t)
2 NT
1
capacity by employing multiple antennas at both the
x(t)
2
dt
transmitter and receiver ends. If knowledge of the NT 0
248
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Since most practical systems deal with discrete-time by A and the number of iterations, denoted by i. In
signals, instead of reducing the continuous-time peak the first step, the iteration is taken as two i.e., i = 2
max |x(t)|, the maximum amplitude of LN samples of and the initial target clipping level is to be taken as A.
x(t) is reduced, where parameter L denotes the
oversampling factor [1]. To evaluate the PAPR The predetermined clipping level, denoted by A, is
performance from statistical point of view of the related to the target clipping ratio, and given as
OFDM signals, the complementary cdf (CCDF) is A 2
used to describe the probability of exceeding a given = 2
threshold PAPR0, i.e., CCDF(PAPR0) = Pr(PAPR > E { x n }
Where, Target
PAPR0) [4].
Clipping Ratio, A Predetermined Clipping Level, xn
Oversampled OFDM signal The clipping of the
III. PAPR REDUCTION TECHNIQUES
peak signal results to distortion of the original OFDM
signal [3]. The signal obtained after filtering the
Several techniques have been proposed to mitigate
the high PAPR of MIMO-OFDM signals. Clipping is
clipped signal
x (i+ 1 )
=x ,
(i)
+ c (i)
used to reduce the peak power by clipping the OFDM
signals to the threshold level. Selected mapping
(SLM) and partial transmit sequence (PTS) schemes Where,
(i)
c
Anti-Peak Signal at the ith iteration,
generate several alternative signal sequences convergence factor. The Convergence Factor (CF),
representing the same OFDM signal sequence and denoted by can be estimated by
select the one with the minimum PAPR among them. [ c (i)
, c (i)
]
Tone reservation (TR), tone injection (TI), and active
constellation extension (ACE) change constellation c (i)
, c (i)
points for some subcarriers to reduce the PAPR [3].
,Where, convergence Factor, Real Part, C(i)
Among various peak-to-average power ratio (PAPR)
reduction techniques, the active constellation
Peak Signal above the Pre-Determined Level, c (i)
extension (ACE) technique is attractive because ACE
allows the reduction of high-peak signals by ,
Anti-Peak Signal at the ith iteration, Complex
extending some modulation constellation points
Inner Part.
toward the outside of the constellation without any
loss of data rate [4]. The basic principle of clipping- The anti-peak signal at the ith iteration generated for
based ACE (CB-ACE) algorithms involves switching
between the time domain and the frequency domain.
the PAPR reduction, denoted by
(i)
, is given by c
However, the CB-ACE algorithms have a low
clipping ratio problem in that they cannot achieve the
minimum PAPR when the target clipping level is set
c ( i ) = T ( i ) c ( i ) Where,
(i)
Anti-Peak c
below an initially unknown optimum value. Our Signal at the ith iteration, T(i) Transfer Matrix at the
approach combines a clipping-based algorithm with ith iteration, C(i) Peak Signal above the Pre-
an adaptive clipping control, which allows us to find Determined Level. The transfer matrix at the ith
the optimal clipping level [4]. Simulation results iteration, denoted by T (i), used for generating the
show that our proposed algorithm can achieve the anti-peak signal is given
minimum PAPR regardless of the low target clipping
level.
T (i)
= Q * (i)
Q (i)
Where, T(i) Transfer
* (i)
IV. PROPOSED PAPR TECHNIQUE: Matrix at the ith iteration, Q Conjugate of
The main objective of the Adaptive Active (i)
Constellation Order,
Q
Constellation Order
Constellation Extension (Adaptive ACE) algorithm
for reducing the Peak-to-Average Power Ratio [4].
(PAPR) [11], is to control both the clipping level and The original Orthogonal Frequency Division
the convergence factor at each step and thereby Multiplexing (OFDM) signal, denoted by xn, is to be
minimize the peak power signal whichever is greater clipped in order to reduce the peak signals. The
than the initial target clipping level [4]. The Adaptive clipping signal is given by
j
Active Constellation Extension (Adaptive ACE) c (i)
n ( x (i)
n e n
,
x (i)
n
algorithm can be initialized by selecting the
x (i)
parameters namely the target clipping level, denoted n
249
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Where,
A Gradient with respect to A, N
^Hs^-1(P^T X I)) ^-1 and is minimized by a training
p matrix P that in general can only be derived through
Number of peak samples larger than A. numerical optimization. But there exist heuristic
solutions with good performance based on water
The Peak-to-Average Power Ratio by the Adaptive filling. As opposed to least-square estimation, the
Active Constellation Extension (Adaptive ACE) [8] , estimation error for spatially correlated channels can
algorithm is to be calculated for the Orthogonal be minimized even if N is smaller than the number of
Frequency Division Multiplexing (OFDM) signal transmit antennas. Thus, MMSE estimation can both
which is obtained after filtering the clipped signal [3]. decrease the estimation error and shorten the required
The clipped and filtered OFDM signal obtained by training sequence [7]. It needs however additionally
using Adaptive Active Constellation Extension the knowledge of the channel correlation
(Adaptive ACE) algorithm is to be transmitted via an matrix R and noise correlation matrix S. In absence
Additive White Gaussian Noise (AWGN) channel, in of an accurate knowledge of these correlation
order to calculate the Signal-to-Noise Ratio (SNR) matrices, robust choices need to be made to avoid
and Bit Error Rate (BER). MSE degradation.
250
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
References:
Packets are distributed in time and frequency domain
as described before, but the packets which are [1]. B. S. Krongold and D. L. Jones,PAR reduction in OFDM via active
constellation extension, IEEE Trans. Broadcast., vol. 49, no. 3, pp. 258
transmitted back to back through same group of 4 sub 268, Sep. 2003.
channel are corrupted due to slowly time varying [2]. L. Wang and C. Tellambura, An adaptive-scaling algorithm for
OFDM PAR reduction using active constellation extension, in Proc. IEEE
nature of fading. We maintain coherence time to be Veh.
more than that of the packet transmission time Technology Conf., Sep. 2006, pp. 15.
through a sub channel, and the channel condition is [3]. E. Van der Ouderaa, J. Schoukens, and J. Renneboog, Peak factor
minimization using a time-frequency domain swapping algorithm, IEEE
fed back for each packet. We simulated block fading Trans. Instrum. Meas., vol. 37, no. 1, pp. 145147, Mar. 1988.
channel with number of sub-bands N = 4 and the [4]. M. Friese, On the degradation of OFDM-signals due to peak-clipping
in optimally predistorted power amplifiers, in Proc. IEEE Globecom, Nov.
coherence bandwidth equivalent to 16 subcarriers (M 1998, pp. 939944.
= 16). QAM is used as modulation scheme. Thus, 128 [5]. J.J. van der Beek, 0. Edfors, M. Sandell, S.K. Wilson, and P. 0.
Borpesson, "On Channel estimation in OFDM systems," Proc. VTC'95, pp.
2 bits per OFDM symbol is transmitted through a 815-819.
sub channel. [6]. V. Kumbasar, O. Kucur, Alamouti Coded Wavelet Based OFDM for
Multipath Fading Channels, Wireless Telecommunications Symposium,
2009.
VII. CONCLUSIONS: WTS 2009, pp. 1 5.
[7]. B. G. Negash and H. Nikookar. Wavelet-Based Multicarrier
Transmission Over Multipath Wireless Channels, IEE Electronics Letters,
The Active Constellation Extension (ACE) vol. 36, (2000) October, pp. 1787-1788.
[8]. K. Abdullah and Z. M. Hussain, Studies on DWT-OFDM and FFT-
Algorithm provides the minimum Peak-to-Average OFDM Systems, IEEE International Conference on Communication,
Power Ratio (PAPR), even when the initial target Computer and Power, (2009) February 15-18.
[9]. W. Saad, N. El-Fishawy, S. EL-Rabaie and M. Shokair, An Efficient
clipping ratio is set below the unknown optimum Technique for OFDM System Using Discrete Wavelet Transform,
clipping point. Hence, the proposed algorithm avoids Springer-Verlag Berlin Heidelberg, (2010), pp. 533541.
the problem of low clipping ratio, which is caused in
251
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
252
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
253
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
its own as well as its companions flying experiences. i = 1,2 ....N , d = 1,2 ....D and N is the number of
Particles are assumed as tap weights in the adaptive particles . The symbols r1 and r2 represent random
channel equalizer. Initially tap weights are randomly numbers between 0 and 1. Similarly c1 and c2 denote
selected. The tap weights are continuously updated acceleration constants that pull each particle towards
by taking number of iterations. The continuous its best and global best positions. The acceleration
updation depends on position and velocity of the constants are usually taken as 2.05 for most
particles. Initial position error is maximum. For each applications.
iteration, position best(pbest) and global best(gbest) is
calculated. The concept of movement of particles of The inertia weight, W is employed to control the
PSO described in Fig. 2 impact of pervious history of velocities on the current
one in order for tradeoff between the global and local
exploitations. At early stage of optimization it is
desirable that the individual particles wonder though
the entire search space, without clustering around the
local optima. On the other a hand, during later stages
it is very important to enhance convergence
toward the global optima so as to find optimum
solution efficiently .
254
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Fig.4.2(a): Error plot obtained using RLS .4.3(a): Error plot obtained using PSO algorithm
Algorithm
MSE is calculated using RLS algorithm for the MSE is reducing by increasing the number
forgetting factor is 0.998 of iterations.
255
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
5. CONCLUSION
REFERENCES
[1] MK Ozdemir, H Arslan, Channel estimation for wireless
OFDM systems. IEEE Commun. Surv. Tutor. 9(2),
1848 (2007)
[2] R Negi, J Cioffi, Pilot tone selection for channel estimation in a
mobile OFDM system. IEEE Trans. Consum. Electron. 44,
11221128 (1998)
[3] L.N.de Castro,J.Timmis,An Introduction to artificial immune
systems:A New Computational Intellegence Paradigm, springer-
verlag, 2002.
[4] B.Widrow and S.D.Stearns,Adaptive Signal
Processing,Second Edition,pearson,2002.
Fig.4.4.Comparison in mean square error plot [5] S.Haykins,Neural Networks:A Comprehensive
of all three algorithms Foundation,Second Edition, Pearson, 2002.
[6] Widrow, B, Walach, E., Adaptive Inverse Control: A Signal
Processing Approach, Wiley-IEEE Press, Nov. 2007.
[7] Qureshi, S.,Adaptive Equalization, IEEE
Communications Magazine, March 1992, pp. 916.
[8] Proakis, J.G., Adaptive Equalization for TDMA Digital Mobile
Radio, IEEE Transactions on Vehicular Technology, Volume 40, No. 2,
May 1991.M. Young, The Technical Writer's Handbook. Mill Valley, CA:
UniversityScience, 1989.
256
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
257
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
good solutions efficiently while requiring minimal connected swarm, meaning that all the particles share
parameterization. The PSO algorithm was first information, any particle knows what is the best
introduced by Dr. Kennedy and Dr. Eberhart in 1995 position ever visited by any particle in the swarm
and its basic idea was originally inspired by
simulation of the social behavior of animals such as The basic steps of MPSO algorithm for the
bird flocking, fish schooling and so on [1]. These design of binary sequences is as follows
roots in natural processes of swarms lead to the Step1: (Initialization): Generate initial particles by
categorization of the algorithm as one of Swarm randomly generating the position and velocity for
Intelligence and Artificial Life. It is based on the each particle.
natural process of group communication to share Step 2: Evaluate each particles cost (E), which is
individual knowledge when a group of birds or amplitude of main peak of the auto correlation
insects search food or migrate and so forth in a to the absolute maximum amplitude in the side
searching space, although all birds or insects do not lobes.
know where the best position is. But from the nature Step 3: For each particle, if the cost (E) is less than
of the social behavior, if any member can find its previous best (Pbest) fitness, update Pbest.
out a desirable path to go, the rest of the members Step 4: For each particle, if the cost (E) is less
will follow quickly. Researchers shown that although than the best one (Gbest) of all the particles, update
PSO finds solutions much faster than most of the Gbest.
contemporary search techniques like Evolutionary Step5: For each
and Genetic Algorithm, it usually do not improve the particle
quality of solutions as the number of iterations a) Generate a new particle according
increase and thus becomes a victim of premature to the following formulae
convergence resulting in a suboptimal solution. This
drawback of PSO is due to the lack of diversity,
which forces the swarm particles to converge to
the local optimum.
In MPSO each particle keeps track of its Step6: If the stop criterion is satisfied, then stop, else
coordinates in the solution space which are associated go to Step 3.
with the best solution (fitness) that has achieved so
far by that particle. This value is called personal best, VI. IMPLEMENTATION OF MPSO
Pbest. Another best value that is tracked by the PSO ALGORITHM
is the best value obtained so far by any particle in
the neighbourhood of that particle. This value is Binary codes having good auto-correlation
called Gbest. In this algorithm we have a completely properties are synthesized by using Particle swarm
258
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
259
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
REFERENCES
V. CONCLUSION
260
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
261
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
262
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
263
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
264
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
265
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
I.
INTRODUC
TION
Fig.1 Block Schematic
PSK modulation is widely used in satellite of I-Q Modulator
data transmission taking into consideration the
spectral efficiency offered by the scheme at a given The data rate of a QPSK modulator is twice the
transmitted power. With the amount of data to be data rate of I and Q data streams. The
transmitted steadily going up, the demand on the data spectrum of a typical QPSK modulated signal
rates to be supported by the modulator is increasing. centered at 2.0 GHz modulated at data rate of
For Low Earth Orbit (LEO) satellites, the 500 Mbps is shown in Fig. 2. In this simulation,
frequency band allotted for payload data transmission PRBS data has been considered for modulation.
lies in the X-band. Realization of a high bit-rate (up to When an alternating sequence of 1 and 0 is
500 Mbps) X- band QPSK modulator and related considered for modulation, the spectrum of the
analysis is presented in this paper. Section II of the QPSK modulated signal would be as shown in Fig.
paper describes the operation of I-Q modulator and I-Q 3. Suppression of power in the carrier and
mixer. Generation of X-band QPSK modulator using symmetry in the amplitude of the frequency
these components is discussed and simulation results components on either side of the suppressed
are presented. Effect of I-Q data skew on QPSK carrier may be noted in this figure.
spectrum asymmetry is discussed in Section III.
Section IV discusses the effect of phase error
introduced in quadrature power divider on the I-Q
mixer output. Practical results of the realized X-band
QPSK modulator supporting 500 Mbps are presented
in Section V. Section VI summarizes the work done and
the results obtained.
266
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Fig.4 Block
Schematic of I-Q
Mixer
267
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
268
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
269
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
270
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
AbstractIn pilot symbol aided channel estimation, LS estimated channel frequency response coefficients
the position of interpolator plays an important at the pilot positions are interpolated in a low-pass
role. In conventional channel estimator, which interpolator. Many types of interpolation schemes are
uses Least Squares (LS) estimation, initially LS discussed in literature [1],[2],[5],[6],[7]. Though low
estimates are obtained at pilot positions, then pass interpolation needs more hardware than linear,
interpolation and denoising operations are low-pass performance is much higher in terms of
performed in order. The most commonly used BER vs. Signal to Noise Ration (SNR) than the
interpolator is the low-pass interpolator. In this others. So it is common in conventional OFDM
paper, it is shown that OFDM systems, which use systems. The basic pilot arrangements are block and
denoising prior to interpolation, have poor system comb types, out of which comb type, suits for both
performance in terms of Bit Error Rate (BER) flat fading and frequency selective fading cases. In
compared to conventional systems which use comb type pilot arrangement, interpolation in an
denoising after interpolation. The theoretical OFDM symbol is required. Based on the
treatment as to why the positioning of the low- interpolation scheme used, the overall system
pass interpolator should be before the denoising performance changes.
block is framed. Further, the simulation results Denoising methods improve the accuracy of LS
obtained in various channel environments provide estimated, interpolated channel response [4], [8], [9].
strong evidence to the proposal. In [4], denoising based on truncation of estimated
channel impulse response is proposed. Thus it is
Keywords-OFDM; interpolation; denoise; understood that in conventional OFDM systems, low-
channel estimation; interpolation; LS; MMSE; pass interpolation is performed on the pilot position
LS channel estimates and then the obtained channel
I. INTRODUCTION AND RELATED WORK
frequency estimates are denoised to obtain the final
Channel estimation in OFDM systems is generally channel response. In literature, most of the channel
done either at the receiver side or at the transmitter estimation schemes based on LS, first use
side. If it is performed at the transmitter side, the interpolation and then denoising rather than the
information is fed back to the receiver so as to design reverse. In this paper, a detailed treatment on why
the equalizer. If the estimation is performed at the this should be the case is given. Simulation results
receiver, the hardware complexity slightly increases are also presented which verify this fact.
but the methods yields better performance and The rest of the paper is organized as follows. In
removes the necessity of feedback. This estimation at section II, the OFDM system is introduced along with
the receiver side is done in many ways, blind, semi- the notations used. The conventional channel
blind or pilot based. In this paper pilot symbol based estimation technique is discussed in section III. The
channel estimation is performed. theoretical analysis on the positioning of
In pilot based estimation, known pilot sequences interpolation filter, is given in section IV. In section
are inserted in the OFDM symbols, at known V, simulation results are shown in different channel
positions and are transmitted along with the data environments. Section VI concludes the paper.
symbols. At the receiver a channel estimation based
on either Least Squares (LS) or Minimum Mean II. OFDM SYSTEM AND SYMBOLS
Square Estimation (MMSE) is commonly used. LS The OFDM system is shown in figure (1). As it is
estimation is simple but it has poor performance understood from the block diagram, the baseband
compared to MMSE [1],[3],[7]. But, LS is more transmission in an OFDM system requires serial to
common than MMSE because MMSE needs prior parallel conversion. The modulation scheme used by
statistics about channel autocorrelation and noise, the mapper may be any of the digital modulation
which are not so easily available in practice. techniques like QAM, BPSK, QPSK, etc. The
271
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
parallel transmissions at a given time instant are said domain symbol. Using an IDFT block the symbol
to be one OFDM symbol at that time instant. A gets converted to time domain. In time domain,
length N OFDM symbol has N different subcarriers. Guard band namely the cyclic prefix is added and is
The OFDM symbol is considered to be a frequency then sent through the channel serially.
S (k ) s (n)
Channel
AWGN
R (k ) r (n)
Parallel to Channel
Binary Estimation Cyclic Serial to
data Demapper Serial
and Symbol FFT Prefix Parallel
Conversion
Estimation Removal Conversion
r(n) is passed through a DFT block and the frequency Pilot LS OFDM OFDM
domain received symbol is obtained. On this symbol, DFT data
symbol estimation Symbols symbol
channel estimation is performed. The channel estimation Retrieval block Interpolation denoising
Block Block block
block is shown in Figure (2). The LS channel estimation
is given by (3).
Figure 2: Conventional Channel estimation
272
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
273
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
F h (n) w(n) I (k ) (15) channel response is similar to h1 ( n) because it is time
LS
limited but not band limited.
H LS (k ) W (k ) I (k )
So, it can be concluded that theoretically H 1 ( ) gives
To make the analysis simple, two facts are assumed: First
is using DTFT instead of DFT for theoretical analysis and better receiver performance than H 2 ( ) and practically
second, assuming a low-pass type of interpolation. With
these assumptions, (14) is re-written as shown in (16),
H 1 ( k ) gives better receiver performance than H 2 (k ) .
This fact is further supported form the simulation results.
using I ( ) , which is the frequency response of the low
pass interpolation filter, given in (17). In (17), A is a V. SIMULATION RESULTS
positive value which is greater than 1 and is chosen based In this section, the conventional channel estimation
on cut-off frequency of low pass filter. Using (17) in scheme is implemented in three different channel
(16), we have (18). environments for two different cases, namely 1)
1 Interpolation first and then denoising 2) Denoising first
H 1 ( )
2 H
LS ( ) I ( )W ( )d (16) and then interpolation. The three different channel
environments are Advanced Television Technology
Center (ATTC), Digital Video Broadcasting for
1
I ( ) A A (17) Television (DVB-T), [1] and Brazil-E [2] channel. In all
the three channel environments, the conventional channel
0 otherwise estimation with case 1, showed superior performance than
case 2. The simulation parameters are shown in table (1).
/A
1
H 1 ( )
2 H
/ A
LS ( )W ( )d , Using comb type of pilot arrangement and lowpass
interpolation, the OFDM system performance results in
three channel environments are shown in figures (6),(7)
(18) and (8) respectively. In DVB-T the channel length is 50.
Similarly using the two assumptions in (15) we have So truncation is done by choosing L as 50. Similarly, in
(19) and using (17) in (19), we get (20). Brazil E channel, L is chosen 25. In ATTC, L is chosen as
128.
I ( )
2
TABLE I: OFDM System Simulation Parameters
H 2 ( ) H LS ( )W ( )d (19) Parameter Value
Mapper QPSK
Number of subcarriers (N) 1024
1
H 2 ( )
2 H
LS ( )W ( )d ,
A
A
Rate of pilot subcarriers
Guard interval (per OFDM symbol)
N/8
256
Channels: Static type ATTC, DVB-T, Brazil-E
(20)
274
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
VI. CONCLUSION
In this paper, a theoretical treatment on the positioning Denoising and then Interpolation
-0.2
of interpolation filter and denoising block in a 10 Interpolation and then denoising
BER
T, Brazil-E and ATTC channel environments. The
-0.4
theoretical proof is given which shows that low pass 10
interpolator should be before the denoising block. The
study in this paper is limited to static impulse responses of
various channels. But it can be extended to slow fading -0.5
10
channels as well as fast fading channels which give similar
results.
5 10 15 20 25
SNR in dB
-0.1
10 Denoising and then interpolation
Interpolation and then denoising
Figure 7: BER Performance of OFDM system in Brazil-E
channel environment
-0.2
10
-0.3
REFERENCES
10
BER
-0.4
[1] Sinem Coleri, Mustafa Ergen, Anuj Puri, Ahmad Bahai, Channel
10
estimation techniques based on pilot arrangement in OFDM
systems, IEEE transactions on broadcasting, vol 48,No. 3,Sep.
-0.5 2002, pp. 223-229
10
[2] J Carlos Augusto Rocha, Luciano Leonel Mendes, Adonias Costa
5 10 15 20 25 da Silveira, Performance Analysis of channel estimation schemes
SNR in dB for OFDM systems, Proc. Of International workshop on
Figure 5: BER Performance of OFDM system in ATTC channel telecommunications-IWT-2007, pp.32-36.
environment [3] Meng-Han Hsieh, Che-Ho Wei, Channel estimation for OFDM
systems based on comb type pilot arrangement in frequency
selective fading channels, IEEE Transactions on Consumer
Denoising and then interpolation
-0.2
Interpolation and then denoising
Electronics, Vol. 44, No. 1, February 1998, pp. 217-225.
10
[4] Jan-Jaap van de Beek, Ove Edfors, Magnus Sandell, Sarah Kate
Wilson, Per Ola Borjesson, On Channel Estimation in OFDM
systems, VTC, IEEE, 1995,pp. 815-819
-0.3
10 [5] Baoguo Yang, K.B. Letaief, Roger S. Cheng, Zhigang Cao,
Windowed DFT based pilot-symbol-aided Channel estimation for
BER
275
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract-This paper provides a survey of privacy- There are number of the applications in the WSN.
preserving techniques for WSNs. The wireless This includes [1]the military applications which
transmission required by a WSN, and the self- controls the monitoring, tracking and surveillance
organizing nature of its architecture, makes privacy of the borders. Other applications include the
protection for WSNs an especially challenging problem. environmental applications, health applications,
In particular, we review two main categories of privacy- home applications, commercial applications.WSN
preserving techniques for protecting content-oriented are capable to collect their data automatically with
and context-oriented privacy, respectively. Thus the help of the sensor devices. Even though there is
protection of the physical location is needed at data great benefit to the users some misuse them so
source and the data sink. Suppose we take the scenario
privacy is a concern there. If we take famous
of the panda hunter where in sensors are being
scenario of the panda and hunter [2], the hunters
deployed in the forest to monitor the endangered
can find physical location of the panda by
pandas. The adversary is quite efficient to monitor the
panda and capture the panda. We present taxonomy of
monitoring the traffic or using sound monitoring
various privacy preserving techniques against a local where in panda is tracked with help of the sound
eavesdropper is being presented. This paper also recognition.
compares the performances of all the privacy preserving
techniques. 1.2 Need for Privacy Preservation In Wireless
Sensor-Networks
Keywords:WSNs, Privacy Preserving Techniques,
Routing protocols, Traffic analysis.
The privacy preservation techniques of wired
networks are not useful in sensor networks,
1. INTRODUCTION firstly due to the fact that network being different
the set of problems are different and secondly
1.1WirelessSensorNetwor because many of the methods pose overheads which
ks are too burdensome for the sensor networks. The
shared wireless medium enhances the chances for
A wireless sensor network (WSN) is a heterogeneous the adversary to locate the origin of the
network composed of a large number of tiny low-cost radio transmission and thus facilitating the hop
devices, denoted as nodes (or motes), and one or few by hop trace back to the origin of the multi hop
general purpose computing devices referred to as communication. Launch of physical attacks and
base stations (or sinks). A general purpose of the node compromises by the adversary thus posing a
WSN is to monitor some physical phenomena (e.g., menace to the whole wireless sensor networks is
temperature, barometric pressure, light) inside an quiet evident due to the miniature size of the sensor
area of deployment. Nodes are equipped with a nodes and very nature of the wireless
communication unit (e.g., radio transceiver), communication environment.
processing unit, battery and sensor(s). Nodes are As we know a wireless sensor network is severely
constrained in processing power and energy, whereas constrained by various resources as computation,
the base stations have laptop capabilities and not storage, and wireless communication bandwidth
severely energy resources. The base stations usually and battery power. The adversary could monitor
act as Gateways between the WSN and other such activities of the sensor as the communication
networks (e.g., Internet). patterns to figure out the energy depletion or
resource usage in order to spot the most
276
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
3.1Data-OrientedPrivacy PreservingTechniques (c) Node E receives multiple (d) Several clusters have been
HELLO messages, and then constructed and the
In data-oriented privacy preserving technique main aggregation randomly selects one to join. tree of cluster
focus is on data which is collected and send to the sink. leaders is formed.
277
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
278
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
location privacy Network Mixing Ring (NMR) [17] the inconsistency of the traffic pattern caused by
is proposed that extends the routing through a hotspots, and camouflage the source node within the
random intermediate node. Logical ring containing group of nodes forming the cloud. The fake packets
large number of nodes is formed around the base also enable the real source node to send the sensed
station to mix messages globally. The actual routing data anonymously to a fake source node selected
has three steps. In the first step message is routed from the clouds nodes to send to the Sink. Crypto-
through a random intermediate node to some node in graphic operations are used to change the packets
the ring, which provides local source location appearance at each hop to prevent packet correlation
privacy. In the second and make the source node indistinguishable because
step, the message circulates in the ring to achieve the the adversary cannot differentiate between the fake
global source location privacy. In the last step, the and real traffic, i.e., the clouds traffic pattern looks
message is simply routed from a random node in the random for the adversary. Moreover, tracing the
ring to the base station. Source location privacy packets back to the source node is nearly impossible
problem was also taken up in [15] to trap an because the real traffic is indistinguishable and the
adversary in a false cyclic path. Several cyclic paths real source node sends its packets through different
are pre-generated in the network which results in fake source nodes.
improved safety period. Once a real message crosses 3.2.1.2 Location privacy for base station
one of the cyclic paths, fake messages start to Since base station collects data from entire network
circulate along this path. Then, adversary which security for a base station is highly essential. If an
backtracks on real messages can be stuck onto the adversary is able to locate the base station, it can
cyclic path. The safety period is improved by this apply several attacks such as Denial of Service
technique while maintaining low latency. attacks [4]. Hence protecting the location of the base
station is extremely important.There are three basic
The idea of fake source and dummy traffic [19] are
traffic analysis techniques [23] for discovering a base
proposed to hide real messages and sources within a
station location: the rate monitoring attack, the time
dummy traffic and fake sources. Two techniques are
correlation attack and the content analysis attack. In the
proposed for above idea namely Short-lived and
rate monitoring attack, an adversary observes node
Persistent Fake Source techniques. Again, these message sending rate and moves closer to a node with
techniques cannot protect the source location against the highest rate. In the time correlation attack, an
a global adversary, because the fake sources are adversary monitors a correlation in sending times
activated by real messages. In this way first source between a node and its neighbors. The adversary tries
can be recognized as the real one by global to detect which node forwards the current message and
adversary. To achieve protection against global traces the path directly to a base station. In the content
adversary; techniques are proposed such as Periodic analysis attack, an adversary tries to acquire valuable
Collection [20] and the Source Simulation [20]. information from message headers and payloads.
