optimization of cooperative spectrum sensing in … · iii declaration i declare that the thesis...

OPTIMIZATION OF COOPERATIVE SPECTRUM

SENSING IN COGNITIVE RADIO

A Thesis submitted to Gujarat Technological University

for the Award of

Doctor of Philosophy

In

Electronics & Communication Engineering

by

Keraliya Divyesh R [129990911004]

under supervision of

Dr. Ashalata Kulshrestha

GUJARAT TECHNOLOGICAL UNIVERSITY

AHMEDABAD

[August 2018]

ii

© [Keraliya Divyesh Rudabhai]

iii

DECLARATION

I declare that the thesis entitled Optimization of Cooperative Spectrum Sensing in

Cognitive Radio submitted by me for the degree of Doctor of Philosophy is the record of

research work carried out by me during the period from June 2012 to August 2017 under

the supervision of Dr. Ashalata Kulshrestha and this has not formed the basis for the

award of any degree, diploma, associateship, fellowship, titles in this or any other

University or other institution of higher learning.

I further declare that the material obtained from other sources has been duly acknowledged

in the thesis. I shall be solely responsible for any plagiarism or other irregularities, if

noticed in the thesis.

Signature of Research Scholar: ……………………… Date: 10/08/2018

Name of Research Scholar: Keraliya Divyesh Rudabhai

Place: Ahmedabad

iv

CERTIFICATE

I certify that the work incorporated in the thesis Optimization of Cooperative Spectrum

Sensing in Cognitive Radio submitted by Mr. Keraliya Divyesh Rudabhai was carried

out by the candidate under my supervision/guidance. To the best of my knowledge: (i) the

candidate has not submitted the same research work to any other institution for any

degree/diploma, Associateship, Fellowship or other similar titles (ii) the thesis submitted

is a record of original research work done by the Research Scholar during the period of

study under my supervision, and (iii) the thesis represents independent research work on

the part of the Research Scholar.

Signature of Supervisor:……………………………. Date: 10/08/2018

Name of Supervisor: Dr. Ashalata Kulshrestha

Place: Ahmedabad

v

Course-work Completion Certificate

This is to certify that Mr. Keraliya Divyesh Rudabhai enrolment no. 129990911004 is a

PhD scholar enrolled for PhD program in the branch Electronics and Communication of

Gujarat Technological University, Ahmedabad.

(Please tick the relevant option(s))

He/She has been exempted from the course-work (successfully completed during

M.Phil Course)

He/She has been exempted from Research Methodology Course only (successfully

completed during M.Phil Course)

He/She has successfully completed the PhD course work for the partial

requirement for the award of PhD Degree. His/ Her performance in the course

work is as follows-

Grade Obtained in Research

Methodology

(PH001)

Grade Obtained in Self Study Course

(Core Subject)

(PH002)

AB BB

Supervisor‟s Sign

(Dr. Ashalata Kulshrestha)

vi

Originality Report Certificate

It is certified that PhD Thesis titled Optimization of Cooperative Spectrum Sensing in

Cognitive Radio submitted by Mr. Keraliya Divyesh Rudabhai has been examined by

us. We undertake the following:

a. Thesis has significant new work/knowledge as compared already published or are under

consideration to be published elsewhere. No sentence, equation, diagram, table, paragraph

or section has been copied verbatim from previous work unless it is placed under

quotation marks and duly referenced.

b. The work presented is original and own work of the author (i.e. there is no plagiarism).

No ideas, processes, results or words of others have been presented as Author own work.

c. There is no fabrication of data or results which have been compiled/analyzed.

d. There is no falsification by manipulating research materials, equipment or processes, or

changing or omitting data or results such that the research is not accurately represented in

the research record.

e. The thesis has been checked using https://turnitin.com (copy of originality report

attached) and found within limits as per GTU Plagiarism Policy and instructions issued

from time to time (i.e. permitted similarity index <=25%).

Signature of Research Scholar:…………………………….. Date: 10/08/2018


Place: Ahmedabad

Signature of Supervisor:…………………………… Date: 10/08/2018


Place: Ahmedabad

vii

Copy of Originality Report

viii

PhD THESIS Non-Exclusive License to GUJARAT

TECHNOLOGICAL UNIVERSITY

In consideration of being a PhD Research Scholar at GTU and in the interests of the

facilitation of research at GTU and elsewhere, I, Keraliya Divyesh Rudabhai having

Enrollment No. 129990911004 hereby grant a non-exclusive, royalty free and perpetual

license to GTU on the following terms:

a) GTU is permitted to archive, reproduce and distribute my thesis, in whole or in part,

and/or my abstract, in whole or in part (referred to collectively as the “Work”) anywhere

in the world, for non-commercial purposes, in all forms of media;

b) GTU is permitted to authorize, sub-lease, sub-contract or procure any of the acts

mentioned in paragraph (a);

c) GTU is authorized to submit the Work at any National / International Library, under the

authority of their “Thesis Non-Exclusive License”;

d) The Universal Copyright Notice (©) shall appear on all copies made under the authority

of this license;

e) I undertake to submit my thesis, through my University, to any Library and Archives.

Any abstract submitted with the thesis will be considered to form part of the thesis.

f) I represent that my thesis is my original work, does not infringe any rights of others,

including privacy rights, and that I have the right to make the grant conferred by this non-

exclusive license.

g) If third party copyrighted material was included in my thesis for which, under the terms

of the Copyright Act, written permission from the copyright owners is required, I have

obtained such permission from the copyright owners to do the acts mentioned in paragraph

(a) above for the full term of copyright protection.

ix

h) I retain copyright ownership and moral rights in my thesis, and may deal with the

copyright in my thesis, in any way consistent with rights granted by me to my University

in this non-exclusive license.

i) I further promise to inform any person to whom I may hereafter assign or license my

copyright in my thesis of the rights granted by me to my University in this non- exclusive

license.

j) I am aware of and agree to accept the conditions and regulations of PhD including all

policy matters related to authorship and plagiarism.

Signature of Research Scholar:………………………………


Date: 10/08/2018 Place: Ahmedabad

Signature of Supervisor:……………………………...


Date: 10/08/2018 Place: Ahmedabad

Seal:

x

Thesis Approval Form

The viva-voce of the PhD Thesis submitted by Mr. Keraliya Divyesh Rudabhai (Enrol

No. 129990911004) entitled Optimization of Cooperative Spectrum Sensing in

cognitive radio was conducted on 10/08/18, Friday at Gujarat Technological University.

(Please tick any one of the following option)

The performance of the candidate was satisfactory. We recommend that he be

awarded the PhD degree.

Any further modifications in research work recommended by the panel after 3

months from the date of first viva-voce upon request of the Supervisor or request

of Independent Research Scholar after which viva-voce can be re-conducted by

the same panel again.

The performance of the candidate was unsatisfactory. We recommend that he

should not be awarded the PhD degree.

-------------------------------------------------- ------------------------------------------------

Name and Signature of Supervisor with Seal External Examiner -1 Name and Signature

-------------------------------------------------- ------------------------------------------------

External Examiner -2 Name and Signature External Examiner -3 Name and Signature

(Briefly specify the modifications suggested by the panel)

(The panel must give justifications for rejecting the research work)

xi

Abstract

Cognitive radio (CR) is a new paradigm in the area of wireless communication system for

effective utilization of radio frequency (RF) spectrum. Cognitive Radio (CR) is used to

maximize the available spectrum utilization. When the primary user (PU) does not access

the channel, at that time cognitive radio (CR) will allow using the spectrum to

communicate with other CR. Spectrum sensing is the principal task of cognitive radio

through which it accurately determines the licensed user‟s existence (signal) and identifies

the available vacant spectrum. Cooperative spectrum sensing (CSS) has been proven to be

an effective method to improve the detection performance and mitigate the impact of

multipath fading, hidden terminal problem and receiver uncertainty issues in local

spectrum sensing.

In cooperative spectrum sensing (CSS), sharing of local spectrum sensing result between

the cognitive radio and the fusion center (FC) is challenging process for which the

performance of cooperative spectrum sensing is decided. The detection performance of

CSS highly depends on the quality of local spectrum sensing result as well as the quality

of the local sensing data collected by fusion center (FC) via reporting channel. The local

spectrum sensing information sent by the CR users is collected at the Fusion center (FC)

by conventional soft decision fusion (SDF) techniques or conventional hard decision

fusion (HDF) techniques. Conventional soft decision fusion (SDF) has excellent detection

performance, but it increases cooperative overhead and consumes high bandwidth of the

control channel for reporting. Similarly, the overhead of conventional hard decision fusion

(HDF) scheme is only one bit, but it has poor detection performance due to less amount of

data used for reporting the sensing result. Therefore, decision logic at the fusion center

(FC) and the size of data used for reporting of sensing result to fusion centre (FC) affect

the system performance.

To reduce this overhead between CR users and FC, we use 2-bit softened hard (quantized)

fusion technique at fusion centre (FC). In conventional HDF scheme which has single

threshold while in this 2-bit scheme, three thresholds and divide the entire range

of the observed energy into four regions of equal in size. In this framework, each CR user

detect the spectrum locally and sends its 2-bit information “quantized observation” in the

xii

form of which is the index of energy level to indicate the location of its observed

energy in that region to the fusion centre (FC) rather than sending the exact sensing

observation to FC. The fusion centre makes final decision as per number of observation in

respective energy level and weight vector of corresponding energy level. Here weight is

given to energy level, not to the CR user.

In this research work, use of teaching learning based optimization (TLBO) algorithm as a

significant method is proposed to optimize the weighting coefficients vector of observed

energy level of sensing information. The TLBO technique evaluate optimal weighting

coefficient vector so that the detection probability is improved for given false alarm

rate under the Neyman-Pearson criteria and minimize overall probability of sensing

error under the Mini-Max criteria. The performance of the proposed TLBO based

cooperative spectrum sensing framework is extensively analysed and compared with

conventional HDF and SDF based CSS schemes as well as other optimization techniques

i.e. particle swarm optimization (PSO) and genetic algorithm (GA) based CSS through

simulations. Simulation result shows that performance of TLBO based method is better

than conventional HDF scheme i.e. AND, OR, MAJORITY etc. and close to conventional

SDF scheme i.e. EGC with low overhead in various fading channel like AWGN, Rayleigh

and Nakagami. Proposed TLBO based CSS scheme is effective and stable and also shows

better convergence output which confirms the computation complexity is lower compared

to GA and PSO based method. Moreover, the strength of proposed method is also

analysed under the false reporting node and imperfect reporting channel.

Finally, an analytical evaluation for performance of proposed method while taking into

consideration of some realistic issues, such as multi-hop case for cooperative spectrum

sensing and the sensing-throughput trade-off is also presented in this research work.

Keywords: Cognitive radio, Cooperative spectrum sensing, TLBO, Softened Hard fusion,

Energy detection, Fading channel etc...

xiii

Acknowledgement

This thesis would have been impossible without the support of many people. I would

express my sincere gratitude to them, for their invaluable support deserves much more

than this short note of appreciation. I would like to thanks, almighty the God giving me

strength and passion for doing research.

I would like to express my appreciation to my supervisor Dr. Ashalata Kulshrestha for

his valuable enthusiastic guidance and encouragement me during every step of my

research. Due to his extensive technical support able complete my research in time. I

whole heartedly give my best wishes to him and his family.

Also, I would like to express my gratitude to Dr. C. H. Vithalani and Dr. K. R. Parmar.

It was a great honor to have them as my thesis Doctoral Progress Committee Member.

Their constructive suggestions made the thesis sound in many aspects.

I also acknowledge Honourable Vice Chancellor, Registrar, Controller of Examination,

Dean Ph.D. section and all staff members of Ph.D. Section of Gujarat Technological

University (GTU) for their assistance and support.

I would like to address special thanks to the reviewers of my thesis, for accepting to read

and review this thesis and giving approval of it. I would like to appreciate all the

researchers whose works I have used, initially in understanding my field of research and

later for updates. I would like to thank the many people who have taught me starting with

my school teachers, my undergraduate teachers, and my post graduate teachers.

Finally, I give the greatest respect and love to my parents, my parents in law, my wife and

my daughters Prexa and Yami. I want to express my highest appreciation for their

support and cooperation. I would like to say thanks to my wife Mrs. Rita for encouraging

me to do research and her moral support. Thanks to Almighty the God for giving me

patiently to complete research.

xiv

This thesis is dedicated to all of my good wishers with my love for assisting me to achieve

the most important stage in my life. I will never let them down and wish them all the

successes in the future.

As it is a prolonged journey, maybe I have forgotten few names to consider, but

nonetheless, they remain a core part of this mammoth task, and I seek forgiveness for the

same and offer my kind respect to every one of them.

Thanking you,

Keraliya Divyesh R

xv

Table of Contents

Abstract ..................................................................................................................................... xi

List of Abbreviations ........................................................................................................... xviii

List of Symbols ........................................................................................................................ xx

List of Figures ......................................................................................................................... xxi

List of Tables ........................................................................................................................ xxiii

1 Introduction .......................................................................................................................... 1

1.1 Introduction .................................................................................................................... 1

1.2 Motivation ...................................................................................................................... 1

1.3 Cognitive Radio Technology .......................................................................................... 3

1.4 Definition of Problem ..................................................................................................... 6

1.5 Objective and Scope of Work ......................................................................................... 7

1.6 Original Contribution by the Thesis ............................................................................... 7

1.7 Thesis Organization ........................................................................................................ 8

2 Background of Spectrum Sensing for CR ........................................................................ 10

2.1 Introduction .................................................................................................................. 10

2.2 Hypothesis Testing ....................................................................................................... 11

2.3 Probability of Detection Over Fading Channel ............................................................ 13

2.3.1 Rayleigh Fading Channel.................................................................................... 14

2.3.2 Rician Fading Channel........................................................................................ 14

2.3.3 Nakagami Fading Channel.................................................................................. 15

2.4 Detection Criteria of Primary User ............................................................................... 16

2.4.1 Neyman-Pearson Criteria .................................................................................... 16

2.4.2 Mini-Max Criteria ............................................................................................... 17

2.5 Sensing Technique ........................................................................................................ 18

2.5.1 Energy Detection ................................................................................................ 18

2.5.2 Matched Filtering ................................................................................................ 19

2.5.3 Cyclostationary Feature Detection ...................................................................... 19

2.5.4 Other Sensing Techniques .................................................................................. 20

2.6 Impact of Wireless Propagation on Spectrum Sensing ................................................. 20

2.7 Cooperative Spectrum Sensing (CSS) .......................................................................... 22

2.8 CSS Related Mathematical Statistics............................................................................ 23

2.9 Data Fusion Schemes for CSS ...................................................................................... 24

xvi

2.9.1 Hard Combining Fusion (HDF) Scheme ............................................................ 24

2.9.2 Soft Decision Fusion (SDF) Scheme .................................................................. 26

2.9.3 Softened Hard (Quantized) Data Fusion ............................................................. 27

2.10 Sensing Errors and Overhead of Cooperation .............................................................. 28

3 Literature Review .............................................................................................................. 29

3.1 Introduction .................................................................................................................. 29

3.2 Literature Reviewed on Spectrum Sensing & CSS ...................................................... 30

3.3 Related Work on Softened Hard Data Fusion Based Cooperative Sensing ................. 33

3.4 Research Gap ................................................................................................................ 34

3.5 Summary of Literature Reviewed ................................................................................. 35

4 Optimization Techniques for Cooperative Spectrum Sensing ....................................... 36

4.1 Introduction .................................................................................................................. 36

4.2 Types of Optimization Problem.................................................................................... 37

4.2.1 Single Objective Problem ................................................................................... 37

4.2.2 Multi Objective Problem .................................................................................... 38

4.3 Classification of Optimization Technique for CSS ...................................................... 40

4.3.1 Classical (Traditional) Optimization Techniques ............................................... 41

4.3.2 Meta Heuristic (Advance) Optimization Techniques ......................................... 41

4.4 Teaching–Learning Based Optimization ...................................................................... 46

4.4.1 Introduction ......................................................................................................... 46

4.4.2 Teacher Phase ..................................................................................................... 48

4.4.3 Learner Phase ...................................................................................................... 50

4.4.4 Unconstrained Optimization Problem................................................................. 50

4.4.5 Comparison of TLBO Result with Other Optimization Techniques .................. 53

5 TLBO Assisted CSS for Maximizing the Probability of Detection ................................ 54

5.1 Introduction .................................................................................................................. 54

5.2 Proposed TLBO Based System Model ......................................................................... 55

5.2.1 Softened Hard Decision Fusion Scheme............................................................. 56

5.2.2 Mathematical Statistics of and for TLBO Based CSS .............................. 58

5.2.3 Mathematical Statistics of and with False Report ..................................... 59

5.2.4 Mathematical Statistics of and for Single Hope Imperfect Channel ........ 60

5.2.5 Mathematical Statistics of and for Multi Hope Imperfect Channel ......... 61

5.3 Optimization Problem ................................................................................................... 61

5.4 TLBO Based Solution .................................................................................................. 62

xvii

5.5 TLBO Assisted Weighted Fusion Algorithm for ..................................................... 63

5.6 Simulation Results and Discussion ............................................................................... 64

5.6.1 System Parameters and Performance Metrics..................................................... 64

5.6.2 Performance Analysis on Under Fading Channels ........................................ 64

5.6.3 Influence of Number of Cooperating Users (N) on ....................................... 69

5.6.4 Influence of Signal to Noise ratio (SNR) on .................................................. 70

5.6.5 Convergence Performance of TLBO Based CSS on ..................................... 71

5.6.6 Performance Analysis on Against False Report ............................................ 77

5.6.7 Influence of Number of False Report on ....................................................... 78

5.6.8 Performance Analysis on Against Imperfect Channel ................................... 79

5.6.9 Influence of Reporting Channel Error on ...................................................... 80

5.7 Chapter Summary ......................................................................................................... 81

6 TLBO Assisted CSS for Minimizing the Cooperative Sensing Error ........................... 82

6.1 Introduction .................................................................................................................. 82

6.2 Proposed TLBO Based System Model ......................................................................... 83

6.3 Mathematical Statistics of for TLBO Based CSS .................................................... 84

6.4 Optimization problem for ......................................................................................... 84

6.5 TLBO Assisted Weighted Fusion Algorithm for ..................................................... 85

6.6 Sensing-Throughput Trade-off Analysis ...................................................................... 86

6.6.1 Constant Primary User Protection (CPUP) ......................................................... 86

6.6.2 Constant Secondary User Spectrum Utilization (CSUSU) ................................. 87

6.7 Trade-off Map for Cooperative Spectrum Sensing Parameter ..................................... 87

6.8 Simulation Result and Discussion ................................................................................ 88

6.8.1 Performance Analysis of TLBO Based CSS on ............................................. 88

6.8.2 Influence of Sensing Time on Throughput .................................................... 90

6.8.3 Convergence Performance of TLBO Based CSS on ...................................... 91

6.9 Chapter Summary ......................................................................................................... 98

7 Conclusions and Scope of Future Work........................................................................... 99

7.1 Conclusion .................................................................................................................... 99

7.2 Scope of Future Work ................................................................................................. 100

List of References .................................................................................................................. 101