Periodic Collection provides optimal source location
privacy. In this technique every sensor node To get protection against the rate monitoring attack,
periodically sends a dummy or real message to the every sensor node sends messages at a constant rate.
sink at pre-defined intervals thus the real messages A child node continuously transmits data packet until
are covered within the dummy ones and the network it is accepted by a parent node. If the child node has
traffic is totally independent from the sensed events. no packet to send, it injects a dummy packet. This
In second technique several fake event sources are technique guarantees constant sending rate over the
added in the network. These fake sources behavior is network at the same time it decreases the lifetime of
identical with the real sources to confuse the the WSN. Different encryption techniques can be
adversary. The biggest challenge of the Source used to protect the base station against content
Simulation technique is to precisely simulate the analysis but they are not adequate. These techniques
source behavior. A statistical framework for do not solve the problem of message identity, which
evaluation of source anonymity in WSNs was remains the same along the path and easy to track.
proposed in [22]. All the above techniques are based Hop-by-hop re-encryption technique [23] can be use
on the fake sources and the dummy traffic that to overcome this problem.
ensures source location privacy in the presence of Different techniques [24, 25] are proposed to counter
global adversary on the cost of huge communication the traffic analysis attacks. The rate monitoring attack
overhead or higher latency. Instead of gaining great can be partially prevented by the multiple parents
source location privacy these techniques cannot be routing technique since traffic spreads along multiple
paths. In this technique, every sensor node has
practically implemented over a long period of time. multiple parent nodes, which route messages to the
Finally, a cloud-based scheme was proposed [21] for base station. Packet is forwarded through any parent
efficiently protecting source nodes location privacy node which is randomly selected by its child nodes.
against Hotspot- Locating attack by creating a cloud
This technique can be extended by the controlled
with an irregular shape of fake traffic, to counteract
random walk. A node forwards a message to one of
279
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
its parent nodes with probability p.with probability1- entities in the network are indistinguishable to the
p the node forwards the message randomly to one of nodes around the sink with regard to communication
its neighbors including the parent nodes. This statistics. Arranging the location of k entities is
technique introduces delivery time delays, which are complex as it affects two conflicting goals: the
proportional to extra hops taken by the messages. routing energy cost and the achievable privacy level,
This technique is still error-prone to the time and both goals are determined by a no analytic
correlation attack. Therefore, Multi-parent routing function.Wemodel such a positioning problem as a
technique with fractal propagation is used to avoid nonlinearly constrained nonlinear optimization
errors. When a node observes that a neighbor problem. To tackle it, we design a generic-algorithm-
forwards a message to the base station, the node based quasi-optimal (GAQO) method that obtains
generates a fake message with some probability and quasioptimal solutions at quadratic time. The
forwards it to one of its neighbors. The main problem obtained solutions closely approximate the optima
with this technique is that it generates a large amount with increasing privacy requirements. Furthermore,
of traffic near the base station, because nodes near to solve k-anonymity sink-location problems more
the base station usually forward more messages. efficiently, we develop an artificial potential-based
quasi-optimal (APQO) method that is of linear time
All the above discussed techniques provide sink
complexity. Our extensive simulation results show
location privacy against the local adversary. A
that both algorithms can effectively find solutions
completely randomized [26] technique is used to
hiding the sink among a large number of network
protect against global adversary. In this technique, a
nodes.
data packet randomly travels through a network for a
predefined number of hops discarding the fact that it 4. COMPARISONS AND CHALLENGES
reaches the base station or not. To improve the
In this we compare all the privacy preserving
probability of data packet delivery, multiple copies of
techniques that have been reviewed in this paper. Table
the same data packet can be sent out by the source or
1 depicts the performance of privacy preservation
a history of already visited nodes can be appended to
techniques in WSNs. We evaluate their performances in
the message to prevent loops in the routing path. In
metrics: privacy, efficiency, location, delay time, and
addition, this technique provides protection against
an internal adversary, because no information on the power consumption, message overhead, scalability.
sink is included in a data packet. This technique may Privacy refers to the degree. The delay time includes
seem to be impractical for larger networks, because both the computation and communication time of data
the probability of delivery decreases rapidly with an transmission at the intermediate sensors. The power
increasing number of nodes. Another scheme based consumption measure focuses on the additional
on local flooding of source and greedy random- messages required for transmission (i.e., additional
walk[26] of sink is proposed to protect the location energy consumed) in the WSN. There are some open
privacy of mobile sinks in sensor networks. In this issues for future research. As the network behavior
scheme, sensor do not know any information about changes so sensor nodes should be designed
sink-location, data are forwarded by local flooding accordingly, along with it there can be improvement in
and stored at pass nodes in the network, the sink every technique with respect to overhead, power,
move in greedy random-walk to collect data from the accuracy.
local nodes occasionally, which prevents the attackers
from predicting their locations and movements. The Table 1: Comparison between different techniques
analysis shows that the scheme can provide location
privacy of mobile sinks effectively, while providing
satisfactory data collection services.
We also use location privacy routing protocol (LPR)
[27]that is easy to implement and provides path
diversity. Combining with fake packet injection, LPR
is able to minimize the traffic direction information
that an adversary can retrieve from eavesdropping.
By making the directions of both incoming and
outgoing traffic at a sensor node uniformly
distributed, the new defense system makes it very
hard for an adversary to perform analysis on locally
gathered information and infer the direction to which
the receiver locates.
To protect the sink-location privacy from a powerful
adversary with a global view, we propose to achieve
k-anonymity [28] in the network so that at least k
280
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
REFERENCES
[4] John Paul Walters, Zhengqiang Liang, Weisong Shi, and Vipin
Chaudhary. Wireless Sensor Network Security: A Survey:Security in
Distributed, Grid, and Pervasive Computing Yang Xiao,(Eds.)
Auerbach Publications, CRC Press.
This paper presents a review of privacy-preserving [12]. Jean-Franois Raymond. Tra_c analysis: Protocols, attacks,
design issues and open problems. In Proceedings of International
techniques for wireless sensor networks (WSN). Two Workshop on Design Issues in Anonymity and
main categories of privacy preserving techniques Unobservability, pages 10{29. Springer- Verlag New York, Inc.,
have been presented; data oriented and context- 2001.
oriented respectively. The existing techniques have
been compared with respect to efficiency, location [13].Pandurang Kamat, Wenyuan Xu, Wade Trappe, and Yanyong
Zhang. Temporal privacy in wireless sensor networks. In ICDCS
overhead, delay, power consumption. th
'07: Proceedings of the 27 International Conference on
Distributed computing Systems, pages 23-30, Washington, DC,
USA, 2007. IEEE Computer Society.
281
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
282
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
283
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Cycle time is shown in Fig2. As per the Service Level In this section we study Variance based DWFQ
Agreement (SLA) between the central office and (VBD- WFQ) and Hybrid based DWFQ
the end user, minimum guarenteed bandwidth is (HBDWFQ) algorithms. These algorithms are aimed
allocated to each ONU. Let N be the no of ONUs. at changing the queue weights dynamically based
Minimum bandwidth is calculated as follows. on the predicted arrivals and measured buffer
occupancy. This change of weights controlls the
average delays of queues by controlling the service
min is the minimum guarenteed amount of rates of the queues. Each diffserv enabled ONU
bandwidth given to each ONU. C is the channel supports three service classes:
capacity. We have seen in the MAC protocol, that
the OLT collects reports before sending the GATE ExpediatedForwarding(EF),AssuredForwarding(AF)
messages to the ONUs. In the current cycle the ONUs ad BestEffort(BE)serviceclasses.TCONT1,TCONT2
send their REPORTs to the OLT. OLT and
recalculates the bandwidth window to each ONU
TCONT3holdsEF,AFandBEservicesrespectively.
based on the current REPORTs.
Inorder
i to calculate the average
i
delays of packets of a
queue, we assume that the packets arrive at
discrete times and the packets that enter in time
interval (k , k) are considered to enter the queue
at the time k . is the cycle time that is shown
in Fig 2. Time interval from k to k is defined as
the nth cycle. In nth cycle DWFQ allocates a
service rate to class i at time k , is denoted as s
k
. Buffer occupancy at time k is bk .
284
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
285
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
286
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
CONCLUSION REFERENCES
The algorithms studied in this paper allocates the [1]ITU-TRecommendationG.984.xSeries
bandwidth in two stages namely Inter and Intra [2]IEEE,IEEE802.3ah,Sept.2004
ONU scheduling. OLT keeps track of the left [3]M.McGarry,M.MaierandM.Reisslein,EthernetPO
s:ASurveyofDynamicBandwidthAllocation(DBA)Algorithms.
over bandwidth by lightly loaded ONUs and it is IEEECom-mun.Mag.,Vol.42,No.8,PagesS8-S15,August2004.
distributed among heavily loaded ONUs in [4]Skubic,B.,JiajiaChen,Ahmed,J.Wosinska,L.andMu
proportion to the requested bandwidth, thus the herjee,B.,AComparisonofDy
increase of the channel utilization and satisfying namicBandwidthAllocationforEPON,GPON,and Next-
GenerationTDMPON, IEEECommun.,Mag.,Vol.47,Issue.3.
SLA needs. ONU distributes the bandwidth 2009.
independent of the OLT, and hence service [5]HussainS.;Fernando,X.,EPON:Anextensiverevie foruptodate
provider can install customized ONUs at customer dynamicbandwidthallocationschemes, IEEEConferences.,2008.
premises by decreasing the OLT computational [6]K.-H.Ahn,K.-E.Han,andY.
C.Kim,Hierarchicaldynamicbandwidth
complexity. An error exists in measuring expected AllocationalgorithmformultimediaservicesoverethernePONs,J.ET
arrivals by using maximum function. To minimize RI, vol.26,pp.321331,2004.
the error better prediction methods can be used, [7]ThuNgoQuynh,HolgerKarl,AdamWolisz,andKlausRebensburg
there by increasing the utilization, especially for .,
RelativeJitterPacketSchedulingforDifferentiatedServies,9thIFIP
VBDWFQ. Conference.,2001.
287
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract This paper details the implementation of of a sinusoidal waveform. A sinusoid is synthesised
Field Programmable Gate Array (FPGA) based sine when the LUT containing digital data is scanned and
wave generation and three phase square wave delivered to the input of a digital to analog converter.
generation technique for the functioning of navigation
sensors. The Sine waveform of 6V and different This paper describes the implementation details
frequencies is required for excitation of rotor angle of Digital Synthesis technique on FPGA platform.
sensor and 3 phase power signal for the wheel rotation. This scheme makes stable waveform with
The Sine wave generation includes counter logic for an programmable frequency. The amplitude and
address generator to read the stored values from a
LUT. The sampled values are converted in to
frequency trimming requirements of conventional
continuous sinusoid using a digital to analog methods can be avoided here. In addition it
converter (DAC). The 3 Phase waveform based on provides good spectral purity. The output
Johnson counter logic is implemented in FPGA. This frequency can be easily changed by changing the
digital technique of wave form generation has many frequency of the reference clock or by
advantages over conventional scheme based on analog reprogramming the LUT.
design. The VHDL program is implemented on FPGA
chip ACTEL A1020. The simulation result of VHDL II. DEVICE SELECTION
is shown in the form of timing diagram obtained using
ModelSim software. The experimental result of
generation of excitation waveform for the sensors is also The design is realized with Actel FPGA A1020. A
presented. single FPGA services both three phase signal for
motor power and sine wave for the Pickoff
I. INTRODUCTION excitation. FPGA is clocked by a Crystal Oscillator,
which has a frequency stability of 50ppm over full
Navigation system use Dynamically Tuned MIL temperature range. Power on Reset circuit is
Gyroscope to acquire the angle/rate information. kept outside the FPGA. DAC AD7845 having
Operation of the gyroscope requires precise and high temperature stability is used to generate sine
stable sine power signal with frequency wave of required frequency from the FPGA output.
between12-17KHz and the voltage amplitude The DAC output goes to current buffer which also
of 6V for exciting the rotor angle measurement does the filtering to remove the sampling noise.
coil (pickoff coil). The spin motor of DTG is a
hysteresis synchronous motor, which is designed Switching amplifier is kept outside the FPGA to
to run between 7800 and 10200rpm and 3 phase provide necessary current drive for the three phase
motor power of 260-340Hz is required for the motor signal for the wheel excitation signal of DTG. The
operation. design utilizes 54% of logic I/O gates and I/O used is
32 out of 60.
Traditionally the sine signal has been generated
byWeinbridge Oscillator with output power stage and III. FUNCTIONAL BLOCK DIAGRAM OF FPGA
amplitudestabilization network. The motor power has
been generated by a digital counter circuit followed
by power switching. Themotor power generator and
Weinbridge Oscillators are freerunning and
harmonics of these signals can beat in
precisioncircuits and degrade navigation
performance. Theperformance of the analog circuits
is also subject to ageingand environmental changes.
digital synthesis is a Look-Up Table (LUT) based
technique to generate any arbitrary waveform using
stored digital samples. The LUT contains digital data
288
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
289
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
CONCLUSION
290
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
ABSTRACT: This paper introduces a very effective external disturbance [8]. If an image is being sent
approach of digital color image segmentation. In electronically from one place to another via satellite
recent days segmentation and edge detection of or wireless transmission or through networked
noisy images is a vast challenge for the computer cables, we may expect errors to occur in the
vision. In this paper Minimum Mean Square Error image signal. These errors will appear on the image
(Wiener) Filtering has been introduced for noise output in different ways depending on the type of
removal from the noisy images and watershed disturbance in the signal. Usually we know what
algorithm using distance transform is applied on the type of errors to expect and the type of noise on
noise free images for segmentation purpose. The the image, hence we investigate some of the
combination of these two strategies produces an standard noise for eliminating or reducing noise in
effective segmentation approach for noisy images. color image [9].Image Noise is classified as
Comparative study is also done in this paper. The Amplifier noise (Gaussian noise), Salt-and-pepper
experimental result ensures the effectiveness of the noise (Impulse noise),Shot noise, Quantization
proposed approach. noise (uniform noise),Film grain, on-isotropic
noise, Speckle noise (Multiplicative noise) and
Key words: Image segmentation, Wiener filtering, Periodic noise.
Watershed algorithm. To perform image segmentation and edge detection
tasks, there are many methods that incorporate region
INTRODUCTION growing and edge detection techniques [10-19] but no
such research work is done with the combination of
Any gray tone image can be considered as a Minimum Mean Square Error (Wiener) Filtering and
topographic surface. If we flood this surface from Morphological image segmentation (Watershed)
its minima and, if we prevent the merging of the algorithm. In this paper a combination of Wiener
waters coming from different sources, we partition filtering and watershed algorithm using distance
the image into two different sets: The catchment transform techniques for image segmentation and
basins and the watershed lines. If we apply the edge detection procedures were used. The wiener
watershed transformation to the image gradient, filtering method was applied to obtain noise free
the catchment basins should theoretically images. Then a watershed algorithm using distance
correspond to the homogeneous gray level regions transform technique is used to segmentation of
of this image. However, in practice, this images.
transform produces an important over-segmentation The structure of this work is the following: Section 3
[1-7] due to noise or local irregularities in the introduces Minimum Mean Square Error (Wiener)
gradient image [8-9]. Filtering. Section 4 is devoted to the segmentation
Images are often degraded by noises. Noise can occur process for edge detection using watershed algorithm
during image capture, transmission, etc. Noise using distance transform. Section 5 presents the
removal is an important task in image processing. In proposed scheme. The results are discussed in section
general the results of the noise removal have a 6 and we nish this paper with some concluding
strong influence on the quality of the image remarks with section 7.
processing and image segmentation technique. Noise
is the result of errors in the image acquisition process 2. WIENER FILTERING
that results in pixel values that do not reflect the true
intensities of the real scene. Noise reduction is the The inverse filtering is a restoration technique for
process of removing noise from a signal. Noise to be deconvolution, i.e., when the image is blurred by a
any degradation in the image signal caused by known lowpass filter, it is possible to recover the
291
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
image by inverse filtering or generalized inverse cascade implementation of inverse filtering and noise
filtering. However, inverse filtering is very sensitive smoothing:
to additive noise. The approach of reducing one
degradation at a time allows us to develop a =
restoration algorithm for each type of degradation and (4)
simply combine them. The Wiener filtering executes
an optimal tradeoff between inverse filtering and The disadvantage of this implementation is that when
noise smoothing. It removes the additive noise and the inverse filter is singular, we have to use the
inverts the blurring simultaneously. generalized inverse filtering. People also suggest the
The Wiener filtering is optimal in terms of the mean power spectrum of the original image can be
square error. In other words, it minimizes the overall estimated based on a model such as the model.
mean square error in the process of inverse filtering
and noise smoothing. The Wiener filtering is a linear 4. WATERSHED SEGMENTATION USING
estimation of the original image. The approach is THE DISTANCE TRANSFORM
based on a stochastic framework. The orthogonality
principle implies that the Wiener filter in Fourier
domain can be expressed as follows:
( , ) ( , )
( , )=| ( )| ( )
, , ( , )
(1)
Where ( , )+ ( , ) are respectively
power spectra of the original image and the additive
noise, and ( , ) is the blurring filter. It is easy to
see that the Wiener filter has two separate part, an
Fig. 1: Watershed segmentation-local minima of gray
inverse filtering part and a noise smoothing part. It
level yield catchment basins, local maxima define the
not only performs the deconvolution by inverse
watershed lines.
filtering (highpass filtering) but also removes the
noise with a compression operation (lowpass
The Watershed method or the watershed transform is
filtering).
an image segmentation technique based on gray-scale
mathematical morphology. It can be classified as a
To implement the Wiener filter in practice we have to
region-based segmentation approach where the main
estimate the power spectra of the original image and
idea behind this method comes from geography. If a
the additive noise. For white additive noise the power
landscape or topographic relief which is flooded by
spectrum is equal to the variance of the noise. To
water, the divide lines of the domains of attraction of
estimate the power spectrum of the original image
rain falling over the region forms watersheds.
many methods can be used. A direct estimate is the
periodogram estimate of the power spectrum
computed from the observation:
= [ ( , ) ( , ) ]
(2)
292
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
An alternative approach is to imagine the landscape In initial stage four noisy color images are chosen and
being immersed in a lake, with holes pierced in local converted into gray scale or black and white image.
minima also called catchment basins (CB), will fill up In next step the noisy gray scale images are de noised
with water starting at these local minima, and, at using Minimum Mean Square Error (Wiener)
points where water coming from different basins Filtering. In fourth step Edge map using Sobel
would meet, dams are built. When the water level has operator has been calculated. In fifth step image
reached the highest peak in the landscape, the process complement is accrued. In next step distance
is stopped. As a result, the landscape is partitioned transform is applied on the image complement and
into regions or basins separated by dams, called finally watershed algorithm is applied to get the final
watershed lines or simply watersheds For digital segmented image map.
image segmentation, the distance transform method is
Noisy Color image
commonly used in conjunction with the watershed
transform. The distance transform is the distance
from every pixel to the nearest pixel of a binary Get greyscale image
image. In distance transform method every 1-valued
pixel has a distance transform value of 0 because its Image denoising using
closest nonzero pixel is itself. In below, figure 4(a) wiener filtering
shows a binary image matrix, and in figure 4(b) Edge map using Sobel
shows the corresponding distance transform. operator
In signal processing, the Wiener filter is a filter used Apply distance transform
to produce an estimate of a desired or target random
process by linear time-invariant filtering an observed Final segmented image using
noisy process, assuming known stationary signal and watershed algorithm
noise spectra, and additive noise. The Wiener filter
minimizes the mean square error between the Figure 3: Flow diagram of the proposed approach.
estimated random process and the desired process.
The goal of the Wiener filter is to filter out noise that 6. EXPERIMENTAL RESULTS AND
has corrupted a signal. This approach often produces DISCUSSION:
better results than linear filtering. The adaptive filter
is more selective than a comparable linear filter,
preserving edges and other high-frequency parts of an
image.
293
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
images of Lena, Laure, ball and frits of 512x512, grayscale images which are shown in figure 5 and
318x554, 256 x 256 and 512x512 dimensions (Figure final segmented images which are obtained by
4) has been chosen to do the experimental work. The proposed approachapplyed on same grayscale noisy
gray scale images of originals are shown in figure 5. images and accordingly shown in figure 8. Image
It has been observed from the segmented images after entropy is a quantity which is used to describe the
using watershed algorithm directly on the noisy gray `business' of an image, i.e. the amount of information
scale images images that they are extremely over which must be coded for by a compression algorithm.
segmented (Figure 6). Image entropy is calculated with the formula:
=
(5)
294
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
295
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Communication, knowledge sharing and learning on to a 2D image plane. Let P(.) represent the camera
information in the global manner is done through projection operator so that the projected 2D position
worldwide networking. Multimedia databases and of the 3D point X is given by x=P(X). Let P-1(.)
digital library are rapidly increasing, so efficient denote the inverse projection operator so that X=P -
1
retrieval algorithms need to be developed. The (x) specifies the 3D position associated with a 2D
necessity of today's world is the retrieval of point x. Then the projected image is related to the 3D
information according to user's choice. Content image by
Based Video Retrieval System is based on retrieving ( P ( X ), t ) (X,t) or ( x, t ) (P-1(x),t)
a video sequence from large database. Textual
Annotations used in Google by user has very high
error rate. In this paper Video Retrieval is based on a The function ( x, t ) is what is known as a video
sample image. The focus of this paper is on the video signal[8].
processing aspects, in particular using texture
information of an image for video retrieval. A unique Architecture of CBVR System:
wavelet transforms based texture features for Content
Based Video Retrieval is presented.
The use of Tree Structured Wavelet is proposed for
extracting texture features such as mean and standard
deviation. The comparison of results is done using
Euclidean distance, Square chord distance and Chi-
Square and found that Squared-Chord distance is
better.
Introduction:
296
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
The main emphasis in video retrieval system is based The algorithm [1] of tree structured wavelet
on similarity function. The similarity function transform is as follows:
compares image with a video for approximate or i) Decompose the given image into four sub images
similar matching. It becomes a primary procedure in ii) The energy of each sub image is calculated
Video retrieval system.In the present system, the iii) If the energy of a sub image is significantly
feature file consists of energy parameters: mean and smaller, stop decomposing the sub image since it
standard deviation. The query feature file is used to contains less information
find the matching feature file from the database, iv) If the energy of the sub image is larger, apply the
computed by distance-measured techniques. The best above decomposition procedure.
matched videos are displayed.
297
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Distance Measures:
Fig5: Distance measures of matched video
Given two feature vectors fx and fy from the same The video with minimum distance is retrieved and
transform, where fx is a query feature vector and fy is displayed as shown in fig.6
the feature vector from database, then the Squared-
chord distance between them is given as:
D( x, y) = n
i 1
2
fx i fy i
where n is the length of feature vector .
The Euclidean distance measure used for retrieval is
given by the formula[2]
n
fxi fyi
D (x, y) = 2 2
i 1
The Chi-Square distance is given by the formula Fig 6: Video retrieval
2
D (x, y) = n fx fy 1
i i Conclusion: Video Retrieval is a major area of
i 1 f f
i f y i
research which is the present need of hour. In most of
the day-to-day applications videos need to be
Experimental Results: retrieved based on the query. Video Retrieval is
The distance measures are calculated between the major challenging area where the retrieval need to be
image query and the video frames of each video using made very fast based on the input from the user.
different distance techniques. The video with Content Based Video Retrieval is
minimum distance is retrieved by using a threshold performed in which a video is retrieved based on a
value. The performance of the three distances is query image. The tree structured wavelet transform is
compared and it was found that squared-chord used to decompose the images. Different distance
distance is the best. The fig. 4 shows the distance measuring techniques are employed for retrieval.
measures of an unmatched video using Euclidean, Further work in this area can be done based on color,
Squared-Chord and Chi-Square. shape, histogram, edges e.t.c. The major drawback of
working with video database is the large number of
video frames which makes the runtime process more.
References:
298
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
N.Anupama1,
Department of CSE,
Acharya Nagarjuna University,
Guntur, Andhra Pradesh, India.
namburianupama@gmail.com
S.Srinivas Kumar2
Department of ECE,
JNT University Kakinada, Kakinada,
samay_ssk2@yahoo.com
E.Sreenivasa Reddy3
Department of CSE,
Acharya Nagarjuna University,
Guntur, Andhra Pradesh, India
edara_67@gmail.com
AbstractMedical image segmentation plays a vital correction of functional imaging data, and computer
role in image processing due to the catering needs of integrated surgery [1]. In this paper, Medical image
the medical images in automating, delineating segmentation & rough sets are briefly discussed in
anatomical structures and diagnosis. Very often the section II. Optimized centroid algorithms for Rough
medical images contain uncertain, vague, and k-means and Rough fuzzy C-means are presented in
incomplete data definition. The concepts of lower and section III. Experimental results are presented in
upper approximations of rough sets effectively handle section IV. Concluding remarks are given in section
this data. In this paper, rough sets based clustering is V.
analyzed with respect to K-means and Fuzzy C-
Medical Image segmentation using Rough sets
means algorithms for MRI images of BrainWeb
database and the centroids are optimized using local Medical image segmentation
maxima of histogram. The performance of proposed
There exist many approaches to image segmentation
Rough K-means(RKM) and Rough Fuzzy C-means
such as histogram based methods, edge detection,
(RFCM) are evaluated and compared with classic k-
region growing, split and merge methods, PDE based
means (KM) and fuzzy C-means (FCM).
methods and clustering methods[3].However, in this
Keywordssegmentation; k-means;C-means;Rough
paper clustering techniques are considered where
k-means; Rough fuzzy C-means.
each image pixels is assigned to a cluster such that all
I. INTRODUCTION members in the same cluster are similar. In Classical
K-means clustering, each object must be assigned to
Images acquired through the techniques like
exactly one cluster. On the other hand, fuzzy C-
Magnetic Resonance Imaging (MRI), Ultra sound
means allows object to belong to a cluster to a degree
imaging, Computed Tomography (CT), Positron
of memberships [2, 5, 12].Fuzzy C-means cannot
Emission Tomography (PET), Single-Photon
recover from noise and outliers. A possibilisty
Emission Computed Tomography (SPECT), and
clustering approach by Krishnapuram and Keller [6,
functional MRI (fMRI) are studied and analyzed by
7] was proposed which uses possibilistic type of
interception and perception of the radiologists [9].
membership function to describe the degree of
There is the need for segmentation algorithms for
membership. Possibilistic C-means generates
delineating of anatomical structures and other region
coincident clusters [11]. To handle the uncertainty
of interest for assisting and automating radiologists
and vagueness in images Rough set theory, developed
tasks. There exist numerous biomedical imaging
by Z Pawlak (1981) [10] gained lot of interest.
applications based on segmentation algorithms such
Lingras and West introduced a new clustering
as the extracting tissue volumes, diagnosis,
method, called rough C-means, which describes a
localization of pathology, study of anatomical
cluster by a prototype (center) and a pair of lower and
structure, treatment planning, partial volume
upper approximations [8].