List of Publications ............................................................................................................... 108

xviii

List of Abbreviations

ABC Artificial Bee Colony

ACO Ant Colony Optimization

AWGN Additive White Gaussian Noise

BER Bit Error Probability

BS Base Station

BSC Binary Symmetric Channel

CFAR Constant False Alarm Rate

CFD Cyclostationary Feature Detection

CR Cognitive Radio

CRN Cognitive Radio Network

CSI Channel State Information

CSS Cooperative Spectrum Sensing

dB Decibel

DE Differential Evolution

DSA Dynamic Spectrum Access

ED Energy Detection

EGC Equal Gain Combining

FC Fusion Centre

FCC Federal Communications Commission

GA Genetic Algorithm

HDF Hard Decision Fusion

HS Harmony Search

IEEE Institute of Electrical and Electronics Engineers

MF Matched filter

MHz Mega Hertz

MOO Multiple-Objective Optimization

MRC Maximum Ratio Combining

NP Neyman-Pearson

NSF National Science Foundation

PDF Probability Density Function

PHY Physical Layer

xix

PSD Power Spectral Density

PSO Particle Swarm Optimization

PU Primary User

QoS Quality of Service

RF Radio Frequency

ROC Receiver Operating Characteristic

SDF Soft Data Fusion

SFL Shuffled Frog Leaping

SNR Signal to Noise Ratio

SNR signal to noise ratio

SOF Single-Objective Function

SU Secondary User

TLBO Teaching Learning Based Optimization

TWS TV White Spaces

WiFi Wireless Fidelity

WiMAX Worldwide Interoperability for Microwave Access

WLAN Wireless Local Area Network

WRAN Wireless Regional Area Network

xx

List of Symbols

Channel gain

Null hypothesis

Alternative hypothesis

Index of quantization level

Nakagami m parameter

Number of cooperative users

The additive white Gaussian noise

Number of observation in respective quantization level

Number of false report

Total probability of error

Probability of detection

Probability of false alarm

Probability of miss detection

Marcum Q-function

Inverse Q-function

Primary user signal

Sensing time

Time bandwidth product

Weighting coefficients vector

Received signal by the secondary user

False alarm constraint of optimization problem or path loss

SNR

Gamma function

Incomplete Gamma function

Inverse upper incomplete Gamma function

Detection threshold

xxi

List of Figures

FIGURE 1.1 : Spectrum utilization .................................................................................. 2

FIGURE 1.2 : Measurement of spectrum utilization (0-6 GHz) ........................................ 2

FIGURE 1.3 : Cognitive Radio cycle ............................................................................... 4

FIGURE 1.4 : Spectrum holes concept ............................................................................. 5

FIGURE 2.1 : Trade-off among probability of miss detection and false alarm ................ 13

FIGURE 2.2 : Relationship between SS methods in terms of accuracy and complexity .. 20

FIGURE 2.3 : A typical wireless propagation environment ............................................ 21

FIGURE 2.4 : Cooperative spectrum sensing in a cognitive radio network. .................... 22

FIGURE 2.5 : Data fusion schemes for cooperative spectrum sensing ............................ 24

FIGURE 2.6 : Energy detection based EGC scheme ....................................................... 26

FIGURE 2.7 : Energy detection based MRC scheme ...................................................... 27

FIGURE 3.1 : Several of features of spectrum sensing ................................................... 31

FIGURE 4.1 : Multiobjective optimization and pareto front solutions ............................ 39

FIGURE 4.2 : Conventional and non-conventional optimization tools and techniques .... 40

FIGURE 4.3 : Schematic diagram of natural evolutionary systems ................................. 42

FIGURE 4.4 : Flow chart of Genetic Algorithm (GA) .................................................... 43

FIGURE 4.5 : Flow chart of Ant Colony Optimization (ACO) ....................................... 45

FIGURE 4.6 : Distribution of marks for student taught by two different teachers ........... 46

FIGURE 4.7 : Model for the distribution of marks obtained for a group of learners ........ 47

FIGURE 4.8 : Teaching–Learning-Based Optimization (TLBO) flow chart ................... 49

FIGURE 5.1 : Proposed system architecture................................................................... 55

FIGURE 5.2 : 2-bit softened hard combination scheme .................................................. 56

FIGURE 5.3 : Binary Symmetric Channel (BSC) ........................................................... 60

FIGURE 5.4 : Flow Chart of TLBO based solution ........................................................ 62

FIGURE 5.5 : Pseudo code of TLBO algorithm ............................................................. 63

FIGURE 5.6 : ROC graph for AWGN channel ............................................................... 65

FIGURE 5.7 : ROC graph for Rayleigh channel ............................................................. 65

FIGURE 5.8 : Comparison of TLBO method with PSO method ..................................... 66

FIGURE 5.9 : C-ROC graph for Rayleigh channel ......................................................... 67

FIGURE 5.10: Comparison of TLBO method with GA and PSO method ........................ 67

FIGURE 5.11: C-ROC graph for Rayleigh channel in semi log axis ................................ 68

FIGURE 5.12: Comparison of TLBO method with other method in semi log axis ........... 68

FIGURE 5.13: versus Nakagami fading parameter (m) .............................................. 69

FIGURE 5.14: Number of CR versus graph under Rayleigh channel .......................... 70

FIGURE 5.15: SNR versus graph under Rayleigh channel ......................................... 71

FIGURE 5.16: Performance of TLBO-based method ...................................................... 72

FIGURE 5.17: Convergence performance on over 2 Iteration ..................................... 73

FIGURE 5.18: Convergence performance on over 5 Iteration ..................................... 73

FIGURE 5.19: Convergence performance on over 10 Iteration ................................... 74


xxii





FIGURE 5.25: ROC graph including false reporting node ............................................... 77

FIGURE 5.26: Number of false reporting nodes versus curve .................................... 78

FIGURE 5.27: ROC graph including imperfect reporting channel ................................... 79

FIGURE 5.28: Reporting channel error versus curve ............................................. 80

FIGURE 6.1 : System model for minimization of sensing error ...................................... 83

FIGURE 6.2 : TLBO algorithm for minimization of sensing error .................................. 85

FIGURE 6.3 : Frame structure for CR networks with spectrum sensing ......................... 86

FIGURE 6.4 : Trade-off diagram ................................................................................... 87

FIGURE 6.5 : Lambda versus total sensing error curve .............................................. 88

FIGURE 6.6 : Lambda versus total sensing error curve .............................................. 89

FIGURE 6.7 : Normalized throughput versus sensing time (ms) curve ........................... 90

FIGURE 6.8 : Iteration versus total sensing error curve............................................. 91

FIGURE 6.9 : Convergence performance on for .............................................. 92

FIGURE 6.10: Convergence performance on for .............................................. 92




FIGURE 6.14: Convergence performance on for ............................................ 94








xxiii

List of Tables

TABLE 1.1: Description of CR Function .......................................................................... 5

TABLE 2.1: Local versus cooperative spectrum sensing ................................................. 23

TABLE 3.1: Various stages of Literature review ............................................................. 29

TABLE 3.2: Summary of element of cooperative spectrum sensing ................................ 32

TABLE 3.3: Summary of work on softened hard data fusion based CSS ......................... 33

TABLE 4.1: Optimization algorithm and its parameter ................................................... 43

TABLE 4.2: Initial population of TLBO ......................................................................... 50

TABLE 4.3: New values of the objective function for teacher phase ............................... 51

TABLE 4.4: Updated value of objective function for teacher phase................................ 51

TABLE 4.5: New values of the objective function for learner phase................................ 52

TABLE 4.6: Updated values of the objective function for learner phase .......................... 52

TABLE 4.7: Comparison between results obtained by various optimization methods ...... 53

TABLE 5.1: Performance matrix of TLBO based CSS on ......................................... 72

TABLE 5.2: Simulation parameter for FR versus curve.............................................. 78

TABLE 5.3: Simulation parameter for versus curve ............................................... 80

TABLE 6.1: Simulation parameter of sensing time versus throughput curve ................... 90

TABLE 6.2: Performance matrix of TLBO based CSS ............................................... 91

Introduction

1

CHAPTER-1

1 Introduction

1.1 Introduction

In the field of wireless communication frequency spectrum is as expensive as gold.

Wireless Service providers must pay a huge amount of money to purchase the right to use

the frequency spectrum for communication. With the advancement of wireless

communication system in this decade, the new wireless communication system has

frequently been used in the same area. Here every user in the system has a requirement of

high data rate because of certain kinds of quality services, so demand for high bandwidth

is the compulsory requirement. The number of subscribers also increases that result

saturation of frequency domain drastically.

1.2 Motivation

The fixed spectrum allotment policy in the wireless network has functioned well in the

former decades. But a recent report shows that there has been an exponential growth in the

use of limited frequency spectrum by the various wireless service operator. Also, many

parts of frequency spectrum are occasionally used as illustrated in Fig 1.1 [1], [2], [3], [4].

It can be seen in the figure that there is heavy usage of a certain portion of spectrum while

a substantial amount of frequency spectrum remain utilized or under-utilized. Government

agency/regulatory bodies are currently working on the modification and implementation of

some technologies with the aim of achieving a higher spectrum efficiency for future

Introduction

2

wireless technologies [5], [6], [7], [8]. A recent report from the Federal Communications

Commission (FCC) has shown that the deviations in the employment of the fixed assigned

spectrum differ from 10% to 85% [1] which makes the proficient employment of these

bands a more substantial problem than the scarcity of the spectrum [2].

FIGURE 1.1: Spectrum utilization

FIGURE 1.2: Measurement of spectrum utilization (0-6 GHz)

Maximum Amplitudes

Frequency (MHz)

Am

plid

ue (

dB

m)

Cognitive Radio Technology

3

While currently, fixed frequency allotment has no scope of bandwidth for future wireless

systems, actual measurements of the spectrum employment by the regulatory body show

that many allotted bands are not being mostly utilized in time and space domain. Fig. 1.2

shows the spectrum employment in the frequency range from 0 to 6 GHz taken by the

Berkeley Wireless Research Centre (BWRC). From the Fig.1.2 it is observable that the

frequency in the ranges 2 to 3 GHz and 5 to 6 GHz, the spectrum consumption is less than

10%, while for 3 to GHz, the spectrum consumption is less than 1%.

1.3 Cognitive Radio Technology

For the solution of the conflicts between the spectrum inadequacy and spectrum under-

consumption, cognitive radio system was proposed [2]. It can improve the performance of

spectrum usages through the access of unused radio spectrum of primary licensed users by

secondary/cognitive users. It is a combination of radio technology and networking so an

intelligent wireless communication system. It picks the wireless communication

parameters like frequency, bandwidth and transmission power to improve the spectrum

utilization efficiency and adjusts its transmission and reception. [9] Gives the following

definition of a cognitive radio.

”Cognitive radio is an intelligent wireless communication system that is alert of its

neighbouring environment, and uses the policy of learning from the environment and

adjust its inside states by changing certain operating parameters i.e, transmission power,

frequency, and modulation method in real-time, with two main goal in mind ”

Highly reliable communications

Efficient utilization of the radio spectrum.

The major function of cognitive radio can then be categorized as [3]:

Radio scene analysis: In this function, the unused frequency band is detected.

Channel state estimation: The task is concentrated on finding the channel.

Spectrum management: The principal aim of this task is effective spectrum sharing

of the free channels detected in the spectrum sensing stage.

Introduction

4

The state diagram of the cognitive radio cycle which shows the connection between the

function of cognitive radio as shown in Fig.1.3 The state diagram illustrates in this figure

shows the action taken by the cognitive radio in reference to change in the environmental

condition. Task 1 and task 2 are normally done at the receiver side, whereas task 3 is done

at the transmitter side. For task 3, the transmitter needs some technical information from

the receiver side which is carried out through a feedback channel.

FIGURE 1.3: Cognitive Radio cycle

The main and most key task of the cognitive radio is the procedure of searching used

spectrum of primary user (spectrum sensing). Once the white spaces are identified, the

cognitive user must select the best available channel that meets quality-of-service (QoS)

requirements and its communication (spectrum management). During the occupation of

the channel by the CR user if licensed user (i.e. PU) want to use this channel, then CR user

immediately terminate their transmission and slightly migrate to another unused channel

due to a lower priority than the primary user (spectrum mobility). Also, In a CR network,

there is some scheduling mechanism to ensure that all CR user get equal opportunities on

accessing the spectrum (spectrum sharing). Brief descriptions of this functionally are

summarized in TABLE 1.1

Cognitive Radio Technology

5

TABLE 1.1: Description of CR Function

One of the major aims of the cognitive radio (CR) is to get the best available spectrum. As

per the frequency regulatory policy of various countries, most of the frequency spectrum is

already allotted to licensed wireless communication technology, so the most challenging

task is to find the spectrum and share this spectrum without making interference with the

transmission of primary user/licensed users. By the spectrum sensing, the CR user is able

to find temporally idle spectrum, which is known as spectrum hole or white space

[10],[11],[12],[13],[14],[15],[16]. If the licensed user comes active for communication

then CR user has to use another spectrum holes or change its transmission parameter to

avoid interference. The spectrum holes theory is illustrated with the help of Fig. 1.4.

FIGURE 1.4: Spectrum holes concept

Po

wer

Frequency

Time

Spectrum Hole/

White Space

Introduction

6

1.4 Definition of Problem

The research aim of this report is to find the techniques to improve the detection

performance and reduce the total sensing error of cooperative spectrum sensing (CSS) in

cognitive radio with low overhead in the presence of fading environment. From the deep

literature review, the facts have been identified that the sharing of local spectrum sensing

result with neighbouring cognitive radio/secondary users or sensing result reported to the

fusion centre (FC) is the most challenging process that decides the performance CSS. The

performance of cooperative sensing heavily depends on the quality of local spectrum

sensing result and quality of the spectrum sensing result collected by the fusion centre via

reporting channel as well as global decision logic. Moreover, It is very difficult to send

whole local spectrum sensing result to fusion centre (FC) through bandwidth-limited

reporting channel which never allows to send this whole information to fusion centre

using complex protocols. Hence, the spectrum sensing nodes must optimize the size

(sensing bits) of their sensing result in an efficient way rather than sending the entire

sensing result to the fusion centre (FC). Furthermore, global decision logic at the FC is

static in nature. Sometimes imperfect reporting channel and false report due to the

malicious secondary user also change the global decision logic which affects the entire

performance so global decision logic must be dynamic for performance improvement.

Many evolutionary optimization techniques i.e. ABC, GA, ACO, ABC, PSO etc. are used

to find the optimal solution for performance improvement of CSS but they are not free

from algorithm parameters i.e. optimal parameter of algorithm is to be determined for the

optimal performance of the algorithm. This aspect is considered in this research work. In

wireless communication, continues parameters setting of these algorithms are a serious

problem due to the malicious secondary user and random nature of fading channel which

affect the performance of optimization technique. Proper selection of the algorithm related

parameters is a very critical issue in the wireless communication which affects the

performance of CSS. The improper selection of algorithm related parameters either

increases the computational time to search the optimal solution or find the local optimum

solution rather than an optimal global solution so optimization technique used to find an

optimal solution should be free from algorithm parameter.

Objective and Scope of Work

7

1.5 Objective and Scope of Work

To examine the effect of various data fusion scheme on cooperative spectrum

sensing(CSS) performance

To analyse the performance of cooperative spectrum sensing in the environment of

different fading channel as well as imperfect reporting channel and false reporting

node.

To reduce the extra overhead and communication bandwidth of control channel

through the optimal threshold and weight vector .

To improve the receiver operating characteristic (ROC) of CSS by improving the

detection probability ( ) for a given false alarm rate ( ) under Neyman-Person

criteria.

To reduce the all over sensing error of Cooperative spectrum sensing by improving

probability of error under Mini-Max criteria.

To compare the performance of various optimization technique for the evaluation

of optimal weight vector in cooperative spectrum sensing(CSS)

To built an optimal cooperative spectrum sensing framework.

1.6 Original Contribution by the Thesis

In this thesis, energy detection based cooperative spectrum sensing (CSS) techniques have

been thoroughly examined and analysed. This CSS techniques use softened hard

(quantized) data fusion method at fusion centre (FC) for reporting. The use of Teaching

Learning Based Optimization (TLBO) algorithm as a substantial method is proposed to

evaluate optimal weighting coefficient vector of sensing information. The original

contributions of this thesis are summarized as:

TLBO based cooperative spectrum sensing framework is proposed which optimize

the weighting coefficients vector of energy level of sensing information so that the

probability of detection is improved and the all over probability of sensing

error is minimized with low overhead.

Introduction

8

The performance of proposed TLBO based cooperative sensing framework is

analysed in various fading channel i.e. Rayleigh, AWGN and Nakagami and also

for the case of imperfect reporting channel.

The performance of TLBO based CSS framework is analysed in various fading

channel i.e. Rayleigh, AWGN and Nakagami and also for the case of false reports

due to malicious secondary user (SU).

The performance of TLBO based cooperative spectrum sensing framework is

compared with conventional soft decision fusion schemes like EGC as well as

hard decision fusion based cooperative spectrum sensing i.e. OR logic, AND

logic, Majority logic etc...

Optimization problems that maximize the probability of detection and

minimize the probability of sensing error is solved

Analytical evaluation is carried out for the performance of TLBO based multi-hop

cooperative spectrum sensing and sensing-throughput trade-off.

The performance of TLBO based CSS method is compared with the other

optimization technique i.e. GA, PSO based CSS for validation

1.7 Thesis Organization

The rest of this thesis is planned as follows. In Chapter 2, we present a brief overview and

some background information on spectrum sensing as well as cooperative spectrum

sensing of cognitive radio. We describe the various local spectrum sensing techniques i.e.

energy detection, match filtering and cyclostationary and discuss the various issue of local

spectrum sensing like hidden node problem, multipath fading and shadowing. We indicate

the most important part of the process of cooperative spectrum sensing that is various data

fusion scheme and its trade off analysis. We also explain some restricting factor of

cooperative spectrum sensing such as overhead and all over sensing error.

In Chapter 3, it consists of literature reviewed on spectrum sensing and cooperative

spectrum sensing. We also discuss the work done on softened hard data fusion based

cooperative spectrum sensing and its limitation. At the end this chapter we find the

research gap in softened hard data fusion based CSS.

Thesis Organization

9

In Chapter 4, we give the overview or optimization technique used for Cooperative

spectrum sensing. We Explain the classification of optimization technique like single

objective, multi objective, conventional (traditional) and con-conventional etc. Finally, we

describe in detail the teaching learning based optimization (TLBO) method which is used

to find the optimal parameter of CSS

Chapter 5 presents the proposed TLBO based cooperative sensing that is based on

softened hard data fusion scheme. The proposed frameworks can be easily applied to

cognitive radio network. The performance of this framework is same with EGC with low

overhead. The detection performance is also analysed for the case of non-ideal reporting

channel and malicious user present in the system. We also illustrate the influence of the

number of cooperating user and SNR on the detection performance as well as imperfect

reporting channel error on the detection performance.

Chapter 6 presents the probability of error analysis for proposed TLBO based CSS. We

discuss in details the sensing throughput trade off and its classification constant primary

user protection (CPUP) and Constant secondary user spectrum Usability (CSUSU). We

analysed the performance of proposed framework on sensing error with different values of

threshold under Mini-Max criteria and compare the proposed techniques with other HDF

and SDF based techniques. We also illustrate the influence of sensing time on through put

of secondary user network.

Finally, conclusions are given with major contribution and possible future scope to extend

this project is outlined in chapter 7

Background of Spectrum Sensing for CR

10

CHAPTER-2

2 Background of Spectrum Sensing for CR

2.1 Introduction

Spectrum sensing is the principal function among other function of the cognitive radio

network. Spectrum sensing deal with the enables the ability of a cognitive radio to sense,

adapt, to measure, acquire, and be aware of surrounding wireless operating environment,

such as the spectrum accessibility and disturbance status. Accessibility of free wireless

spectrum differs based on the instance, frequency and space which lead to acquiring

spectrum opportunities. Secondary users (SU) or cognitive radio (CR) can use the existing

free spectrum during the inactive period of the primary user (PU).

Spectrum sensing supports the cognitive user to achieve this objective by finding the free

spectrum in reliable and quick manner. Spectrum sensing also helps cognitive users (CR)

to sense the existence of licence user signals to defend the licence user's signal

transmission. It also helps in rapidly scanning if the primary users have become active for

signal transmission in this band acquired by cognitive users so that those bands can be

evacuated instantly. This is very essential for confirming that the interference should be

below the accepted level due to PU‟s signal transmissions. Moreover, the existence of

other cognitive radio is also necessary and coordination mechanism has to be established

between this secondary users. The detail survey of spectrum sensing and concern issues

can be found in [17]–[26].