299
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
300
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
move the boundary region elements to clusters hence, Wlower is the weights for lower approximation and
reducing the no of computations. Wupper is the weights for Upper approximation.
Let dij be minimum similarity distance of an element Algorithm for Rough K-Means:
Xi to clusters, by dividing all other distances dik, if
Input: image data(X); number of cluster K, threshold,
the result is less than threshold then Xi belongs to
last_Cj(Image clusters), new_Cj(Image clusters),
A(Vj) and (Vj) and Xi cannot be a member of any
Wlower, Wupper, stop criterion (here =0.001,
lower approximation belongs to (Vj) such that dij is
threshold =2, Wlower=0.5 and Wupper=0.5).
minimum over the k clusters.
Output: new_Cj( segmented image data).
Let
Chose initial centroids as described in section III.A.
T = {j: dik/dij threshold and ij} Vj; 1 <= j<=k; for the k clusters.
(2) Calculate the similarity distance between the data
point Xi to all cluster means Vj. Update the last_Cj
Then we get: with the data points Xi which belong to same
If T then at least one other centroid is close to clusters.
the element. Let dij be the distance from Vj to Xi data point and
dik be the distance between the Xi and the Vk.
If T = then no other centroids are similarly close Determine the nearest centroid for Xi let it be dij.
to the element. Check for other centroid closer to the nearest centroid
if (T ) using equation 1:
if (T )
then Xi (Vj) and Xi (Vk),Xi T then Xi (Vj) and Xi (Vj),Xj T
else Xi (Vj) and Xi A(Vj) else Xi (Vj) and Xi A(Vj)
Compute new centroids Vj based on equation 2.
Calculate the similarity distance between the data
Hence, the centroid calculation is modified to include point Xi to all cluster means Vj. Update the new_Cj
the effects of lower as well as upper approximation. with the data points Xi which belong to same
The value of centroid is given by: clusters.
Repeat step 2 to 6 until |last_Cj-new_Cj<=stop
criteria|.
Stop.
P, if A(V j ) and A(V j ) A(V j ) ,
Rough Fuzzy C-means Rough K-means assign an
object definitely to a cluster based on the interval of
lower and upper approximation where as Rough
Q, if A(V j ) and A(V j ) A(V j ) Fuzzy C-means follows a membership function for
Vj
the degree of belonging of object to a cluster based
on the boundary of rough set. The membership value
Wlower P Wupper Q, if A(V j ) and A(V j ) A(V j ) for the objects belonging to the lower approximations
is 1 and the objects belonging to boundary region has
(3) a membership of [0,1]. Figure 3.1 presents the details.
Where Figure 3.1 shows the membership values at lower and
boundary approximation
x( A (V j ) A (V j ))
Xi
P
A(Vj ) A(V j )
(3.1)
x( A (V j ) A (V j ))
Xi
Q
A(V j ) A(V j )
(3.2) The membership of the objects present in the
boundary is defined by membership function [15].
301
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
U
c
1
xi v j
m
Vj ij
Update the last_Cj with the data points Xi which
n i 1 x j X belong to same clusters based on the Uij.
Determine the nearest centroid for Xi based on
(5) maximum Uij.
The rough fuzzy C-means partition the objects into Check for other centroid closer to the nearest centroid
clusters by minimizing the objective function. The using equation 7:
objective function is given by equation 5. if (T )
then Xi (Vj) and Xi (Vk),Xi T
else Xi (Vj) and Xi A(Vj)
P1 ,if A(Vj ) and A(Vj ) A(Vj ) , Calculate Uij using equation 4.
Compute the objective function JRF based on
J RF Q1 ,if A(Vj ) and A(Vj ) A(Vj ) equation 6.
Wlower P1 Wupper Q1 ,if A(Vj ) and A(Vj ) A(Vj ) Repeat step 3 to 8 until | last_JRF-new_JRF<= (stop
(6) criteria)|.
stop
where IV. EXPERIMENTAL RESULTS
c 2
If T then at least one other centroid is close to The evaluation of segmentation results are calculated
Xi. using the quantitative comparison with the ground
truth. A number of similarity coefficients methods [4]
If T = then no other centroids are close to the Xi.) are used to specify how well segmentation A
IF (T ) matches a referenced ground truth B.
( )
Algorithm for Rough Fuzzy C-means: Jaccard coefficient =
( )
302
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Specificity =
Precision =
Segmentation Accuracy=
303
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
V.CONCLUSION REMARKS
In this work the segmentation frame work based on
Rough sets is proposed and the results show
significant improvement in accuracy for segmenting
complex MR images of brain.
The experimentation on MR images prove
the efficiency of proposed method compared to
classic models such as k-means and Fuzzy C-means.
The performance is evaluated by visual perception
and quantitative measurements of sensitivity,
specificity, precision, accuracy and similarity.
However, the results depend on the initial centroids
for K-means and initial membership values for fuzzy
c-means. The random initialization of centroid leads
to uncertain clusters and increase computation time.
In this work the optimization of the centroids and
member ship values using histogram local maxima
based method provide accuracy and closer proximity
of centroids to ground truth.
304
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
REFERENCES
305
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
N.V.Lalitha
Department of ECE
GMR Institute of Technolog
Rajam, INDIA.
lalitha.nv@gmrit.org
306
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
The 1-D wavelet transform is given by : Step 2: Perform DWT transformation on original
audio signal. This operation produces Two sub-
bands: A, D. The D represents the Details sub-band,
and A represents the approximation sub-band.
Step3: Apply Arnold Transform and DWT to
watermark image.
Wavelet transform decomposes a signal into a set of
basis functions. These basis functions are called Step 4: Embed the transformed watermark
wavelets. Wavelets are obtained from a single coefficients into the DWT transformed original audio
prototype signal.
wavelet y(t) called mother wavelet by dilations Step 5: Apply the inverse DWT operation to obtain
and each watermarked audio frame.
shifting:
a = scaling parameter
b = shifting parameter
C. ARNOLD TRANSFORM
Arnold transformation is periodic and when it is Step 1: Sample the watermarked audio signal at a
iterated some times the original signal will be sampling rate (f ) and partition the sampled file into
obtained. frames each having certain samples.
307
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
6. Imperceptibility Test
5. Expermental Results Each listener was presented five times with the pairs
of original signal and the watermarked signal and
In the experiments a gray image of size 256x256 is asked to report whether any difference could be
used as watermark shown in Fig.3b & original audio detected between the two signals. The MOS criteria
signal in Fig.3a. are listed in Table 1. and the MOS values are
tabulated in Table2.
308
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
309
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
7. Conclusion
310
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
AbstractPropose a system which uses techniques that gradient image decomposition to separate the
provide visual quality of images and displays under image in to base layer and detail layer. After that
Low backlight conditions for saving the battery. This using median-preserving gamma correction output
method first proposes a morphological skeleton is levels of a monitor is altered. Using JND-based
a skeleton representation of a shape of an image, Boosting for detail . Enhancement of an image and
computed by means of operators. Then using the results show that this approach performs better
constrained L0 gradient image decomposition to
than the existing methods.
separate the image in to base layer and detail layer.
Using median-preserving gamma correction output II. PROPOSED SYSTEM
levels of a monitor is altered and using JND-based
Boosting for detail enhancement of an image will be
done and results show that this approach performs
better than the existing methods.
I. INTRODUCTION
( B 1 ( x , y )) ( x ) and ( B 1 ( x , y )) ( y )
Fig. 2. The figures shows : (a) Input gray scale image, (b) Binary Now the L0 gradient minimization is represented as:
V1x, y B1x, y .Cx, y
image, 2
(c) Skeleton of binary image. min x, y
B1,ax ,ay
So finally after applying the skeletonization operation
to the illumination signal of V(x,y) it will get the B x, y 2
B1x, y
2
. ax x, y ay x, y
1
V1(x,y) which is used for further process. x y
(6)
D. Constrained L0 Gradient Image Decomposition
Where C(x,y) is represented in equation (5) and is
Now V1(x,y) is the signal after skeletonization
an automatic tuning parameter to control the
operation. So first this method Decompose the
similarity between variables ax(x,y) and ay(x,y) and
illumination signal V1(x,y) in to base layer B1(x,y)
their corresponding gradients. The parameters when
and detail layer D1(x,y) using the constrained L0
doing the L0 smoothing while getting the base layer
gradient minimization. The original L0 gradient
the smoothing weight is set to 0.01, while is set to
minimization is represented as follow
0.02 and is multiplied by 2 during iterations, kappa
Parameter that controls the rate, kappa = 2 is
|V1 ( x, y ) B1 ( x, y ) | .C ( x, y )
2
min ( 4) suggested for natural images. max has the fixed value
B ( x, y ) ( x, y ) of 1e5 [2].Equation (6) is solved by alternatively
minimizing B1, ax, ay. We fix one set of variables
In L0-gradient minimization C(x,y) represents the while obtain another set of variables.
312
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
logmedianB1 x, y
Where the L0-norm of gradient can be modeled as:
(12)
logmedians.B1 x, y
C x, y Edge x, y .H ax x, y a y x, y (9)
This method adopt gamma correction and let the
median of the scaled brightness approximate to the
median of the original brightness this design forces
Where the H(|ax(x,y)|+|ay(x,y)|) is the binary function. the final scaled base layer to maintain the brightness
Returning 1 when (|ax(x,y)|+|ay(x,y)|) 0 and of the original full backlight.
returning 0 otherwise.By calculating (7) and (8) final Suppose if <1 the image is weighted toward higher
Base layer B1(x,y) is obtained from the L0-gradient (brighter) output values. If >1 the image is weighted
minimization.The illumination signal is V1(x,y) from toward lower (darker) output values. If = 1 the
the skeletonization operator. So the detail layer is transformation has no effect on the image.An
obtained by the difference between illumination underexposed photograph can be corrected using
signal V1(x,y) and the base layer B1(x,y). gamma correction with <1. An overexposed
photograph can be corrected using gamma correction
i.e. D1(x,y) = V1(x,y) B1(x,y) with the > 1.Using the equation (12) the value of
(10) =0.25 is obtained. By this if the < 1 the image is
weighted toward higher (brighter) output values. The
The detail layer which is useful for the JND for final scaled base layer will come out after applying
extracting the details and boosting the low intensity scaling of 40% backlight which is denoted as denoted
regions of the image. as Bs(x,y).
The final scaled base layer can be obtained:
E. Median-Preserving Gamma Correction
As this method compress the range of illumination to B ( x, y) J max
Bs ( x, y) s.W . 1 (13)
simulate low backlight that is applying the 40 % W
scaling, adopt a Gamma Correction - controls the
overall brightness of an image and is a setting that Where W is the white value (equal to 255 in 8-bits
determines how bright the output of the display will image or 1 in normalized images) and J max is the
be. Therefore, "gamma correction" is used to alter the maximum value of human JND. J max= + *W
output levels of a monitor. The gamma setting affects referring [2]. Where and are constants.
both the brightness and the contrast of the display. F. JND Based Detail Boosting
The reason gamma correction is used is because the
input signal, or voltage, sent to a monitor is not high Just-noticeable-difference (JND) is the smallest
enough to create a bright image. Therefore, if the stimulus for human vision to perceive the difference
313
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
JNDx, y . Bs x, y (14)
314
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
REFERENCES
1. Soo-Chang Pei, Fellow, IEEE, Chih-Tsung Shen, and Tzu-
Yen Lee Visual Enhancement Using Constrained L0
Gradient Image Decomposition for Low Backlight Displays
IEEE Signal Processing Letters, VOL. 19, NO. 12,
December 2012
2. L. Xu, C. Lu, Y. Xu, and J. Jia, Image smoothing via l0
gradient minimization,ACM Trans Graph., vol. 30, no. 6,
2011.
3. D. Choi, I. Jang, M. Kim, and N. Kim, Color image
enhancement based on single-scale retinex with a JND-based
nonlinear filter, in IEEE Symp. Circuit and Syst., 2007, pp.
39483951.
4. T.-H. Huang, C.-K. Liang, S.-L. Yeh, and H. H. Chen, JND-
based enhancement of perceptibility for dim images, in
IEEE Conf. Image Process., 2008, pp. 17521755
5. P.-S. Tsai, C.-K. Liang, T.-H. Huang, and H. H. Chen,Image
enhancement for backlight-scaled TFT-LCD displays, IEEE
Trans. Circuits Syst. Video Technol., vol. 19, no. 4, pp. 574
583, 2009
6. C. Lee, C. Lee, Y.-Y. Lee and C.-S. Kim,Power-constrained
contrast enhancement for emissive displays based on
histogram equalization,IEEE Trans. Image Process., vol. 21,
no. 1, pp. 8093, Jan. 2012.
315
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
AbstractFace recognition is one of the most intensively reduction algorithm or a robust classification
studied topics in computer vision and pattern technique may yield more accurate recognition
recognition, but few are focused on how to robustly results, they usually require multiple training images
recognize faces with expressions under the restriction of for each subject. How- ever, multiple training images
one single training sample per class. A con- strained
optical flow algorithm, which combines the advantages
per subject may not be available in practice.
of the unambiguous correspondence of feature point Some authors have proposed different approaches to
labeling and the flexible representation of optical flow deal with facial expression variations in face
computation, has been developed for face recognition. These algorithms can be roughly
recognition from expressional face images. In this divided into two main categories: morphable model
paper, we propose an integrated face recognition system based and optical flow based. The basic idea in the
that is robust against facial expressions by
combining information from the computed intraperson former category is to warp images to similar
optical flow and the synthesized face image in a global face geometries as the ones used for training.
probabilistic framework. Our experimental results show The concept of separately modeling texture and
that the proposed system improves the accuracy of geometry information has been applied in active
face recognition from expressional face images. shape model and active appearance model
Index TermsConstrained optical flow, face (ASM/AMM) [3], [4]. Face geometry is defined
recognition. via a set of feature points in ASM, while face
texture can be warped to the mean shape in AAM.
INTRODUCTION Ramachandran et al. [15] presented preprocessing
FACE recognition has been studied for several steps to convert a smiling face to a neutral face. Li
decades. Comprehensive reviews of the related works et al. [9] applied a face mask for face geometry
can be found in [14]. Even though the 2-D face normaliza- tion and further calculated the
recognition methods have been actively studied in the eigenspaces for geometry and texture separately, but
past, there are still some in- herent problems to be not all images can be well warped to a neutral
resolved for practical applications. It was shown that image because of the lack of texture in certain
the recognition rate can drop dramatically when the regions, like closed eyes. Moreover, linear warping
head pose and illumination variations are too large, was usually applied, which was not consistent to the
or when the face images involve expression nonlinear characteristics of facial expression
variations. Pose, illumination, and expression movements.
variations are three essential issues to be dealt with in The other category is to use optical flow to compute
the research of face recognition. To date, there was not the face warping transformation. Optical flow has
much research effort on overcoming the expression been used in the task of expression recognition [5].
varia- tion problem in face recognition, though a number However, it is difficult to learn the local motion in
of algorithms have been proposed to overcome the the feature space to determine the expression
pose and illumination vari- ation problems. change for each face, since different persons have
To improve the face recognition accuracy, researchers expressions in different motion styles. Martinez [12]
have applied different dimension reduction proposed a weighting method that independently
techniques, including principle component analysis weighs the local areas which are less sensitive to
(PCA) , linear discriminant analysis (LDA) [10], expressional changes. The intensity variations due to
independent component analysis (ICA) [1], expression may mislead the calculation of op- tical
discriminant common vector (DCV) [2], kernal-PCA, flow. A precise motion estimation method was
kernal-LDA, kernal-DCV [7], etc. In addition, proposed in [11], which can be further applied for
several learning techniques have been used to train expression recognition. However, the proposed
the classifiers for face recognition, such as SVM. motion estimation did not consider in- tensity
Although applying an appro- priate dimension changes due to different expressions.
316
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
In this paper, we focus on the problem of face brightness constancy constraint, however, is not valid
recognition from a single 2-D face image with facial in many circumstances. Therefore, with the
expression. Note that this paper is not about facial generalized dy- namic image model (GDIM)
expression recognition. For many practical face proposed by Negahdaripour and Yu [13], Teng et al.
recognition problem settings, like using a passport [16] generalized the optical flow constraint to
photo for face identification at custom security or
identifying a person from a photo on the ID card, it is where the subscript denotes the location, is the
infeasible to gather multiple training images for each concate- nation vector of all the flow components an
subject, especially with dif- ferent expressions. and all the where and denote the multiplier
Therefore, our goal is to solve the expressive face and offset factors of the scene brightness variation
recognition problem under the condition that the field,is the image intensity function, the subscripts ,
training database contains only neutral face images
, anddenote the spatiotemporal partial derivatives,
with one neutral face image per subject. In our
is a point in the spatiotemporal domain, and
previous work [8], we combined the advantages of
the above two approaches: the unambiguous is the motion vector at the point . They
correspondence of feature point labeling and the pro- posed to minimize the following discrete energy
flexible represen- tation of optical flow computation. function, which is defined with an adaptive
A constrained optical flow algorithm was proposed, smoothness adjustment scheme to constrain both flow
which can deal with position move- ments and components ( and) and all the bright- ness
intensity changes at the same time when handling the variation multiplier and offset factors (and)
corresponding feature points. With our proposed constrained
optical flow algorithm, we can calculate the
expressional motions from each neutral faces in the
database to the input test image, and estimate the
likelihood of such a facial expression movement.
Using the optical flow information, neutral images in
the database can be further warped to faces with the
exact expression of input image. In this paper, we
propose to exploit the two different types of
information, i.e., the computed optical flow and the
synthesized image, to improve the accuracy of face
recognition. Experimental validation on the brightness variation multiplier and offset factors,
Binghamton University 3-D Face Expression (BU- and are the parameters controlling the degree of
3DFE) Database is given to show the improved smoothness in the motion and brightness fields, is the
performance of the proposed face recog nition system. set of all the discretized locations in the image
Since we do not attempt to solve the automatic facial domain, is the weight for the data constraint, and
land mark localization problem in this paper, the , , , and are the weights for the
facial land-mark points in our experiment are labeled corresponding components of the smooth constant.
manually. In our application, we regard a neutral face image
The remainder of this paper is organized as and an expressional face image as two adjacent time
follows. We briefly review the constrained optical instance.
flow computational technique in Section II. The
proposed expression-invariant face recognition However, the motion vectors of the facial feature
system is presented in Section III. Section IV gives points, which were used only as references for
the experimental results by applying the proposed interpolating compu- tation in ICPCG, cannot be
expres- sion-invariant face recognition algorithm. guaranteed to be consistent in the final converged
Finally, we conclude the paper in the last section. optical flow. In order to guarantee the computed
optical flow to be consistent with the motion vectors
II. CONSTRAINED OPTICAL FLOW at these corresponding feature points, we modify
COMPUTATION the unconstrained optimization problem in the
original formulation of the optical flow estimation to
The computational algorithms of traditional optical a constrained optimization problem [8] given as
flow cannot guarantee that the computed optical flow follows:
corresponds to the exact pixels in different images,
since the intensity varia- tions due to expression may
mislead the computation of optical flow. The
317
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
A. Basic Concept
We treat the expression-invariant face recognition
system as a probabilistic maximum a posteriori
(MAP) classification problem. To do this, we
formulate the problem as follows:
Fig. 1. Symbolizations of
operators
318
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
319
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
tip and the other six around the mouth region. With
the labeled points, the distance between the outer
corners of both eyes is used as the reference to
normalize face images (0.5, 0.5, 0.5, and 1.5 times to
left, right, top, and bottom, respectively), which is
depicted
Fig. 7.
B. Benchmark Test
320
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
321
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
without noise; (b) normalized image using feature points with computational complexity of the proposed face
small noises; (c) size difference between normalized images with
and without noises. recognition approach.
322
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
323
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
324
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
cipher text block through a XOR operation transformed image as has been shown in [3].
before Therefore the transformed chaotic image is richer in
being encrypted with the AES. If we denote the frequency content, a desirable property in
encrypted version of a block Bi and the previous watermarking, as there are more suitable candidate
encrypted block, s thus given b coefficients for manipulation and information
embedding. Voyatzis and Pitas in [3] presented a
watermarking scheme based on a two dimensional
Where K is the encryption key. The reader may refer chaotic function, called toral auto orphism. This
fora complete description of the AES. AES cyclic chaotic function, each time applied on a square
Encryption in CBCmode, Bi, denote the plaintext image rearranges its pixels. After applied T times,
block, theencrypted block, andthe encryption key, where T is the period of the function, the pixels are
respectively. Iv is a random vectorinitialization found in their initial location. If (x, y) are the initial
coordinates of a pixel, the outcome coordinates of the
A.2.Mapping Primitives: chaotic function (x, y) are given by
325
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
The purpose of our system is to verify the reliability IV. IMPLEMENTATION OF JOINT ENCRYPTION AND
of the image within the special domain as well as the WATERMARKING METHOD
encrypted domain. The watermark is passes through
the c-map for protection of the original watermark Earlier, our implementation works with AES and
through which the encryption process is done and chaotic-maps encryption methods and the blind and
then the watermark is placed in the original aerospace non-blind watermarking methods. The methods and
image to be transmitted. This encrypted watermark their implementation are stated as follows.
image is again subjected to the encryption through Choatic map Implementation:
the AES method for the extra protection. Now this
encrypted
watermarked image is allowed for the watermarking Of all the chaotic maps we prefer the Arnold chaotic
mapping or toral auto orphism for the present
followed by the decryption which result the image
encryption. In this toral auto orphism the actual
that can be transmitted. Thus the encryption and
watermark is applied a toral transform say Arnold
watermarking can be joined together which gives
transform by using the formula
justification for the title joint encryption and
watermarking.
326
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
The watermark embedding position is obtained by In this non-blind watermarking algorithm the actual
using the quantization tables as shown below: image pixels are added by the watermark pixels by
multiplying them with a constant factor say a. The
implementation can be given as
Image Distortion:
Now this blocks are applied through the toral auto
orphism for the security (developed by scholars of
We are using Peak Signal to Noise Ratio (PSNR) in
G.Yoatis and I.Pitas) before watermarking
order to measure the distortion between the image (I)
embedding this is done using the formula..
and itswatermarked and deciphered version (I)wd
The formulae to calculate the distortion is given as
follows
327
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
REFERENCES
[1] A Joint Encryption/Watermarking System for Verifying
the
Reliability of Medical Images Dalel Bouslimi, Member, IEEE,
Gouenou Coatrieux, Member, IEEE, Michel Cozic, - IEEE
TRANSACTIONS ON INFORMATION TECHNOLOGY IN
BIOMEDICINE, VOL. 16, NO. 5, SEPTEMBER 2012
[2] Chaotic-Correlation Based Watermarking Scheme for
Still
Images E.Chrysochos V.Fotopoulus M.Xenos Pro ceedings
328
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
329
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
330
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
331
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
DM642EVM DSP. Thus the designed Simulink checked accordingly. Later the incremental building of
model facilitates to generate C code automatically the Simulink model is done. Finally the code is
corresponding to the desired problem with the generated and downloaded as a project including all
help of embedded target and Embedded Coder board specific header files and source files. This
facility provided in MATLAB/Simulink. However project in CCStudio is build and a.out file generated
the DSP development software, Code Composer is loaded in to DM642 DSP chip and the project is
Studio can accept either C or assembly code to run for results. The complete porting procedure of
generate output (.out) file, which can be load on DSP developed Simulink model on to the DM642 EVM DSP
chip. So Code composer studio version 3.3 used to processor to make is given in Figure 6.
generate .out file. This file now can be load into our
DM642EVM processor. Thus the Simulink model
developed for standalone application can be run on the
processor continuously without fail making it a real
time product. In addition this approach has an
advantage of verification of the developed model once
if the system Fails.
332
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
333
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract -Global Navigation Satellite System (GNSS) coded modulation characteristics uniquely qualify SS
is a specific term for any satellite navigation system that for navigation from the standpoints of both range
provides autonomous geo-spatial positioning with global measurements and direct fading. DS-SS is the most
coverage.Some of the GNSS are already in service, but preferred for ranging.
some of them still under development. These systems
rely on the code division multiple access (CDMA) in A. Properties of Spreading Codes: The code used for
which Pseudo noise (PN) sequences are generated and Spreading should possess some properties such as
widely used in GNSS. This paper analyses the spread 1. It must appear deterministic.
spectrum codes and verifying some interesting 2. It must appear random to a listener.
properties of PN sequence: Balance Property, Auto 3. The correlation of two different codes (Cross -
Correlation, and Cross Correlation through which the Correlation) should be small.
criteria of selecting PN Sequence out of various 4. The correlation of a code with a time-delayed
sequences is done.
version of itself (auto-correlation) should be equal
to 0 for any time delay other than zero in order to
I. INTRODUCTION
reject multi-path interference.
GNSS is used to determine the position of a 5. The code must have a long period.
receiver on land, at sea, or in space by means of
constellation of multiple artificial satellites. The past B. Choice of Code:
decade has seen a rapid development of
severalGNSSs. Some of them are already in service In BPSK transmission, bandwidth is a direct
(e.g. GPS by USA). However, some GNSSs still function of the code chip rate. Code repetition rate is
under planning or partial operational phase (e.g., simply given as
Compass by China and Galileo by European Union).
There are two types of GNSS, one is Global RRep=Clock rate(chips/sec)/Code length
Navigation and second is Regional Navigation.In this (chips)(1)
paper we overview the GNSS systems like GPS, This repetition rate determines the line spacing in the
Beidou-I (open service). The Section II talks about RF output spectrum. Hence, it is an important
the various spreading codes used for the Spread consideration in a system design. Another important
spectrum communication. Section III gives the criterion is that the period of the code must be very
knowledge of the PN sequence properties. Section IV large. Table I lists various code lengths for a 1-Mcps-
makes you understand about the code selection chip rate. Other consideration should be that the
criteria and the results we obtained. This paper ends choices of code rate and code length determine the
with Section VI which presents you our conclusion relationship of the repetition rate to the information
on this paper baseband and use of the system for ranging. The code
II. SPREADING CODES used in a direct sequence system should be adjusted
The method of Spread Spectrum communication for repetition rate and length such that noise does not
involves the spreading of the signal spectrum to make pass into the demodulator especially under jammed
its appearance noise-like. The signal will thus be conditions. A properly chosen code can not only
hidden and any intentional jammer would not be able improve resolution but also can ease range
to detect it. The method usually employed for measurement problem. if the chip rate is chosen in
spreading the signal spectrum is to multiply it with such a way that an integral number of code offset can
spreading sequences.Some ways to spread the be used to measure range. Clock rates that are
bandwidth of the signal: Frequency Hopping,Time integral multiple of 161,875Hz-produce code at rates
Hopping, Direct sequence (DS),Hybrid system.Due that are integrally related to the speed of propagation
to the LPI, AJ characteristics, the most popular SS are of RF signal.
in fields where security is an important consideration.