Hypothesis Testing

11

2.2 Hypothesis Testing

A fundamental job of spectrum sensing task is to judge whether the PU‟s spectrum is

inactive or active. Binary hypothesis test [27],[28] is the process by which spectrum

sensing issue is conventionally expressed. The null hypothesis indicated by shows that

there is no PU‟s transmission, i.e., only channel noise is received by cognitive radio. On

the other hand, another hypothesis indicated by shows that the PU is active for

communication, i.e., primary user signal and channel noise is received by cognitive radio.

If there are cases where the hypotheses have no unspecified parameters, then it is called

simple hypotheses. If there are unspecified parameters, then it is called complex

hypotheses. The example of a binary hypothesis test for spectrum sensing the licence

user's signal in the environment of additive white Gaussian noise (AWGN) channel is

given by following.

{

(2.1)

Where is the signal received by SU and is primary user‟s transmitted signal,

is the additive white Gaussian Noise (AWGN) and is the channel gain. We also

represented by γ the signal-to-noise ratio (SNR).

and are the detecting condition for availability and unavailability of primary user‟s

signal respectively. In another word; is the null hypothesis which specifies that primary

user does not transmit their signal and is the alternative hypothesis that specifies that

the primary user transmit their signal and active in this duration. There are four possible

cases for the sensing signal:

1. Detecting under hypothesis which is called Probability of Detection

2. Detecting under hypothesis which is called Probability of Missing

3. Detecting under hypothesis which is called Probability of False Alarm

4. Detecting under hypothesis

If is detected under hypothesis, then it reference to probability of miss detection,

, that is probability of declaring that there is no primary user signal but actually primary

signal present. In another word it is the probability of missing the signal. If is declared


12

while is actually observed then it reference to the probability of false alarm which

specifies to declare primary user signal present while there is no primary user

communicating. Thus false alarm error related to utilization of spectrum inefficiently.

Missed detections of primary user signal are the major problem for spectrum sensing

because it related to interference with the primary system. It is recommend to maintain

false alarm probability as down as possible, so that the system can employ all possible

utilization of free channel.

A function is having a period and bandwidth can be approximately represented by

number of sample values [29]. Using this background, detector decision variable

statistics is a sum of , zero and non-zero mean, Hence, the PDF can be written as [30],

{

(

)

(√ )

(2.2)

The average probability of detection, probability of false alarm and probability of missed

detection in AWGN Channel environment are given, respectively, by [31].

{ | } (2.3)

{ | }

⁄

(2.4)

(2.5)

In (2.3) is the threshold value of energy detector, is the instantaneous signal to noise

ratio (SNR) of Cognitive radio. While in (2.4) is the time-bandwidth product. The

function is the gamma function, is the incomplete gamma and is

generalised Marcum Q-function defined as follow

∫

(2.6)

∫

(2.7)

Probability of Detection Over Fading Channel

13

FIGURE 2.1: Trade-off among probability of miss detection and false alarm

The trade-off among the probability of miss-detection and false alarm is illustrated in Fig.

2.1. The normal likelihood distributions for the availability and unavailability of PU‟s

signal are assumed with their two different respective mean and variances of probability

distributions function are taken to be same. It is obvious that we should have detection

probability should be kept high value as it specifies the amount of isolation

of the PU‟s signal from the disturbing the SU‟s signal. On the other side, the value of false

alarm rate should be necessarily low to main high secondary user network throughput,

since a false alarm probability is related to avoid the free vacant spectrum from being

acquired by cognitive users associated to ineffective spectrum usage.

2.3 Probability of Detection Over Fading Channel

It can be noticed from (2.4) that is independent of γ so false alarm probability under

fading environment remains the same as given by (2.4). Average value of detection

probability can be evaluated by averaging the respective probability distribution function

(PDF) over γ which shows the respective fading model

∫ (2.8)

In (2.8) indicates the probability distribution function (PDF) over SNR under the

environment of fading.


14

2.3.1 Rayleigh Fading Channel

In the wireless channel, if the received signal has a random amplitudes and phase angle

due to scattered properties of waves propagation, then this types of received signal can be

modelled as Rayleigh distribution.

Under Rayleigh fading, PDF of γ have an exponential distribution given by following

(

) (2.9)

The formula for is evaluated by taking averaging over in (2.8)

∫ (2.10)

∑

(

)

(

)

(

∑

(

)

) (2.11)

2.3.2 Rician Fading Channel

The received signal has line of sight path in some type or wireless environment. The

received signal can be modelled as Rician distribution. Under Rician fading, PDF is given

by the following equation

(

) ( √

) (2.12)

In the (2.12) is the Rician factor and formula for is evaluated taking averaging over

in following equation

∫ (2.13)

(√

√

) (2.14)

Probability of Detection Over Fading Channel

15

2.3.3 Nakagami Fading Channel

This fading mode is frequently used to characterize when signal is travels through

different path which is known as a multi path signal transmission environment and also

covers the wide range for the long distance high-frequency channel by the use of

Nakagami fading parameter m. For example, if Nakagami fading parameter then

channel can be approximated as Rayleigh distribution as special case. Similarly, when

Nakagami fading parameter is higher than one, then Nakagami-m distribution can be

approximated as Rician distribution. The Nakagami m distribution is most suitable to

characterize the urban and indoor multipath signal propagation

The following equation gives probability distribution function under Nakagami fading

(

)

(

) (2.15)

In the (2.15) is the Nakagami parameter and formula for is evaluated taking

averaging over in equation

∫ (2.16)

[ ∑ ⁄

(

)] (2.17)

Where is the confluent hyper geometric function and and is given by

(

)

(2.18)

(

)

(2.19)

, is the confluent hyper geometric function defined by

∑

(2.20)


16

2.4 Detection Criteria of Primary User

In cognitive radio systems of the secondary user, spectrum sensing function is tasked

intermittently performed to detect the free spectrum by monitoring of primary user activity

in a reliable way. Thus, precise spectrum detection of primary user signal is challenging

requirement for both sides: PUs side and SUs side. Actually, the spectrum sensing is often

expressed as a pure signal detection theory and problem. In this section, two basic

viewpoints of exploiting the detecting white spaces are expressed and can addresses which

are given by following.

2.4.1 Neyman-Pearson Criteria

Neyman-Pearson criteria were projected based on the issue of the most effective tests of

statistical hypotheses. Neyman-Pearson criteria are concern with minimum possible

interference which is produced by Secondary user to primary user in active position. The

decision statistics aims to maximize probability of detection for a given false alarm

rate. If we the probability of missed detection is , the decision method can

also be expressed as to minimize miss detection probability for given false alarm rate

Pf. This situation is known as constant false alarm rate (CFAR). The basic philosophy of

Neyman-Pearson criteria are based on minimizing latent interference caused by SU to

active PU while preserving constant rate of acquiring vacant spectrum. The generalized

optimization problem can be formulated under the Neyman–Pearson (NP) criteria is given

by following equation

(2.21)

The threshold value of CR as per the Neyman-Pearson criteria is evaluated as

( ) (2.22)

Put this threshold value in the equation of detection probability gives receiver operating

charactertics (ROC) for given probability of false alarm which is given by following.

{ | } (2.23)

Detection Criteria of Primary User

17

2.4.2 Mini-Max Criteria

A Mini-Max criterion is able to trade-off between effective utilization of spectrum and

interference in Primary user. In another way, we minimize total sensing probability of

error and probability of miss detection which are unwanted in CR system. The

Bayes‟ risk for a binary hypothesis statistics is given by following

| | |

| (2.24)

The probability of different decisions in terms of the probability of false alarm , the

probability of miss and the probability of detection is expressed as

| ∫ |

(2.25)

| ∫ |

(2.26)

Total sensing error probability is defined by following.

(2.27)

In the (2.27) and represent the probabilities of primary user under null

hypothesis and active hypothesis correspondingly. While threshold value of

decision in Neyman-Pearson criteria is kept at a fixed value for a given probability of false

alarm , the decision threshold based on Mini-Max criteria is adaptively changed as per

the statistical characteristic of the probability distribution function under and .

(2.28)

In the above equation, we need to minimize the maximum sensing error in cooperative

spectrum sensing (CSS) which ensured to called Mini-Max criteria. If the space between

the average value of the two distribution function is far, the decision threshold might be

fixed to be comparatively large value so that false alarm rate can be minimized while

still preserving no miss detection of the primary user signal. For simplicity, the value of

false alarm rate and probability of miss detection are assumed to be equal to get

closed form expression for the decision threshold.


18

2.5 Sensing Technique

In order to achieve the interference-free operation of the primary user, the secondary users

need to sense the allocated spectrum periodically and if the primary user becomes active

during communication, the secondary user immediately release the band to avoid the

interference. For example in the IEEE 802.22 standard, secondary users periodically detect

the Television signal and wireless microphone signals and if the signal is detected, they

have to vacate the channel within very short time [31][32]. Therefore, sensing of spectrum

process plays a challenging task in CR network to avert the interference to the primary

users and to constantly and rapidly point out the white spaces in the spectrum and utilize

the chance.

Lots of local spectrum sensing methods, including energy detection, cyclostationary

detection, matched filtering, waveform-based sensing and radio identification have been

used to achieve this task. Among these methods, energy detection is the most commonly

used technique due to its simplicity, easy implementation and little computational

complexity. The accuracy and complexity comparison of these techniques is shown in Fig.

2.3 and a detail explanation of these techniques are found in [32]–[34].

2.5.1 Energy Detection

Energy detection techniques can be implemented to sensing the availability of the primary

user signal for a Gaussian noise model and when noise power is identified to secondary

user. This is very simple detection technique which collects the energy of the received

signal during the spectrum sensing period and compares this collected energy with the

predefined threshold declares the band is occupied by the primary user if the collected

energy exceeds a certain threshold. The value of the threshold is fixed based on the desired

probability of false alarm [29], [30], [35]–[39]. Energy detection does not need any

knowledge about the primary user signal and channel gains as compared to other detection

technique. It is strong to unknown fading in the wireless channel. It is easier to implement

and hence is less expensive compared to another method. Therefore, energy detection

method is mostly adopted for spectrum sensing in [29], [30], [35]–[39].

Sensing Technique

19

2.5.2 Matched Filtering

When there is knowledge about the primary user signal to the CR, matched filter is the

optimum techniques. The basic principle behind the matched filter is to take correlation

between primary user signal and an unknown signal which is to be sense. This process is

equivalent to convolving the indefinite signal with a template having a reversal of time.

Matched filtering-based spectrum sensing techniques are most compatible for stationary

Gaussian noise situations which maximize the SNR of received signal. For the best

performance, they require prior perfect information of the channel reactions from the

primary user to the secondary user and the formation and waveforms of the primary user

signal i.e. modulation technique, frame diagram, pulse form as well as accurate

coordination at the SU [40]–[46]. In cognitive radios, this type of knowledge is not easily

accessible to the secondary users. It increases the employment cost and complexity of

detector when the number of primary user‟s signal bands is very large. Therefore, this

local spectrum sensing technique is not practical and appropriate to CR technology.

2.5.3 Cyclostationary Feature Detection

The cyclostationary feature detection is based on the basic properties of communication

signal. This basic property of signal is available in all wireless communicated signals from

transmitter. It increases the slight overhead of signalling. This method has no in-band

interference and also sensing and differentiates the different types of signal. This spectrum

sensing techniques can discriminate between modulated signals and noise [47]–[53]. This

detection techniques works on the basis of the fact the licence user modulated signals are

cyclostationary in nature which have spectral correlation. This primary user signal has a

redundancy of signal periodicity. The examples of this are sine wave carrier signal, series

of pulse and cyclic prefixes. The noise have wide-sense stationary signal characteristic

which have no correlation [47]–[53]. Therefore, cyclostationary detectors are strong to the

ambiguity in noise power [47]–[53]. cyclostationary detectors analyse the spectral

correlation function with extreme computational complexity and long sensing times.

Moreover, the information of the cyclic frequencies of the primary user is required in this

spectrum sensing method which is normally not available to the secondary users.


20

2.5.4 Other Sensing Techniques

There are also other spectrum sensing methods used by secondary which includes

waveform-based sensing and radio identification based. The comparison between all these

methods is given in Fig.2.2

FIGURE 2.2: Relationship between SS methods in terms of accuracy and complexity

2.6 Impact of Wireless Propagation on Spectrum Sensing

Irrespective of the spectrum sensing method used, fruitful detection of the primary user

signal highly depends on the quality of the received primary user signal and its strength.

This received signal strength is directly proportional to the wireless propagation

environment. Fig.2.3 shows the practical wireless communication environments in which

two inherent effects called multipath fading and shadowing, which reduce the quality and

strength of the received primary user signals[54]–[56]. Multipath fading also called the

small-scale fading occurs when the free space electromagnetic wave travels from

transmitter to receiver via many paths with dissimilar delays and deviation of the

amplitude of signal due to a random characteristic of the wireless propagation channel.

Energy

Detector

Radio

Identification

Match

Filtering

Cyclostationary

Waveform-Based

Sensing

Complexity

Acc

ura

cy

Impact of Wireless Propagation on Spectrum Sensing

21

FIGURE 2.3: A typical wireless propagation environment

In some fading environment, a substantial amount of attenuation of the propagated signal

weakens the received signal to noise ratio to a very low level in such a way that the

detection of the primary user signal at the receiver becomes difficult. Another effect

known as shadowing also called large-scale fading, occurs when there is large obstacles

comes between the transmitter-to-receiver track and a consistent signal reception at the

receiver side is not guaranteed. This is also called the hidden terminal problem [55], [56]

in which the CR cannot detect the PU existence and blindly assume that PU is absent and

starts to transmit the signal, hence creating the interference with the Primary user signal.

This hidden terminal problem and fading effect are clearly illustrated in Fig. 2.3. This CR

receiver blindly assumes that PU is out of its range due to shadowed by high building, so it

uses the active channel and creates the interference to the PU. Similarly, another CR

receiver is subject to fading due to a high building and high raise trees. Both of this effect

strongly affects the performance of the spectrum sensing reliability. The deployment of

spatial diversity in spectrum sensing also known as cooperative spectrum sensing is one of

the effective solutions to these problems.


22

2.7 Cooperative Spectrum Sensing (CSS)

The performance of a local spectrum sensing reduces due to the occurrence of wireless

propagation characteristic such as shadowing and fading caused by multipath propagation

of the signal. These type of wireless channel state may also result in the issue of hidden

node, in which cognitive transceiver is very far from the scanning range of a primary user

signal but close enough to the primary receiver to generate disturbance. These issues can

be resolved using cooperative spectrum sensing (CSS). In this cooperative spectrum

sensing adjacent secondary user cooperate in spectrum sensing a joint Primary user (PU)

signal transmission by sharing local sensing observation between them before making a

final decision.

FIGURE 2.4: Cooperative spectrum sensing in a cognitive radio network.

There is two methodologies to implement this CSS, centralized and distributed detection

[27], [57]–[59]. Fig. 2.4 shows this centralized cooperative spectrum sensing (CSS)

scheme, where N number of SUs are in cooperation to detect the channels for the PU

signal and send the sensing observation via bandwidth limited reporting channels to the

fusion centre (FC). FC makes the ultimate decision, whether the primary user is present or

not. It is next to impossible that all the channels between the Primary user and secondary

user will be in a high fade environment simultaneously. Thus cooperative spectrum

detection helps to mitigate the effect of hidden node problem and multipath fading through

the cooperative diversity [55], [60]–[62]. Another advantage of cooperative spectrum

sensing is the performance improvement of detector, enlarged coverage, easy to design the

detector and better toughness to non idealities. Therefore, Cooperative spectrum sensing

(CSS) gives lots of attraction in the cognitive radio literature. The various literature and

CSS Related Mathematical Statistics

23

other details on CSS and related problem along with detail references can be found in [17],

[18], [22], [61]–[64]. TABLE 2.1 tabulates the advantages and disadvantages of local

versus cooperative sensing schemes

TABLE 2.1: Local versus cooperative spectrum sensing

Sensing

Scheme Advantages Disadvantages

Local

sensing

Simplicity for Computational

and implementation

Hidden node problem

Multipath and shadowing

Cooperative

sensing

Higher accuracy Complexity

Reduced sensing time. Overhead

hidden node problems can be prevented The requirement of control channel.

When energy detection techniques is used for cooperative spectrum sensing, cooperative

nodes send the sensing observation via reporting channel to fusion centre, in either the soft

data fusion(SDF)or the hard decision fusion(HDF). In SDF, each cooperative node simply

forwards their exact observation for primary user to the fusion centre(FC) without making

any decision and FC decide whether primary user signal is present or not. In HDF, each

cooperative node makes its own decision for primary user activity send their individual

decisions in the form of 1 bit to the fusion centre FC decides based on some decision logic

i.e. AND, OR, MJORITY whether primary user signal is absent or present.

2.8 CSS Related Mathematical Statistics

One of the main crucial issues of spectrum sensing is the hidden terminal problem for the

case when the cognitive radio is shadowed or in deep fade. To mitigate this issue, multiple

cognitive radios can be cooperative work for spectrum sensing via cooperative diversity so

cooperative spectrum sensing can greatly improve the probability of detection in fading

environment. In cooperative spectrum sensing fusion centre (FC) calculates average

probability of false alarm and probability of detection probability with the references of

the probability of each CR. The false alarm probability is given by [56]

∑ (

)

( )

{

}

(2.29)


24

Also, Detection probability is given by

∑ (

)

{

}

(2.30)

2.9 Data Fusion Schemes for CSS

FIGURE 2.5: Data fusion schemes for cooperative spectrum sensing

The procedure of combining locally sensed observation of independent secondary users is

knows as a fusion scheme in cooperative spectrum sensing (CSS). Based on the types of

local sensing observation reported via reporting channel to fusion centre, there are data or

decision fusion schemes in cooperative spectrum sensing. In soft decision fusion (SDF)

schemes (data fusion), cognitive radio(SU) share their exact local observations. While in

the hard decision fusion (HDF) schemes (decision fusion), cognitive radio (SU) only share

their individual sensing decisions.

2.9.1 Hard Combining Fusion (HDF) Scheme

In the hard combining scheme, the final decision at the fusion centre (FC) depends on the

individual local decisions of secondary user reported via reporting channel. When the

local decisions are in the form of 1 bit send to the fusion centre (FC), it is suitable to put

into practice a linear combination of all fusion rules to get the cooperative decision. The

Soft Decision Fusion (SDF)

CR User Fusion Center (FC)

0 = PU absent

1 = PU present Hard Decision Fusion (HDF)

Data Fusion Schemes for CSS

25

benefit of this scheme is the saving of communication overhead. The literature on Hard

decision fusion (HDF) scheme for cooperative spectrum sensing is available in [65]–[70].

The examples for frequently used fusion rules are AND Logic, OR Logic and MAJORITY

Logic which are typical cases of the general K-out-of-M rule. These decision fusion rules

in hard combining scheme are summarized as below [56].

AND LOGIC: In this fusion scheme, if all the cooperative node for spectrum sensing

detect presence of primary user signal, then fusion centre(FC) make the decision that PU

signal is present. Fusion centre‟s decision is evaluated by logic AND of the individual

hard decision of cognitive user received at FC. The cooperative probability of detection

and false alarm can be evaluated by setting in (2.29), (2.30) respectively.

(2.31)

And

(2.32)

OR LOGIC: In this fusion scheme fusion centre‟s decision is evaluated by logic OR of

the individual hard decision received at FC. The cooperative probability of detection and

false alarm can be evaluated by setting in (2.29),(2.30) respectively.

(2.33)

And

(2.34)

MAJORITY LOGIC: In this fusion scheme fusion centre‟s decision is evaluated by logic

AND of the individual hard decision received at FC. Cooperative probability of detection

and false alarm can be evaluated by setting in (2.29), (2.30) respectively.

∑ (

)

⁄

(2.35)

And

∑ (

)

( )

⁄

(2.36)


26

2.9.2 Soft Decision Fusion (SDF) Scheme

In decision fusion (SDF) scheme, CR users forward the exact sensing result to the fusion

centre without making any local decision. The decision is made by combining these results

at the fusion centre by using appropriate combining rules such as EGC, MRC etc., Soft

combination provides better performance than hard combination, but it requires a larger

communication bandwidth of the control channel for reporting [20][28][71]. It also

generates more overhead than the hard combination scheme [28].

EGC: Equal gain combining (EGC) is one of the simplest soft data fusion combining

schemes. In this scheme, the sensing energy in each secondary user forwards their local

observation to fusion centre (FC) where this energy will add together then collective

energy is compared to a predefined threshold. If this collective energy is above the

threshold value, then FC decide the presence of PU otherwise it decide the absence of the

PU. The performance of EGC scheme is poor if CR user is subject to deep fading. The a

decision statistic is given by [27] [72].

∑

(2.37)

FIGURE 2.6: Energy detection based EGC scheme

Data Fusion Schemes for CSS

27

MRC: The basic difference between Maximum ratio combining(MRC) and the EGC is

that in this method a normalized weight is given to the sensing energy received at FC from

each secondary use and then combined and compare with a threshold. In this method, the

weight is depending on the received SNR of the each CR. The decision statistic for this

scheme is given by[73].

∑

(2.38)

FIGURE 2.7: Energy detection based MRC scheme

2.9.3 Softened Hard (Quantized) Data Fusion

In this data fusion scheme, the effort has been made to realize a trade-off between the

overhead and the detection performance. In quantized data fusion scheme, it divides the

whole observed energy space into multiple regions with multiple thresholds. Rather than

sending the exact observation or sending the one-bit decision to FC, each secondary user

sends the index of its observed energy to fusion centre(FC) in this way better detection

performance is achieved. In [65], expression for a two bit scheme is given in (2.39).


28

∑

(2.39)

In (2.39) is weight given to corresponding observed energy, number of observation

in respective energy level and is the threshold level.

2.10 Sensing Errors and Overhead of Cooperation

Sensing errors are depending on the probability of false alarms or miss-detections which

happen due to noise and fading environment. False alarms is due to the declaration

channels as busy state by the fusion centre(FC) when the channel is actually free and miss-

detections is due to the declaration channels as a free state by the fusion centre(FC) when

the channel is actually busy. When the false alarm occurs, a spectrum usages chance is

minimized by the secondary user, and finally spectrum utilization efficiency will degraded

at sensing outcome. On the other side, miss-detections are concern with the disturbance

with the PUs transmissions. Therefore, it is essential to minimize the sensing error in

spectrum sensing. This probability of sensing error is the summation of the probability of

miss-detection and probability of false alarm that reduces interference with primary users'

signal transmissions and increases the utilization of vacant spectrum. An optimal detection

threshold can minimize a spectrum sensing error which gives the sufficient protection for

interference with primary users' signal communication and raise the spectrum utilization.

Some literature shows that the probability of sensing error can also minimize by selecting

an optimal number of user who cooperates for spectrum sensing.

The utilization of spatial diversity in cooperative spectrum sensing results in substantial

amount of improvement in detection performance. However, cooperative spectrum

sensing (CSS) among secondary users also adds various overhead that restrict the

improvement progress in detection performance. The overhead that are concerned with all

elements of CSS is known as cooperation overhead. An example of cooperation overhead

is extra transmission cost, extra delay, energy, additional sensing time and any extra

process related to cooperative spectrum sensing.

Introduction

29

CHAPTER-3

3 Literature Review

3.1 Introduction

A literature review is an unremitting procedure which was carried out during the course of

my research work. Major objective of this section is to offer the sufficient background

information on softened hard (quantized) data fusion based CSS in cognitive radio.

Literature reviews of my research work are divided into five different stages Table 3.1

shows component wise explanation about this stage with their objectives.

TABLE 3.1: Various stages of Literature review

Stage Topic of Literature

Review Objective Outcomes

1 Cognitive Radio To understand concept of cognitive

radio.

Brief knowledge about the

concept of cognitive radio is

gained.

2 Spectrum Sensing To understand the function of

spectrum sensing in cognitive radio

Various spectrum sensing

techniques and its limitation is

studied.

3 Cooperative

Spectrum sensing

To find research gap in CSS and its

current challenges

Sharing the local sensing result

between CR and FC as well as

dynamic global decision logic is challenging task

4 Data Fusion Method

in CSS

To explore the data fusion method in

CSS with low overhead

Softened hard data fusion

techniques in CSS have low

overhead and good detection

performance is identified

5 Optimization

Techniques

To find efficient and stable

optimization technique for

Cooperative spectrum sensing

TLBO a based optimization

technique is identified which is

free parameter of algorithm which

is most suitable for wireless

environment. It is efficient and

stable.

Literature Review

30

Following are the literatures that we referred during my research work

Peer review Journal

Research papers on cooperative spectrum sensing (CSS)

Conference proceeding on CSS

Report on spectrum policy and regulation for various country

Patent databases

Online Video lectures delivered by well-known universities on CSS

Theses and Dissertation database on CSS

Books

Methodical literature reviewed during my research work is very useful to find the

following information

The concept of cognitive radio

Various types of the spectrum sensing techniques and its merits and demerits

Major challenges of cooperative spectrum sensing (CSS)

Possible solutions to fill the research gap in CSS

The concept of Optimization technique

Search the best optimization techniques to find optimal solution

Obtain the result through simulation

3.2 Literature Reviewed on Spectrum Sensing & CSS

Dynamic spectrum access (DSA) method is used as a key that can solve the spectrum

scarcity problem. In wireless communications, some frequency ranges are mostly used i.e

10 MHz and 6 GHz. However, some experiment and some literature shows that spectrum

scarcity arises due to allocated band are not utilized properly. As explained in chapter 2,

spectrum sensing is a key necessity of CR. Reliable and efficient sensing scheme is a

challenging task to ensure negligible interference to PU during the utilization of unused

frequency bands. From the various literature reviewed, the various features of spectrum

sensing summarized in the form of diagram and are shown in Fig.3.1.

Literature Reviewed on Spectrum Sensing & CSS

31

FIGURE 3.1: Several of features of spectrum sensing

Literature Review

32

Conventional cooperative spectrum sensing (CSS) has a three main process: (1) Local

spectrum sensing (2) Reporting (3) Data fusion. In this section, the state-of-the-art detail

literature survey of cooperative sensing is given to review the research challenges of the

fundamental components of CSS Table 3.2 describe the detailed summary of this element

of cooperative spectrums sensing

TABLE 3.2: Summary of element of cooperative spectrum sensing

Sr.

No

Element of

CSS Types Research challenges Ref.

1 Cooperation

models

Parallel

fusion model Most of the existing model is performance

centric only few of them consider the

overhead

Existing models assume that the PUs are

non-cooperative but in some cases it is not true

[27],[74]

[75] Game theoretical

model

2 Sensing

techniques

Energy detection Performance will degrade due to hidden

terminal problem

Knowledge of primary signal is required in

some method.

Some method is facing difficulties to

implement due to complexity

[30],[47]

Cyclostationary

Compressed sensing

Match filtering

3 Hypothesis

testing

Binary hypothesis

testing In Bayes test the prior knowledge of some

un known parameter is required

Fixed sample size is required in NP, GLRT

and MLE testing method

Implementation issue in sequential testing

due to complexity

[76],[77] Composite

hypothesis testing

Sequential testing

4 Control channel and

reporting

Common Control

Channel

There is bandwidth requirement in control channel

There is reliability issue due to imperfect

reporting channel

Most of CSS method assumes a dedicated

data reporting channel, but in some

application, channel should be dynamic

allocated

[78],[79]

Dedicated control

Channel

5 Data fusion

Hard decision

fusion rules(HDF) There is a trade-off between overhead and

quality.

[80],[75]

[81]

Soft decision

rules(SDF)

6 User selection

Centralized

selection Knowledge of cooperation footprint is

required

It is very difficult to select cluster

head(user)and It increases the overhead

[82],[83]

[84] Cluster-based

selection

7 Knowledge

base

Radio

environmental maps There is security issue for sharing this

knowledge base data

Remote knowledge base access:

For remote knowledge base, There is quick, secure, energy efficient access to data is

required

[85],[86]

[87]

Spatial received

signal strength

profiles

Power spatial

density maps

Channel gain maps

Related Work on Softened Hard Data Fusion Based Cooperative Sensing

33

3.3 Related Work on Softened Hard Data Fusion Based Cooperative

Sensing

In cooperative spectrum sensing, if the amount of sensing data shared between cognitive

radio (CR) user and Fusion centre (FC) is increased, the requirement of communication

bandwidth of control channel is increased. On the other hand, if CR user sends a very less

amount of data to FC then strength and reliability of CSS decrease so there is trade-off

between overhead and accuracy. The approach in this research work is to provide a

balance between the amounts of sensing data send to FC and accuracy of the decision.

Here CR user sends a softened hard (quantized) data to FC which is known as quantized

cooperative spectrum sensing. Cooperative spectrum sensing schemes with quantization

have been proposed in various literature. Table 3.3 gives the detailed summary of work

done on softened hard (quantized) data fusion based CSS.

TABLE 3.3: Summary of work on softened hard data fusion based CSS

Sr

No Publication Key Contribution Ref.

1 Elsevier Research challenges and classification of cooperative spectrum is given [64]

2 IEEE

Trans.

Cooperative spectrum sensing with 1-bit or hard decision fusion

techniques are evaluated for multiple primary bands.

[88]

3 Wiley Performance of cooperative spectrum sensing due to the imperfect reporting channel for hard decision logic is analyzed

[89]

4 IEEE

Trans.

Optimal quantizer for local signal detection is designed for cooperative

spectrum sensing.

[90]

5 NRSC,

Springer

2-bit and 3-bit quantized cooperative spectrum sensing having uniform

quantization is proposed

[91],

[92]

6 ICASSP

Bit Error Probability (BEP) wall concept is launched for M-out-of-N

hard descision fusion rules in cooperative spectrum sensing and also

analyze the reduction in SNR loss because of BEP

[93]

7 IEEE

Trans. Author design the optimal design of the quantizer

[94]

[95]

8 IEEE

Trans.

In cooopeaative spectrum sensing author introduced and implement the

censoring scheme.

[96]

[97]

9 IEE

conf.

Spectrum sensing which use the Welch‟s periodogram techniques for

the case of AWGN channel is proposed

[98]

10 IEEE

Trans.

Spectrum sensing technique is proposed which based on Dempster-

Shafer theory of evidence.

[44]

11 IEEE

Letter

Author design the optimal quantizer which is based on Lloyd-max

algorithm to minimize mean-squared error of quantization.

[99]

12 IEEE

Trans. Author design the optimal quantizer which is based on Log-likelihood

[100]

13 IEEE Trans.

two-bit quantization scheme is planned to maximize the probability of detection for a given false alarm.

[66]

14 IEEE

Trans.

Author proposed a linear test based on the LRT detector in cooperative

spectrum sensing

[77]

15 IEEE

Trans.

Proposed the maximum eigenvalue-based Likelihood Ratio Test

(LRT) in the environment of unknown noise of primary user signal

[101]

16 IEEE

Trans.

Studied a distributed Likelihood Ratio Test detector for spectrum

sensing

[102],

[103]

Literature Review

34

17 IEEE

Trans.

Author proved that the likelihood ratio quantizers (LRQs) is the

optimal quantizers for signal detection

[57]

[90]

18 IEEE

Trans.

Author proposed the local optimum quantizer based on minimum

average error (MAE)

[104]

19 IEEE

Trans.

Author address the issue of multibit distributed detection of weak

random signals for the case of correlated signal observations at sensors.

[105]

20 IEEE

Trans.

Author consider the case of optimum local decision partitioning in

distributed detection environment.

[106]

21 IEEE

Conf.

Different method have been review to quantize the decision

mathematical statistics

[107]

22 IEEE Trans.

Author design the optimal quantizer which is based on Lloyd-Max quantization

[99]

23 IEEE

Trans.

Author implemented the quantized versions of SNR or energy values

as Soft data fusion(SDF) and compare their proposed detection method

with HDF method

[65],

[66]

3.4 Research Gap

In this research work, we found following research gap after our literature reviewed of all

the phases

Performance of cooperative sensing heavily depends on quality of local spectrum

sensing result and quality of the spectrum sensing result collected by the fusion

centre via reporting channel as well as global decision logic but there is trade-off

between overhead and performance so node should optimize size of sensing bits

and global decision logic should be dynamic

In wireless communication, continues parameters setting of these algorithms are

required due to the malicious secondary user and random nature of fading channel

which affect the performance of optimization technique. The improper selection of

algorithm related parameters either increases the computational time to search the

optimal solution or find the local optimum solution rather than a global optimal

solution so optimization technique used to find an optimal solution should be free

from algorithm parameter.

None of the researcher or inventor has worked on optimization techniques for CSS

which is free from the algorithm parameters, i.e. no algorithm parameters

The research gaps mentioned above have been addressed in the proposed optimal

cooperative spectrum sensing model.

Summary of Literature Reviewed

35

3.5 Summary of Literature Reviewed

In this research work, the qualitative and exploratory approaches have been used and

followed the research methodology steps. Very first, during the literature review, we

referred various research papers, journal and other articles on optimization cooperative

spectrum sensing in cognitive radio (CR) for wireless communication. Our research work

has been carried out using simulation approach. During this initial phase of literature

review, we found researchers had done work on cooperative spectrum sensing but very

few of them worked on decision fusion technique in cooperative spectrum sensing for

centralize architecture. Major reasons for this gap are that the most of the optimization

algorithms have difficulty in determining the optimum controlling parameters (parameter

of algorithms). If any change in the algorithm parameters occurs, It will change the

effectiveness of the algorithm and hence affect the performance of CSS and also very less

work have been done for dynamic decision fusion techniques in cooperative spectrum

sensing for the centralized environment.

During our literature review, we found there will be research works on quantized

cooperative spectrum sensing that evaluate the optimal weight vector to form the global

decision for centralized architecture. Therefore, our second phase of the literature review

was mainly focused on optimization of softened hard (Quantized) cooperative spectrum

sensing. By studying and analysing various approaches, we found the estimation of

optimal weight vector for global decision are challenging task. We also found none of the

researchers has worked on optimization techniques for CSS which is free from the

algorithm parameters, i.e., no algorithm parameters which evaluate this weight vector for

global decision logic. As a result of both phases of literature review, we proposed the use

of TLBO based cooperative spectrum sensing system model (framework and algorithms)

with two main objectives 1)Improve the receiver operating characteristic (ROC) of the

cooperative spectrum sensing and 2) Reduce the all over sensing error of Cooperative

spectrum sensing. The proposed system model and TLBO based algorithm is available in

the further chapter

Optimization Techniques for Cooperative Spectrum Sensing

36

CHAPTER-4

4 Optimization Techniques for Cooperative

Spectrum Sensing

4.1 Introduction

In a cognitive wireless network, the accessible radio resources such as bandwidth are very

limited. On the other side, each and every user wants to communicate in a wireless

network with the requirement of the high data rate. Not only the number of wireless users

increased exponentially but also has their requirement of some special kind of wireless

services which work on high bandwidth. The example of this new services are a video on

demand, TV on request, the Internet on wireless medium and wireless gaming. Finding a

technique to adjust all these requirements on limited bandwidth become a difficult

research issue in wireless communication system. Cooperative spectrum sensing and its

optimization are general methods to improve spectral utilization efficiency, but there are

some trade-offs and limitation so it‟s better implementation can be put into practice to

achieve this goal.

This chapter will focus on how to formulate cooperative spectrum sensing problems as

optimization problems from the perspective of detection performance and sensing error.

Precisely, this chapter discusses what the optimization theories, its parameters, practical

limitation, and performance are. In addition, we describe in detail the evolutionary

optimization technique that is teaching-learning based optimization (TLBO) that can be

put into practice to find optimal weighting coefficient vector for the improvement of

detection performance and reduce the sensing error in the cooperative sensing.

Types of Optimization Problem

37

Optimization can be formulated as evaluating the solution of given problem which might

be of single-objective or multi-objective with reference to the number of criteria. It is

essential to maximize or minimize objective functions within a search space which consist

of some acceptable values of variables. This acceptable variable also satisfied the

constrain put in it. There might be a large number of acceptable variables in the search

space that maximize or minimize the objective function in which is called the acceptable

solutions. The best solution among them is called the optimum solution of a given

problem. An objective function is a principal aim which is either to be maximized or to be

minimize. For example in cooperative spectrum sensing the main aim to maximize the

detection probability and minimize the sensing error.

4.2 Types of Optimization Problem

Optimization is the procedure to make something better. An engineer or scientist invented

new concept/theory and optimization makes an improvement on this idea. Optimization

algorithm gain the information and makes the improvement. A computer program is used

to implement this optimization algorithm. The standard equation of maximization of

objective function can be written as

(4.1)

In (4.1) belong to is the vector of n decision variables, is an objective function

is the ith constraint of m total constraint form the feasible search space

4.2.1 Single Objective Problem

There are only one objective function in optimization problems is called single objective

problem. This problem can be formulated as a single objective optimization problem.

Following is the example of a constrained single objective optimization problem.

(4.2)


38

In (4.2), the objective function is the linear equations but in practical scenario

this objective function can be formed as nonlinear equations. From the above range of

variable given by (4.2), it can be the detected that the variables can have any value within

specified ranges.

Here the optimization problem is to find the best combination of and that minimize

the objective function z. The single optimization problem can be solved either by

traditional techniques (linear programming, dynamic programming etc.) or through the

nature inspired advanced optimization techniques (particle swarm optimization (PSO),

artificial bee colony algorithm (ABC), differential evolution (DE), genetic algorithm

(GA), Harmony search (HS) etc.)

4.2.2 Multi Objective Problem

The optimization problem that consists of more than one objectives function which is to be

optimized simultaneously is known as a multi objective problem. In this type of case, it is

not essential to have a solution that is optimal with reference to all objectives. In

optimization theory, one solution might be best for one objective function but sup-optimal

or even worst for another objective function. Therefore, there is normally a pair of feasible

solutions for given multi-objective problem. These solutions are usually known as Pareto

front solutions. For this types of solutions, no further improvement is possible in any

objective without losing at least one objective functions. It required higher computational

complexity and effort to find the feasible solution in the multi objective problem as

compared to the single objective optimization problem.

To explain this idea of multi-objective optimization theory along with Pareto solutions,

consider the following optimization problem and its solution [108].

√

√

(4.3)

Types of Optimization Problem

39

In the above example, these two objective functions are directly proportional to variable

but inversely proportional to variables , and . Fig.4.1 shows that graph of

versus for 103 arbitrary combinations of , , , and which indicated by blue line.

Green region shows the graph of another 106 arbitrary combinations also known as

feasible region. This feasible region denotes all achievable solutions for given objective

function. The solutions that jointly minimize and are called optimal solution among

this obtainable solution which are located on the south-western border of feasible region

by red line shown in Fig.4.1. This red line is called as Pareto optimal solutions. The

solutions which are lower end of the Pareto region are best in minimizing but they are

best in minimizing . Same thing is happen for the upper ends of the Pareto region are

good for but poor for . Thus, these solutions cannot be further best in any movement.

Thus, the principal aim of multi-objective optimization is to find pair of optimal solution

rather than evaluating the optimal solutions of individual objectives. Multi-objective

optimization can often be used in design of frame the work, mathematical modelling

signal and resources sharing in the field of cognitive radio network.

FIGURE 4.1: Multiobjective optimization and pareto front solutions


40

4.3 Classification of Optimization Technique for CSS

FIGURE 4.2: Conventional and non-conventional optimization tools and techniques

Engineering design problem can be modelled as a goal-centric and constrained decision-

making process. The brief classification of conventional and non-conventional

optimization techniques are given in the Fig.4.2. The engineering design problem and

optimization terminology are related by following

Design parameters = design variables

Design goals = objective functions

Search method = optimization method

Search spaces = feasible solutions

The optimization algorithms find the optimal solution from all available feasible solution

in search space. For example, in wireless communication the system designer include

optimization technique in which he/she consider certain objectives related to their design

such as the probability of detection, sensing error, throughput, signal to noise ratio,

sensing time, no of user, etc. In real wireless communication design problems, design

variables may vary from one to very large.