Also, multiple access enables many users to operate The relationship may be derived as follows:Speed of
at the same time without much interference.The light in vacuum is c = 2.997926 x 108 m/s, 1 nautical
334
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
335
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Walsh codes of the same set (matrix) is zero, Mathematically, the correlation between two
when system is perfectly synchronized. sequences x (k) and y (k) as a function of the time
Walsh codes are not maximal length or PN type delay m is expressed as
codes for spread spectrum. Although the members
R(m)xy= ( ) ( + ) (4)
of the set are orthogonal, they do not give any
spreading. The correlation equation for the digital bit sequence
Walsh code spreading can be used if all users of can thus be written as
the same channel are synchronized in time,
because the cross-correlation between different R (m) = total number of 1s / total number of bits
shifts of Walsh codes is not zero. (5)
6) Barker and Williams Sequence:An important These three properties make PN sequences efficient
sequence designed to achieve the minimum aperiodic for speechencryption. However, particularly due to
correlation values when l0 is known as barker the third property, adjacent bits correlation becomes
sequence. These sequences are used primarily for considerably less, thereby making the PN sequences
synchronization of data streams, such as frame more effective to be used in systems like CDMA.
synchronization. A barker sequence is characterised Therefore, useful PN sequences must have very good
by the fact that its aperiodic auto correlation function auto-correlation and cross-correlation properties as
satisfies wellas maintaining some randomness properties.
|Ca(l) | 1 for l 0..(2)
Welch bound: Designing codes optimized for any of
7)Bent Sequence:Bent sequences are nonlinear the potential application is practically impossible,
binary sequences that achieve Welchs bound on using code centric metric is more appropriate. This is
cross and auto correlation. The period of the sequence the reason why the Welch bound has gained
is 2m-1 with m the number of stage in the shift importance in recent years as suitable metric for
register. This value m must be multiple of 4 (n= 0 evaluating PRN codes.The Welch bound is the
mod 4).The number of sequences in the set is 2m/2 theoretical minimum of the maximum value of cross-
and the auto correlation function has 3 values and correlation that can be obtained for a given code
max=2m/2+1. length L within a set of M codes [3].
The Welch bound for a set of K sequences with
III. PROPERTIESOF PN-SEQUENCE eachsequence of length N (N K) is defined as
336
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
A: The carrier amplitude They are verified with the welch bound condition
Ci(t) : C/A code for the ith satellite given by
Di(t) : Navigational data for the ith satellite
: L1 carrier with frequency of 1575.42MHz {Cross Correlation maximum peak value} x {Code
Length} Code Length
Gold codes are constructed by taking the modulo-2
sum of two maximal length sequences of length 2n-1,
although the two sequences are required to be related
to one another in a particular way. An explanation for
the theory of Gold codes is beyond thescope of this
thesis, although an excellent summary may be found
in the Appendix of(Spilker, 1996).
In the case of the C/A codes, the pair of maximal-
length sequences are constructed using generating
polynomials G1 and G2, where G1 and G2 are given
by
G1= 1+X3+X10
G2= 1+X2+X3+X6+X8+X9+X10
With the G1 and G2 generators reset to all 1s after
generation of 1023 pseudo-randombinary chips. The
Gold code is constructed as the modulo-2 sum of the
G1 sequenceand a delayed version of the G2 Fig 3. C/A code Generator of GPS
sequence, with each satellite number allocated TABLE II
adifferent delay quantity. As there are 1025 different OBTAINED AUTO CORRELATION VALUES
sequences in the family of codes,but only 513 codes Auto Correlation Off peak value in
are balanced codes where the number of 1s and 0s Sl Satellite
Off-peak dB
differ by 1, theassignment of codes to satellites is no Number
maximum value (10*log10(value))
from the set of balanced codes.In the case of GPS,
two different taps of the 10-bit G2 register are added
1 1 78 18.920
thereby allowing a maximum of 45 different codes to
be generated.Alternative approaches that are more
widely used are to reset the G2 register to aspecified 2 2 85 19.294
initial value rather than all 1s, or to insert a
programmable delay that delaysthe reset of G2 until a 3 3 81 19.084
specified number of chips in the sequence have
elapsed. Theadvantage of these methods is that not 4 4 80 19.030
only do they permit all of the 1023 codes in thefamily
to be generated, but they also offer a simpler 5 5 72 18.573
implementation3. In particular,they also permit
generation of the codes used by the WAAS and 6 6 92 19.637
EGNOS satellite basedaugmentation systems
(SBAS),Figure 3 illustrates the implementation based 7 7 68 18.325
on LFSR
8 8 83 19.190
All the 37 Satellite vehicles have different code phase
assignments so as to generate the Pseudo random 9 9 71 18.512
noise codes. From the generated PN sequence the
various properties mentioned in Section III are
10 10 72 18.573
performed and they are verified. The balanced
property is verified for all the satellites.
11 11 83 19.190
The Auto correlation values obtained for the
generated PRN code is tabulated along with the
waveforms generated. The Cross Correlation is 12 12 83 19.190
generated and tabulated along with the waveforms.
337
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
338
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
45 23&24(Max) 0.2659 -5.75 272.01 As mentioned above from the condition {Cross
Correlation maximum peak value} x {Code Length}
46 23&27(Min) 0.2630 -5.80 269.04 Code Length (which is 34.684) is must satisfy for
all satellites Therefore Code Length =1023 chips;
47 24&26(Max) 0.2650 -5.75 272.09 Correlation Max Peak Value * Code length=
272.0157
48 24&31(Min) 0.2639 -5.78 269.96 (272.0157 45.23237) Satisfies Welch bound
339
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
G1(X)=1+X+X7+X8+X9+X10+X11
SV-15 & SV-23
G2(X)=1+X+X2+X3+X4+X5+X8+X9+X11
Fig.6:Cross-correlation Waveforms for various SVs Fig 7. C/A Code Generator for Beidou-1
340
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
and G1, a ranging code is generated for each Satellit No. Cross Correlation
satellite. e of maximum peak
(Max) x (Min) x
Combi Co value
(Code (Code
TABLE IV nations mbi
Length) Length)
nati
AUTO CORRELATION RESULTS THROUGHMAT LAB ons
Max Min
Satellite Auto 1 37 36 0.2581 0.2571 528.0726 526.0266
Off peak
number Correlation
value in dB 2 37 35 0.2581 0.2571 528.0726 526.0266
Sl No and Off-peak
(10*log10(val
Signal maximum 3 37 34 0.2581 0.2571 528.0726 526.0266
ue`))
Type value 4 37 33 0.2581 0.2576 528.0726 527.0496
1 1 114 20.56904851
5 37 32 0.2581 0.2576 528.0726 527.0496
2 2 102 20.08600172
3 3 100 20 6 37 31 0.2581 0.2571 528.0726 526.0266
4 4 99 19.95635195 7 37 30 0.2581 0.2571 528.0726 526.0266
5 5 89 19.49390007 8 37 29 0.2581 0.2571 528.0726 526.0266
6 6 104 20.17033339
9 37 28 0.2581 0.2576 528.0726 527.0496
7 7 112 20.49218023
8 8 105 20.21189299 10 37 27 0.2581 0.2576 528.0726 527.0496
9 9 95 19.77723605 11 37 26 0.2581 0.2571 528.0726 526.0266
10 10 95 19.77723605 12 37 25 0.2581 0.2576 528.0726 527.0496
11 11 96 19.82271233
13 37 24 0.2581 0.2571 528.0726 526.0266
12 12 112 20.49218023
13 13 89 19.49390007 14 37 23 0.2581 0.2566 528.0726 525.0036
14 14 93 19.68482949 15 37 22 0.2581 0.2576 528.0726 527.0496
15 15 2047 33.11117843 16 37 21 0.2581 0.2576 528.0726 527.0496
16 16 95 19.77723605
17 37 20 0.2581 0.2571 528.0726 526.0266
17 17 96 19.82271233
18 18 100 20 18 37 19 0.2581 0.2576 528.0726 527.0496
19 19 97 19.86771734 19 37 18 0.2581 0.2576 528.0726 527.0496
20 20 96 19.82271233 20 37 17 0.2581 0.2571 528.0726 526.0266
21 21 116 20.64457989 21 37 16 0.2581 0.2576 528.0726 527.0496
22 22 104 20.17033339
23 23 96 19.82271233 22 37 15 0.2581 0.2566 528.0726 525.0036
24 24 94 19.73127854 23 37 14 0.2581 0.2571 528.0726 526.0266
25 25 99 19.95635195 24 37 13 0.2581 0.2571 528.0726 526.0266
26 26 102 20.08600172 25 37 12 0.2581 0.2571 528.0726 526.0266
27 27 93 19.68482949
28 28 93 19.68482949 26 37 11 0.2581 0.2576 528.0726 527.0496
29 29 96 19.82271233 27 37 10 0.2581 0.2576 528.0726 527.0496
30 30 98 19.91226076 28 37 9 0.2581 0.2576 528.0726 527.0496
31 31 94 19.73127854 29 37 8 0.2581 0.2576 528.0726 527.0496
32 32 108 20.33423755
30 37 7 0.2581 0.2576 528.0726 527.0496
33 33 95 19.77723605
34 34 115 20.6069784 31 37 6 0.2581 0.2571 528.0726 526.0266
35 35 99 19.95635195 32 37 5 0.2581 0.2576 528.0726 527.0496
36 36 106 20.25305865 33 37 4 0.2581 0.2576 528.0726 527.0496
37 37 102 20.08600172
34 37 3 0.2581 0.2581 528.0726 528.0726
TABLE V 35 37 2 0.2581 0.2576 528.0726 527.0496
CROSS CORRELATION RESULTS THROUGHMAT 36 37 1 0.2581 0.2581 528.0726 528.0726
LAB
As mentioned above from the condition {Cross
Correlation maximum peak value} x {Code Length}
341
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
2000 2000
1500 1500
1000 1000
500 500
0 0
Fig 9. Cross Correlation Waveforms for various SVs
-500 -500
0 500 1000 1500 2000 2500 3000 3500 4000 4500 0 500 1000 1500 2000 2500 3000 3500 4000 4500
SV-23 SV-29
V. CONCLUSION
Auto-Correlation Auto-Correlation
2500 2500
2000 2000
1500 1500 Out of all the codes the PN sequence is an ideal test
1000 1000
signal for GNSS, as it simulates the random
500 500
characteristics of a digital signal and can be easily
0 0
-500 -500
generated. If the PN sequence is chosen to minimize
0 500 1000 1500 2000 2500 3000 3500 4000 4500 0 500 1000 1500 2000 2500 3000 3500 4000 4500
342
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
It is designed to provide position accuracy better than III. IRNSS SIGNAL MODULATION SCHEME
20m over India and the region extending about 1500
Kms around India. IRNSS system consists of Space Standard Positioning Service
Segment, Ground Segment & User Segment. IRNSS The Standard Positioning Service (SPS) signal is
will transmit two types of signals in L5 & S band. BPSK(1) modulated on L5 and S bands. The
navigation data at data rate of 50 sps (1/2 rate FEC
L5 band centre frequency is 1176.45 MHz and S encoded) is modulo 2 added to PRN code chipped at
band centre frequency is 2492.028 MHz. Both L5 and 1.023 Mcps identified for SPS service. The CDMA
S band consists of two downlinks. IRNSS provides modulated code, modulates the L5 and S carriers at
two basic services such as Standard Positioning 1176.45MHz and 2492.028 MHz respectively.
Service (SPS) for common users and Restricted
Service (RS) for special authorized users. The SPS Restricted Service
and RS signal are BPSK(1) and BOC(5,2) The Restricted Service (RS) is mainly for the
respectively modulated on L5 and S bands. authorized users. The RS signal is transmitted on L5
343
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
and S bands using Binary Offset Coding (BOC) IV. SIGNAL PERFORMANCE ASSESSMENT
modulation. It has two channels 1) Data channel and
2) Pilot or data less channel. In the data channel the
The performance assessment of IRNSS signals (L5
navigation data of 25 bps is modulo2 added with
& S) with modulation scheme of BPSK(1) and
designated PRN code chipped at 2.046 Mcps. The
BOC(5,2) has been carried out. The parameters
CDMA bit stream modulates the L5 and S carriers
considered for this analysis are Power Spectrum
using BOC(5,2). The pilot channel is transmitted
Density (PSD), Spectral Separation Coefficient
using Primary code and Secondary code without data
(SSC), Correlation function, Root Mean Square
modulation. The primary codes are chipped at 2.046
(RMS) bandwidth, Code Tracking jitter, Multipath
Mcps. The pilot carrier is in phase quadrature with the
performance etc. The above functionalities are
data channel.
simulated in the in-house developed MATLAB
codes.
Interplex Modulation
A. Power Spectral Density
IRNSS modulation scheme is shown in figure-1.
Each carrier is modulated by three signals namely
The Power Spectral Density (PSD) describes the
BPSK (1), data channel BOC(5,2) and Pilot channel
distribution of power within the signal with regard to
BOC(5,2). These when passed through power
frequency. The PSD for BPSK signal is given by
amplifier or TWTA operated at saturation will
produce a non constant envelope. Hence
additional fourth signal namely interplex signal is
added in order to have a constant envelope at the
output of TWTA RF spectrum.
A BOC signal may be viewed as being the product of
a
BPSK-R signal with a square wave sub-carrier, it is
represented by BOC(n,m), where n is the sub-
carrier
frequency (fs) in multiples of 1.023 MHz and m is
chip rate (fc) in multiples of 1.023 Mcps. The energy
of the signal is allocated around sub-carrier
frequency, so the maximum of the power spectrum is
shifted with respect to the centre frequency. PSD for
BOC(n,m) is given as
344
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
345
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
346
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
E. Tracking Jitter
Performance
347
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
CONCLUSIONS
ACKNOWLEDGMENT
REFERENCES
348
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
AbstractIndian Regional Navigation Satellite System (PNT) to the user with an accuracy of better than 20m
(IRNSS) is ISROs initiative to build an independent within the IRNSS service volume.
satellite navigation system based on a constellation of
GEO and GSO satellites. The IRNSS is a regional IRNSS space segment is formed by a constellation of
navigational system providing position accuracy better 7 satellites, which will broadcast ranging signals along
than 20m in and around the Indian region. IRNSS will with time. The satellites will be placed in three
have three main segments, namely the space segment, different orbital planes; 3 satellites in the Geo-
the ground segment and the user segment. The ground
stationary orbit (GEO) and 4 satellites in two Geo-
segment consists of a network of one-way ranging
stations and a network of two-way CDMA ranging
synchronous orbit (GSO) with an inclination of 29.
stations a precise timing facility along with control For enabling precise orbit determination of the IRNSS
centers for navigation and satellite control. It is also satellites, one-way CDMA ranging data from IRIMS,
planned to carry out Laser ranging of IRNSS satellites two-way CDMA ranging data from IRCDR and laser
on a campaign basis for calibration. In view of this a ranging data from ILRS is planned to be used.
suitable laser ranging payload was designed for IRNSS Satellite laser ranging technique, using a pulsed laser
satellites. This paper addresses the design and a telescope, measures the range of the retro-
considerations of retro reflectors for IRNSS, link reflectors from the ground with an achievable
margin analysis that was carried out and the strategy
accuracy of few centimeters. By fitting range
that is planned for ILRS network selection to achieve
the mission goals.
measurements from multiple laser ranging stations,
the orbit of the satellite can be determined precisely to
Nomenclature an accuracy of few meters. Laser tracking may be
used periodically to calibrate the IRNSS orbit
CDMA: Code Division Multiple Access determined by the other techniques.
ECI : Earth Centered Inertial
GEO : Geostationary Equatorial Orbit A suitable laser array was designed and fitted on the
GSO : Geosynchronous orbit earth-viewing surface of IRNSS satellites. This paper
ILRS : International Laser Ranging Service brings out the analysis that was carried out for sizing
IRNSS : Indian Regional Navigation Satellite System the corner cube and the laser array respectively and
LARA : LASER Retro reflector Array the link margin study with respect to ILRS stations.
NPE : Number of Photo Electrons This work is supported by mathematical modeling and
NNE : Number of Noise Electrons analysis.
SLR : Satellite Laser Ranging
SNR : Signal to Noise Ratio II.OBJECTIVES
The objective of the mathematical model is to study
the following
I. INTRODUCTION
1. Sizing of the individual retro-reflector cube
Indian Regional Navigation Satellite System (IRNSS)
for IRNSS
is ISROs initiative to build an independent satellite
2. Sizing of the Laser Retro Reflector Array
navigation system based on a constellation of 3 GEO
(LARA) for IRNSS
and 4 GSO satellites. The objectives of IRNSS are to
3. Computation of received power at SLR
provide accurate Position, Navigation and Time
station in terms of NPE, NNE and SNR
349
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
4.Sensitivity Analysis of various parameters This vector is also the direction of the laser beam
like vector in the inertial frame. Orbit Reference Frame is
Station-Satellite Geometry a coordinate system represented about the
Cube/Array Dimensions
SLR Uplink power
instantaneous position of the spacecraft with XO
Transmitter Efficiency etc towards the earth center and Z 0 towards the
A model for the above needs was developed in
MATLAB and it includes the following negative orbit normal while YO completes the right-
1. Satellite Orbit Propagator handed triad. They are computed as given below.
2. Satellite Attitude Profile Model
3. Link Margin Calculator X O = unit vector along the -ve spacecraft position
vector = - SCPOSVEC _ UNIT
III.MATHEMATICAL FORMULATION
Z0 = - SCPOSVEC_UNIT x
The aim of the modeling is to develop the required SCVELVEC_ UNIT
mathematical model to compute
The angle of incidence of the laser beam and YO = Z 0 x X O
X Y Z is
from a given station on the laser array
mounted on the IRNSS satellite The spacecraft body reference frame b b b
The number of photoelectrons that will defined about the spacecraft center of mass such that
arrive at the SLR for a transmitted pulse this frame is coincident with yaw, roll and pitch axes
of the body. When attitude bias is not introduced,
Modeling and Analysis ensures a flawless approach spacecraft body frame coincides with orbit reference
in operations and also cautions about possible frame. In the presence of attitude biases like the yaw
steering bias , the body frame is computed from the
X Y Z as
deviations in the future. The development of
mathematical model requires spacecraft position and knowledge of the orbit reference frame o o o
velocity vectors for every instant T.
given below [3]
STN 2SCVEC _ Unit = unit ( STN 2SCVEC ) Where the transformation matrix A is given by
350
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
351
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
nnoise t gate
e
D
h
d
4
352
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
REFERENCES
CONCLUSION
A model has been developed for computing the laser
link margins for IRNSS satellites and the sizing of
the individual cube and the array have been arrived
at. Sensitivity analysis with respect to the cube
dimensions, array size and station characteristics have
also been performed. The following are the
conclusions.
1. A LARA of 40 cubes with each cube having
an approximate diameter of 38 mm was sized
for providing an adequate link margin for
IRNSS Laser ranging.
2. Based on the link margin analysis, a network
of ILRS stations was also proposed for
IRNSS.
353
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
AbstractPulse compression techniques are widely used coefficients vary from code to code, and change as a
and active research topic in radar systems. There were function of code and filter length. There are several
few demerits like masking with the existing methods. techniques used to generate a mismatched filter to
In order to overcome these limitations , in this project minimize side lobes.
a R-G filter is used to suppress the sidelobes of the
radar coming out from the matched filter and BLMS
algorithm is used to calculate the filter coefficients. There are two methods which have been usually used
A weighting function is utilized to shape the sidelobe for side lobe reduction. Designing a mismatched
energy in an iterative manner that will yield more filter directly from codes. Employing an additional
sidelobe reduction. Comparison of the (R-G) filter using weighting network after the matched filter.
BLMS algorithm and the matched filter shows that at
the expense of insignificant loss in signal-to-noise ratio One possibility is conventionally called the (R-
(LSNR), adequate mainlobe-to-peak-sidelobe ratio G) filters which were introduced by Rihaczek and
(MSR) can be achieved. Golden [3]. The (R-G) filters have advantages of
simple filter structure and relatively suitable
I. INTRODUCTION performance.
In this work an adaptive (R-G) filter is used and
High range-resolution is important for many radar the Block LMS algorithm is used to calculate the
applications. Pulse compression technique is filter unknown coefficients. The filter yields a
employed to achieve high range resolution with satisfactory mean square sidelobe level (MSSL) and
enough low power[1]. Radar range resolution an adequate MSR.
depends on the Bandwidth of the received signal.
Bandwidth of a pulse is inversely proportional to the II. DESIGN ISSUES
pulse duration. So, short pulses are better for
range resolution. But
354
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
The continuous part i.e., u(a, b) essentially Hence, the transfer function of sidelobe reduction
determines the output pulse shape, whereas the filter is suggested as the following parameterized
discrete part i.e., op reveals the relative output form
amplitude. Therefore, they can be evaluated
separately [5] and the designing can be performed in
discrete domain. The true output of matched filter can
be also be written as
355
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
IV. PERFORMANCE
EVALUATION
To demonstrate the acceptability of the
filters performance, the (R-G) filter is compared Fig. 5. MSR results of matched filter and one, two (R-G)WF
with the matched filter for three cases. In the first filters
356
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
CONCLUSION
References
357
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract In this paper a novel method for solving the optimization of each objective. Many, or even most, real
multi- function optimization problems proposes. Radar engineering problems actually do have multiple-
is mainly used to detect the targets in different objectives, i.e., minimize cost, maximize performance,
environments by analyzing the echo signal from the maximize reliability, etc. These are difcult but realistic
target. One method of distinguishing multiple-time- problems. GA are a popular meta-heuristic that is
around echoes from unambiguous echoes is to operate particularly well-suited for this class of problems.
with a varying pulse repetition frequency. The different
pulse repetition frequencies also allow eliminating
Traditional GA are customized to accommodate multi-
the blind speeds in finding the moving targets objective problems by using specialized tness functions
information. The use of more than one PRF offers and introducing methods to promote solution diversity.
additional flexibility in the
design of MTI radars. The detection level of the target The general approach to multiple-objective optimization
can be decided by the performance factors called merit is to determine an entire Pareto optimal solution set or a
factor and discrimination factors. The performance representative subset. A Pareto optimal set is a set of
factors are varying with the variation of PRI sequence. solutions that are non-dominated with respect to each
Hence we are generating the non-uniform PRI sequence other. While moving from one Pareto solution to
to maximize the both performance factors using multi another, there is always a certain amount of sacrice in
objective genetic algorithm optimization technique.
one objective(s) to achieve a certain amount of gain in
the other(s).
Index Terms Merit Factor, Discrimination Factor,
MOGA.
II. MULTI-OBJECTIVE OPTIMIZATION
I. INTRODUCTION FORMULATION
Consider a decision-maker who wishes to
The objective of this paper is to an optimize non- optimize K objectives such that the objectives are
uniform PRI parameters of Radar signals by multiple- non-commensurable and the decision-maker has no
objective optimization methods using genetic algorithms clear preference of the objectives relative to each
(GA). For multiple-objective problems, the objectives other. Without loss of generality, all objectives are
are generally conicting, preventing simultaneous of the minimization type--a minimization type
optimization of each objective. Many, or even most, real objective can be converted to a maximization type
engineering problems actually do have multiple- by multiplying negative one. A minimization multi-
objectives, i.e., minimize cost, maximize performance, objective decision problem with K objectives is
maximize reliability, etc. These are difcult but realistic dened as follows:
problems. GA are a popular meta-heuristic that is
particularly well-suited for this class of problems. Given an n-dimensional decision
Traditional GA is customized to accommodate multi-
variable vector x {x1 ,...., xn } in the solution
objective problems by using specialized tness functions
and introducing methods to promote solution diversity. space x, find avector x* that minimizes a given
The general approach to multiple-objective optimization set of K
is to determine an entire Pareto optimal solution set or a
representative subset. A Pareto optimal set is a set of
solutions that are non-dominated with respect to each
other. While moving from one Pareto solution to
another, there is always a certain amount of sacrice in
one objective(s) to achieve a certain amount of gain in
the other(s). The objective of this paper is to an In many real-life problems, objectives under
optimize the parameters of Radar signals by multiple- consideration conict with each other. Hence,
objective optimization methods using genetic algorithms optimizing x with respect to a single objective often
(GA). For multiple-objective problems, the objectives results in unacceptable results with respect to the
are generally conicting, preventing simultaneous other objectives. Therefore, perfect multi- objective
358
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
solution that simultaneously optimizes each of GA may exploit structures of good solutions with
objective function is almost impossible. A respect to different objectives to create new non-
reasonable solution to a multi- objective problem is dominated solutions in unexplored parts of the Pareto
to investigate a set of solutions, each of which front. In addition, most multi-objective GA do not
satises the objectives at an acceptable level require the user to prioritize, scale, or weigh
without being dominated by any other solution. If objectives.
all objective functions are for minimization, a
feasible Therefore, GA have been the most popular heuristic
approach to multi-objective design and optimization
problems. Jones et all reported that 90% of the
approaches to multi objective optimization aimed to
approximate the true Pareto front for the underlying
is said to be a pareto optimal if it is not problem. A majority of these used a meta-heuristic
dominated by any other solution in the solution technique, and 70% of all met heuristics
space. A Pareto optimal solution cannot be approaches were based on evolutionary approaches.
improved with respect to any objective without
worsening at least one other objective. The set of all
IV. DESIGN ISSUES AND COMPONENTS OF MULTI-
feasible non-dominated solutions in X is referred OBJECTIVE
to as the Pareto optimal set, and for a given Pareto GA:
optimal set; the corresponding objective function
values in the objective space are called the Pareto Assign no. of chromosomes(n),
front. For many problems, the number of Pareto
Assign no. of bits(b).
optimal solutions is enormous (perhaps innite).
Assign no. of variables to optimize(m).
Initialization of population:
The ultimate goal of a multi-objective optimization
Generate population by using a random function
algorithm is to identify solutions in the Pareto
and this random function values multiply with the
optimal set. However, identifying the entire Pareto
R max and Rmin values such that the generated
optimal set, for many multi-objective problems, is
random values become with in specified range.
practically impossible due to its size. In addition, for
r= rand(n,m*b)-0.5;
many problems, especially for combinatorial
P=(Rmin)+((Rmax-
optimization problems, proof solution optimality is
Rmin)/1023)*r;
computationally infeasible. Therefore, a practical
The generated population acts as
approach to multi-objective optimization to
chromosomes.
investigate a set ofsolutions (the best-known Pareto
Calculate the fitness values i.e., merit factor
set) that represent the Paretooptimal set as well as
and discrimination factor,
possible. With these concerns in mind, a
Assign the rank based on fitness values.
multi-objective optimization approach should achieve
Calculate crowding distance
the following three conflicting goals.
III. MULTI-OBJECTIVE GA
359
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
V. RESULTS:
360
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
VI. CONCLUSION
This paper consists non uniform PRI pcw radar signal .as
the non-uniform PRI avoids the blind speeds. Using multii
objective genetic algorithm merit factor and discrimination
factors have been set as two object functions using MOGA
approach better results have been obtained for PSLR,
ISLR, with compared to conventional uniform and
nonuniform(ascending, descending, and random) PRI
strategies
REFERENCES
361
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
AbstractIn recent years interest on vision based calculate the position of interested target through
tracking systems for applications like navigation, sensory information obtained through a variety of
surveillance, robotics, augmented reality, automotive, sensors such as position, velocity, and vision sensor.
etc., has grown. Vision based tracking techniques face The information obtained is processed through on
the challenge of detecting and tracking the moving board system or in ground station to give the target
object for extracting accurate target position
location, position co-ordinates, etc. Earlier seperate
measurements under dark conditions. The problem
becomes more complex to track a moving object using a
ground computer was used for processing the
camera while keeping the object within the image. This information obtained through sensors. This system
work presents the development of a vision based real depends heavily on the communication between the
time tracking system with the use of a laser point platform and the ground computer and a
and gimbal mounted camera. A better measure of a communication failure will result in tracking system
targets position is obtained through laser because it is failure. The solution for this type of tracking system
capable of determining the object at long range and is on-board visual tracking for platform (gimbal)
helps the vision based tracking system to track the stabilization[1].This onboard tracking system
objects under dark conditions. Stabilization unit of the
eliminates the dependency on the communication with
gimbal mounted camera is used to keep the object
within the image by tracking the laser point and
the ground station and makes the system less prone to
controlling the gimbal position. The algorithm for failure. Tracking and targeting a target with visual
tracking the laser needs detection of laser point and servo control simply requires recognizing the target
calculation of X and Y co-ordinates of the laser point. within the field of view, capture this target,
Segmenetation and intensity based algorithm is used for following it, and finally giving the order to fire on
detection and centroid tracking algorithm is used to the target. This actually requires an optical sensor to
calculate the co-ordinates of laser point. The video based obtain the field of view, a signal processing to
detection and tracking algorithm was prototyped and recognize the target, a control system for the target
validated using MATLAB with a USB connected camera
tracking loop[3]. Compared to the traditional vision-
mounted on the gimbal system interfaced to the PC
using RS232. The Atmega128 micro- controller on
based tracking system, as a new kind of
board the gimbal system is programmed for receiving measurement instrument, the laser range scanner has
the co-ordinate data and controlling the motors. The received the increasing attention for solving tracking
validated algorithm is ported to TMS320DM642 DSP problems in recent years[4]. To keep the target within
and evaluated for its real-time performance. The the field of view, stabilized platform approach is
embedded DSP based system was able to track a laser employed. The stabilized platform is a two axis pan-
point for camera stabilisation in real time at 30 fps and tilt platform which recieves sensor error signals for
720x480 resolution. stabilization.