Classification of Optimization Technique for CSS

41

4.3.1 Classical (Traditional) Optimization Techniques

Optimization problems are very important for research and scientific world. The

optimization algorithm gives weight on minimizing or maximizing certain objective

function. Traditional or conventional optimization techniques are frequently used to find

the optimum solution of continuous as well as differentiable functions. The traditional or

conventional optimization method puts some limitations on solving some complex

problems. While most of the research designs optimization problems require solution of

some complex problem. The following are the limitation associated with classical

(conventional) methods

The efficiency of this method is very much depends on the size of feasible

solution, variable size and constraint in the design problem

This method do not gives generic solution

Most of the method tends to get hang to a sub-optimal point

Classical method cannot be effectively applied on a parallel machine.

Classical optimization methods are very much depends on class of constraint

functions i.e. linear, nonlinear etc.

Conventional methods are very much depends on nature of variables used in the

optimization problem i.e., linear, nonlinear etc.

Mainly due to these reasons, classical/conventional optimization techniques are not

appropriate or challenging to practice for the operative optimization problem solution.

4.3.2 Meta Heuristic (Advance) Optimization Techniques

With a motive to solve some of the standard limitation of the conventional optimization

method, evolutionary-based optimization method (known as nonconventional optimization

procedure) mostly inspired from nature which has been invented by researcher‟s

community. These optimization procedures are goal oriented and it is independent to

model. These Meta heuristic optimization techniques are efficient and flexible. Most of the

researchers are working on these optimization techniques and many new modified and


42

improved versions are continually invented in the scientific research. The advantages of

these optimization algorithms are given by following.

The optimization method is robust for dynamic changes in the environmental.

Optimization method can be practiced to any problem.

Hybridization (Mixing) with other optimization methods can be possible.

These methods can solve problem having no solution

These methods are easily modified to create its improved version.

FIGURE 4.3: Schematic diagram of natural evolutionary systems

The evolutionary optimization techniques include differential Evolution (DE) harmony

search (HS), Ant colony optimization (ACO), genetic algorithm (GA), artificial bee

colony (ABC) and particle swarm optimization (PSO) etc. These optimization techniques

are very useful for single objective as well as multi objective optimization problems. The

Fig. 4.3 shows the diagram of main four optimization method [109]. In these optimization

techniques, the proper tuning of the parameter setting is required in the optimization

algorithm. Otherwise it creates a serious problem which affects the entire performance of

optimization technique. Table 4.1 show these optimization algorithms and its parameter.

Few of them are discussed by following.


43

TABLE 4.1: Optimization algorithm and its parameter

Sr. No Optimization

Algorithm Parameter of algorithm

1 GA Crossover rate, mutation and operator selection

2 PSO Velocity, position, social factor, cognitive factor and inertia weight

3 ABC Bees number and the limit

4 ACO Reward factor, parameters for exponent and evaporation rate

5 HS Number of improvisations, pitch rate and harmony rate

6 DE Differential amount of weight and crossover probability

7 SFL Memeplexes number and iteration by memeplexes

Genetic Algorithm (GA): The basic concept of Genetic Algorithm (GA) is designed to

simulate processes of evolution. Here the evaluation is the process of optimization. This

algorithm obeys the principles of the survival of the fittest. It is normally used in case

where the search space is large and cannot be effectively solved by classical

(conventional) optimization method. Unlike the classical method here we put the series of

a solution called population to the objective function rather than making a single solution.

FIGURE 4.4: Flow chart of Genetic Algorithm (GA)

{Worst}

Discard

{Best}

Mutation

Crossover

Initialization

Evaluate Fitness

Termin

ate

Selection

End

Yes

No

pop_size

}

pop_size


44

Initial population is given which made of from different feasible solution. After given this

population, the optimization algorithm finds the best chromosomes and rejects the unfits.

A fitness score by means of the maximization of our objective function is evaluated to find

which generations are most appropriate for surviving to the forthcoming generation. The

generation who gives highest score are nominated for surviving. Successive generations

are formulated in the following three steps. Flowchart of GA shown in Fig.4.4

Selection

Crossover

Mutation

The last two processes are repeated most of the time. This evaluation and generation

process is done iteratively to raise the proportion of fit members. When the stop criteria is

satisfied either by number of iteration or achieved some predefined fitness then algorithm

stop and resultant value gives optimal solution.

PSO: Particle swarm optimization (PSO) is a population based optimization techniques

which inspired by the behavior of flock of bird or fish for search of food [110]–[112]. This

behavior of bird can be useful to design the optimization algorithms. Here each

particle/bird fly in the sky, those particle/bird are close to food, it simply chirp loudly and

other bird follow this bird. If another bird which are more close to food as compared to

first one, it again chirp loudly and remaining all bird circulating around it. In this way the

objective (food) is achieved. Each particle is a latent solution to the optimization problems

and travels in the direction of the best optimal solution based on past experience. PSO

algorithm is very simple, high converge capability and easier to implement [113],[114].

PSO have two main equations expressive the velocity and position of the particle at

particular time. After the each iteration, the position and velocity is modified until

terminating conditions satisfied. The termination conditions can either a number of

iteration of some predefined fitness value. In general, the velocity and position of the

particle at t +1 can be given by following equation.

( ) (4.4)

(4.5)

In (4.4) and are social and cognitive factor, and are random values from [0 1]


45

ACO: Ant colony optimization (ACO) algorithm was derived by the movement of the

ants for finding the food sources. In ACO, the ants search the food and take back to the

nest. The ant then leaves a substance known as pheromones when go back to the nest. The

amount of pheromone placed, which may be subject to amount and quality of the food

sources, this pheromone will guide other ants to the food. The other ants obey the paths

where pheromone is larger. Following are the three main function of ACO optimization

method.

1. Auto solution construct

2. Pheromone Update

3. Daemon Action

FIGURE 4.5: Flow chart of Ant Colony Optimization (ACO)


46

4.4 Teaching–Learning Based Optimization

4.4.1 Introduction

Teaching–Learning-Based Optimization (TLBO), is proposed by Rao et al [115],[116] and

later on improved by Puralachetty [117] which is based on the impact of a teacher on the

outcomes of learners. The outcomes can be in the form of marks, grade or reward. In the

classroom, the teacher shares his knowledge, concept and idea with a student. The

performance of the teacher is reflected to a student in the form of their result, marks or

grade. Obviously, good teacher trains the student property for improvement of their result.

FIGURE 4.6: Distribution of marks for student taught by two different teachers

Let us assume that we have two different teachers and who teaches a subject of same

topic to same quality level student to two separate classes. The Fig. 4.6 shows the

distribution of mark obtained by the student of two separate classes teaches by two

different teachers. Curve 1 is for teacher and curve 2 are for teacher . We have

assume that there is normal distribution for the obtained marks. It has been observed from

Fig. 4.6 that curve-2 denotes better output than curve-1so it can be commented that teacher

teaches in the class better than teacher . The good teacher makes input an to

produced better average value of result. Here is higher than . The Learners can also

learn through the interaction among them which will affect their results

Obtained Marks

𝑴𝟏 𝑴𝟐

Pro

bab

ility

Den

sity

Teaching–Learning Based Optimization

47

FIGURE 4.7: Model for the distribution of marks obtained for a group of learners

A mathematical model is developed based on the above teaching-learning process and

practiced for the optimization of objective function. This innovative optimization method

is known Teaching–Learning-Based Optimization (TLBO). Let us Consider Fig. 4.7 in

which curve-A show the distribution of marks obtained for learner in division A having

their mean . The teacher in the class has more knowledge than learner so those learner

more knowledge than other learner called best learner is mimicked as teacher which is

indicted by in Fig. 4.7. The teacher in the class are share their idea, concept and

information among learners, which will upsurge the performance of whole class and help

learner to get better output in the form of grades, marks or point so teacher raise the

average value of result of class according to his or her effort.

From the Fig. 4.7, teacher will increase mean toward their own input through his or

her capability, thereby growing the new mean . Teacher will tries to increase the

mean of class but student the will achieved knowledge according to quality of teaching

and level of student in the class-room. The level of student is decided by the mean value of

class. Teacher will increase the mean of class to , at which stage the student

needs new teacher of higher quality than themselves, i.e. here the new teacher is .

Hence, there will be a new teacher having curve-B

𝑴𝑨 𝑴𝑩

Pro

bab

ility

De

nsi

ty

𝑻𝑨 𝑻𝑩

0 10 20 30 40 50 60 70 80 90

Obtained Marks


48

TLBO is also an evolutionary nature-inspired population based optimization algorithm.

The process of teaching-learning and optimization terminology are related by following

Population = class or group of learners

Design variables = subjects

Fitness value= learners‟ result.

The working of TLBO process has two phases. (1) „Teacher Phase‟ (2) „Learner Phase.'

The „Learner Phase‟ concern with learning through the mutual communication between

students and the „Teacher Phase‟ concerned with learning from the teacher. The flow chart

is shown in Fig. 4.8

4.4.2 Teacher Phase

From the Fig. 4.7, the teacher will increase the mean of class to subject to good

teacher. A good teacher will increase the knowledge level of student as per his or her

capability. But in reality this is not possible and teacher will raise mean of class up to

some level based on the knowledge gain ability of class.

At any iteration let be the teacher and be the mean of learners. will try to

increase mean as per his or her ability. After the any iteration there will be a new

mean, say . The solution is updated as per the difference between the new mean and

the existing mean given by the following expression

_ (4.6)

In (4.6) is a teaching factor that decides the value of mean to be change, is a random

number between [0, 1] which is given by following.

{ } (4.7)

The difference evaluated by (4.7) and updated the existing solution by (4.8)

_ (4.8)


49

FIGURE 4.8: Teaching–Learning-Based Optimization (TLBO) flow chart


50

4.4.3 Learner Phase

Learners raise their knowledge by two different approaches: one from the learning through

teacher and the other through mutual interaction among the learners. A learner interacts

randomly with each other through discussions in the group, active presentations, normal

communications with each other, etc. It learns something novel if the other learner has

additional knowledge than him or her. In this phase randomly two learners say and

are selected where . Learner alteration is represented by following is expressed as

( ) (4.9)

( ) (4.10)

4.4.4 Unconstrained Optimization Problem

Himmelblau function is the standard function which is used to test the performance of

various optimization techniques. It is the unconstrained function taken as benchmark

function for validation of the working of TLBO method. The objective is to evaluate the

value of and so that the function (Himmelblau) is minimized.

Objective function: Unconstrained Himmelblau function

(4.11)

In Teaching learning based optimization (TLBO), we set the population size is five,

number of iteration is one and two design variable and . The randomly created

initial population are given by Table 4.2.

TABLE 4.2: Initial population of TLBO

Learner

1 3.2 0.4 13.2

2 0.2 2.2 77.8

3 3.1 0.3 14.0

4 1.7 4.6 261.6

5 2.2 .87 43.7

Mean 2.0 1.7


51

In this TLBO the lowest value of the objective function is referred to as the best learner

who is the teacher. The mean value is improved through the effort from the teacher. We

assume , and for and respectively. The differences

of mean for this two variable is calculated by

(4.12)

(4.13)

These differences of mean are added to and for all value of respective columns of

Table 4.2. Table 4.3 shows this new value of and and the respective value of the

objective function .

TABLE 4.3: New values of the objective function for teacher phase

3.5 -0.2 13.3

0.4 1.7 94.2

3.4 -0.2 12.8

1.9 4.0 136.4

2.5 0.3 39.3

These values of objective function of Table 4.1 and Table 4.2 are equated and

minimum value (best value) of are put in Table 4.4. The teacher phase is over for the

TLBO algorithm.

TABLE 4.4: Updated value of objective function for teacher phase

3.2 0.4 13.2 0.2 2.2 77.7 3.4 -0.2 12.7 1.9 4.0 136.4 2.4 0.3 39.3

Here, in the learner phase, any student will interact randomly with other student through

formal communication, presentation, group discussion etc. After interactions, values of

and for learner are updated shown by Table 4.5.

Considering , for and respectively. These new values of

and are estimated by following.


52

After the interaction between learner 1 and learner 2, the transfer of knowledge from

learner 1 to learner two occurs because the value of objective function is better for

learner 1 than learner 2. This updated new values of and for the case of learner 1

are estimated as,

(4.14)

(4.15)

Similarly, after the interaction between learner 5 and learner 2, the knowledge is transfer

from learner 5 to learner 2 because the value of objective function is better for

learner 5 than learner 2. This updated new values of and for the case of learner 2

are estimated as,

(4.16)

(4.17)

TABLE 4.5: New values of the objective function for learner phase

4.6 -9.2 112.3 1&2

1.2 1.6 69.2 2&5

3.5 -0.4 20.6 3&1

2.2 2.8 20.6 4&5

2.8 -0.9 30.2 5&4

These values of objective function of Table 4.4 and Table 4.5 are equated and

minimum value (best value) of objective function is put in Table 4.6. The learner

phase is over for the TLBO algorithm.

TABLE 4.6: Updated values of the objective function for learner phase

3.2 0.4 13.2

1.2 -0.4 69.2

3.5 -0.4 12.4

2.2 2.8 20.7

2.8 -0.9 30.3

After the one iteration, the values of objective function down from 13.2 to 12.4. It

will further slowdown through increasing the number of iterations


53

4.4.5 Comparison of TLBO Result with Other Optimization Techniques

In this section, we perform the experiment to verify the effective ness of TLBO algorithm

as compared to other optimization method. Here we take an example of nonlinear

objective function for maximization having 10 design variables and one constraint which

is given by following

(√ ) ∏

∑

(4.18)

The maximum value of global solution is at

√

√ having objective function

value . We set the TLBO parameter as: population size is 50; maximum number

of iteration is 2000. The simulation result is shown in Table 4.7.

TABLE 4.7: Comparison between results obtained by various optimization methods

Technique Worst Mean Best Function

Evaluations

TLBO 1 1 1 100000

M-ES 1 1 1 240000

CDE 0.63 0.78 0.99 100100

ABC 1 1 1 240000

PESO 0.46 0.76 0.99 350000

It can be noticed from Table 4.7 that global solution evaluated by TLBO have better value

and lower function calculation than solution obtained by other optimization solutions i.e.,

PESO, ABC CDE and M-ES.

TLBO Assisted CSS for Maximizing the Probability of Detection

54

CHAPTER-5

5 TLBO Assisted CSS for Maximizing the

Probability of Detection

5.1 Introduction

This chapter presents the utilization of teaching-learning based optimization (TLBO)

method for optimizing the performance of softened hard(Quantized) data fusion based

cooperative spectrum sensing(CSS). Here the optimization problem is to maximize the

probability of detection for given probability of false alarm based on Neyman-person

criteria and is formulated as an objective function on which TLBO works, the TLBO-

based optimization system is evaluated, and the optimal solutions obtained by TLBO are

analysed to decide on the effectiveness of the proposed methodology.

This chapter begins with system model of proposed TLBO based cooperative spectrum

sensing (CSS) framework in which system model working is explained. Mathematical

equations i.e., probability of detection, probability of false alarm are extensively analysed

and its various cases like equation for false reporting node, imperfect reporting channel,

multi hop cooperative spectrum sensing with the identical and non-identical channel are

evaluated in section II. The details of TLBO algorithm and optimization problem for

maximization of the probability of detection is explained in section III and IV. Finally, this

chapter ends with simulation work carried out to judge the performance proposed TLBO

based CSS under various fading channel, under the imperfect reporting channel and false

reporting CR user. The effect of SNR and CR user on performance as well as convergence

performance of TLBO techniques is compared with the other optimization techniques.

Proposed TLBO Based System Model

55

5.2 Proposed TLBO Based System Model

FIGURE 5.1: Proposed system architecture

The system model for the proposed TLBO based softened hard (quantize) CSS method is

depicted in Fig. 5.1. In this framework, each CR user senses the spectrum locally and

sends its 2-bit information “quantized observation” in the form of index to the fusion

centre (FC) rather than sending the exact local spectrum sensing observation to FC. This

index indicates the energy level for the local observation of CR user. The fusion centre

(FC) makes a global decision based on the index of energy level and weight vector

of corresponding energy level. Here weight is given to corresponding energy level not to

the cognitive radio (CR) user.

𝐻 𝐻

𝑥 𝑘

CR-1

∑|𝑥 𝑘 |

𝑀

𝑘

4 – Level Quantizer

𝑥 𝑘

CR-1

𝑥𝑛 𝑘

CR-1

∑|𝑥𝑛 𝑘 |

𝑀

𝑘

∑|𝑥 𝑘 |

𝑀

𝑘



Sensing Channel

∑

Reporting Channel

𝐿 𝐿𝑛 𝐿

Global Decision

(H0/H1)

FC

PU


56

In Soft combination based data fusion scheme, detection performance is obtained by

allocating different weights to different CR users according to their SNR. In the

conventional one-bit hard combination based data fusion scheme, there is only one

threshold dividing the whole range of the observed energy into two regions. As a

result, all of the CR users above this threshold are allocated the same weight regardless of

the possible significant differences in their observed energies. softened hard(quantized)

two-bit combination based data fusion scheme achieve the better detection performance

and less complexity with two-bit overhead by dividing the whole range of the observed

energy into four regions of energy and allocate a different weights to each region.

5.2.1 Softened Hard Decision Fusion Scheme

FIGURE 5.2: 2-bit softened hard combination scheme

The principle of dividing whole observed energy into different observed space is shown in

Fig. 5.2 which is the case of 2-bit softened hard (quantized) data fusion based cooperative

spectrum sensing. Fig. 5.2 shows this the principle of the two-bit softened hard (quantized)

combination based data fusion scheme. Here, three thresholds and divide the

whole observed energy space into four regions. Each CR user senses the spectrum locally

and sends its 2-bit information “quantized observation” in the form of index (energy

level) to the fusion centre (FC). The fusion centre (FC) makes a global decision based on

the index of energy level and weight vector of corresponding energy level of 2-bit

TLBO based quantized cooperative spectrum sensing (CSS)

Energy(Y) Region 4 W

4

Region 3 W3

Region 2 W2

Region 1 W1

λ3

λ

2

λ1


57

The probability of detection in respective region under hypothesis and using [118]

AWGN channel are given by following

{

(5.1)

In the proposed method, the global decision depends on the number of observation in each

level and the weight vector. For this 2-bit softened hard data fusion scheme, fusion centre

receives the quantized measurements and counts the number of users in each quantization

level which is given by following equations.

] (5.2)

] (5.3)

The decision function is estimated with the help of the number of users in the each energy

level and corresponding weight.

{

(5.4)

The summation of weighted number of user is given by following

∑

(5.5)

Where = Number of observation in region i.

= Weight given in region i.

In (5.5) the weighted summation is formed from the multiplication between the number of

node having their observation in each that region and the weight given in the respective

region. This weighted summation is divided into two group of equal number. The total

number of weighted summation is four for the 2-bit scheme, so two group is having two

weighted summations in each group. If the weighted summation of first two levels is

higher than weighted summation of remaining level, the true state is otherwise true

state is . From (5.5) it has been observed that weight vector permits us to adjust the

system performance and its behaviour. For example, the weight given to each energy level

is equal then it can be form the equal gain combining (EGC) decision fusion rule.


58

5.2.2 Mathematical Statistics of and for TLBO Based CSS

In softened hard combination based data fusion strategy, the cooperative detection

probability and cooperative false alarm probability are evaluated using [91], [119], [66]

which is given by following.

∑ { } (5.6)

∑ ∑ |

(5.7)

We divide the entire range of observed energy into four equal regions. From (5.7), number

of observation for local sensing result in the first region is . Similarly, and are

for region 2, 3 and 4 respectively. In first region there are number of observation out of

total observation so probability of detection is evaluated by taking factorial of .