1 INTRODUCTION Specific objects can be recognized relatively easily
using vision-based techniques, but these techniques
experience difficulties in extracting accurate target
Object tracking is an important field that as attracted position measurements. A better measure of a targets
lot of interest in computer vision. Tracking has vast position is obtained through laser. Laser is capable of
range of applications in different areas such as determining distances to objects at long range,
surveillance, traffic monitoring, gaming, gesture with high accuracy.The paper aims at developing a
recognition, augmented reality, defence, human laser assisted vision based tracking system with
activity, robotics, industrial automation, etc. Object stabilization unit to keep the target within the field
tracking can be defined as the process of of view.
segmenting an object of interest from a video scene
and track its motion and orientation in order to
extract useful information. The information In this paper, video based detection and tracking
obtained can be used for activity control. Visual algorithm was developed. For detection of laser point,
target tracking is to identify the location and to segmentation and intensity algorithm was used
362
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
and for tracking centroid algorithm was employed. Image acquisition setup: It consists of webcam, video
The developed algorithm was prototyped and camera with suitable interface connected to processor
validated using MATLAB with USB connected or personal computer.
camera mounted on the gimbal system interfaced to Image analysis: The tools which are necessary for
the PC using RS232. The co-ordinates calculated analyzing the content in the video or image captured
are transmitted to the gimbal system for controlling for detecting and tracking the laser point.Processor:
its azimuth and elevation motors. The Atmega128 The Digital signal processor(TMS320DM642) with
micro-controller on board the gimbal system is a dedicated video processing unit is used for
programmed for receiving the co- ordinate data and processing the video for detecting and tracking the
controlling the motors. The validated algorithm is laser point.
ported to TMS320DM642 DSP and evaluated for its
real-time performance. Gimbal Control: After obtaining the co-ordinate
pixel values, the mechanical action is taken for
2 DESIGN OF LASER ASSISTED VISUAL controlling the motors of a gimbal using serial port
TRACKING SYSTEM of a PC.
A detailed description of each module and the 2.3 Detection and Tracking Algorithm
algorithm for the real-time visual tracking system will
be given followed by the experimental results and Processing algorithm is used for the extracted frame
discussion on the performance of the system to detect and track the laser point. The Flowchart for
algorithm is shown in Figure 2.
2.1 Problem Statement
363
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Laser X = leftmost+Rightmost/2
Laser Y = Topmost+Bottomost/2
364
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
connected to video in port of the board captures live The serial communication is used to transmit X and
video where the laser point. The video is processed to Y co- ordinate values from the PC to the micro-
give necessary co-ordinates for tracking the laser controller in the Gimbal system for controlling the
point. The co-ordinates from the processor is azimuth and elevation motor. The PC with Matlab is
transmitted serially to micro-controller in the Gimbal interfaced to Gimbal system through RS232 cable. In
to keep the target within the field of view . the PC, serial com port 3 is intialized with the baudrate
of 9600 and databits of 8. After intialization, the serial
2.5 Gimbal Control and Processing Unit com port is opened. Once the com port is opened, the
X and Y co-ordinate values are sent to controller in
The camera is focused on background where the form of packet as show in Figure 5.
laser is illuminated. The image with laser point is
captured and processed to give the X and Y co-
ordinate values. The X and Y co-ordinate values from
PC or DSP is sent to controller in the Gimbal system
through serial port i.e. RS232. The gimbal was
controlled from two ways,
1. The processing results from the algorithm 2.7 Hardware Implementation
validated in MATLAB
The software design implemented in PC and processor
2. The processing results from the DSP is interfaced to gimbal system. The obtained X and Y
co- ordinates of the laser point are fed to controller via
RS232. Based on these values, the motors of the
gimbal system take necessary action to keep the
camera focussed to the laser point.
365
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
The implementation and analysis of developed The TMS320DM642 DSP is chosen for
algorithm has been compared for its computation time implementation. For this, a simulink model is
on various platforms like MATLAB, TMS320DM642 developed using specific DSP blockset. The captured
DSP. Initially the algorithm was tested with dark video is processed in TMS320DM642 DSP and the
background with red laser source. The algorithhm was tracking result is displayed on TFT LCD monitor. The
able to detect and track the laser spot. The tracking DSP setup and results are shown in Figure 13.
result is represented with the intersection of two red
lines as shown in Figure10.
4.CONCLUSIONS
Figure 11: Track result with Lighting
Conditions
Vision based tracking system for tracking the laser
point has been implemented for gimbal control to
The tracking algorithm was modelled in simulink and
keep the target within the field of view. The
was tested with different backgrounds. The algorithm
developed algorithm was tested on MATLAB and
was able to track the laser spot and was giving exact
DSP for its functionality. The co-ordinates calculated
X and Y co-ordinates of laser spot to the gimbal for
from the tracking algorithm was serially transmitted
control. This is shown in Figure12.
to Gimbal mounted camera for controlling and
keeping the object within the field of view. Laser
pointing mechanism can be employed to enhance the
detection and tracking of moving objects under low
light conditions and with low resolution cameras.
The tracking algorithm has been developed for
static camera movements. In future, the work can
be extended to track moving object with dynamic
camera movements . The algorithm will be improved
Figure 12: Simulink to adapt with different surrounding conditions.
366
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
ACKNOWLEDGMENTS
This work was supported by System Controls
Technologies Pvt Limited. I wish to thank our
guidesMr. K. Kanakaraju, Technical Director,
System Controls Technology Solutions Pvt and Mr.
Naveen K. S, Senior Design Engineer at System
Controls Technology Solutions Pvt. Ltd. for their
support and co-operation during the project.
REFERENCES
367
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract- In the last few years the work towards the receiver in a single package [5,8,9]. The basic idea
integration of the Inertial Navigation System (INS) with of this system is to assure an optimal usage of the
Global Navigation Satellite System (GNSS) is being GPS+GAGAN and GLONASS measurements in
actively carried out by multiple R&D groups across the
two operating modes. A nineteen-state Extended
globe. Presently the investigations are going on to
evaluate new integration schemes using MEMS inertial Kalman Filter implemented and tested under Ultra
sensors with GNSS sensors towards the possible tightly coupled scheme to improve both INS and
development of new generation low cost, small size and GNSS performance on many fronts as discussed in
low weight system by targeting multiple navigation the end of this paper [2,4,5]. Further analysis on this
applications in the industry. To meet the prime scheme is in progress.
requirement of this class of work in various defence
navigation applications, a novel architecture for Ultra
Tightly integrated navigation scheme for MEMS INS II. MEMS INERTIAL SENSORS
with miniaturized GPS-GLONASS-GAGAN receiver TECHNOLOGY
has been realized. The partial work in this paper is
focused on the design and development of PLL Doppler The micro electromechanical system based on the
aid to Scalar based Receiver tracking loop for achieving
micro electronics technology is an integration of
more accurate navigation solution, improved tracking
ability in dynamics and enhanced anti-jamming
silicon micro technology, LIGA technology and
performance of the combined system. precision mechanical technology. It is also an
intelligent combination system of micro sensors,
Keywords: MEMS, GoM, INS, ECEF, UTC, PLL etc micro actuators, signal conditioning and control
circuits, interfaces, communication and power
supply with the modern information technology fig
I. INTRODUCTION
1[5]. Micro inertial devices, consisting of micro
accelerometers and micro gyroscopes, have the
Combined MEMS and GNSS/INS optimum character of smaller dimensions, lower costs and
solutions offers low power, low cost and light available integration on one chip over the traditional
weight navigation sensors for autonomous vehicle devices.
navigation. INS in itself is a self-contained device
which operates independently of any external signals
or inputs, providing a complete set of navigation
parameters, including position velocity and attitude,
with a high data rate [1]. However, one of the main
drawbacks of INS especially with MEMS class
sensors when operated in standalone mode leads
rapid growth of systematic errors with time [5]. In
contrast to INS short-term positioning accuracy,
satellite based GNSS navigation techniques can
offer relatively consistent accuracy if sufficient
GNSS signal can be tracked during the entire course
of navigation [8].
This paper describes the design of a integrated
navigation systems for short and medium long range
navigation applications, which comprises of MEMS
Fig.1Basic MEMS Structure
IMU and a miniaturized multi ASICs based G3oM
368
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
369
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
370
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
GPS+GLONASS+GAGAN and their respective raw user; the other from the receiver clock drift. The
measurements to integration with INS[1]. Doppler aided PLL can be modeled as a control
system, as shown in fig.6. In order to facilitate the
IV. ARCHITECTURE OF ULTRA TIGHTLY analysis of the phase errors induced by Doppler
GNSS-INS WITH AN AID TO RECEIVERS aiding error, the true Doppler, Fd(s), is removed from
SCALAR BASED TRACKING LOOP the block diagram of Doppler-sided PLL, resulting in
In this architecture the PLL of the GNSS receiver is figure NN, the block diagram of Doppler aiding error
aided with the Doppler shift calculated from the propagation. In this figure, phase error
navigation solution of the GNSS-INS system ( and r(s)= r(s)- Fd(s)/s 3.2
satellite information) as indicated in fig.6, the noise
bandwidth of the loop filter of the PLLs can be and Doppler aiding error
narrowed more than in a GNSS-only case. Since the Fd(s) = Fdc(s) - Fd(s) 3.3
loop noise bandwidth cannot be made arbitrarily
small and is limited by the navigation solution error are inputs and
of the GNSS-INS and the receivers oscillator errors,
(s)=(s)- Fd(s)/s 3.4
methods of choosing the noise bandwidth for the PLL
loop filter are investigated in order to improve signal is output, where r(s) is PLL input phase, (s) is
tracking, signal reacquisition and overall navigation output phase, Fdc(s) is the calculated Doppler aiding.
solution performance. Specifically, the bandwidth can The phase error induced by Doppler aiding error is
be selected according to the performance of the obtained from fig .7 as follow
integration system and the quality of the receivers
Total Phase Error in Doppler aided PLL
oscillator. By selecting the adaptive loop filter(ALF)
for Doppler aided PLL loop the ALF noise The total phase error (1-sigma) of the Doppler-aided
bandwidth can be adjusted according to the PLL can be expressed as
performance of the integration system, the quality of
PLL =( 2tPLL+ 2v+ Arx+ Asx+ GM )+ RR / 3 3.6
the receivers oscillator, and the satellites information.
With the ALF, both the navigation performance and Where tPLL , v, GM, and RR are the thermal
the PLL tracking ability of the UT GNSS-INS can be noise, oscillator vibration phase jitter, Doppler aiding
improved [4]. GM phase jitter, and Doppler aiding RR phase error
respectively and Arx and Asx are respectively the
receiver and satellite Allan phase jitters. The RR
phase error is considered as a dynamic stress error
(much like the dynamic stress error in a standard
tracking loop), so it appears outside the square root in
equation 3.6 [4,5]. In equation 3.6 the thermal noise,
oscillator Allan phase jitter oscillator vibration phase
jitter, and Doppler siding error-induced phase error
are considered to be independent. Furthermore, the
Doppler aiding GM phase jitter and Doppler aiding
It is noted in the fig. 6 that the tight integration box phase error are assumed to be independent to simplify
contains an extended Kalman Filter to integrate both the analysis of Doppler aiding error- induced phase
the GNSS error. It is noted that equation 3.6 is applicable to any
Doppler aided PLL. Furthermore, if the parameters
and INS data. The Doppler aiding rate for the for calculating the total phase error with equation 3.6
tracking loop is equal to the INS measurement rate are known, according to the PLL tracking threshold
i.e. the output rate of the Klamn filter for Doppler PLL 15 (degree) (Ward et al 2006), the PLL is
aiding calculation. The Doppler aiding value is considered to be locked.
calculated from as:
The minimum noise bandwidth of Doppler-aided
fd=-fT ( (V-vu )./c) - fT tu +d 3.1 PLL can be obtained with equation 3.6. Since PLL is
Where fd is in Hz, V is the satellite velocity vector, vu a nonlinear function of the noise bandwidth, as shown
is the user velocity vector, is the user-to-satellite in equation 3.6 there are no analytical solution for the
line of sight unit vector, fT is the transmitted signal noise bandwidth. Therefore, a search strategy is
frequency, tu is receiver clock drift rate and d is needed to obtain the minimum noise bandwidth
noise.It consists of two parts: one comes from the solution. If the noise bandwidth solution is obtained
relative motion of the satellite with respect to the in real time and applied to the PLL loop filter
371
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
372
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
[4] Debo Sun, Ultra Tight GPS/Reduced IMU for land Navigation
UCGE Reports Number 20305, March 2010 ,University of
Calagry, Alberta,
[5] Kedog Wang,Young Li, Chirs Rizos, The Feasibility of MEMS
inertial sensors for deep integration of GPS and INS Natinal
Technical Meet U.S ION, Fort Worth, Texas,12-14 June 2011
,2889-2895
[6] Brown, Robert Grover, Hwang,Patrick Y.C, Introduction to
Random Signals and Applied Kalman Filtering.
[7] Kaplan,ElliottD,Ed., Understanding GPS Principles and
Applications,Artech House, Norwood,MA,1996.
[8] M.S Grewal,Lawrence R. Weill, Angus P. Andrews, Global
Positioning Systems, Inertial Navigation, and Integration, john
Wiely& Sons,ISBN 0-471-20071-9
[9] G.Sateesh Reddy, Manjit Kumar INS- GPS-GLONASS Navigation
fusion scheme for high dynamics guided projectilesCo-
ordinates,pg . 35-42 August, 2012.
[10] A. Soloview, S Gunawardena, F Graas, Deeply integrated
GPS/Low-cost IMU for low CNR signal processing. Concept
description and in-flight demonstration Navigation: Journal of
the Institute of Navigation, 2008, 55(1): 1-13.
373
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
I. INTRODUCTION
374
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
inFigure 2 One additional design consideration is the Ionosphere terms are collected in 18 s. In addition, the
change in the subframe slotting following UTC and PDOP issue highlighted above with 7 satellites
Ionosphere transmission. For example, for SV1 after disappears in a global scenario with an assumption of
T0 + 18 s, instead of transmitting subframe 2 for SV1 a minimum of 8 or more satellite available at any
commensurate with the other satellites, subframe 1 is point on the globe.Finally, the UTC and Ionosphere
transmitted. This ensures that the maximum time terms shall be straddled in multiple satellites to ensure
required to collect ephemeris (from the satellite) is users across the globe obtain it from at least one
restricted to 18 s. The merit of this scheme is satellite.
Tephreduces to 12 s with Ionosphere and UTC
parameters transmitted periodically every 18 s. The
proposed scheme of Three Subframe Fixed (TSF) is as
shown inFigure 2.
One limitation with this approach is for 6 s (for
example, T0 + 12 to T0 + 18,) ephemeris from six
satellites, as opposed to all seven, is available (Figure
2).For 6 s (till T0 + 30 s), the Position Dilution of
Precision (PDOP measure of satellite geometry [3]
will be relatively high as shown inFigure 3.
375
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
376
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
simulator and receiver hardware used in the [2]with and the ionosphere is dropped. This is under the
the modified versions of the software developed are assumption that the receivers are operating in dual
used. The software change in the simulator is frequency mode and thus, with measurements, will
according to the flowchart specified in Figure 6. be able to estimate directly the ionosphere effect.
With the simulator and the receiver software in
OCM configurations, the receiver data was logged in
GGVISION (a graphical tool) to profile the T eph
component. Three runs were conducted to establish
the results in different configurations as follows:
1. First, to profile SPS only mode of operation
2. Second, restricted direct (hybrid, only in
OCM)
The results of the proposed scheme are as shown in
Figure 8.
ACKNOWLEDGEMENTS
377
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
378
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
AbstractThe implementation of OFDM system In an OFDM system the input bit stream is
suffers from the effect of phase noise generated by multiplexed into N symbol streams each with symbol
the local oscillator which disturbs the period T and each symbol is modulate with sub-
orthogonality among sub carriers and causes inter carriers.A serial to parallel converter groups the
carrier interference (ICI).The ICI may also suffers stream of input bits from the source encoder into
due to insufficient cyclic prefix at the transmitter groups of bits. Where M is the alphabet size
Self cancellation is the method is suggested for of digital modulation scheme
reducing the ICI. The frequency offsets between
the transmitter and receiver is measured in terms A typical discrete-time baseband OFDM transceiver
of carrier to interference ratio (CIR) and bit error system. A serial to parallel converter groups the
rate (BER). stream of input bits from the source encoder into
Index Terms Orthogonality, self cancellation, Inter groups of log2M bits. Where M is the alphabet size
Carrier Interference (ICI), Carrier to Interference Ratio of the digital modulation scheme employed on each
(CIR). sub-carrier a total of N such symbols Xm are
created. The N symbols are mapped to bins of an
1. INTRODUCTION IFFT. The IFFT bins corresponding to the
orthogonal sub-carriers in the OFDM system.
OFDM is emerging as the preferred modulation
scheme in the modern high data rates wireless Therefore OFDM symbol can be expressed as
communication systems [1, 2.]. OFDM is a special
case of MC (Multicarrier modulation).MC is the
concept of splitting a signal into a no. of signals,
modulating each new signal to several frequency
channels and combining the data received on the
multiple channels at the receiver. OFDM has been
used in many communication systems such as The Digital to Analog converter creates an analog
WLAN (wireless LAN), DVB (Digital video time- Domain signal which is transmitted through the
broadcasting), etc. However one of the major channel. At the receiver, the signal is converted back
problems in OFDM is its vulnerability to to a discrete N point sequence y(n), corresponding to
frequency offset which leads to loss of orhogonality each sub-carrier. The discrete signal is demodulated
resulting into ICI [3, 4]. For suppressing ICI there is using an N-point FFT operation at the receiver
self cancellation, (SC) [5, 6, and 7] techniques in
OFDM. Self- cancellation is a two stage technique The demodulated symbol stream is given by
that uses predefined weighting coefficients to reduce
ICI for OFDM systems [8,9, 10, and 11].
OFDM by the insertion of a cyclic prefix between all sub- carrier indices
successive OFDM symbols. This cyclic prefix is
discarded at the receiver to cancel out ISI. The complex ICI Coefficient S (l-k) are plotted for
all sub- carrier indices
3. ANALYSIS OF INTER-CARRIER
INTERFERENCE
380
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
381
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
REFERENCES
382
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract When the data is transmitted at high bit cells that share the same frequency channel, leading
rates, over mobile radio channels, the channel impulse to distortion of the desired signal and also low system
response can extend over many symbol periods, which performance. Therefore, wireless system must be
lead to Inter-Symbol Interference (ISI). Orthogonal designed to mitigate fading and interference, to
Frequency Division Multiplexing (OFDM) is one of the
guarantee a reliable communication. A successful
promising technologies to mitigate the ISI. In an OFDM
signal the bandwidth is divided into many narrow sub- method to improve reliable communication over a
channels which are transmitted in parallel. Each sub- wireless link is to use multiple antennas and another
channel is typically chosen narrow enough to eliminate promising scheme is using orthogonal frequency
the effect of delay spread. Analysis of the performance division multiplexing (OFDM), which can effectively
of 2x2 MIMO-OFDM system with Successive mitigate inter symbol interference (ISI) induced by
Interference Cancellation (SIC) and 64-QAM wireless multipath fading channels..
modulation for higher data rate has been carried out In wireless communications, it is well known that the
assuming flat fading Rayleigh channel. Combining channel capacity can linearly increase with the
different linear detection techniques (ZF, MMSE) and
number of antennas (provided that the numbers of
nonlinear detection techniques (SIC, OSIC) improves
the quality of the received signal in high interference transmit and receive antennas are the same) [1], [2].
environment. 64-QAM constellation transfers more bits Thus, to increase the channel capacity, the transmitter
per symbol and the number of bits per symbol impacts and receiver can be equipped with multiple antennas,
the accuracy of OSIC. Using Concatenating ZF or and the resulting channel becomes a multiple-input
MMSE with OSIC equalization, better cancellation gain multiple-output (MIMO) channel.
can be achieved. In this paper, MMSE-OSIC has been The growing demand for multimedia services in
presented, whose performance is better in high wireless communications has developed methods to
interference environment when compared to the earlier increase system capacity and reliability. In multiple-
detection schemes like ZF, ZF-SIC, ZF-OSIC, MMSE,
inputmultiple-output(MIMO) systems, capacity
MMSE-SIC etc,.
increase is brought about by using the spatial
Index terms: Successive Interference Cancellation (SIC), multiplexing mode, which offers capacity
Orthogonal Frequency Division Multiplexing (OFDM), proportional to the number of parallel transmit
Multiple-Input Multiple-Output (MIMO), Quadrature streams that can be created (i.e., the minimum
Amplitude Modulation (QAM), Inter symbol number of transmit and receive antennas) [3][6].
interference (ISI). There are MIMO receivers for spatial multiplexing
schemes, such as the Maximum-Likelihood (ML)
I. INTRODUCTION receiver, the linear receiver, the Successive
Interference Cancellation (SIC) receiver, etc. [3]. The
Wireless communication is highly challenging due to ML receiver is an optimal receiver, but it is difficult
complex time varying propagation medium. If we to implement due to high complexity arising from
consider a wireless link with one transmitter and one exhaustive searches over all candidate vector
receiver the transmitted signal that is launched into symbols.
wireless environment arrives at the receiver along a On the other hand, linear receivers, such as zero-
number of diverse paths referred to as multipath. forcing (ZF) or minimum mean square error (MMSE)
These paths occur from scattering and rejection of receivers, have low decoding complexity, but
radiated energy from objects (buildings, hills, trees detection performance decreases in proportion to the
etc.) and each path a different time-varying delay, number of transmit antennas. Therefore, there has
angle of arrival and signal amplitude. As a been a study on a practical nonlinear receiver,
consequence the received signal can vary as a namely, the ordered SIC (OSIC) receiver, which
function of frequency, time and space. These successively decodes data streams through nulling
variations are referred as fading and cause detoriation and canceling [6].
of the system quality. Furthermore wireless channel Although the OSIC receiver requires higher
suffer from co-channel interference (CCI) from other complexity than the linear receiver, it outperforms
383
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
linear receivers, and its performance is improved due where r is the received vector of size NrX1, H is
to the selection diversity arising from canceling the the channel matrix of size NrXNt s is the
detected signal out of the received signal [3]. SIC transmitted vector of size NtX1, and n is the noise
algorithm relies on a sequential detection of the vector of size NrX1. Each noise element is typically
received block. At each step, one symbol is detected modeled as independent identically distributed white
before being subtracted from the received block. This Gaussian noise [5], [6] with variance Nt/(2*SNR) [3].
introduces successively freedom degrees which To prevent correlation due to spacing they are
enable to reduce the noise/interference influence for typically spaced at least _c/2 where _c is the
the next users to be detected and therefore increases wavelength of the carrier frequency [5]. The second
the reliability of the decision process. reason correlation can occur is due to lack of
multipath components. It is for this reason that rich
Wireless communication networks design has the multipath is desirable in MIMO system. The
limitation on spectrum resources is an important multipath effect can be interpreted by each receive
restriction for achieving high bit rate transmissions. antenna being in a different channel. For this reason
The use of M-ary Quadrature Amplitude Modulation the rank of MIMO channel is defined as the number
(M-QAM) is considered an attractive technique to of independent equations offered. It is important to
overcome this restriction due to its high spectral note that
efficiency, and it has been studied and proposed for Rank (H) min (N r, Nt) (2)
wireless systems by several authors [8], [9]. And the maximum number of streams that a MIMO
system can support is upper bounded by min (Nr, Nt).
II .SYSTEM MODEL Current MIMO system includes MISO and SIMO
system that uses MIMO technique to improve the
MIMO systems are composed of three main performance of wireless system can be divided into
elements, namely the transmitter (Tx), the channel two kinds. One is spatial multiplexing which provides
(H), and the receiver (Rx). In this paper, Nt is denoted a linear capacity gain in relation to the number of
as the number of antenna element at the transmitter transmitting antenna and the other is spatial diversity
and Nr is denoted as the number of elements at the schemes which can reduce the BER and improve the
receiver. Fig1depicts the Physical MIMO channel. reliability of the wireless link.
The channel with Nr outputs and Nt inputs is denoted
as Nr x Nt matrix. A. Spatial Multiplexing
384
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
385
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
386
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
x
receiver decides to subtract effect of 2 from the
received vector y1 and y2, and then re-estimate
x1 .
x x
Once the effect of either 1 or 2 is removed, the
new channel becomes a one transmit antenna, 2
receive antenna case and the symbol on the other
spatial dimension can be optimally equalized Fig. 3. 2X2 MIMO-OFDM System, ZF equalizer,
by Maximal Ratio Combining (MRC). BER vs Eb/No for 64-QAM modulation with and
without SIC
II.SIMULATION RESULTS
Fig 3 shows the simulation result for a 2x2 MIMO-
In all simulation results the 2x2 MIMO-OFDM OFDM system with a ZF-equalizer for the case of
system in Rayleigh fading channel is taken and the pure equalization and the combination with
two modulation schemes are applied i.e. BPSK and Successive Interference Cancellation. SIC is
387
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
simulated with random and optimal order for 64- MIMO-OFDM systems are gaining much more
QAM Modulation to improve the transmission rate. attention in wireless communication research due to
their potential to increase considerable capacity in
mobile cellular communication.
MMSE-SIC-SORTING detector for 64-QAM MIMO
system performance is better than ZF, ZF-SIC, ZF-
SIC-SORTING, MMSE and MMSE-SIC. The MIMO
system performance is improved especially by using
higher order modulation scheme such as QAM than
BPSK modulation.
IV.FUTURE SCOPE
388
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
In this paper, we design the spreading code MIMO-CDMA with Parity bit selected spreading is
permutations based on Space-Time Block Codes discussed in [2]. In parity bit selected spreading
(STBC).Compared to the results presented in [2], the sequence technique, based on the calculated parity bits
STBC-based design can improve the bit error rate the spreading waveform is Selected for each antenna. In
(BER) performance over the flat fading channel this approach, for each user 2(n- k) spreading sequences
without increasing the system complexity[3] will be assigned. Where n-k is the number of parity bits
389
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
of the code. The system MIMO-CDMA with t-design The above expression finds the squared Euclidean
permutation spreading is similar to parity bit selected distance between the received vector and all possible
spreading. remaining vectors and the vector which gives the
Now M is a set of all possible message vectors, contains smallest distance that is the transmitted message.