Likewise, second region have a number of observation out of remaining so

detection probability is evaluated by taking factorial of . In this way the final

cooperative probability of detection is given by following equation

∑ (

) (

) (

) (

)

(5.8)

Similarly, equation for cooperative false alarm probability in softened hard data fusion

based Cooperative spectrum sensing is evaluated by

∑ { } (5.9)

∑ ∑ |

(5.10)

∑ (

) (

) (

) (

)

( )

( )

( )

( )

(5.11)


59

5.2.3 Mathematical Statistics of and with False Report

When a secondary user (SU) reports the existence of primary user signal due to malicious

user or some unavoidable circumstances, then detection performance of proposed method

will degrade. For example, OR logic based hard decision fusion (HDF) rule always

decides the existence of primary user (PU) signal. In this case, = 1 so cognitive user

never utilize the channel at any condition and spectrum utilization efficiency will be zero.

Let us assume that there are false reports in the wireless system. The probabilities of

cooperative detection and false alarm are given by following.

∑ { } (5.12)

∑ ∑ |

(5.13)

∑ (

) (

) (

) (

)

(5.14)

Similarly, equation for false alarm probability in softened hard data fusion based CSS is

evaluated by

∑ { } (5.15)

∑ ∑ |

(5.16)

∑ (

) (

) (

) (

)

( )

( )

( )

( )

(5.17)

In (5.17), N is the total number of cooperative CR, and is the number of

users having their sensing result in respective energy level without taking into

consideration of false reporting node. If the number of false reporting node increases in

the system, it will degrade the performance of cooperative sensing.


60

5.2.4 Mathematical Statistics of and for Single Hope Imperfect Channel

FIGURE 5.3: Binary Symmetric Channel (BSC)

If there is reporting channel with an error at the fusion centre (FC), errors may occur due

to the channel on the data collected by fusion centre (FC) for local spectrum sensing

result. For the case of hard decision fusion (HDF) based CSS, It can be demonstrated as

Binary Symmetric Channel (BSC) with crossover probability as shown in the Fig.5.3

When the PU signal is active (i.e., under the hypothesis ), the FC gets bit „1‟ sent by CR

user when (1) one-bit decision is sent by CR user is „1‟ and the FC receive bit „1‟ through

imperfect reporting channel with probability or (2) the one-bit decision is sent

by CR user is „0‟ and the FC receive bit „1‟ through imperfect reporting channel with

probability On the other hand, when the primary When the PU signal is absent

(i.e., under the hypothesis ), the FC gets bit „1‟ sent by CR when(1) one-bit decision is

sent by CR user is „1‟ and the FC receive bit „1‟ through imperfect reporting channel with

probability or (2) the one-bit decision is sent by CR user is „0‟ and the FC

receive bit „1‟ through imperfect reporting channel with probability .

Therefore, the probability of detection and false alarm for reporting channel with error can

be evaluated as

(5.18)

(5.19)

Optimization Problem

61

5.2.5 Mathematical Statistics of and for Multi Hope Imperfect Channel

In a multi-hop wireless network system, each cooperative node sensing the spectrum

locally for the absence or presence of the primary user (PU) signal and send the result to

the forthcoming hop. Each hop can be formulated as BSC. For example, there are M hops

between the primary user and FC then a channel with (M − 1) non-identically series

connected BSC is represented into equivalent one BSC with equivalent cross-over

probability is given by

( ∏( )

) (5.20)

Channel with (M − 1) identically series connected BSC is represented into equivalent

single BSC with equivalent cross-over probability is given by

] (5.21)

The probability of detection and false alarm for reporting channel with error under a multi-

hop cooperative system can be given as

(5.22)

(5.23)

5.3 Optimization Problem

The main goal of the spectrum sensing is to decide between the two hypotheses,

channel state is empty and channel state is used. The most suitable optimality criteria

for this decision are the Neyman-Pearson criteria that maximize probability of detection

for a given probability of false alarm [120]. Proposed TLBO based system model

needs to optimize the weight vector so our optimization problem is given by following.

Optimization Problem 1:


62

Different values of weight vector affect the value. Teaching learning based

optimization (TLBO) algorithm is used to optimize the weighting coefficients vector of

observed energy level of sensing information. The TLBO optimization technique evaluate

optimal weighting coefficient vector so that the probability of detection is improved for

given probability of false alarm under the Neyman-Pearson criteria.

5.4 TLBO Based Solution

FIGURE 5.4: Flow Chart of TLBO based solution

A novel teaching-learning based optimization (TLBO) method is first proposed by prof.

Rao [115],[116],[121] which is based on the process of teaching and learning. TLBO

method is also population based evolutionary techniques which mimic the impact of a

teacher on learners (student). The class of learner can be considered as a population and

different design variables are different subjects. Learners‟ outcome is similar to the fitness

value of the objective function. In the entire population, the best solution is considered as

the teacher. TLBO method is effective and stable and also gives better optimum solutions

than other method in wireless communication system. The flowchart of TLBO based

solution for given optimization problem is shown Fig. 5.4. The detail description of

teaching-learning based optimization (TLBO) technique is given in chapter-4.

Initialization

TLBO Optimizer

Is Terminate?

𝑃𝑑 , Optimal vector ��

Yes

No

Channel, SNR, N, λ, Iteration,

Population size

TLBO Assisted Weighted Fusion Algorithm for

63

5.5 TLBO Assisted Weighted Fusion Algorithm for

Teaching-learning-based optimization (TLBO) algorithm is used to find optimal weight

vector for softened hard (quantized) data fusion based cooperative spectrum sensing

(CSS). The Pseudo of TLBO algorithm is shown in Fig. 5.5 in which TLBO algorithm

find the optimal solution until the predefined iteration for TLBO reached using WHILE

loop. The FOR loop is executed to implement the teacher phase and student phase and the

best solution is obtained through IF loop. Teaching-learning based optimization (TLBO)

algorithm is robust and effective algorithm [122],[123]. It gives better optimum solutions

than another evolutionary optimization algorithm i.e. Genetic Algorithm (GA), Ant colony

optimization (ACO) and Particle swarm optimization (PSO) etc.

{ }

Algorithm 1: Weight Optimization Using TLBO for

Input: Channel, SNR, N, λ, Iteration, Population size

Output: , Optimal vector Initialize the population size and Generation

While

{ } Find the mean of each design variable

Identify the best solution as teacher

] For

Calculate { }] ]

Calculate ( )

If ( ) then

End If { }

Select a learner randomly such that

If ( ) then

Else

End If

If ( ) then

End If { }

End For End While

FIGURE 5.5: Pseudo code of TLBO algorithm


64

5.6 Simulation Results and Discussion

5.6.1 System Parameters and Performance Metrics

The probability of detection is very important parameter which is useful to judge and

compare the performance of proposed TLBO based cooperative spectrum sensing with

other method. is inversely proportional to probability of miss detection . Having low

probability of detection mean high chance of miss detection which consequences the

misguidance of CR user in leaving the channel during primary user want to access the

channel. On the other hand, probability of false alarm indicates the spectral efficiency

performance. High results in low employment of free channel. Therefore, the trade-off

between and must be weighted. Receiver Operating Characteristic (ROC) is used for

the performance examination between and .In this simulation, we consider the

system parameter: average signal to noise ratio (SNR), time into bandwidth product (TW)

and number of cooperating users (N).

5.6.2 Performance Analysis on Under Fading Channels

The performance has been access by taking simulation of proposed TLBO based softened

hard data fusion based cooperative sensing. The optimal weights obtained by TLBO

algorithms are used to plot the receiver operating characteristics (ROC) curve shown in

Fig. 5.6 for AWGN channel and Fig. 5.7 for Rayleigh channel. In this simulation we have

set time-bandwidth product , the channel is AWGN and Rayleigh, the number of

received signal samples , average signal to noise ratio = 5 dB and number of

cooperating user is 10. In TLBO, we have used the number of particles and

. We have assumed perfect reporting channels and there is no false

reporting node. Simulation result shows the comparison between proposed TLBO based

CSS and hard decision fusion technique i.e. MAJORITY, OR, AND rules etc. as well as

the soft data fusion techniques i.e. EGC. TLBO algorithm is effective and powerful in

wireless communication system. From the simulation result, it is observable that the

performance of TLBO based method is out perform with convention hard decision fusion

technique and almost same to EGC with low overhead.

Simulation Results and Discussion

65

FIGURE 5.6: ROC graph for AWGN channel

FIGURE 5.7: ROC graph for Rayleigh channel

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

PF

PD

NoCoop

OR Logic

Majority Logic

AND Logic

TLBO

EGC

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

PF

PD

NoCoop

OR Logic

Majority Logic

AND Logic

TLBO

EGC


66

FIGURE 5.8: Comparison of TLBO method with PSO method

The optimal weights obtained by TLBO as well as particle swarm optimization (PSO)

algorithms are used to plot the receiver operating characteristics (ROC) shown in Fig. 5.8

which illustrates a comparison between TLBO based method and PSO based method. In

this simulation we have taken simulation parameter: time-bandwidth product , the

channel is Rayleigh, SNR = 5dB and number of cooperating user . It is noticeable

that TLBO based Cooperative spectrum sensing have an improvement in performance as

compared to particle swarm optimization (PSO) based CSS for given false alarm .

The TLBO, as well as particle swarm optimization (PSO) and genetic algorithm (GA)

algorithms, are used to plot the complementary receiver operating characteristics (C-ROC)

curve shown in Fig. 5.9 and Fig. 5.10 respectively. From the comparison between TLBO-

based, GA-based and PSO-based scheme, it is observable that TLBO based method

outperforms all other optimization methods with a considerable deference for performance

improvement for given probability of false alarm . The C-ROC curve and comparison

with various optimization methods are also plotted on semi log axis shown in Fig. 5.11

and Fig. 5.12 respectively.

10-3

10-2

10-1

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

PF

PD

PSO

TLBO


67

FIGURE 5.9: C-ROC graph for Rayleigh channel

FIGURE 5.10: Comparison of TLBO method with GA and PSO method

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

PF

Pm

NoCoop

OR Logic

Majority Logic

AND Logic

TLBO

EGC

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 10

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

PF

Pm

GA

PSO

TLBO


68

FIGURE 5.11: C-ROC graph for Rayleigh channel in semi log axis

FIGURE 5.12: Comparison of TLBO method with other method in semi log axis

10-2

10-1

100

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

PF

Pm

NoCoop

OR Logic

Majority Logic

AND Logic

TLBO

EGC

10-2

10-1

100

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

PF

Pm

GA

PSO

TLBO


69

FIGURE 5.13: versus Nakagami fading parameter (m)

Analysis for the effect of Nakagami fading parameter for a given =0.01 on the

performance of proposed TLBO based method is illustrated in Fig. 5.13. The above graph

shows that when the channel between the PU and the CR is Nakagami frequency selective

channel (NFS), probability of detection improves with so minimum interference with

CR user is more confirmed. It is observed from Fig. 5.13 that initially increase

potentially during m from 1 to 5. After that it improved very slowly and then became

stable for high values .

5.6.3 Influence of Number of Cooperating Users (N) on

Analysis for the effect of a number of user on the performance of proposed TLBO based

method is illustrated in Fig. 5.14. In this simulation we have set time-bandwidth product

, the channel is Rayleigh, SNR = 5dB, received signal samples and

=0.01. In TLBO, we have used the number of particles and .

1 2 3 4 5 6 7 8 9 100

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Nakagami fading parameter (m)

PD

NoCoop

Majority Logic

AND Logic

TLBO

EGC


70

FIGURE 5.14: Number of CR versus graph under Rayleigh channel

Growing the number of the cooperating users, increase performance. From the Fig. 5.14 it

has been observed that if a number of cooperative user is 12 then the performance of OR

logic and Majority Logic is same. Thus, if number of user increases, the demarcation

between the hypotheses and raises and the performance of the ROC graph

progresses accordingly

5.6.4 Influence of Signal to Noise ratio (SNR) on

The Effect of SNR for a given =0.01 on performance of proposed method is analysed in

Fig.5.15. In this simulation we have taken the simulation parameters , the channel

is Rayleigh, signal samples and number of cooperating user = 10. In TLBO

algorithm, we have used the number of particles and . From this

graph it has been shown that as SNR increases, performance of proposed TLBO based

method increases. From the graph it has been observed that when SNR value is 10 dB then

performance of AND logic and no cooperation are same.

6 7 8 9 10 11 120

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Number of CR(s)

PD

NoCoop

OR Logic

Majority Logic

AND Logic

TLBO

EGC


71

FIGURE 5.15: SNR versus graph under Rayleigh channel

5.6.5 Convergence Performance of TLBO Based CSS on

The comparison of convergence speed of first 30 iterations between TLBO based softened

hard data fusion based cooperative spectrum sensing(CSS) method and particle swarm

optimization(PSO) based softened hard data fusion CSS method for a given =0.01 are

shown in Fig. 5.16. In TLBO, we have used the number of particles and

. The TLBO based method achieve probability of detection = 0.6 for a

given =0.01 with 10 iteration while PSO based method required 15 iteration to achieve

the same probability of detection for given false alarm rate. It is noticeable that TLBO-

based optimization technique converges after the 10 iterations while the convergence for

PSO-based optimization is achieved after 15. TLBO algorithm is robust and effective in

wireless communication system. From the simulation result, it can be conclude that the

TLBO based method has fast convergence speed as compared to PSO based method which

ensures to meet real time requirement of cooperative spectrum sensing.

0 1 2 3 4 5 6 7 8 9 100

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

SNR (dB)

PD

NoCoop

OR Logic

Majority Logic

AND Logic

TLBO


72

FIGURE 5.16: Performance of TLBO-based method

The performance evaluation of TLBO algorithm for achievement the probability of

detection are shown in next figures for a given =0.01. In this simulation result, and

number of iteration is taken in x and y axis respectively. It has been obviously observed

that increasing the number of iteration improve that detection performance for a given

. TLBO method offers a less computational complexity compared to other method.

Table 5.1 shows the summary of the achievement of the detection probability after

iteration. The convergence performances on probability of detection for given false alarm

rate over the various iterations are shown in Fig. 5.17 to Fig. 5.24. There is not significant

improvement in detection probability after the 25 iteration shown in this figures.

TABLE 5.1: Performance matrix of TLBO based CSS on

0 5 10 15 20 25 300

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Iteration

PD

PSO

TLBO

Sr. No Iteration Detection probability( )

1 2 0.5

2 5 0.61

3 10 0.62

4 15 0.662

5 20 0.665

6 30 0.67

7 40 0.67

8 50 0.67


73

FIGURE 5.17: Convergence performance on over 2 Iteration


0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2-0.5

0

0.5

1

1.5

2

Number of Iteration

PD

1 1.5 2 2.5 3 3.5 4 4.5 50.55

0.56

0.57

0.58

0.59

0.6

0.61

0.62

Number of Iteration

PD


74



1 2 3 4 5 6 7 8 9 10

0.57

0.58

0.59

0.6

0.61

0.62

0.63

Number of Iteration

PD

0 5 10 150.59

0.6

0.61

0.62

0.63

0.64

0.65

0.66

0.67

Number of Iteration

PD


75



0 2 4 6 8 10 12 14 16 18 20

0.58

0.59

0.6

0.61

0.62

0.63

0.64

0.65

0.66

0.67

Number of Iteration

PD

0 5 10 15 20 25 300.654

0.656

0.658

0.66

0.662

0.664

0.666

0.668

0.67

0.672

Number of Iteration

PD


76



0 5 10 15 20 25 30 35 400.625

0.63

0.635

0.64

0.645

0.65

0.655

0.66

0.665

0.67

0.675

Number of Iteration

PD

0 5 10 15 20 25 30 35 40 45 50

0.58

0.59

0.6

0.61

0.62

0.63

0.64

0.65

0.66

0.67

Number of Iteration

PD


77

5.6.6 Performance Analysis on Against False Report

FIGURE 5.25: ROC graph including false reporting node

In cooperative spectrum sensing the false reporting node due to malicious secondary user

change the global decision logic which affects the whole performance of cooperative

spectrum sensing (CSS). The effect of false reporting CR user on performance of TLBO

based cooperative spectrum sensing is analysed in Fig. 5.25.

In this simulation we have set time-bandwidth product , the channel is AWGN,

received signal samples , average signal to noise ratio = 5 dB and no of

cooperating user is 10. In teaching learning based optimization (TLBO) algorithm, we

have taken the number of particles and . In ROC graph for the

case of false reporting node, and false alarm rate is taken in x and y axis respectively. It

has been observed that the false reporting node will decrease the cooperative detection

probability of the system for a given false alarm rate due to malicious secondary user

(SU). Similarly, higher number of false reporting nodes FR will change the global

decision logic at fusion centre (FC) and also degrade the system performance of CSS.

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

PF

PD

TLBO, FR=0

TLBO, FR=2


78

5.6.7 Influence of Number of False Report on

FIGURE 5.26: Number of false reporting nodes versus curve

The analysis of the effect of number of false reporting nodes on the detection performance

of proposed TLBO based Cooperative spectrum sensing is shown in Fig.5.26. In this

simulation, we set the following simulation parameter.

TABLE 5.2: Simulation parameter for FR versus curve

Sr. Simulation parameter Value

1 Time-Bandwidth product(TW) 5

2 Channel AWGN

3 Received signal samples (M) 2u

4 SNR 5 dB

5 Number of CR Node (N) 10

6 False alarm rate( 0.01

7 Reporting Channel Error ( 0

From the above simulation result, and number of false reporting node is taken in x and

y axis respectively. In TLBO algorithms, we have used the number of particles is 15 the

total number of iteration is 25. It has been obviously observed that increasing the number

of false reporting nodes degrade the detection performance for a given .

0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 50

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

FR User

PD

TLBO


79

5.6.8 Performance Analysis on Against Imperfect Channel

FIGURE 5.27: ROC graph including imperfect reporting channel

The receiver operating receiver operating characteristics (ROC) curve including imperfect

reporting channel is shown in Fig. 5.27 for Rayleigh channel. In this simulation we have

set time-bandwidth product , the channel is AWGN, received signal samples

, average signal to noise ratio = 5 dB and no of cooperating user N is 10, the

probability of error in reporting channel = 0.01. In TLBO algorithms, we have used the

number of particles and .

If there is reporting channel with an error at the fusion centre (FC), this error will mainly

affect the global decision logic for the case of hard decision fusion (HDF) because of the

single sensing information. It also affects the performance of soft data fusion (SDF) based

CSS. From the above simulation result, and false alarm rate including imperfect

reporting channel is taken in x and y axis respectively. It is observable that when the

imperfect reporting channel error is 0.01 the probability of detection will be approximately

decrease 0.03. It has been concluded that the detection performance of TLBO based

method is degraded in imperfect reporting channel.

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

PF

PD

TLBO Reporting Channel with Error

TLBO Reporting Channel without Error


80

5.6.9 Influence of Reporting Channel Error on

FIGURE 5.28: Reporting channel error versus curve

The analysis of the reporting channel error probability on the detection performance of

proposed TLBO based Cooperative spectrum sensing (CSS) is shown in Fig.5.28. In this

simulation, we have taken the following simulation parameter.

TABLE 5.3: Simulation parameter for versus curve



2 Channel AWGN


4 SNR 5 dB


6 False alarm rate( 0.01

7 Number of false report 0

From the above simulation result, and reporting channel error probability is taken in x

and y axis respectively. In TLBO algorithms, we have used the number of particles is 15

the total number of iteration is 25. It has been obviously observed that increasing reporting

channel error probability degrade the detection performance for a given .

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.090

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Reporting Channel Eroor Probability

PD

Majority Logic

AND Logic

TLBO

Chapter Summary

81

5.7 Chapter Summary

In this chapter, we used the teaching-learning based optimization (TLBO) algorithm to

optimize the weighting vector of observed energy level of sensing information. The

TLBO algorithm evaluates the best weighting vector so that the detection probability is

improved under constant false alarm rate (CFAR). The performance of the proposed

TLBO based method is analysed and compared with conventional HDF and SDF based

CSS schemes as well as PSO and GA based CSS through simulations. Simulation result

shows that performance of TLBO based method is better than conventional HDF scheme

and close to SDF scheme with low overhead in various fading channel. Proposed TLBO

based CSS scheme is effective and stable and also shows better convergence output which

confirms the computation complexity is lower compared to GA and PSO based method.

The better convergence output meets the real time requirement of cooperative spectrum

sensing in cognitive radio network. Finally the strength of proposed method is also

analysed under the false reporting node and imperfect reporting channel.