16 elements. M is divided into 8 possible cosets. The
MIMO- CDMA system is designed with Nt=Nr=4
antennas, and N=8. On signaling interval, the message to V.RESULT ANALASIS
be transmitted is {m1, m2,mNt}.The cosets are
M1={0000 1111},M2={00011110}, M3={0010 MIMO-CDMA system with 4 transmit and 4 receive
1101}, M4={0011 1100} M5={01001011}, antennas are considered in this proposed technique
M6={0101 1010}, M7={0110 1001}, and existing techniques .Simulation results for BER
M8={01111000}. And the spreading sequences are (bit error rate) performances are presented for 4x4
assigned to the antennas as shown in TABLE I. for antenna system. From Fig.2 we see that MIMO-CDMA
example, if the message comes from M1, then spreading system employing STBC permutation spreading giving
waveforms for four antennas the better BER performance as compared to t-designs
are c1, c3, c5 and c7correspondingly that is shown in the method, and conventional spreading also taken as
first row of the TABLE I.The received vector at the reference. And here the proposed technique
output of matched filters of the system is outperforms the conventional and t-designs. From
TABLE I, we can see that in t-design permutation
IV. SPACE-TIME BLOCK CODE APPROACH spreading table, there is no code symmetry between
FOR BIT ERROR PROBABILITY ANALYSIS different cosets, that leads to some degrees of freedom
are lost, when MLD (maximum likelihood detection) is
The MIMO-CDMA system employing STBC performed at the receiver. So that cause to slight
permutation spreading [4] is shown in Fig 1. The increment of BER. But the system with STBC spreading,
input bit stream is converted into Nt parallel data the permutation spreading table provides code symmetry
streams. And these bits are between different cosets, so that gives better BER
given to spreading sequence selector that selects Nt performance.
spreading waveforms, then these data bits are modulated
with BPSK modulation, then each bit is spread by Fig. 2. BER performance for conventional-designs and
using spreading STBC for 4x1 and 4x4 antenna system.
waveform. On a given signaling interval Nt spreading
waveforms will be selected from N orthogonal spreading
waveforms. And at the receiver side, the output of
each
antenna is given to a bank of matched filters, and
according to matched filter decision, the transmitted data
is estimated.
The spreading assignment is done based on 8x8
STBC
matrixes. And the matrix [6] is given as:
390
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
CONCLUSIONS
REFERENCES
391
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
393
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
394
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
395
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
CONCLUSIONS
396
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
AbstractUplink power control plays a key role on and closed loop power control has been studied in
the perfor- mance of uplink cellular network. In this [2], [3]. It has been shown in [4] that the
work, the power control factor1 ( [0, 1]) is evaluated fractional path loss compensation factor with
based on three parameters namely: average transmit closed loop power control can greatly improve the
power, coverage probability and average rate. In other system performance. The impact of fractional
words, we evaluate power control factor such that
average transmit power should be low, coverage
power control on the SINR and interference
probability of cell-edge users should be high and also distribution has been studied in [5] and also a sub-
average rate over all the uplink users should be high. optimal configuration is proposed for the fractional
We show through numerical studies that the power power control.
control factor should be close to 0.5 in order to achieve
an acceptable trade-off between these three It has been shown in [6] that fractional path loss
parameters. compensation is advantageous than the full path loss
compen- sation in terms of cell-edge capacity and
Power control is an important consideration for also battery life time. A modified fractional power
the uplink cellular networks. It has two modes of control utilizing the path loss difference between
operation: closed loop and open loop power control serving cell and strongest interfering cell is
[1]. In closed loop power control, the base station proposed in [7] to improve the cell-edge bitrate and
(BS) compare the received Signal-to-noise- plus- overall spectral efficiency. Recently, authors of [8],
Interference-ratio (SINR) to the desired target [9] proposed an analytical approach to this
SINR. If the received SINR is lesser than the problem and they have provided the insight for
desired target SINR a transmit power control choosing the fractional power control. However,
command is transmitted to the MS to increases although insightful design guidelines are provided,
the transmit power. Otherwise, transmit power they have not provided the specific parameter value
control command is transmitted to decrease the which is at the end, the interest of cellular operator.
transmit power. On the other hand in open loop
power control power there is no feedback path. In
this work, an open loop power control is
considered.
398
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Fig. 5 plots the average rate for the FR1 and FR3
with respect to power control factor for different
path loss expo- nents. Firstly, it can be observed
that as power control factor increases average rate
decreases for both the reuse systems and both the
path loss exponents. Secondly, it can be observed
that as power control factor increase from 0 to 0.25
average rate does not decrease significantly
(especially for FR3 system).
By observing Fig. 2, 3, 4 and 5 and behaviour of
these plots as discussed before, one can conclude
that at power control factor = 0.5, the average
transmitted power is decreased by 2, the coverage
probability of cell-edge users increases significantly
and also the average rate does not decrease
significantly and hence the power control factor
should be chosen close to 0.5.
We have one another way to evaluate the power
control factor. We develop a cost function J which
take care of all three parameter: average rate, edge
coverage probability and average transmitted power.
399
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
V. CONCLUSION
400
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Dr. P. Siddaiah 2
Professor & DEAN, IEEE member
University College of Engineering & Technology,
Acharya Nagarjuna University,
Guntur, Andhra Pradesh, India
Abstract- A planar phased-array antenna has been need for any mechanical rotation. These unique
constructing from a 15x15 square grid of z-directed capabilities have found phased arrays a broad range
monopoles with a length of 0.475 , element spacing of of applications since the advent of this technology.
0.29 , average directivity of 20.0 dBi across all scan Phased arrays have been traditionally used in military
angles, an average H-plane HPBW of 37 degrees, an
applications for several decades. Recent growth in
average E-plane HPBW of 23 degrees, and an efficiency
of 99.6%. Two current distributions are tested: uniform civilian radar-based sensors and communication
and binomial. The binomial current distribution is systems has drawn increasing interest in utilizing
selected to minimize the side lobe level. For this phased array technology for commercial applications.
implementation we are using HFSS 14.0. Phased array antennas are common in
communications and radar and offer the benefit of
Keywords- Phased array, Electronically steerable, far-field beam shaping and steering for specific, agile
Current Distribution, HFSS. operational conditions. They are especially useful in
modern adaptive radar systems where there is a trend
I. INTRODUCTION toward active phased arrays and more advanced space
time adaptive signal processing. In phased arrays all
In antenna theory, a phased array is an array of the antenna elements are excited simultaneously and
antennas in which the relative phases of the the main beam of the array is steered by applying a
respective signals feeding the antennas are varied in progressive phase shift across the array aperture.
such a way that the effective radiation pattern of the
array is reinforced in a desired direction and
suppressed in undesired directions. An antenna array II. PHASED ARRAY ARCHITECTURE
is a group of multiple active antennas coupled to a
common source or load to produce a directive The phased array antenna has an aperture that is
radiation pattern. Usually, the spatial relationship of assembled from a great many similar radiating
the individual antennas also contributes to the elements, such as slots or dipoles, each element being
directivity of the antenna array. Use of the term individually controlled in phase and amplitude.
"active antennas" is intended to describe elements
whose energy output is modified due to the presence
of a source of energy in the element (other than the
mere signal energy which passes through the circuit)
or an element in which the energy output from a
source of energy is controlled by the signal input.
401
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
(courtesy from skatelescope.org) antennas, the beam will experience a time delay equal
Accurately predictable radiation patterns and beam to Eqn. 1 in reaching successive antennas.
pointing directions can be achieved. A phased array is = 2 d sin ()/ (1)
an array antenna whose beam direction or radiation
pattern is controlled primarily by the relative phases Here, is the wavelength of the signals. Hence, if
of the excitation coefficients of the radiating the incident beam is a sinusoid at frequency with
elements. Physically it is composed of a group of amplitude of A, the signals received by each of the
individual elements that are arranged in a linear or antennas can be written as Eqn 2.
two dimensional (typically planar) spatial
configurations. Si = A e-jn (2)
The plane wave incident at an angle upon the
The first phased array antennas were developed phased array experiences a linear delay progression at
by the Nobel laureate Karl Ferdinand Braun in the the successive antenna elements. Therefore, the
beginning of the 20th century. He demonstrated the variable delay circuits must be set to a similar but
possibility to enhance antenna transmission in a with reverse delay progression to compensate for the
specific direction. Quite a number of applications of delay of the signal arrived at the antenna elements. In
phased array transmissions have been developed, linear arrays, variable time delays are designed to
ranging from radar systems with marine vessels to provide uniform phase progression across the array.
airplane navigation systems in which antennas are Therefore, the signal in each channel at the output of
placed in matrix configurations. The technique has the variable delay block can be written as Eqn. 3.
been refined considerably so as to allow for multi
functional solid state radar systems in which ships Si' = A e-jn e-jn (3)
and airplanes can be detected and in which
communication between moving objects can be In this equation, denotes the difference in phase
realized in an efficient manner. Other applications of shift provided by two successive variable time delay
phased array techniques can be found in radio blocks. Therefore, array factor which is equal to the
astronomy, your mobile phone and with military sum of all the signals normalized to the signal at one
applications. path can be written as Eqn. 4.
N
A. Phased Array Principle F = e-jn (-) (4)
n=1
402
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
can be controlled by electronically tuning the variable that meets the directive angle requirements is
time delays eliminating the need for any mechanical required. Therefore, we consider an arrangement of
rotation of antenna array. It should be also noted that concentric circles, a 2dimentional rectangular planar
the benefits of using phased array is increased as the array, and a 3dimentional placement of dipoles as
number of elements in the phased array is increased. possible element arrangements. We were not able to
As mentioned before, the maximum value of array consider the concentric circles or 3dimensional
factor N2 is directly proportional to the number of arrangement because the required mathematical
array elements. Furthermore, the beam width of the analysis involved is lengthy and our design time is
array can be decreased by increasing the number of limited. In addition, 3 dimensional arrays add extra
array elements in order to enhance the spatial mechanical complexity to the design, and a goal of
selectivity of phased array. this project is to trade mechanical complexity for
electrical complexity.
B. Classification of Phased Arrays
Therefore, only the 2dimensional square array is
(1) Scanning methods considered. Two dimensional rectangular arrays are
(2) Radiator feed methods much easier to analyze. Although we expect to lose
(3) Positioning of radiators in the array some degree of accuracy azimuthally, a rectangular
array seemed like an effective trade off between
The main scanning methods are phase scanning design complexity and performance.
and frequency scanning. It is the phase scanned
arrays that are referred to as phased arrays. From the B. Element Spacing
viewpoint of feed methods, arrays are divided into
constrained feed arrays and space fed arrays, the In general, increasing element spacing allows for
latter taking the form of reflect arrays or transmission finer beam widths, but element spacing greater than
arrays. With regard to element positioning, phased halfwavelength results in undesired grating lobes.
arrays are divided into uniformly spaced and These grating lobes are shown for a separation of 0.8
unequally spaced arrays. They provide the radar with wavelengths along the x and yaxis elements: We
flexibility and adaptation to the assigned task, ability chose a uniform element spacing of 0.29 wavelengths
to change beam position in space almost such that the beam width is moderately fine, does not
instantaneously (electronic scanning), generation of produce grating lobes, and allows sufficient number
very high powers from many sources distributed of elements to fit on the circular mounting structure.
across the aperture, high directivity and power gain
possibility of synthesizing any desired radiation This element spacing determines the total number of
pattern (including formation of pattern nulls in the array elements due to restrictions in the maximum
directions of undesired interference sources) allowable array area. Since the maximum length of
capability of combining search, track, and recognition the square array is 0.707 the allowable diameter (50
functions when operating in multiple target and inches), the number of elements allowed with a 0.29
severe interference environments (including wavelength element separation equals the following:
jamming), enhanced target throughput capability, and
compatibility with digital computers and digital N or M max=
signal processing algorithms.
403
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
VI. CONCLUSION
404
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
of the phased array systems alleviate the problem of [14] R. J. Mailloux, Phased array antenna handbook. Artech
House, Boston, 1994.
multipath fading and co-channel interference by
[15] A. K. Bhattacharyya, Phased Array Antennas. Wiley-
suppressing signals emanating from undesirable Interscience, New York,2006.
directions. Unfortunately, side lobe levels on the [16] A. Ishimaru, R. Coe, G. Miller, and W. Geren, Finite
uniform distribution are above the 20dBi periodic structure approach to large scanning array
problems, Antennas and Propagation, IEEETransactions
requirement. on, vol. 33, no. 11, pp. 12131220, Nov 1985.
[17] A. Skrivervik and J. Mosig, Analysis of nite phase arrays
The uniform distribution experiences significantly of microstrip patches, Antennas and Propagation, IEEE
reduced beam width than the binomial distribution; Transactions on, vol. 41, no. 8,pp. 11051114, Aug 1993.
[18] A. Neto, S. Maci, G. Vecchi, and M. Sabbadini, A truncated
however, directivity for each case exceeds the
Floquet wave diraction method for the full wave analysis of
specified requirement (20dBi). By using Dolph large phased arrays. I. Basic principles and 2-D cases,
Chebyshev current distribution we can reduces Antennas and Propagation, IEEE Transactions on, vol. 48,
HPBW while maintaining -20 dBi side lobes relative no. 4, pp. 594600, Apr 2000.
to the main beam. [19] C. Craeye, A. Thuis, and D. Schaubert, An e cient MoM
formulation for nite-by-innite arrays of two-dimensional
antennas arranged in a threedimensional structure, Antennas
ACKNOWLEDGMENT and Propagation, IEEE Transactions on, vol. 52, no. 1, pp.
271282, Jan. 2004.
[20] D. M. Pozar, General relations for a phased array of printed
We would like to thank the Organization of antennas derived from innite current sheets, Antennas and
Acharya Nagarjuna University, Guntur, special Propagation, IEEE Transactions on, vol. 33, no. 5, pp. 498
504, May 1985.
thanks to the Dr. P. Sidaiah garu, for support to use
R&D Laboratories .
AUTHOR BIOGRAPHY
REFERENCES
Smt. M. Vasujadevi obtained B.Tech
degree in Electronics and
[1] Danial Ehyaie, Novel Approaches to the Design of Phased communication Engineering from
Array Antennas, Doctor of Philosophy(Electrical S.V.V.S.N Engineering College,
Engineering) thesis, University of Michigan,2011 Acharya Nagarjuna University, Guntur
[2] Kyle Howen , Andrew Huard,Design of Planer Phased in 2008, M. Tech Degree from BSIT,
array antenna, unpublished. JNT University, Hyderabad in 2011.
[3] R. J. Mailloux and I. Books, Phased array antenna handbook:
Artech House,2005.
She is currently working as Assistant Professor, Dept. of
[4] R. J. Mailloux, "Electronically scanned arrays," Synthesis ECE in Chalapathi Institute of Engineering and
Lectures on Antennas, vol. 2, pp. 1-82, 2007. Technology, affiliated to Acharya Nagarjuna University,
[5] Joseph Spradley, A Volumetric Electrically Scanned Two- Guntur, India. Her interesting fields are Antennas, Signal
Dimensional Microwave Antenna Array, IRE National Processing, Communications, VLSI Design.
Convention Record, Part I - Antennas and Propagation; e-mail: vasujadevibhavanam@gmail.com
Microwaves, New York: The Institute of Radio Engineers,
1958, 204-212.
[6] P. D. Trinh, S. Yegnanarayanan, F. Coppinger and B. Jalali
Silicon-on-Insulator (SOI) Phased-Array Wavelength
Dr. P Siddaiah obtained B.Tech degree
Multi/Demultiplexer with Extremely Low-Polarization in Electronics and communication
Sensitivity, IEEE Photonics Technology Letters, Vol. 9, No. Engineering from JNTUA college of
7, July 1997 engineering in 1988. He received his
[7] "YIG-sphere-based phase shifter for X-band phased array M.Tech degree from SV University
applications". Scholarworks. Tirupathi. He did his Ph.D program in
[8] Worlds Most Complex Silicon Phased Array Chip JNTU Hyderabad. He is the Chief
Developed at UC San Diego in UCSD News (reviewed 02. Investigator for several outstanding Projects sponsored by
November 2007)
[9] A. Balanis, Antenna theory, 3rd ed. Wiley-Interscience, New
Defense Organizations, AICTE, UGC & ISRO. He is
York, 2005. currently working as Professor & DEAN, Department of
[10] P. Patel, D. Kant, E. Wal, and A. van Ardene, Phased array ECE in University College of Engineering and Technology,
antennas demon- strator as a radio telescope - embrace, Acharya Nagarjuna University, Guntur, India. He has
Antennas and Propagation Society International Symposium, taught a wide variety of courses for UG & PG students and
2008. AP-S 2008. IEEE, pp. 14, July 2008. guided several projects. Several members successfully
[11] S. Hay, J. OSullivan, J. Kot, C. Granet, A. Grancea, A. completed their Ph.D under his guidance. Several members
Forsyth, and D. Hayman, Focal plane array development for pursuing their Ph.D degree. He has published several
ASKAP (Australian SKA pathnder), Antennas and
Propagation, 2007. EuCAP 2007. The Second
papers in National & International Journals & Conferences.
EuropeanConference on, pp. 15, Nov. 2007. He is the life member of FIETE, IE & MISTE.
[12] N. Amitay, V. Galindo, and C. P. Wu, Theory and analysis of
phased arrayantennas. Wiley-Interscience, New York, 1972.
[13] R. J. Mailloux, Phased array antenna handbook. Artech
House, Boston, 1994.
405
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract - Employment of powerful Direction-of-Arrival The very purpose of DOA estimation is to use the
(DOA) estimation and beam-forming algorithms with data received by the array to estimate the direction of
smart antenna systems enables continuous tracking of arrival of the signal. The results of DOA estimation
desired signal irrespective of the signal environment
are then used by the array to design the adaptive
there by, enhancing the quality of wireless services. In
this paper emphasis is laid mainly on the performance beam-former [4], which is used to maximize the
of the various powerful high-resolution subspace power radiated towards users, and to suppress
algorithms like MUSIC, root-MUSIC and another fast interference. As a result, we can infer that a
DOA estimation algorithm making use of pseudo- successful design of an adaptive array depends highly
covariance matrix, taking into the consideration, the on the performance of the DOA algorithm. Direction
number of snap-shots required to build an array of arrival algorithms are usually complex and their
correlation matrix. The aim is to compare their performance depends on many parameters such as
performances by varying the number of input signal number of mobile users and spatial distribution, the
snap-shots and to find the best algorithm that gives
number of array elements and their spacing and the
optimum performance with reduced number of snap-
shots from the results obtained. Their performance in number of signals samples.
cases of high signal correlation and rapidly changing
signal DOAs is also discussed. In this paper, the performance comparison of two
DOA Estimation algorithms is done based on number
Keywords - array correlation matrix, Direction-of- of snap-shots required for each of them to give a
Arrival (DOA) estimation, MUSIC, pseudo-covariance good resolution. The number of snap-shots required
matrix, root-MUSIC, subspace algorithms. determines the effectiveness of the system to rapidly
update DOAs of the signals and enabling the
I. INTRODUCTION continuous tracking of the targets. In the following
sections, the algorithms considered for comparison
In wireless communication system, the design,
are analyzed in detail and the results obtained are
orientation and performance of the antenna decides
discussed.
the quality of the wireless services. And so, the
performance of the antenna systems deployed for II.MUSIC and root-MUSIC DOA ESTIMATION
such purposes is of prime importance. Hence the ALGORITHMS
Omni-directional[1] were replaced by the smart (Sub-Space based Methods)
antenna array systems [2] to reduce multipath and co-
channel interference and to offer preferential gain for In Subspace based method [5], the observed
the signals of served users by rejecting the signals covariance matrix is decomposed into two orthogonal
that interfere with those of desired ones. This enables spaces: signal subspace and noise subspace. The
a higher capacity, in addition to spectral efficiency DOA estimation is calculated from any one of the
and frequency reuse [3]. This is achieved by focusing subspaces. The subspace based DOA estimation
the radiation only in the desired direction and algorithms MUSIC root-MUSIC and ESPRIT
adjusting itself to changing traffic conditions or provide high resolution, more accuracy and are not
signal environments. Smart antennas employ a set of limited to physical size of array aperture. The steering
radiating elements arranged in the form of an array vectors corresponding to the directional sources are
(ULA, in this case). The signals from these elements orthogonal to the noise subspace. As the noise
are combined to form a movable or switchable beam subspace is orthogonal to the signal subspace, these
pattern that follows the desired user. steering vectors are contained in the signal subspace.
406
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
(2.1) (2.5)
The denominator expression can be simplified by
Where defining the matrix which is Hermitian.
is an M x D array steering matrix, and This leads to the root-MUSIC expression
(2.2)
is D x D source correlation matrix. (2.6)
Rxx has D eigenvectors associated with signals and In case of a Uniform Linear Array (ULA), the
M D eigenvectors associated with the noise. We element of the array steering vector is given by:
can then construct the M x (M-D) subspace spanned
by the noise eigenvectors such that
where m=1, 2, .M
(2.3)
The denominator argument of the root-MUSIC
Noise eigen-vectors are orthogonal to steering vectors expression can be written as
at the angles of arrival .
Because of this orthogonality condition, the
Euclidean distance
for each and every arrival angle. The noise subspace
eigenvectors are orthogonal to array steering
vectors at the angles of arrivals
and placing this distance
expression in the denominator gives sharp peaks at
the angles of arrival. The MUSIC Pseudo-spectrum (2.7)
is given as:
407
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
408
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
(3.1)
(3.4)
X (k) v = v (3.8)
409
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
except for the steering vector for the look direction In the proposed algorithm, there is no signal
. cancellation problem, even though the signals are
correlated. The output power of array is
The constraint condition is determined as
(3.9)
(3.13)
Where
Where K is the number of snapshots required to
observe the output power. If a signal impinges
and ( is the steering vector for the look on the array within a small level of deviation from the
direction with Mx1 dimension. look direction, the bearing response can be
given by
The weight vector can be obtained from the eigen
vector v and the steering vector as (3.14)
+ (3.16)
410
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
411
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
REFERENCES
412
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
About Authors:-
413
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract - The development in communication systems In this paper co-axial feed rectangular microstrip
requires the development of low cost, minimal weight and antenna is designed for Radar applications and is
low profile antennas that are capable of maintaining expected to operate within 5.45GHz
high performance over a wide spectrum of 5.85GHz frequency span.
frequencies. This technological trend has focused much
effort into the design of a Microstrip patch antenna. This
paper deals with design aspects of Rectangular
Microstrip patch Antenna with co-axial feed and its
properties. In this paper simulations are carried out on
C- band Microstrip patch antenna that suits for Radar
applications. With the adjusted parameters it
exhibits maximum 6.88 dBi Gain and VSWR=1.3 at the
centre frequency 5.65GHz. Desired patch antenna design
was simulated by IE3D Software
I. INTRODUCTION
414
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
electric field perpendicular to the radiating edge constructive and deconstructive interference effect
(this is shown in The Figure 2). While the in the far field.
radiating edges are perpendicular to the E field
plane the size of the patch is also very D. Co-axial Feed
important due to the fact that a mode will be set
The Coaxial feed or probe feed is a very common
up in the cavity beneath the patch which
technique used for feeding Microstrip patch
determines the shape of the currents and the
antennas. As seen from Figure 3 the inner conductor
shape of the effective radiators.
of the coaxial connector extends through the
dielectric and is soldered to the radiating patch,
B. Uses of Patch Antennas
while the outer conductor is connected to the
ground plane The main advantage of this type of
Patch antennas are becoming more common because
feeding scheme is that the feed can be placed at any
more complicated designs are continually being
desired location inside the patch in order to match
created allowing patch antennas to enter new
with its input impedance. This feed method is easy
niches of technology. The most common use of a
to fabricate and has low spurious radiation.
patch antenna today is in the cell phone. Patch
However, a major disadvantage is that it provides
antennas are cheaper to manufacture and can provide
narrow bandwidth and is difficult to model since a
Equal or better gain performance as compared to the
hole has to be drilled in the substrate and the
standard dipole and monopole antennas traditionally
connector outside the ground plane.
utilized in cell phones in the past. Also patch
antennas, being integrated into the PWB, can very
easily be created with active structures which create
opportunities for active and smart antennas in small
packages. This means that as time goes on, it is
feasible more and more antennas will utilize the
patch antenna methodology because very high
directional gains are achievable with the added
capability of being steerable while still in a small
package. Patch antennas are also utilized in many
other locations such as airplanes and space
applications because of their very light weight and
small size. It is also quite easy to create a patch
antenna which can handle harsh environments
because of their simple structure. Future work in
patch antennas will include smart active antenna
designs as well as multiple dimensional arrays, which
utilize the multi-layer PWB technology to create Fig 3. Rectangular Microstrip Patch Antenna with
advanced microstrip antennas. Co-axial feed
415
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
performance whose dielectric constants is in the on Zeland program Manager. The Zeland
lower end of the ranges because they provide better window will be opened then click on
efficiency and larger bandwidth but at the cost of file and select MGRID.
larger element size. RT/Duroid 5880 is used the
purpose of selecting this particular substrate is New window will be opened and in
because; it possesses lowest electrical loss tangent that click on insert the substrates symbol
and has a better thermal conductivity that are very and insert the dielectric constant, Height
much needed for aerospace application antennas. and thickness of the substrate.
A. Physical Parameters of the Patch Antenna
After entering all the details go to
The antenna parameters of this antenna can be Entity then click on rectangle a
calculated by the transmission line method (Balanis, dimensions window will appears then
2005), as exemplified below. insert our length and width of the patch
L=17mm and W=21mm.
3. SIMULATION RESULTS
Click on Start Go to Programs then click Fig 5. VSWR of the Patch antenna
416
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
4. CONCLUSIONS
5 .FUTURE SCOPE
REFERENCES
rd
[1]. Antenna Engineering Handbook part 1 &2 (3
edition) by Richard C. Johnson.
[2]. Constantine. A Balanis, Antenna Theory and Design,
Second edition, 1997, John Wiley Publications.
[3]. Antenna and Wave propagation by K.D. Prasad
[4]. J.D Krauss and Ronald J Marhefka, Antennas for all
applications, 2000,Third Edition
Fig 8. Frequency properties of the patch [5]. Microwave Engineering by D.M.Pozar
[6]. CAD of Microstrip Antennas for Wireless
Applications by Robert A.Sainati
[7]. Handbook of Microstrip Antennas, Volume-1 by
J R James & P S Hall.
[8]. Radar Handbook, Second Edition by Merrilli. Skolnik
[9]. IE3D Users manual , Release 9, Zeland software.
[10]. Microstrip antenna using MStrip 40 Lab
Manual by Mr. Martin Leung.
[11] I.Govardhani , M.Venkata Narayana Published paper in
International Journal of Computer Science & Communication
Networks,Vol 2(1), 375-380, ISSN:2249-5789 on Rectangular
Patch Array Antenna with Liquid Crystal Substrate for Ka and Q
Band Applications
[12] Dr.K.S.N Murthy, Venkata Raviteja.K, I.Govardhani,
M.Venkata Narayana Published a paper on Multi-band Ladder-
shape Microstrip Patch Antenna IJSER Volume 3, Issue 3.
Fig 9. Gain vs Frequency display
417
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
418
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
The length of patch (L) is about g/2 (g is effective III.CO-AXIAL PROBE FEED
wave length) and substrate height(h) is of order of
g/20.Due to very small space between radiating This is a common feed technique here outer
element and ground plane main power is radiated conductor is connected to ground plane and the inner
towards broad side. The fringing fields effectively conductor of co-axial connector is extends through
increase the length(L) of patch need to be accounted dielectric and soldered to patch. Inner conductor of
in determine resonance frequency[1]. co-axial cable transfers the power from strip line to
microstrip antenna from slot in the ground plane.
Placing of feed position is important in order to have
best matching with input impedance. Here fees is
applied at (0mm, 9.2mm). It provides narrow
bandwidth performance and it is difficult to design
for thick substrate.[1]
Fig 2: Electrical Field Lines(Side View) Present paper we have considered different dielectric
materials with same permittivity and different loss
The most commonly used design equations of tangent are listed bellow.
antenna
a) Effectivedielectricconstant TABLE I
DIELECTRIC LOSS TANGENT
Material Name Permittivity Loss
Tangent
b) Lengthextensionis
Rubber_hard 3 0
Roger RO3003 3 0.0013
Arlon AD300A 3 0.002
Neltec NH 9300 3 0.0023
c) EffectiveLength Rogers Ultralam 3 0.003
1300
d) Actuallengthofpatch
V. NUMERICAL DESIGN
e) Patchwidth
The proposed co-axial feed rectangular micro strip
patch antenna operating at 1.7616GHz having
physical dimensions of radiating patch length
f) Groundplanedimensions L=47.4mm along Y-axis, width W=56.5mm along X-
axis (here width of patch is maintained higher than
length because of wider operating band width). The
co-axial feed is provided at 0mm, 9.2mm along X, Y-
axis respectively. The substrate height is 1.56mm.
419
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
TABLE II
RETURN LOSS
Fig 4: Coaxial feed micro strip antenna Substrate Return loss Operating
Frequency(GHz)
Rubber_hard -18.1618 1.7616
VI. SIMULATION SETUP Roger RO3003 -20.9417 1.7616
Arlon AD300A -22.5992 1.7616
Current proposal of coaxial feed rectangular micro
Neltec NH 9300 -23.3269 1.7616
strip antenna. And results of return loss, gain and
Rogers Ultralam -24.8394 1.7616
band width were presented. The Ansoft HFSS
1300
software is utilized here for execution of
One thing we can say that as the loss tangent of
dielectric material increases the return loss value also
decreasing. Indicates more amount of power is
forwarded and very less amount of power is reflected
back.
B. Bandwidth:
It can be defined as the range of frequencies over
which gain is constant. In software simulation this
value is taken at intersecting point s of -10db line and
return loss curve.
TABLE III
BAND WIDTH
Fig 5: Simulated Antenna Model
Substrate Band width Operating
(MHz) Frequency(GHz)
VII. RESULTS & DISCUSSION
Rubber_hard 22.5 1.7616
A. Return loss Roger RO3003 22.5 1.7616
Arlon AD300A 22.5 1.7616
According to maximum power transfer theorem, Neltec NH 9300 22.5 1.7616
maximum amount of power will be transferred when Rogers Ultralam 22.5 1.7616
there is a perfect matching between input and output. 1300
If load is mismatched the whole power is not
delivered to load and there is a return of power that is The loss tangent of substrate does not effect the
called loss, since this loss is returned hence is called operating bandwidth of antenna.
return loss is -20log. Where is reflection C.Impedance:
coefficient. The response of magnitude of S11 verses
frequency curve clearly explains return loss Here only one impedance diagram is shown which
indicates all the substrate materials have same
420
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
impedance, means loss tangent is not effected the Accepted 0.009 0.009 0.009 0.009 0.009
impedance. Power 7394 8149 8422 8515 867
Incident 0.009 0.009 0.009 0.009 0.009
Power 9012 9012 9012 9012 9012
Radiation 1.002 0.930 0.895 0.881 0.849
Efficiency 3 23 53 39 91
FBR 26.57 26.57 26.57 26.57 26.43
3 7 2
IX. CONCLUSIONS
421
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
422
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract: In this paper, we present a video requires the use of the original image in the extraction
watermarking algorithm to resist the video attacks and process then it is called Non-Blind watermarking
noise attacks.. Each video is divided in to frames. Each technique. The watermarking scheme requires the
frame is subdivided into non-overlapping 8*8 blocks.
watermark data and/or the parameters used to embed
Each block is transformed 2D-DCT. After performing
the same transformation to all the blocks and frames, the data, then it is called Semi-Blind watermarking.
1D DCT is applied along time axis. Now, the
quantization index modulation is used to embed the In this paper, we present an invisible watermarking
secret logo. Embedding is performed to DCT technique. The watermarking logo is embedding is
intermediate frequency coefficients of each block. The transform domain of the transmitted video. There are
DCT intermediate coefficients are adjusted by using a so many transformation techniques (DCT, DWT,
proposed algorithm that can resist the attack DFT, DHT etc). We use a discrete cosine
transform and quantization technique. The DCT gives
Index Terms Discrete cosine Transform, SQM, good PSNR.
Resistance to attacks
II. DCT
423
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
(2 y 1)v
cos (EQ 3)
2N
N 1 N 1
( 2 x 1)u
f ( x, y ) (u ) ( v )C (u, v ) cos Fig .1 Illustration of the Proposed DCT Method
u 0 v 0 2N
424
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Watermarking embedding:
425
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Experimental Results:
Video resize
Frame removal
426
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
About Authors:-
REFERENCES
427
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
AbstractX-ray imaging and Nuclear Imaging (PET, very important to remove the noise from these
SPECT) are helpful in medical diagnosis. The images images.
acquired from these modalities are suffering from In medical Image processing applications image
data dependent noise named as Poisson noise. The
presence of noise will hide the diagnostic details in
denoising is the important preprocessing task before
the images. So the images corrupted with the noise performing image analysis tasks such as image
is not suitable for automatic analysis by the segmentation, image feature extraction, image
computers. Many algorithms in the literature are registration, storage and image retrieval etc. The
using Variance Stabilization Transforms (VST) to noise present in the images will lead to wrong
convert the Poisson noise into Gaussian noise and diagnois and create problems in image guided
then apply the denoising algorithms developed for surgery. So the signal to noise ratio (SNR) of the
Gaussian noise on the transformed image. This method
is suitable when the image is having high photon count images must be improved before performing image
i.e. the noise level is low. In the low photon count analysis tasks. The major challenge in designing the
situations i.e. when the noise level is high, this denoising algorithms is removing the noise
approximation will not hold good. In this paper we are effectively without removing the image features
proposing a novel approach based on Blockwise Non- such as edges, lines etc. from the images [1].
Local (NL-) means filter, Variance Stabilization
Transform and Dual Tree Comlex Wavelet Transform
(DTCWT) for images corrupted with poisson noise.
The noise may be additive or multiplicative
Quantitative validation of the proposed method was depending on the modality used for medical
carried out using several quality metrics. The results image acquisition. The noise due to electronic
show that the proposed filter performed well than the components in the acquisition hardware will be
recently proposed filters based on Discrete Wavelet modeled with Gaussian noise which independent
Transform, NL-means filters. The proposed filter
removes noise effectively while preserving fine
of data, the data dependent noise such as quantum
structures such as edges, lines etc. even for very noisy noise in X-ray imag- ing is modeled with Poisson
cases. distribution, the speckle noise in ultrasound imaging
is modeled with Rayleigh distribution and the noise
Index TermsDualtree Complex Wavelet in MRI is modeled with Rician distribution. Here in
Transform, NL- means Filtering,Variance Stabilization this paper we are attempting to denoise the images
Transform,Universal Quality Index (UQI), Structural corrupted with quantum noise.
Similarity Index (SSIM)
II. POISSON NOISE
I.INTRODUCTION
The noise in X-ray imaging and Nuclear Imaging
Medical imaging became an integral part of medical (PET, SPECT) is modeled with Poisson noise. X-ray
di- agnosis in present days. Various medical imaging photons inci- dent on a receptor surface in a
modalities are developed for various applications
random pattern. We cannot force them to be evenly
since last few decades. These modalities are used, to
distributed over the receptor surface. One area of
acquire the images of the anatomical structures within
the receptor surface may receive more photons
the body to be examined, without opening the body.
than another area, even when both the areas are
X-rays, Computed Tomography, Ultrasound, Magnetic
resonance Imaging and Nuclear imaging are the exposed to the same average x-ray intensity. In all
popular modalities at present to diagnose the various medical imaging procedures using gamma or x-ray
diseases. However these modalities are suffering with photons most of the image noise is produced by the
a big problem called noise. Every modality is random behaviour of the photons that are distributed
suffering from noise in image acquisition and within the image. This is generally designated
transmission stage such as Quantum noise in X- quantum noise. Each individual photon is a quantum
rays and Nuclear imaging, speckle noise in ultrasound (specific quantity) of energy. It is the quantum
imaging, Rician noise in Magnetic resonance structure of an x-ray beam that creates quantum
imaging etc. The noise present in the images will noise.
degrade the contrast of the image and creates A Poisson model assume that each pixel x of an
problems in the diagnostic phase. So denoising is image f (x) is drawn from a Poisson distribution of
428
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
The Fig.1. gives the Poisson distribution for various After the preprocessing we will apply the inverse
values of as the increases the Poisson distribution
turns towards the Gaussian distribution variance stabilization transform to get back the
original distributions. In this work we are using
The denoising problem was attempted by many Anscombe transforms. The mathematical
researchers from many directions such as representation of the transform is given below. x
spatial filtering using local statistics, Transform is the input image pixel, Z is the transformed
domain approaches (Multiscale & Geometrical pixel and I is the inverse transform of the
transforms) [2], statistical modeling of transformed pixel.
images(Gaussian Scale Mixtures, Hidden Morkov
models etc.), Scale space theory and dictionary based IV. NON LOCAL MEANS FILTERING
approaches [3],[4]. Recently mathematical image
processing techniques such as Partial differential The nonlocal means filtering is also considered
equations (PDEs), Calculus of variations, Kernel as the extension of the fundamental neighborhood
Regression and patch based algorithms became
filter (Yaroslavsky filter) where the pixel intensity is
effective in solving the denoising problem [5], [6].
decided by averaging the intensities of its
The spatial filtering approaches were easy to
neighborhood [1],[16]. Here in this method the
implement but tend to remove image features along
pixels which are having similar neighborhood are
with noise. The image structures were preserved by
the algorithms based on Wavelet Transforms, PDE found and averaged in deciding the intensity of a
based methods such as anisotropic diffusion [7], Total pixel. Normally similarity between the pixels is
Variation problem of Poisson denoising is found by observing their intensities, if the two
transformed into a classical Gaussian denoising pixels having the same intensity we can say that
problem by applying a VST, a large variety of they are similar otherwise they are different.
standard denoising approaches can be used. In the Here in the nonlocal means algorithm the similarity
case of very low SNR images, the two major is found by considering the neighborhoods of the
problems are: VST robustness in low count situations pixels. Two pixels are similar when the
and the efficiency of the selected denoising filter neighborhoods of the two pixels are having same
under very noisy conditions. intensities. So we can define the neighborhood of
pixel i is a set of pixels j whose
III. VARIANCE STABILIZATION neighborhoods are similar to the neighborhood of
TRANSFORM i. Fig.2. shows the similarity windows in the image
429
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
430
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
431
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
B.Universal image quality index (UQI) words, the measuring of image quality based on an
initial uncompressed or distortion-free image as
It is a measure used to find the image reference. SSIM is designed to improve on traditional
distortion. It is mathematically defined by making methods like peak signal-to-noise ratio (PSNR)
the image distortion relative to the reference image and mean squared error (MSE), which have proved to
as a combination of three factors: Loss of be inconsistent with human eye perception.
correlation, Luminance distortion and contrast
distortion. This measure was developed by Alan
C.Bovik. If two images f (x, y) and g(x, y) are
considered as a matrices with M column and N
rows containing pixel values f (x, y) and g(x, y)
respectively the universal image quality index Q
may be calculated as a product of three
components
432
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
means algorithm.
REFERENCES
433
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Abstract: . In medical and health care areas, because of such as scans, different types of X-rays, MRI images
regulations and the availability of computers, a large and other radiological techniques. With all this
amount of data is becoming available .Data mining is a effort done in the research field, there is still a lot of
relative important objective to acquire knowledge from
large amounts of data. . On one side, practitioners are
place for improvements and the medical image
expected to use all this data in their work but, in a short processing is a domain in continuous expansion.
time it is difficult to process such a large amount of data Data mining is the non trivial extraction of implicit
by humans to make diagnosis, prognosis and treatment previously unknown and potentially
schedules. Evaluateing data mining tools in health care useful information about data[1]. Data mining
and medical applications to make timely and accurate technology provides a user- oriented approach to
decision is the aim of this work . The conventional method novel and hidden patterns in the data.
in medicine for brain MR images classification and tumor
detection is by human inspection. Operator-assisted The discovered knowledge can be used by the
classification methods are impractical for large amounts healthcare administrators to improve the quality of
of data and are also non-reproducible. MR images service. The discovered knowledge can also be used
also always contain a noise due to operator performance by the medical practitioners to reduce the number of
which can lead to serious errors in classification. The use adverse drug effect, to suggest less expensive
of artificial intelligent techniques , such as neural therapeutically equivalent alternatives.
networks, fuzzy logic, neuro fuzzy have play important
role in this field. In this paper we are going to consider
image registration technique as soft computing technique. II. DISCOVERY OF KNOWLEDGE FROM
Here we will consider geometry of the image and process MEDICAL DATABASES
the images. The dominant features which we got from
registration scheme are given to neural network for Data mining is an important aspect for acquiring
robust training and appropriate result analysis on medical knowledge . In Information now a days it has
images. attrected great deal of interest . [13]. Knowledge
Key Word: Treatment schedules, health care application, discovery process consists of an iterative sequence
MR images, tumor detection, soft computing technique, of, data selection data cleaning, data integration,
registration scheme data mining pattern recognition and knowledge
presentation. In particulars, data mining may
I. INTRODUCTION accomplish class description, association,
classification, clustering, prediction and time series
Automated classification and detection of tumors in analysis.
different medical images demands high accuracy
since it deals with human life. Also, computer
assistance is highly sought in medical institutions
due to the fact that it could improve the results of
humans in such a domain where the false negative
cases must be at a very low rate. It has been proven
that double reading of medical images could lead to
better tumor detection. But the cost incurred in
double reading is very high; therefore good software
to assist humans in medical institutions is of great
interest nowadays. Different approaches that can
produce medical images must be studied. Also, the
technique that produces those Fig. 1: Data extracting cycle
images is very important in order to know what Data mining in contrast to traditional data analysis is
to apply to a certain medical image in order to get discovery driven. Data mining is a young
better results. A lot of methods have been proposed interdisciplinary field closely connected to data
in the literature for CT (Computed Tomography), warehousing, statistics, machine learning, neural
434
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
networks and inductive logic programming. Data a number of basic problems concerning document
mining provides automatic pattern recognition and retrieval, data mining in the databases, and
attempts to uncover patterns in data that are difficult visualization of high-dimensional data.
to detect with traditional statistical III.PROBLEM STATEMENT
methods. Without data mining it is difficult to
realize the full potential of data collected within The conventional method in medicine for brain MR
healthcare organization as data under analysis is images classification and tumor detection is human
massive, highly dimensional, distributed and inspection. Operator-assisted classification methods
uncertain. are impractical for large amounts of data and are
also non-reproducible. MR images also always
Massive healthcare data needs to be converted contain a noise caused by operator performance
into information and knowledge, which can help which can lead to serious inaccuracies classification.
control, cost and maintains high quality of patient The MR images data is by nature, a huge, complex
care. Healthcare data includes Patient centric data and cognitive process. Accurate diagnosis of MR
and Aggregate data. images data is not an easy task and is always time
Recent progress in data storage and acquisition has consuming. In some extreme scenario, diagnosis
resulted in a growing number of enormous with wrong result and delay in delivery of a correct
databases. The information contained in these diagnosis decision could occur due to the
databases can be extremely interesting and useful; complexity and cognitive process of which it is
however the amount is too large for humans involved. The use of artificial intelligent techniques
to process manually. These databases store for instance, neural networks, fuzzy logic, neuro
information covering every aspect of human fuzzy has shown great potential in this field. With
activity. In the business world, the data bases are the involvement of soft computing, the pattern
created for marketing, investment, manufacturing, matching, classification and detection of algorithms
customers, products or transactions, and such which have direct applications in many medical
information can be stored both for accounting and problems have become much easier to be
analysis purposes. The domain of scientific research implemented and diagnosed. Hence, this project
is another area which makes heavy use of databases tries to find out how the neuro registration system or
to store information about, for example patients, the hybrid combination of neural network and
patient histories, surveys, medical registration techniques can be applied in
investigations. Medical databases contain data in a the classification and detection purposes,
variety of formats: images in the form of X-rays or particularly on the medical field.
scans, textual information to describe details of
diseases, medical histories, psychology reports, IV.DATA MINING
medical articles or various signals like EKG, ECG,
EEG, etc. This data does not need to be located on Data mining is a new discipline within the field of
the same system. It may be distributed amongst information retrieval that extracts interesting and
various disparate computers depending on the useful material from large data sets. In the broader
source of the data. Such heterogeneous data make sense, data mining is defined as part of knowledge
the process of information retrieval an even more discovery in databases (KDD) and draws on the
complex process. There is a substantial need for fields of statistics, machine learning,
signal/image processing and data mining tools pattern recognition and database management. The
in medical information processing systems and in general approach of KDD process is presented on
telemedicine that are flexible, effective and used figure 2. The KDD process involves the following
friendly. Telemedicine is defined as: The steps:
investigation, monitoring and management of 1. Selecting the target data (focusing on the part of
patients and education of patients and staff the data on which to make discoveries). The data
using systems which allow ready access to expert may consist from different type of information.
advice and patient information no matter of where 2. Data cleaning and preprocessing includes
the patient or relevant information is located. Such removal of noise, outliers, missing data (if they are
systems serve a dual role: they can assist medical not an object of the discovery).
professionals in medical research and in the clinic 3. Transforming data includes data reduction and
medical diagnosis and secondly, they can be used as projection in order to find and investigate useful
information sources for patients. Telemedicine faces features and reduce the data feature space
435
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
VI.DESIGN METHODOLOGY
436
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
used for filling missing values and elimination of categories: area- and feature based methods [8]
duplicates in the database. [11]. Area-based methods, a.k.a. template matching,
STEP 4:- Transformation of data in this step can be combine the feature-detection and feature-matching
defined as decreasing the dimensionality of the data steps. These methods first define a window in the
that is sent for data mining. Usually there are cases input image as the template and then search for a
where there are a high number of attributes in the corresponding template in the reference image. The
database for a particular case. With the reduction centers of matched image templates are used as
of dimensionality we increase the efficiency of the CPs. There are many ways to find a matching
data-mining step with respect to the accuracy and template, and one of the most popular criteria is
time utilization. cross correlation. Commonly used correlation-based
STEP 5:- The data mining step is the major step in methods include normalized cross
data KDD. This is when the cleaned and correlation (NCC), sequential similarity detection,
preprocessed data is sent into the intelligent and Fourier methods [8], [12]. Template-matching
algorithms for classification, clustering, similarity approaches using correlation based criteria, because
search within the data, and so on. Here we chose the of their ease of use and high efficiency, A
algorithms that are suitable for discovering patterns disadvantage of the correlation-based approach,
in the data. Some of the algorithms provide however, is that it only works well if the two images
better accuracy in terms of knowledge discovery have rather similar gray-level
than others. Thus selecting the right algorithms can characteristics. Therefore, the approach is not suited
be crucial at this point for co registration of images from different sensors.
STEP 6:- Interpretation. In this step the mined data For multi sensor images, the use of mutual
is presented to the end user in a human-viewable information which relies on statistical
format. This involves data visualization, which the dependences between the two images is proposed
user interprets and understands the discovered [10]. Feature-based registration methods rely on
knowledge obtained by the algorithms. salient structures in the images to be matched and
can be based on regions, lines, or points. Two
VII.IMAGE REGISTRATION METHODS critical steps in feature-based methods are feature
extraction and feature matching. There are
In most studies, multi angle image registration relies many kinds of feature-extraction algorithms, and
on the manual selection of homologue points in among them, point based feature-detection methods
each image [1], carried out by a skilled operator receive the most attention. Commonly used point-
familiar with the study area. However, the manual based feature-detection algorithms include the
selection of CPs is tiring, time consuming, prone to Harris operator, the Gabor filter or Gabor wavelet,
inconsistency, and may result in low and the scale-invariant feature transform (SIFT)
accuracy. Manual selection is also impractical in [11], [13][15]. Detected points are expected to be
situations where a large number of images need to distinct and ideally evenly spread over the image.
be registered. Thus, there are pressing needs for After features have been detected in the pair of
automatic image registration methods. A typical images, they need to be matched. Important feature
registration method can be divided into four steps. matching methods include invariant descriptor and
1) Feature detection: Detect features of salient and spatial relation and relaxation methods [8]. Most of
distinctive objects in the input image (the image to these methods are equipped with filtering criteria to
be registered) and the reference image. reject possible outliers. Generally speaking, area-
2) Feature matching: Establish the correspondence based methods attain higher accuracies than feature-
between features in the input image and reference based methods [14] and are particularly suited for
image. the registration of images acquired from the
3) Transformation model estimation: Set up a same sensor. However, it is difficult to apply area-
transformation model and estimate the model based methods directly on multi angle imagery.
parameters by the use of matched CPs. Geometric distortion is usually high for images
4) Image re sampling: Transform the input image acquired at large view angles. Hence,
according to the transformation model and the rigid transformation models such as affine and
chosen re sampling function [8]. projective transforms are not suitable. A viable
Many registration methods have been proposed. alternative is a non rigid registration model.
Depending on whether feature detection [step 1)] is However, for a non rigid model to work well,
involved, methods can be broadly classified into two an adequate number of quality CPs and a good
437
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
estimation of the model parameters are needed. developed to cope with real-world situations. They
Furthermore, because CPs are selected by first are mainly considered in the situations such as ill-
defining a window in the input image and then defined and noisy natural data.
searching for a match in the reference image,
one faces the problem of having to deal with a very Under these conditions, ANN computing methods
large search space (in the reference image). In order are more effective and economical than the
to solve this problem, a coarse-to-fine hierarchical traditional computation methods. In order to deal
strategy is usually adopted [14]. Using a pyramid with non-stationary data, the properties of the ANNs
approach, candidate matching locations are first should be made adaptive. In this case, the
identified from the images that have been performance of the system should improve with use,
downscaled to coarser resolutions. Then, the and the system should be able to capture and store
accuracy of the matched locations is gradually information, or to perform learning.
improved by moving up to finer
resolutions. Compared with area-based approaches, In addition, ANNs should be able to perform
feature-based methods are more robust to variation generalization, that is, able to deal with subsets of
in view angles and are therefore more suited for the problem domain that are not yet to be
multi angle image registration [16]. The main encountered. All these properties must be
disadvantage of feature-based methods is that appreciated and inherited in the context of ANNs.
the number of detected CPs may sometimes be low, Without these properties, ANNs are similar to mere
which will cause problems in the estimation of a look-up tables.
complex non rigid transformation such as a TPS or a
piecewise linear model. To tackle the IX.RESULTS
aforementioned problems, we propose an
automatic registration method that integrates the
merits of both area- and feature-based methods.
438
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Extracted Region
439
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Region filled Image [9] Ramiro Castellanos and Sunanda Mitra, Segmentation of
magnetic resonance images using a neuro-fuzzy algorithm. IEEE
Symposium on Computer-Based Medical Systems, 2000.
[10] K. Johnson, A. C. Rhodes, I. Zavorin, and J. Le Moigne,
Mutual information as a similarity measure for remote sensing
image registration, in Proc. SPIE Aerosense, Geo-Spatial Image
Data Exploitation II, Orlando, FL, Apr. 2001, vol. 4383, pp. 51
61.
[11] Y. Bentoutou, N. Taleb, K. Kpalma, and J. Ronsin, An
automatic image registration for applications in remote sensing,
IEEE Trans. Geosci. Remote Sens., vol. 43, no. 9, pp. 21272137,
Sep. 2005.
[12] J. P. Lewis, Fast template matching, in Proc. Vis. Interface,
1995, pp. 120123.
[13] L. Yu, D. Zhang, and E. J. Holden, A fast and fully
automatic registration approach based on point features for multi-
source remote-sensing images, Comput. Geosci, vol. 34, no. 7,
pp. 838848, Jul. 2008.
[14] G. Hong and Y. Zhang, Wavelet-based image registration
technique for high resolution remote sensing images, Comput.
X.CONCLUSION Geosci, vol. 34, no. 12, pp. 17081720, Dec. 2008.
440
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
441
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
processing time is required for the conversion. Due to authentication codes are accordingly computed and
quantization the retransformation may not result in inserted into selected random pixels.
the original image. So we are using spatial domain In Tzeng and Tsais method [8], randomly
transformation. generated authentication codes are embedded into
The given gray image is transformed using image blocks for use in image authentication, and a
radon transform. Radon is a spatial domain transform, so-called code holder is used to reduce image
where the pixels intensity will be oriented in a distortion resulting from data embedding.
particular angle. Since the transformed image is Lee et al. [9] proposed a Hamming-code-
different from the original image the confidentiality based data embedding method that flips one pixel in
will be maintained. each binary image block for embedding a watermark,
This gives the first level of confidentiality to yielding small distortions and low false negative rates.
the secret image. Then the byte run length code is Lee et al. [10] improved the method later by using an
applied to get the compressed transformed image. edge line similarity measure to select flippable pixels
This gives second level of confidentiality. Then apply for the purpose of reducing the distortion.
Shamirs secret sharing on this encoded data to get n
number of shares. This gives third level of Disadvantages:
confidentiality. With water marking only we are performing
security.
1.4 Model Diagram Pixels neighboring factor is too less here,
that means with association among pixels.
Working with document images is not
generating stego-images with desired quality.
3. Proposed System
In proposed system, a method for the
authentication of document images with an additional
self-repair capability for fixing tampered image data
is proposed. The input cover image is assumed to be a
binary-like grayscale with two major gray values.
After the proposed method is applied, the cover image
is transformed into a stego-image in the Portable
Network Graphics (PNG) format with an additional
Fig. 1.4 Model Diagram alpha channel for transmission on networks or
archiving in databases. The stego-image, when
2. Existing System received or retrieved, may be verified by the proposed
Several methods for binary image method for its authenticity. Integrity modifications of
authentication have been proposed in the past. Wu the stego-image can be detected by the method at the
and Liu [4] manipulated the so-called flippable pixels block level and repaired at the pixel level. In case the
to create specific relationships to embed data for alpha channel is totally removed from the stego-
authentication and annotation of binary images. image, the entire resulting image is regarded as
Yang and Kot [5] proposed a two-layer inauthentic, meaning that the fidelity check of the
binary image authentication method in which one image fails. The proposed method is based on the so-
layer is used for checking the image fidelity and the called -threshold secret sharing scheme proposed by
other for checking image integrity. In the method, a Shamir [11] in which a secret message is transformed
connectivity-preserving transition criterion for into shares for keeping by participants, and when of
determining the flippability of a pixel is used for the shares, not necessarily all of them, are collected,
embedding the cryptographic signature and the block the secret message can be losslessly recovered. Such a
identifier. Later, Yang and Kot [6] proposed a pattern- secret sharing scheme is useful for reducing the risk
based data hiding method for binary image of incidental partial data loss.
authentication in which three transition criteria are Conventionally, the concepts of secret
used to determine the flippabilities of pixels in each sharing and data hiding for image authentication
block, and the watermark is adaptively embedded into are two irrelevant issues in the domain of information
embeddable blocks to deal with the uneven security. However, in the proposed method, we
embeddability condition in the host image. In the combine them together to develop a new image
method proposed a set of pseudorandom pixels in a authentication technique. The secret sharing scheme
binary or halftone image are chosen and cleared, and is used in the developed technique not only to carry
442
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
authentication signals and image content data but also III. Generation, transmit of stego image
to help repair tampered data through the use of shares.