TLBO Assisted CSS for Minimizing the Cooperative Sensing Error

82

CHAPTER-6

6 TLBO Assisted CSS for Minimizing the

Cooperative Sensing Error

6.1 Introduction

In this chapter, TLBO method is used for optimizing the performance of Cooperative

spectrum sensing (CSS). Here the optimization problem is to minimize the probability of

sensing error under Mini-Max criteria and is targeted as objective function on which

TLBO algorithm works, solutions obtained by TLBO are analysed to judge effect of the

proposed methodology.

There are two types of the detection errors in CR system first one is the probability of miss

detection and second one is probability of false alarm , which reduce the sensing

performance. Evaluation of the optimal weight vector for the minimization of sensing

error is challenging task that affect the performance. In our approach we used TLBO

algorithm which evaluate the best weight vector that minimize the all over sensing error

based on Mini-Max criteria.

This chapter starts with system model of proposed TLBO based CSS framework for

minimization of sensing error. Mathematical equations, i.e., probability of error is

extensively analysed and evaluated for proposed method. The details of TLBO algorithm

and optimization problem for minimization of the probability of detection error is

explained in section III and IV. Finally this chapter ends with Sensing-Throughput Trade-

off case and simulation work carried out to judge the performance proposed TLBO.


83

6.2 Proposed TLBO Based System Model

FIGURE 6.1: System model for minimization of sensing error

The system model for the TLBO based softened hard (quantize) CSS method is depicted

in Fig. 6.1. In this framework, each CR user senses the spectrum locally and sends its 2-bit

information “quantized observation” in the form of index to the fusion centre (FC).

This index indicates the energy level for the local observation of CR user. The fusion

centre (FC) makes a global decision based on the index of energy level and weight

vector of corresponding energy level.

The basic difference between this framework and convention one-bit hard fusion is the

division of entire sensing energy through three thresholds and . In 2-bit scheme

we divide whole energy into four weighted level through three thresholds while in one-bit

hard decision fusion (HDF) based cooperative spectrum sensing scheme, there is only one

threshold dividing the whole range of the observed energy into two regions. As a

result, all of the CR users above this threshold are allocated the same weight regardless of

the possible significant differences in their observed energies. The details description

about this framework is given in chapter 5.

Rep

ortin

g C

han

nel

∑

FC

Global Decision

(H0/H1)

𝐿

𝐿

𝐿𝑛

Lis

ten

ing

Ch

an

nel

Local Sensing ∑|𝑥 𝑘 |

𝑀

𝑘

4 – Level

Quantizer

Local Sensing

∑|𝑥 𝑘 |

𝑀

𝑘

4 – Level

Quantizer

Local Sensing

∑|𝑥𝑛 𝑘 |

𝑀

𝑘

4 – Level

Quantizer

PU

Tx


84

6.3 Mathematical Statistics of for TLBO Based CSS

The probability of cooperative detection and probability of cooperative false alarm are

evaluated using [115], [121] which is given by

∑ (

) (

) (

) (

)

(6.1)

∑ (

) (

) (

) (

)

( )

( )

( )

( )

(6.2)

The overall probability of error can be represented as

(6.3)

(6.4)

(6.5)

It is observable that the probability of detection and probability of error is highly

dependent on . Therefore, the optimal solution is the weighting vectors that minimize

the total probability of error.

6.4 Optimization problem for

The main aim of the spectrum sensing is to decide between the two hypotheses,

channel state is empty and channel state is used. The most suitable optimality criteria

for this decision is the Mini-Max criteria that minimize the probability of error. Proposed

TLBO based system model needs to optimize the weight vector so our optimization

problem is given by following.

Optimization Problem 2:

TLBO Assisted Weighted Fusion Algorithm for

85

In the optimization problem 2, we need to minimize the maximum sensing error in

cooperative spectrum sensing (CSS) which ensured to called mini-max criteria. Different

values of weight vector in respective energy level affect the value. Teaching-learning

based optimization (TLBO) algorithm is used to optimize the weighting coefficients

vector of observed energy level of sensing information. The TLBO optimization technique

evaluate optimal weighting coefficient vector so that the probability of error is

minimized under the Mini-Max criteria.

6.5 TLBO Assisted Weighted Fusion Algorithm for

Algorithm 2: Weight Optimization Using TLBO for

Input: Channel, SNR, N, λ, Iteration, Population size

Output: , Optimal vector Initialize the population size and Generation

While

{ } Find the mean of each design variable

Identify the best solution as teacher

] For

Calculate { }] ]

Calculate ( )

If ( ) then

End If { }

{ } Select a learner randomly such that

If ( ) then

Else

End If

If ( ) then

End If { }

End For End While

FIGURE 6.2: TLBO algorithm for minimization of sensing error


86

6.6 Sensing-Throughput Trade-off Analysis

FIGURE 6.3: Frame structure for CR networks with spectrum sensing

The frame structure of cognitive radio with sensing time τ is shown in the Fig.6.3. The

frame made up from sensing time following by transmission time T. If the sensing time is

τ and the frame time is then and is the throughput of the secondary network under

the present and absent of PU respectively. The cognitive radio network use the primary

user channel if the fusion centre(FC) decide that the channel is free, this will happens in

two cases

1. When the licence user does not use the channel and FC decide that channel is free,

the probability of that case can be given by

| (6.6)

2. When the licence user use the channel and FC decide that channel is free, the

probability of that case can be given by

| (6.7)

The Throughput of system can be expressed as

(

) ( ) ] (6.8)

6.6.1 Constant Primary User Protection (CPUP)

In constant primary user protection, the objective is to find optimal sensing time so that

throughput of secondary network is maximizing which is given by following.

Trade-off Map for Cooperative Spectrum Sensing Parameter

87

6.6.2 Constant Secondary User Spectrum Utilization (CSUSU)

In constant secondary user spectrum usability, the objective is to find optimal sensing time

so that throughput of secondary network is maximizing which is given by following.

6.7 Trade-off Map for Cooperative Spectrum Sensing Parameter

FIGURE 6.4: Trade-off diagram


88

The trade-offs map[124] is shown in Fig. 6.4. The following example explains how to

operate this map. Study the number of samples, N. Then if N rises, there are three cases

1. will decrease due to inversely proportional to it therefore it is good for system

2. will increase so It is directly proportional to N therefore it is good for system.

3. τ will increase, so it is for bad for the system

6.8 Simulation Result and Discussion

6.8.1 Performance Analysis of TLBO Based CSS on

FIGURE 6.5: Lambda versus total sensing error curve

The curve of λ versus total sensing error for different data fusion scheme is shown in

Fig. 6.5. It can be clearly observed, the TLBO-based method search the best weighting

coefficients vector which lead to minimized probability of error for CSS compared to

0 5 10 15 20 25

0.4

0.5

0.6

0.7

0.8

0.9

1

Lambda

Pe

NoCoop

Or Logic

Majority Logic

AND

TLBO

EGC

Simulation Result and Discussion

89

other schemes. On the other hand, conventional hard decision fusion (HDF) based

spectrum sensing provides the worst error performance because of very less amount of

data received from secondary user (SU) in the network.

FIGURE 6.6: Lambda versus total sensing error curve

The Teaching learning based optimization (TLBO) as well as particle swarm optimization

(PSO) algorithm is used to plot lambda versus total sensing error curve which is shown

in Fig. 6.6. From the comparison between TLBO-based and PSO-based scheme, it is

observable that TLBO based method outperforms the PSO based with a considerable

differences of probability of total sensing error. From the above simulation result, lambda

and sensing error is taken in x and y axis respectively. In the above result, the TLBO

based method have a sensing error = 0.39 for a given lambda=10 while PSO based

method have a sensing error = 0.39 for the same value of given lambda. TLBO

algorithm is robust and effective algorithm and also gives better optimum solutions than

other evolutionary optimization algorithm. From the Fig. 6.6, it can be conclude that the

TLBO technique search optimal weight vector compared to PSO method so that total

sensing error is minimized which meets the one of the prime requirement of cooperative

spectrum sensing in cognitive radio network.

0 5 10 15 20 25

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1

Lambda

Pe

PSO

TLBO


90

6.8.2 Influence of Sensing Time on Throughput

FIGURE 6.7: Normalized throughput versus sensing time (ms) curve

The performance analysis of the effect of sensing time in on the normalize throughput

of a secondary user (SU) network is shown in Fig.6.4. From the graph it has shown that

when sensing time is increasing then the corresponding throughput of secondary user

network increases and after some optimum point the throughput decreases. It is observed

that when the sensing time is 5 then maximum throughput put is achieved. Long

sensing time causes the reduction of effective throughput of secondary user network due

to the probability of arrival of primary user signal during the sensing time. In this

simulation we set the following simulation parameter.

TABLE 6.1: Simulation parameter of sensing time versus throughput curve



2 Channel AWGN


4 SNR 5 dB


6 Detection probability( 0.9

7 Frame duration T 100ms

1 2 3 4 5 6 7 8 90

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

Sensing Time

Th

rou

gh

pu

t


91

6.8.3 Convergence Performance of TLBO Based CSS on

FIGURE 6.8: Iteration versus total sensing error curve

The comparison of the convergence speed between TLBO based softened hard decision

fusion scheme and PSO based softened hard data fusion for CSS for a given is

shown in Fig. 6.8. This is noticeable that TLBO-based optimization technique converges

after the 10 iterations while the convergence for PSO-based optimization requires larger

iteration time for the same value of lambda. TLBO based method have fast convergence

speed as compared to PSO based method which ensure to meet real time requirement CSS

with low sensing error. The achievement the probability of error for different value of

lambda is shown in next figures. Table 6.2 shows the summary of this result.

TABLE 6.2: Performance matrix of TLBO based CSS

Sr.

No Lambda

Sr.

No Lambda

1 0 0.46 8 14 0.39

2 2 0.43 9 16 0.4

3 4 0.41 10 18 0.44

4 6 0.4 11 20 0.48

5 7 0.38 12 22 0.50

6 10 0.38 13 24 0.57

7 12 0.39

0 5 10 15 20 25 300

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Iteration

Pe

PSO

TLBO


92

FIGURE 6.9: Convergence performance on for


0 5 10 15 20 25 300.46

0.47

0.48

0.49

0.5

0.51

0.52

0.53

Number of Iteration

Pe

0 5 10 15 20 25 300.425

0.43

0.435

0.44

0.445

0.45

0.455

0.46

Number of Iteration

Pe


93



0 5 10 15 20 25 300.4

0.45

0.5

0.55

0.6

0.65

Number of Iteration

Pe

0 5 10 15 20 25 300.35

0.4

0.45

0.5

0.55

0.6

0.65

0.7

Number of Iteration

Pe


94



0 5 10 15 20 25 300.35

0.4

0.45

0.5

0.55

0.6

Number of Iteration

Pe

0 5 10 15 20 25 300.35

0.4

0.45

0.5

0.55

0.6

Number of Iteration

Pe


95



0 5 10 15 20 25 300.38

0.4

0.42

0.44

0.46

0.48

0.5

Number of Iteration

Pe

0 5 10 15 20 25 300.39

0.4

0.41

0.42

0.43

0.44

0.45

0.46

0.47

0.48

0.49

Number of Iteration

Pe


96



0 5 10 15 20 25 300.4

0.45

0.5

0.55

0.6

0.65

0.7

0.75

Number of Iteration

Pe

0 5 10 15 20 25 300.4

0.45

0.5

0.55

0.6

0.65

0.7

0.75

Number of Iteration

Pe


97



0 5 10 15 20 25 300.48

0.49

0.5

0.51

0.52

0.53

0.54

0.55

0.56

0.57

Number of Iteration

Pe

0 5 10 15 20 25 30-0.5

0

0.5

1

1.5

2

Number of Iteration

Pe


98


6.9 Chapter Summary

In this chapter, we used the teaching learning based optimization (TLBO) algorithm to

optimize the weighting coefficients vector of observed energy level of sensing

information. The TLBO algorithm estimates the optimal weight vector so that probability

of sensing error is minimized under the Mini-Max criteria. The performance of the

TLBO method is analysed and compared with conventional HDF and SDF based CSS

schemes. Simulation result shows that proposed TLBO based method have a low

probability of sensing error as compared convention HDF and SDF scheme. Simulation

result confirm that proposed method have better convergence output as compared to GA

and PSO based method and have a lower computation complexity. Finally, an analytical

evaluation for sensing-throughput trade-off is also presented in this chapter.

0 5 10 15 20 25 300.55

0.6

0.65

0.7

0.75

0.8

0.85

0.9

0.95

Number of Iteration

Pe

Conclusion

99

CHAPTER-7

7 Conclusions and Scope of Future Work

7.1 Conclusion

The cognitive radio is developed to overcome the radio spectrum scarcity issue that puts

the limitation of wireless communication application. Dynamic spectrum allocation is

enabled by cognitive radio (CR) in which spectrum holes or idle channel is sense and

opportunistically utilized when the primary user are idle. Once the primary user want to

use the channel, then secondary user (SU) immediately release the occupied channel and

again repeat this process for new channel. This is achieved by adapting transmission

parameter of CR based on awareness about surrounding wireless environment and

preserving minimum interference level to the primary user (PU), Consequently, spectrum

sensing has becomes a key function of cognitive radio, in order to improve this spectrum

utilization efficiently. This fact has motivated the research in the development of well-

organized spectrum sensing framework.

The optimal cooperative spectrum sensing framework is propsed for cognitive radio which

maximizes the detection performance and minimized the cooperative overhead and also

maintaining the minimum sensing errors. We studied conventional data fusion techniques,

i.e., hard decision fusion (HDF) and soft data fusion (SDF) of cooperative spectrum

sensing and show up that there is trade-off between overhead and performance through

simulation. To reduce the overhead, we use energy detection based softened hard

(quantized) cooperative spectrum sensing (CSS) technique which quantizes its local

observations and sends to FC to prepare the final decision

Conclusions and Scope of Future Work

100

In this research work, use of teaching learning based optimization (TLBO) algorithm as a

significant method is proposed to optimize the weighting coefficients vector of observed

energy level of sensing information so that probability of detection is maximized and

minimized the sensing error with low overhead. The performance of the proposed TLBO

based cooperative spectrum sensing (CSS) framework is extensively analysed and

compared with conventional HDF and SDF based CSS schemes in the environment of

various fading channel. The performance of TLBO algorithm for CSS is also compared

with other optimization techniques i.e. PSO, GA for CSS through simulations.

From the simulation result, it can be concluded that proposed TLBO based CSS scheme is

effective, stable and robust. It gives better performance than conventional HDF based CSS

and close to SDF based CSS with low overhead. It gives the better convergence output

compared to genetic algorithm (GA) and particle swarm optimization (PSO) based method

which confirms the lower computation time and complexity of TLBO based framework.

7.2 Scope of Future Work

There are some recommendations and potential research directions that extend this work

as the scope of future work. The summaries for future scope of this work are following

It is also worth to apply other optimization algorithms which may give better

convergence output and free from algorithm parameter.

This work can be extended with the optimization of other parameter i.e. sensing

time, cooperative user etc.

Proposed TLBO based cooperative spectrum sensing framework can be extended

with multiple cognitive networks

Performance of proposed method can be evaluated with user mobility

Sensing-throughput trade-off problem can be extended to the case of the primary

user network by including the effect of the Noise uncertainty (NU) on the primary

user‟s receiver.

List of References

101

List of References

[1] D. Čabrić, S. M. Mishra, D. Willkomm, R. W. Brodersen, and A. Wolisz, “A Cognitive Radio

Approach for Usage of Virtual Unlicensed Spectrum,” 14th IST Mob. Wirel. Commun. Summit.,

Dresden, Ger., pp. 1–4, 2005.

[2] FCC, “ET Docket No. 02- 135 November 2002 Federal Communications Commission,”

Communications, no. 02–135, pp. 1–73, 2002.

[3] D. A. Roberson, C. S. Hood, J. L. LoCicero, and J. T. MacDonald, “Spectral Occupancy and

Interference Studies in support of Cognitive Radio Technology Deployment,” in 2006 1st IEEE

Workshop on Networking Technologies for Software Defined Radio Networks, 2006, pp. 26–35.

[4] Q. Zhao and B. M. Sadler, “A Survey of Dynamic Spectrum Access,” IEEE Signal Process. Mag.,

vol. 24, no. 3, pp. 79–89, May 2007.

[5] F. C. Commission and others, “Second Report and Order and Memorandum Opinion and Order In

the Matter of Unlicensed Operation in the TV Broadcast Bands,” Addit. Spectr. Unlicensed Devices

Below, vol. 900.

[6] C. Of, T. H. E. American, S. For, and H. Engineering, “The American Society for Healthcare

Engineering of the American Hospital Association („ASHE‟), pursuant to Section 1.415 of the

FCC‟s rules, hereby submits comments on the,” vol. 9589, no. 8, pp. 1–5, 2009.

[7] F. Granelli et al., “Standardization and Research in Cognitive and Dynamic Spectrum Access

Networks : IEEE SCC41 Efforts and Other Activities Standardization and Research in Cognitive and

Dynamic Spectrum Access Networks : IEEE SCC41 Efforts and Other Activities,” Commun. Mag.

IEEE, vol. 48, no. 1, pp. 71–79, 2010.

[8] S. H. Hwang, J. S. Um, M. S. Song, C. J. Kim, H. R. Park, and Y. H. Kim, “Design and Verification

of IEEE 802.22 WRAN Physical Layer,” in 2008 3rd International Conference on Cognitive Radio

Oriented Wireless Networks and Communications (CrownCom 2008), 2008, pp. 1–6.

[9] S. Haykin, “Cognitive radio: Brain-empowered wireless communications,” IEEE J. Sel. Areas

Commun., vol. 23, no. 2, pp. 201–220, 2005.

[10] M. Wellens, A. D. Baynast, and P. Mahonen, “Performance of dynamic spectrum access based on

spectrum occupancy statistics,” IET Commun., vol. 2, no. 6, pp. 772–782, Jul. 2008.

[11] M. A. McHenry, “NSF spectrum occupancy measurements project summary,” Shar. Spectr. Co.,

2005.

[12] M. A. McHenry, P. A. Tenhula, D. McCloskey, D. A. Roberson, and C. S. Hood, “Chicago

Spectrum Occupancy Measurements & Analysis and a Long-term Studies Proposal,” in Proceedings

of the First International Workshop on Technology and Policy for Accessing Spectrum, 2006.

[13] F. Miguel Lopez-Bebtiez, R. Casadevall, Anna Umbert and Jordi Perez-Romero, J. P. Hachemani,

and M. Christophe, “Spectral occupation measurements and blind standard recognition sensor for

cognitive radio networks,” Proc. 2009 4th Int. Conf. Cogn. Radio Oriented Wirel. Networks Commun. CROWNCOM 2009, 2009.

[14] M. Matinmikko, M. Mustonen, M. Hoythtya, T. Rauma, H. Sarvanko, and A. Mammela,

“Distributed and directional spectrum occupancy measurements in the 2.4 GHz ISM band,” in 2010

7th International Symposium on Wireless Communication Systems, 2010, pp. 676–980.

[15] M. Wellens and P. Mähönen, “Lessons Learned from an Extensive Spectrum Occupancy

Measurement Campaign and a Stochastic Duty Cycle Model,” Mob. Netw. Appl., vol. 15, no. 3, pp.

461–474, Jun. 2010.

List of References

102

[16] T. M. Taher, R. B. Bacchus, K. J. Zdunek, and D. A. Roberson, “Long-term spectral occupancy

findings in Chicago,” 2011 IEEE Int. Symp. Dyn. Spectr. Access Networks, DySPAN 2011, pp. 100–

107, 2011.

[17] E. Biglieri, L. J. Greenstein, N. B. Mandayam, and H. V. Poor, Principles of cognitive radio.

Cambridge University Press, 2013.

[18] T. Yucek and H. Arslam, “A Survey of Spectrum Sensing Algorithms for Congnitive Radio

Applications,” Proc. IEEE, vol. 97, no. 5, pp. 805–823, 2009.