Here we are generating the stego-image for
Advantages: binarization to receiver. The stego-image, when
Loss less data is able to be transmitted received or retrieved, may be verified by the proposed
through the network. method for its authenticity. Integrity modifications of
Here authenticity operation is done with the stego-image can be detected by the method at the
stego image. block level and repaired at the pixel level. In case the
Repair operation is done successfully here alpha channel is totally removed from the stego-
and generates PNG Image image, the entire resulting image is regarded as
dynamically. inauthentic, meaning that the fidelity check of the
image fails.
Four Modules Description
IV. Binarization and verify image at receiver side
I. Gray scale image binarization from
sender side
After performing the binarization at the receiver side,
II. Appending alpha channel phase
the image is to be filtered from the alpha channel.
III. Generation , transmit of stego image
After stego image generation if there is no authentic
IV. Binarization and verify image at
process repair the tampered image blocks then
receiver side
remove the alpha channel. If the authentication is
success directly receive the PNG Image at the
I. Grayscale image binarization from sender side
receiver side.
If an image has had its background
normalized to a constant value, it can be binarized by
V.DISCUSSIONS
a global value chosen to be less than the background
by an appropriate amount. It must be less because the
In addition to being capable of data repairing and
background cannot be made exactly constant,
being blind in nature (requiring no overhead other
especially in situations where it was changing rapidly
than the stego-image), the proposed method has
or where there is bleed through from text or image on
several other merits, which are described in the
the other side of the page. Here we are performing
following.
binarization on an input document image.
5.1 Providing pixel-level repairs of tampered
II. Appending alpha channel phase
image parts: As long as two untampered partial
Appending the alpha channel to the binarized
shares can be collected, a tampered block can be
image at the time the pixels in the binarized is
repaired at the pixel level by the proposed method.
performed the associativity the neighbouring pixels.
This yields a better repair effect for texts in images
Alpha channel is an additional channel that can be
because text characters or letters are smaller in size
added to an image. It contains transparency
with many curved strokes and need finer pixel-level
information about the image, and depending on the
repairs when tampered with.
type of alpha it can contain various levels of
transparency. The alpha channel basically controls the
5.2 Having higher possibility to survive image
transparency of all the other channels.
content attacks: By skillfully combining the Shamir
scheme, the authentication signal generation, and the
By adding the alpha channel to an image you control
random embedding of multiple shares, the proposed
the transparency of the red channel, green channel
method can survive malicious attacks of common
and the blue channel. There are two types of alpha
content modifications, such as superimposition,
channels, pre-multiplied alpha and straight alpha.
painting, etc., as will be demonstrated by
Which alpha type you use depends on the software
experimental results subsequently described.
being used.
5.3 Making use of a new type of image channel for
data hiding: Different from common types of
images, a PNG image has the extra alpha channel
plane that is normally used to produce transparency to
the image. It is differently utilized by the proposed
method for the first time as a carrier with a large
space for hiding share data. As a comparison, many
Fig.5.1 Appending Alpha channel phase other methods use LSBs as the carriers of hidden data.
443
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
5.4 Causing no distortion to the input image: Hierarchical watermarking for secure image authentication
with localization, IEEE Trans. Image Process., vol. 11, no. 6,
Conventional image authentication methods that
pp. 585595, Jun. 2002.
usually embed authentication signals into the cover [3] Z. M. Lu, D. G. Xu, and S. H. Sun, Multipurpose image
image itself will unavoidably cause destruction to the watermarking algorithm based on multistage vector
image content to a certain extent. Different from such quantization, IEEE Trans. Image Process., vol. 14, no. 6, pp.
822831, Jun. 2005.
methods, the proposed method utilizes the pixels
[4] M. Wu and B. Liu, Data hiding in binary images for
values of the alpha channel for the purpose of image authentication and annotation, IEEE Trans. Multimedia, vol.
authentication and data repairing, leaving the original 6, no. 4, pp. 528538, Aug. 2004.
image (i.e., the grayscale channel) untouched and thus [5] H. Yang and A. C. Kot, Binary image authentication with
tampering localization by embedding cryptographic signature
causing no distortion to it. The alpha channel plane
and block identi-fier, IEEE Signal Process. Lett., vol. 13, no.
may be removed after the authentication process to 12, pp. 741744, Dec. 2006.
get the original image. [6] H. Yang and A. C. Kot, Pattern-based data hiding for binary
images authentication by connectivity-preserving, IEEE
Trans. Multimedia, vol. 9, no. 3, pp. 475486, Apr. 2007.
6. CONCLUSION [7] H. Y. Kim and A. A?f, Secure authentication watermarking
for halftone and binary images, Int. J. Imag. Syst. Technol.,
A new blind image authentication method with a data vol. 14, no. 4, pp. 147152, 2004.
repair capability for binary-like grayscale document [8] C. H. Tzeng and W. H. Tsai, A new approach to
authentication of binary images for multimedia
images based on secret sharing has been proposed.
communication with distortion reduc-tion and security
Both the generated authentication signal and the enhancement, IEEE Commun. Lett., vol. 7, no. 9, pp. 443
content of a block have been transformed into partial 445, Sep. 2003.
shares by the Shamir method, which have been then [9] Y. Lee, J. Hur, H. Kim, Y. Park, and H. Yoon, A new
binary image au-thentication scheme with small distortion and
distributed in a well-designed manner into an alpha
low false negative rates, IEICE Trans. Commun., vol. E90-
channel plane to create a stego-image in the PNG B, no. 11, pp. 32593262, Nov. 2007.
format. The undesired opaque effect visible in the [10] Y. Lee, H. Kim, and Y. Park, A new data hiding scheme for
stego-image coming from embedding the partial binary image authentication with small image distortion, Inf.
Sci., vol. 179, no. 22, pp. 38663884, Nov. 2009.
shares has been eliminated by mapping the share
[11] A. Shamir, How to share a secret, Commun. ACM, vol. 22,
values into a small range of alpha channel values near no. 11, pp. 612613, Nov. 1979.
their maximum transparency value of 255. [12] C. C. Lin and W. H. Tsai, Secret image sharing with
steganography and authentication, J. Syst. Softw., vol. 73,
no. 3, pp. 405414, Nov./ Dec. 2004.
7. FUTURE ENHANCEMENT
[13] W. H. Tsai, Moment-preserving thresholding: A new
approach, Comput. Vis. Graph. Image Process., vol. 29, no.
In the process of image block authentication, a block 3, pp. 377393, Mar. 1985.
in the stego-image has been regarded as having been
tampered with if the computed authentication signal
does not match that extracted from corresponding
partial shares in the alpha channel plane. For the self-
repairing of the content of a tampered block, the
reverse Shamir scheme has been used to compute the
original content of the block from any two
untampered shares. Measures for enhancing the
security of the data embedded in the alpha channel
plane have been also proposed. Experimental results
have been shown to prove the effectiveness of the
proposed method. Future studies may be directed to
choices of other block sizes and related parameters
(prime number, coefficients for secret sharing,
number of authentication signal bits, etc.) to improve
data repair effects. Applications of the proposed
method to the authentication and the repairing of
attacked color images may be also tried.
REFERENCES
444
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
445
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
3. CLASSIFICATION OF ALGORITHMS
where N is the side of the macro bock, Cij and Rij are
the pixels being compared in current macro block and
reference macro block, respectively.
Fig 2: Classification
4. SET UP PROCESS
peak to peak value of original data 2
PSNR 10 log10
MSE
The idea behind block matching is to divide the
current frame into a matrix of macro blocks that are
then compared with corresponding block and its 5. DIFFERENT ALGORITHMS
adjacent neighbors in the previous frame to create a
vector that stipulates the movement of a macro block
5.1 EXHAUSTIVE SEARCH (ES)
from one location to another in the previous frame.
Usually the macro block is taken as a square of side
16 pixels, and the search parameter p is 7 pixels.
446
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
This algorithm calculates the cost function at each algorithm is as follows. As mentioned above, NTSS
possible location in the search window. As a result of differs from TSS by full search assuming a center
which it finds the best possible match and gives the biased checking point pattern in the first step and
highest PSNR amongst any block matching NTSS incorporating a half-way stop technique for
algorithm. The obvious disadvantage to ES is that the stationary or quasi stationary blocks. Details are as
larger the search window gets the more computations follows.
it requires.
1. In addition to the original checking points
5.2 THREE STEP SEARCH (TSS) used in TSS, eight extra points are added,
which are the eight extra neighbors of the
It is a block based search technique and uses a search window (center-bias).
maximum of three steps. It is mainly used for Motion 2. A half-way stop technique is used for
Estimation in low bit rate video compression stationary and quasi stationary block in
applications such as videophone and video order to fast identify and then estimate the
conferencing. motions for these blocks.
a. If the minimum MBD in the first
step occurs at the search window
The TSS algorithm is a simple and effective method
enter, stop search (first-step-stop).
used for Motion Estimation. When compared with
Exhaustive Block Matching Algorithm (EBMA) it
has a saving factor of greater than a 100. It has a
fixed number of three search steps and a maximum
and minimum number of search points of 25. The
disadvantage of this technique is that it uses a
uniformly allocated checking point pattern in the first
step which makes it inefficient for the estimation of
small motion.
NTSS algorithm is a modification of the well-known For the search window , in the first step the N3SS
TSS algorithm. Since most real world motion algorithm utilizes a 3x3 grid search pattern at the
sequences are smooth and gentle, detecting small center in addition to the larger 9x9 grid in the 3SS.
motions becomes an important issue. One solution to
this could be to make the search adaptable to the If the minimum BDM point is found at the center of
uncertainty and motion scale. This is achieved by the search window, the search will stop. If the
employing a center-biased checking point pattern in minimum BDM point is one of the eight points on the
the first step. 3x3 grid, only additional five or three points will be
checked, otherwise the search window center will be
Flow chart moved to the winning point and the following
procedure is the same as in 3SS. The details are as
The block diagram shown in Fig. 5 shows the steps follows.
involved in the NTSS algorithm. The NTSS
447
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Step 1 If MAD (A) MAD (B) and MAD (A) < MAD (C),
II is selected;
Totally points are checked including the central nine
points on the 3x3 grid and the eight neighboring If MAD (A) < MAD (B) and MAD (A) < MAD (C),
pointsonthe 9x9grid.Iftheminimum BDM point is the III is selected;
search window center, the search will be terminated;
otherwise go to Step 2. If MAD (A) < MAD (B) and MAD (A) MAD (C),
IV is selected.
Step 2
Step 3
5.4 SIMPLE AND EFFICIENT SEARCH (SES) With a search pattern of +/-7, the FSS algorithm
utilizes a center-biased search pattern with nine
checking points on a 5 X 5 window in the first step,
The main idea behind the algorithm is that for a
instead of a 9 X 9 window as in the TSS. The center
unimodal surface there cannot be two minimums in
of the window is then shifted to the point with the
opposite directions and hence the 8 point fixed
minimum BDM. In addition to that, FSS reduces the
pattern search of TSS can be changed to incorporate
worst case search points from 33 to 27. Similar to
this and save on computations
NTSS, 4SS [6] also employs center biased searching
and has a halfway stop provision. 4SS sets a fixed
Consider the nth step. pattern size of S = 2 for the first step, no matter what
the search parameter p value is. Thus it looks at 9
In the rst phase, we compute the MADs of A, B, locations in a 5x5 window.
and C, where A is the center location, B and C are the
locations s(n) pixels away from A along the The FSS algorithm can be summarized as follows
horizontal and vertical directions, respectively.
Furthermore, we denote MAD (X) as the MAD of
1. A minimum BDM point if found from a nine
location X and label four quadrants I, II, III, and IV
checking point pattern on a 5 X 5 window
as shown.
located at the center of the 15 X 15
searching. If the minimum BDM is found at
The rule for determining a search quadrant can be the center of the search window, then go to
described as follows. step 4 else, go to step 2.
2. The search window size is maintained 5 X 5.
If MAD (A) MAD (B) and MAD (A) MAD (C), I However search pattern will depend on the
is selected; position of the previous minimum BDM. a If
the previous minimum BDM point is located
448
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
at the comer of pervious search window, performance that is close to that of full search with a
five additional checking points are used. significant speedup.
449
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Fig 10: Video 1:Comparison of no. of computations Fig 14: Video 2: Comparison of PSNR values
450
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
New
Three Step 26.0855 29.9879 29.5212
Search
Simple
and
Efficient
Search 16.1873 17 17
Four-Step
Fig 16: Video3: Comparison of no. of computations Search
19.9055 24.5909 23.9121
Diamond
18.8418
Search
28.7636 27.5242
Adaptive
Rood
Pattern
Search 10.3127 14.3364 15.5303
451
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Three [8] Chun-Ho Cheung, and Lai-Man Po, A Novel Small Cross-
Diamond Search Algorithm for Fast Video Coding and Video
Step Conferencing Applications, Proc. IEEE ICIP, September 2002.
Search 27.2403 31.4614 24.1004
New [9] Chun-Ho Cheung, and Lai-Man Po, A Novel Cross-Diamond
Three Search Algorithm for Fast Block Motion Estimation, IEEE Trans.
Step Circuits And Systems For Video Technology, vol 12., no. 12, pp.
Search 28.9353 31.9846 23.8800 1168-1177, December 2002.
Simple
and
Efficient
Search 27.011 30.8501 23.5710
Four-Step
Search
29.3966 31.6058 24.1317
Diamond
Search
29.4049 31.8737 23.9882
Adaptive
Rood
Pattern
Search 29.3871 31.8748 24.5474
REFERENCES
[7] Yao Nie, and Kai-Kuang Ma, Adaptive Rood Pattern Search
for Fast Block-Matching Motion Estimation, IEEE Trans. Image
Processing, vol 11, no. 12, pp. 1442-1448, December 2002.
452
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
I.INTRODUCTION
V. RESULTS
454
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
REFERENCES
[1] Ms. I.Suneetha and Dr.T.Venkateswarlu,
Enhancement Techniques for Gray scale Images in
Spatial Domain, International Journal of Emerging
Technology and Advanced Engineering, website:
www.ijetae.com(ISSN 2250-2459) Volume 2, Issue 4,
April 2012, pp.13-20.
[2] Ms. I.Suneetha and Dr.T.Venkateswarlu,
Enhancement Techniques for True Color Images in
Spatial Domain, International Journal of Computer
Science & Technology (IJCST), Website:
www.ijcst.com(ISSN 0976-8491) Volume 3, Issue 2,
Version 5, April - June 2012, pp. 814-820.
[3] Ms. I.Suneetha and Dr.T.Venkateswarlu, Image
Enhancement Through Noise Suppression Using
Nonlinear Parameterized Adaptive Recursive Model,
International Journal of Engineering Research and
Applications (IJERA), Website: www.ijera.com (ISSN
2248-9622), Volume 2, Issue 4, July-August 2012, pp.
1129-1136.
[4] Ms. I.Suneetha and Dr.T.Venkateswarlu, Image
Visual inspection of subjective results Enhancement Through Contrast Improvement Using
indicates that, the Local Parametric Hybrid Model Parameterized Gradient Intercept Model, ARPN Journal
works very well by enhancing gray scale and true of Engineering and Applied Sciences (ARPN-JEAS),
color images that are corrupted by decreased contrast Website:www.arpnjournals.com (ISSN 1819-6608),
Volume 7, No. 8, August 2012.
and unwanted noise. Visual inspection of objective [5] J Rafael C Gonzalez, Richard E. Woods, and Steven L.
results shows that, MSE was decreased and PSNR Eddins, Digital Image Processing Using MATLAB
was increased for the gray scale images and also for (Second Edition, Gates mark Publishing, 2009).
the R, G, and B components of true color images. The [6] J. Y. im, L. S. Kim, S. H Hwang, An advanced Contrast
Enhancement Using Partially Overlapped SubBlock
limitation in the proposed model is small decrement Histogram Equalization, IEEE Transactions on Circuits
in mse and small improvement in PSNR for and Systems for Videc Technology, Vol. 11, No. 4,
enhancing gray scale and true color images that are pp.475-484,2001.
corrupted by increased contrast and unwanted noise. [7] Prof. A. Senthilrajan, Dr. E. Ramaraj, High Density
Impulse Noise Removal in Color Images Using Region of
The reason is, while increasing contrast for getting Interest Median Controlled Adaptive Recursive Weighted
simulation results some pixel values reach maximum Median Filter, Proceedings of the International
value which are then treated as salt during denoising. MultiConferenceof Engineers and Computer Scientists
Therefore resultant images are not having very good (IMECS), Vol. II, March 17-19, 2010, Hong Kong.
contrast. This problem can be overcome by slight
change in LPHM algorithm.
455
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
I.INTRODUCTION
456
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
moving video object. Figure 2.2 shows an example of 5. If we determine the locations of curvature zero
a frame from a video shot and its corresponding crossings of every during evolution. We can
extracted shape contour. display the resulting points in (u, ) plane.
Where u is normalized arc length and is width
or standard deviation of the Gaussian kernel.
6. The process starts with =1 & at each level is
increased by , chosen as 0.1. As increases,
repeat steps 3 to 5, then shrinks and becomes
smoother & the number of curvature zero
crossing points on it decreases.
7. Finally when is sufficiently high will be a
convex curve with no curvature zero crossing
Figure 2.2. (a) Sample frame from a video shot (b) points.
Extracted shape contour from frame (a). 8. The result of this process can be represented as a
binary image called the CSS image.
After feature extraction, the shape and motion 9. We find the peaks of both branches of a contour
trajectory of the video object are represented using in the CSS image and consider the midpoint of
two different techniques standardized in MPEG-7: the line segment joining the pair as a maximum
Curvature Scale Space for shapes and Polynomial of the CSS image. The maxima of CSS image is
Curve fitting for motion trajectory. The shape and indicates shape representation. In our system, if a
motion features are stored in the video database. For maximum is less than 1/6 of the largest
a query video, shape and motion features are maximum of the same CSS image, it is
extracted from the query. The represented query considered as noise or small ripples.
shape is matched with the shape feature of a model
from the database using CSS based matching. The CSS image
features of the object in the query are matched against
the features of the stored objects in the database. The The CSS representation is robust with respect to
features of the object in the query are matched against scale, noise and change in orientation. A rotation of
the features of the stored objects in the database. The the object usually causes a circular shift on its
video shots corresponding to the video objects are representation, which is easily determined during the
rank ordered and retrieved according to their matching process (compare Fig.3.1a and b). Note that
computed mathcosts. the effect of a change in the starting point is also the
same. Due to arc length normalization, scaling does
III.SHAPE REPRESENTATION AND MATCHING not change the representation, and as Fig. 3.1 shows,
noise may create some small contours on the CSS
Algorithm for Shape Representation image, but the main contours and therefore the
corresponding maxima remain unaffected.
1. Every object contour is represented by x and y Compactness is another aspect of the CSS
coordinates of its boundary points representation. A shape is represented by about less
2. To normalize the arc length, the boundary is than ten pairs of integer values which can be
resample and represented by 200 equally distant determined without any ambiguity.
points.
3. The curve is then smoothed by convolution of
contour with a Gaussian function. The smoothed
curve is called , where denotes the width of
the Gaussian kernel g(u, ) here u is the
normalized arc length, is the width of the
Gaussian kernel
g(u, ) = e
(1)
4. To locations of curvature zero crossings on
are determined at different levels of scaling using
Fig 3.1: a) A boundary and its CSS image. b) Change
( ,) ( ,) ( ,) ( ,)
k(u, ) = in orientation causes a circular shift in CSS image. c)
( ( ,) ( ,) )
Noise creates small contours in CSS
(2)
457
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
The matching algorithm which compares two sets of 5. To expand a node, select the largest scale image
representations and assigns a match value as the curve CSS maximum (which is not in the first list)
measure of similarity between the shapes is also and apply that nodes shift parameter to map that
simple and fast. Another property of the CSS image maximum to the model CSS image. Locate the
is that it retains the local properties of the shape. nearest model curve CSS maximum (which is not in
Every contour of the CSS image corresponds to a the second list). If the two maxima are in a reasonable
concavity or a convexity of the shape. A local horizontal distance (0.2 of the maximum possible
deformation of the shape mainly causes a change in distance), define the cost of the match as the straight-
the corresponding contour of the CSS image. Using line distance between the two maxima. Otherwise,
this property, one can include more local information define the height of the image curve CSS maximum
about the shape in the CSS image. For example, it is as the cost of the match. If there are no more image
possible to use the average curvature of a segment curve CSS maxima left, define the cost of match as
together with the maximum of its CSS contour. the height of the highest model curve CSS maximum
Alternatively, one can find the corresponding corners not in the nodes second list. Likewise, if there is no
of each pair of curvature zero-crossings. In shape more models curve CSS maxima left; define the cost
similarity retrieval, the task is not to accurately of match as the height of the selected image curve
recognize the input among the existing models, but it maximum. Note that these cases may occur when the
is to find the most similar models to the input and number of image maxima is different from the
rank them in terms of a similarity measure. The CSS number of model maxima. Add the match cost to the
representation is a reliable tool to handle this task. node cost. Update the two lists associated with the
node.
CSS matching 6. Select the lowest cost node. If there are no more
model or image curve CSS maxima that remain
The complete matching algorithm which compares unmatched within that node, then return that node as
the two sets of maxima, one from the image and the the lowest cost node. Otherwise, go to step 5 and
other from the model is as follows. expand the lowest cost node.
1. Create a node consisting of the largest scale 7. Reverse the place of the image and the model and
maximum of the image and the largest scale repeat steps 1 to 6 to find the lowest cost node in this
maximum of the model. Initialize the cost of this case.
node to the absolute difference of -coordinates of 8. Consider the lowest node as the final matching cost
the image and the model. Compute a CSS shift between the image and the model. Using this
parameter for each node: algorithm and considering its amendment, the system
= Um Ui ; where U is the horizontal coordinate of associates a matching value to every candidate and
a maximum, and i and m refer to image and model, then displays the n best matched as its output, where
respectively. This parameter is used to compensate n has already been selected by the user.
the effect of different start points or change in
orientation. IV.TRAJECTORY REPRESENTATION
2. If there are more than one maximum in the model
which have a -coordinate close (within 80%) to the The motion trajectory of a video object is extracted
largest scale maximum of the image, create extra from the video shot by finding the centroid of the
nodes consisting of the largest scale maximum of the moving object. A Polynomial curve is fitted for every
image and that respective additional maximum of the sub trajectory [7]. The sequence of steps involved for
model. Also create the same nodes for the second trajectory representation and matching are briefly
largest scale maximum of the image and the explained below.
respective maxima of the model. Initialize the cost Algorithm for Trajectory Representation
and compute the CSS shift parameter for each node
accordingly. 1. A sampling method is first used to detect all
3. Create two lists for each node obtained in steps 1 key points with high curvatures from the
and 2. The first list will contain the image curve trajectory for capturing its geometrical features.
maxima and the second list will contain the model 2. High curvature points are points on the
curve maxima matched within that node at any point trajectory whose angle () is less than a certain
of the matching procedure. Initialize the first and threshold (T). Angle () at a key point is
second list of each node by the corresponding computed by selecting two specified points,
maxima determined in the first two steps. which lie on either side within a certain range
4. Expand each node created in steps 1 and 2 using of distance (dmin and dmax) from the keypoint.
the procedure described in step 5. 3. As we start from the initial key point of the
trajectory, whenever there is a clockwise
458
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
direction change followed by an anticlockwise into two sub trajectories (c) Key points regenerated
direction change or vice versa, the trajectory is (d) Polynomial curve fitted on the input trajectory.
divided into sub trajectories at that key point.
4. To improve the accuracy of curve fitting, if These key points are used as control points for curve
distance between any two control points are fitting. To improve the accuracy of curve fitting, if
greater than some threshold (> Td) in a sub distance between any two control points are greater
trajectory, their middle point is selected as a than some threshold (> Td) in a sub trajectory, their
new control point. middle point is selected as a new control point. A
5. A polynomial curve is fitted through the polynomial of second order is fitted through the
control points for every sub trajectory. control points for every sub trajectory as shown in
V.IMPLEMENTATION OF POLYNOMIAL dotted lines in Fig. 4d. The coefficients of the
CURVE FITTING polynomials along with their temporal occurrences
are stored for every sub trajectory.
The motion trajectory of a video object is extracted
from the video shot by finding the centroid of the Trajectory matching between the query (q) and model
moving object. Polynomial curve fitting is used for (m) in the database is done based on their position,
trajectory representation as standardized in MPEG-7 velocity and size differences. For similarity
[6]. Trajectory is divided into sub trajectories based comparison, the sampled data points from fitted
on the key points detected as described in [7]. A polynomial curve are normalized to maintain scale
Polynomial curve is fitted for every sub trajectory. and shift invariance. Position difference (Cp) is
The sequence of steps involved for trajectory calculated from the Euclidean distance between the
representation and matching are briefly explained positions of query and model trajectories as:
below. A sampling method is first used to detect all
key points with high curvatures from the trajectory Trajectory matching
for capturing its geometrical features. High curvature
points are points on the trajectory whose angle () is Trajectory matching between the query (q) and model
less than a certain threshold (T). Angle () at a key (m) in the database is done based on their position,
point is computed by selecting two specified points, velocity and size differences.
which lie on either side within a certain range of
distance (dmin and dmax) from the keypoint. The Position difference (Cp) is calculated from the
high curvature points are further eliminated for better Euclidean distance between the positions of query
curve fitting such that two key points lie far from and model trajectories as:
each other (> dmin). The direction change at the key N 1
1
points is detected to divide the trajectory into sub Cp
N
{( x
K 0
q (k ) xm (k )) 2 ( y q (k ) ym (k )) 2 }
trajectories. Clockwise or anticlockwise direction
change is determined using the two adjacent key (1)
points which lie on both sides of the key point. As we Similarly, velocity difference (Cv) is given by,
N 1
1
{(v (k ) v
start from the initial key point of the trajectory,
whenever there is a clockwise direction change Cv x
q
x
m (k )) 2 (vqy (k ) vmy (k )) 2 }
N K 0
followed by an anticlockwise direction change or vice
versa, the trajectory is divided into sub trajectories at (2)
that key point. This segmentation is done for accurate Where vx and vy are the x and y components of the
representation and easier partial trajectory matching. velocity vector and N is the number of sampled
Figure 4.1 shows an example trajectory. High points. Size difference (Csz) is given by,
curvature points are marked on the input trajectory in 2 | size(q ) size(m) |
Csz
Fig. 4.1 a. The segmented sub trajectories with their size(q ) size(m)
key points are shown in Fig. 4.1 b.
(3)
Where size (q) = max (max (xq) min (xq), max (yq)
min (yq)). The three distances are integrated to give
the dissimilarity between the trajectories as follows:
Ct = wt1 Cp + wt2 Cv + wt3 Csz
(4)
Figure 4.1. (a) Input trajectory with key (high
curvature) points marked (b) Trajectory segmented
459
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
Integration of shape and motion features is done to The precision-recall graph averaged over 200 queries
retrieve video shots with similar shape and motion. for the three methods is shown in Fig. 5.3. The
The matchcosts obtained from CSS matching for precision-recall graph of our method outperforms the
shape (Cs) and trajectory matching (Ct) are other two methods based on only shape [6] or
normalized and summed up with appropriate weights trajectory [7].
as follows:
Precision: In the field of video retrieval, precision is
C = ws *Cs + wt * Ct the fraction of video that are relevant to the search. A
(5) good retrieval system should only retrieve relevant
items.
ws and wt are weights assigned for shape and
trajectory respectively.
Figure 5.1 : boundary and its CSS image for different (b)
values
Figure 5.2: a) Query video b) Retrieved videos
460
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
REFERENCES
461
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
462
International Conference on Navigational Systems & Signal Processing Applications, December 13th & 14th,2013
University College of engineering & Technology, Acharya Nagarjuna University, Nagarjuna Nagar-522510,A.P.,India
463
With Best Complements from
With Best Complements from
With Best Complements from
With Best Complements from