[19] R. Tandra, S. M. Mishra, and A. Sahai, “What is a Spectrum Hole and What Does it Take to

Recognize One?,” Proc. IEEE, vol. 97, no. 5, pp. 824–848, May 2009.

[20] Y. Zeng, Y.-C. Liang, A. T. Hoang, and R. Zhang, “A Review on Spectrum Sensing for Cognitive

Radio: Challenges and Solutions,” EURASIP J. Adv. Signal Process, vol. 2010, p. 2:2--2:2, Jan.

2010.

[21] D. D. Ariananda, M. K. Lakshmanan, and H. Nikookar, “A survey on spectrum sensing techniques

for cognitive radio,” in 2009 Second International Workshop on Cognitive Radio and Advanced

Spectrum Management, 2009, pp. 74–79.

[22] J. Lundén, “Spectrum sensing for cognitive radio and radar systems,” Helsinki University of

Technology (TKK), Espoo, Finland, 2009.

[23] S. Chaudhari and others, “D3. 1--Spectrum sensing algorithm evaluation,” Webpage, May, 2010.

[24] K. Kokkinen and others, “D3. 2--Test report on sensing algorithms,” Webpage, Sept., 2010.

[25] L. Bixio, M. Ottonello, M. Raffetto, and C. Regazzoni, “Comparison among Cognitive Radio

Architectures for Spectrum Sensing,” EURASIP J. Wirel. Commun. Netw., vol. 2011, no. 1, p.

749891, Feb. 2011.

[26] E. Axell, G. Leus, E. G. Larsson, and H. V Poor, “Spectrum Sensing for Cognitive Radio : State-of-

the-Art and Recent Advances,” IEEE Signal Process. Mag., vol. 29, no. 3, pp. 101–116, May 2012.

[27] P. K. Varshney, Distributed detection and data fusion. Springer Science & Business Media, 2012.

[28] Y. C. Liang, Y. Zeng, E. C. Y. Peh, and A. T. Hoang, “Sensing-Throughput Tradeoff for Cognitive

Radio Networks,” IEEE Trans. Wirel. Commun., vol. 7, no. 4, pp. 1326–1337, Apr. 2008.

[29] H. Urkowitz, “Energy detection of unknown deterministic signals,” Proc. IEEE, vol. 55, no. 4, pp.

523–531, Apr. 1967.

[30] F. F. Digham, M. S. Alouini, and M. K. Simon, “On the energy detection of unknown signals over

fading channels,” in Communications, 2003. ICC ’03. IEEE International Conference on, 2003, vol.

5, pp. 3575–3579 vol.5.

[31] W. Zhang, R. K. Mallik, and K. B. Letaief, “Optimization of cooperative spectrum sensing with

energy detection in cognitive radio networks,” IEEE Trans. Wirel. Commun., vol. 8, no. 12, pp.

5761–5766, Dec. 2009.

[32] H. V Poor, An Introduction to Signal Detection and Estimation. Springer New York, 1988.

[33] S. A. Kassam and J. B. Thomas, Signal detection in non-Gaussian noise. Springer-Verlag, 1988.

[34] S. M. Kay, Fundamentals of Statistical Signal Processing: Practical Algorithm Development.

Prentice Hall, 2013.

[35] R. Tandra and A. Sahai, “SNR Walls for Signal Detection,” IEEE J. Sel. Top. Signal Process., vol.

2, no. 1, pp. 4–17, Feb. 2008.

[36] J. Lehtomäki, “Analysis of energy based signal detection,” 2005.

List of References

103

[37] V. I. Kostylev, “Energy detection of a signal with random amplitude,” in 2002 IEEE International

Conference on Communications. Conference Proceedings. ICC 2002 (Cat. No.02CH37333), 2002,

vol. 3, pp. 1606–1610 vol.3.

[38] A. Ghasemi and E. S. Sousa, “Collaborative spectrum sensing for opportunistic access in fading

environments,” in First IEEE International Symposium on New Frontiers in Dynamic Spectrum

Access Networks, 2005. DySPAN 2005., 2005, pp. 131–136.

[39] A. Ghasemi and E. S. Sousa, “Impact of User Collaboration on the Performance of Sensing-Based

Opportunistic Spectrum Access,” in IEEE Vehicular Technology Conference, 2006, pp. 1–6.

[40] D. Cabric, A. Tkachenko, and R. W. Brodersen, “Spectrum Sensing Measurements of Pilot, Energy,

and Collaborative Detection,” in MILCOM 2006 - 2006 IEEE Military Communications conference,

2006, pp. 1–7.

[41] M. Muterspaugh, H. Liu, and W. Gao, “Thomson proposal outline for WRAN,” IEEE 802.22-

05/0096r1, 2005.

[42] C. Cordeiro, K. Challapali, and M. Ghosh, “Cognitive PHY and MAC Layers for Dynamic

Spectrum Access and Sharing of TV Bands,” in Proceedings of the First International Workshop on

Technology and Policy for Accessing Spectrum, 2006.

[43] L. Chen, Q. Peng, J. Wang, and S. Li, “Sensing schemes for DVB-T in cognitive radio context,” in

2007 International Conference on Communications, Circuits and Systems, 2007, pp. 314–318.

[44] N. Nguyen-Thanh and I. Koo, “Evidence-Theory-Based Cooperative Spectrum Sensing With

Efficient Quantization Method in Cognitive Radio,” IEEE Trans. Veh. Technol., vol. 60, no. 1, pp.

185–195, Jan. 2011.

[45] D. Danev, E. Axell, and E. G. Larsson, “Spectrum sensing methods for detection of DVB-T signals

in AWGN and fading channels,” in 21st Annual IEEE International Symposium on Personal, Indoor

and Mobile Radio Communications, 2010, pp. 2721–2726.

[46] S. V Nagaraj, “Entropy-based Spectrum Sensing in Cognitive Radio,” Signal Process., vol. 89, no.

2, pp. 174–180, Feb. 2009.

[47] W. A. Gardner, A. Napolitano, and L. Paura, “Cyclostationarity: Half a Century of Research,”

Signal Process., vol. 86, no. 4, pp. 639–697, Apr. 2006.

[48] E. Serpedin, F. Panduru, I. Sari, and G. B. Giannakis, “Bibliography on Cyclostationarity,” Signal

Process., vol. 85, no. 12, pp. 2233–2303, Dec. 2005.

[49] J. Lunden, V. Koivunen, A. Huttunen, and H. V Poor, “Collaborative Cyclostationary Spectrum

Sensing for Cognitive Radio Systems,” IEEE Trans. Signal Process., vol. 57, no. 11, pp. 4182–4195,

Nov. 2009.

[50] W. Gardner, “Spectral Correlation of Modulated Signals: Part I - Analog Modulation,” IEEE Trans.

Commun., vol. 35, no. 6, pp. 584–594, Jun. 1987.

[51] W. A. Gardner and C. M. Spooner, “Detection and source location of weak cyclostationary signals:

simplifications of the maximum-likelihood receiver,” IEEE Trans. Commun., vol. 41, no. 6, pp.

905–916, Jun. 1993.

[52] P. Marques, J. Bastos, and A. Gameiro, “Sensing UMTS Bands Using Cyclostationary Features and

Cooperation Between Opportunistic Terminals,” in 2008 3rd International Conference on Cognitive

Radio Oriented Wireless Networks and Communications (CrownCom 2008), 2008, pp. 1–5.

[53] L. P. Goh, Z. Lei, and F. Chin, “Feature Detector for DVB-T Signal in Multipath Fading Channel,”

in 2007 2nd International Conference on Cognitive Radio Oriented Wireless Networks and

Communications, 2007, pp. 234–240.

[54] V. R. S. Banjade, “Analysis of Non-Coherent Detectors for Opportunistic Spectrum Access in

List of References

104

Broadband Wireless Networks,” University of Alberta, 2016.

[55] D. Cabric, S. M. Mishra, and R. W. Brodersen, “Implementation issues in spectrum sensing for cognitive radios,” in Conference Record of the Thirty-Eighth Asilomar Conference on Signals,

Systems and Computers, 2004., 2004, vol. 1, p. 772–776 Vol.1.

[56] K. B. Letaief and W. Zhang, “Cooperative Communications for Cognitive Radio Networks,” Proc.

IEEE, vol. 97, no. 5, pp. 878–893, May 2009.

[57] J. N. Tsitsiklis and others, “Decentralized detection,” Adv. Stat. Signal Process., vol. 2, no. 2, pp.

297–344, 1993.

[58] R. Viswanathan and P. K. Varshney, “Distributed detection with multiple sensors I. Fundamentals,”

Proc. IEEE, vol. 85, no. 1, pp. 54–63, Jan. 1997.

[59] R. S. Blum, S. A. Kassam, and H. V Poor, “Distributed detection with multiple sensors I. Advanced

topics,” Proc. IEEE, vol. 85, no. 1, pp. 64–79, Jan. 1997.

[60] S. M. Mishra, A. Sahai, and R. W. Brodersen, “Cooperative Sensing among Cognitive Radios,” in

2006 IEEE International Conference on Communications, 2006, vol. 4, pp. 1658–1663.

[61] R. Viswanathan and B. Ahsant, “A REVIEW OF SENSING AND DISTRIBUTED DETECTION

ALGORITHMS FOR COGNITIVE RADIO SYSTEMS.,” Int. J. Smart Sens. Intell. Syst., vol. 5, no.

1, 2012.

[62] R. Viswanathan, “Cooperative spectrum sensing for primary user detection in cognitive radio,” in

2011 Fifth International Conference on Sensing Technology, 2011, pp. 79–84.

[63] V. K. Bhargava and E. Hossain, Cognitive Wireless Communication Networks. Secaucus, NJ, USA:

Springer-Verlag New York, Inc., 2007.

[64] I. F. Akyildiz, B. F. Lo, and R. Balakrishnan, “Cooperative Spectrum Sensing in Cognitive Radio

Networks: A Survey,” Phys. Commun., vol. 4, no. 1, pp. 40–62, Mar. 2011.

[65] J. Ma and Y. G. Li, “Soft Combination and Detection for Cooperative Spectrum Sensing in

Cognitive Radio Networks,” in IEEE GLOBECOM 2007 - IEEE Global Telecommunications

Conference, 2007, pp. 3139–3143.

[66] J. Ma, G. Zhao, and Y. Li, “Soft Combination and Detection for Cooperative Spectrum Sensing in

Cognitive Radio Networks,” IEEE Trans. Wirel. Commun., vol. 7, no. 11, pp. 4502–4507, Nov.

2008.

[67] F. E. Visser, G. J. M. Janssen, and P. Pawełczak, “Multinode Spectrum Sensing Based on Energy

Detection for Dynamic Spectrum Access,” VTC Spring 2008 - IEEE Veh. Technol. Conf., vol. 0, pp.

1394–1398, 2008.

[68] R. Viswanathan and V. Aalo, “On counting rules in distributed detection,” IEEE Trans. Acoust., vol. 37, no. 5, pp. 772–775, May 1989.

[69] E. Peh and Y. C. Liang, “Optimization for Cooperative Sensing in Cognitive Radio Networks,” in

2007 IEEE Wireless Communications and Networking Conference, 2007, pp. 27–32.

[70] H. Zhang, D. Le Ruyet, and M. Terre, “Signal detection for OFDM/OQAM system using cyclostationary signatures,” in 2008 IEEE 19th International Symposium on Personal, Indoor and

Mobile Radio Communications, 2008, pp. 1–5.

[71] Z. Quan, S. Cui, and A. H. Sayed, “Optimal Linear Cooperation for Spectrum Sensing in Cognitive

Radio Networks,” IEEE J. Sel. Top. Signal Process., vol. 2, no. 1, pp. 28–40, Feb. 2008.

[72] M. Ranjeeth, “Cooperative spectrum sensing with square law combining diversity reception,” in

2015 3rd International Conference on Signal Processing, Communication and Networking (ICSCN),

2015, pp. 1–6.

List of References

105

[73] S. Atapattu, C. Tellambura, and H. Jiang, Energy detection for spectrum sensing in cognitive radio.

Springer, 2014.

[74] W. Saad, Z. Han, M. Debbah, A. Hjorungnes, and T. Basar, “Coalitional Games for Distributed

Collaborative Spectrum Sensing in Cognitive Radio Networks,” in IEEE INFOCOM 2009, 2009, pp.

2114–2122.

[75] B. Wang, K. J. R. Liu, and T. C. Clancy, “Evolutionary cooperative spectrum sensing game: how to

collaborate?,” IEEE Trans. Commun., vol. 58, no. 3, pp. 890–900, Mar. 2010.

[76] S. Zarrin and T. J. Lim, “Composite Hypothesis Testing for Cooperative Spectrum Sensing in

Cognitive Radio,” in 2009 IEEE International Conference on Communications, 2009, pp. 1–5.

[77] S. Zarrin and T. J. Lim, “Cooperative Spectrum Sensing in Cognitive Radios With Incomplete

Likelihood Functions,” IEEE Trans. Signal Process., vol. 58, no. 6, pp. 3272–3281, Jun. 2010.

[78] I. F. Akyildiz, W.-Y. Lee, M. C. Vuran, and S. Mohanty, “NeXt Generation/Dynamic Spectrum

Access/Cognitive Radio Wireless Networks: A Survey,” Comput. Netw., vol. 50, no. 13, pp. 2127–

2159, Sep. 2006.

[79] B. F. Lo, I. F. Akyildiz, and A. M. Al-Dhelaan, “Efficient Recovery Control Channel Design in

Cognitive Radio Ad Hoc Networks,” IEEE Trans. Veh. Technol., vol. 59, no. 9, pp. 4513–4526,

Nov. 2010.

[80] Q. Qin, Z. Zhimin, and G. Caili, “A Study of Data Fusion and Decision Algorithms Based on

Cooperative Spectrum Sensing,” in 2009 Sixth International Conference on Fuzzy Systems and

Knowledge Discovery, 2009, vol. 1, pp. 76–80.

[81] D. Teguig, B. Scheers, and V. Le Nir, “Data fusion schemes for cooperative spectrum sensing in cognitive radio networks,” in 2012 Military Communications and Information Systems Conference

(MCC), 2012, pp. 1–7.

[82] A. C. Malady and C. R. C. M. da Silva, “Clustering methods for distributed spectrum sensing in

cognitive radio systems,” in MILCOM 2008 - 2008 IEEE Military Communications Conference,

2008, pp. 1–5.

[83] C. Guo, T. Peng, S. Xu, H. Wang, and W. Wang, “Cooperative Spectrum Sensing with Cluster-

Based Architecture in Cognitive Radio Networks,” in VTC Spring 2009 - IEEE 69th Vehicular

Technology Conference, 2009, pp. 1–5.

[84] J. Wei and X. Zhang, “Energy-Efficient Distributed Spectrum Sensing for Wireless Cognitive Radio

Networks,” in 2010 INFOCOM IEEE Conference on Computer Communications Workshops, 2010,

pp. 1–6.

[85] Y. Zhao, J. Gaeddert, K. K. Bae, and J. H. Reed, “Radio environment map enabled situation-aware

cognitive radio learning algorithms,” in Software Defined Radio Forum (SDRF) technical

conference, 2006.

[86] A. W. Min and K. G. Shin, “An Optimal Sensing Framework Based on Spatial RSS-profile in

Cognitive Radio Networks,” in 2009 6th Annual IEEE Communications Society Conference on Sensor, Mesh and Ad Hoc Communications and Networks, 2009, pp. 1–9.

[87] S. J. Kim, E. Dall‟Anese, and G. B. Giannakis, “Cooperative Spectrum Sensing for Cognitive

Radios Using Kriged Kalman Filtering,” IEEE J. Sel. Top. Signal Process., vol. 5, no. 1, pp. 24–36,

Feb. 2011.

[88] P. Kaligineedi and V. K. Bhargava, “Sensor Allocation and Quantization Schemes for Multi-Band

Cognitive Radio Cooperative Sensing System,” IEEE Trans. Wirel. Commun., vol. 10, no. 1, pp.

284–293, Jan. 2011.

[89] D.-C. Oh and Y.-H. Lee, “Cooperative spectrum sensing with imperfect feedback channel in the

cognitive radio systems,” Int. J. Commun. Syst., vol. 23, no. 6–7, pp. 763–779, 2010.

List of References

106

[90] B. Picinbono and P. Duvaut, “Optimum quantization for detection,” IEEE Trans. Commun., vol. 36,

no. 11, pp. 1254–1258, Nov. 1988.

[91] H. Sakran, M. Shokair, E. S. El-Rabaie, and A. A. El-Azm, “Three bits softened decision scheme in

cooperative spectrum sensing among cognitive radio networks,” in 2011 28th National Radio

Science Conference (NRSC), 2011, pp. 1–9.

[92] H. Sakran and M. Shokair, “Hard and Softened Combination for Cooperative Spectrum Sensing over

Imperfect Channels in Cognitive Radio Networks,” Telecommun. Syst., vol. 52, no. 1, pp. 61–71,

Jan. 2013.

[93] S. Chaudhari, J. Lunden, and V. Koivunen, “BEP walls for collaborative spectrum sensing,” in 2011

IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2011, pp.

2984–2987.

[94] C. W. Helstrom, “Improved multilevel quantization for detection of narrowband signals,” IEEE

Trans. Aerosp. Electron. Syst., vol. 24, no. 2, pp. 142–147, Mar. 1988.

[95] W. A. Hashlamoun and P. K. Varshney, “Near-optimum quantization for signal detection,” IEEE

Trans. Commun., vol. 44, no. 3, pp. 294–297, Mar. 1996.

[96] C. Rago, P. Willett, and Y. Bar-Shalom, “Censoring sensors: a low-communication-rate scheme for

distributed detection,” IEEE Trans. Aerosp. Electron. Syst., vol. 32, no. 2, pp. 554–568, Apr. 1996.

[97] R. Jiang and B. Chen, “Fusion of censored decisions in wireless sensor networks,” IEEE Trans.

Wirel. Commun., vol. 4, no. 6, pp. 2668–2673, Nov. 2005.

[98] M. Mustonen, M. Matinmikko, and A. Mammela, “Cooperative spectrum sensing using quantized

soft decision combining,” in 2009 4th International Conference on Cognitive Radio Oriented Wireless Networks and Communications, 2009, pp. 1–5.

[99] N. Nguyen-Thanh and I. Koo, “Log-likelihood Ratio Optimal Quantizer for Cooperative Spectrum

Sensing in Cognitive Radio,” IEEE Commun. Lett., vol. 15, no. 3, pp. 317–319, Mar. 2011.

[100] S. J. Zahabi, A. A. Tadaion, and S. Aissa, “Neyman-Pearson Cooperative Spectrum Sensing for Cognitive Radio Networks with Fine Quantization at Local Sensors,” IEEE Trans. Commun., vol.

60, no. 6, pp. 1511–1522, Jun. 2012.

[101] R. Ujjinimatad and S. R. Patil, “Spectrum sensing in cognitive radio networks with known and

unknown noise levels,” IET Commun., vol. 7, no. 15, pp. 1708–1714, Oct. 2013.

[102] N. Reisi, S. Gazor, and M. Ahmadian, “Distributed Cooperative Spectrum Sensing in Mixture of

Large and Small Scale Fading Channels,” IEEE Trans. Wirel. Commun., vol. 12, no. 11, pp. 5406–

5412, Nov. 2013.

[103] N. Reisi, S. Gazor, and M. Ahmadian, “A Distributed Average Likelihood Ratio Detector for

Detection of Signals in Frequency-Selective Nakagami Channels,” IEEE Wirel. Commun. Lett., vol.

3, no. 3, pp. 245–248, Jun. 2014.

[104] S. Kassam, “Optimum Quantization for Signal Detection,” IEEE Trans. Commun., vol. 25, no. 5, pp.

479–484, May 1977.

[105] R. S. Blum and M. C. Deans, “Distributed random signal detection with multibit sensor decisions,”

IEEE Trans. Inf. Theory, vol. 44, no. 2, pp. 516–524, Mar. 1998.

[106] C. C. Lee and J. J. Chao, “Optimum local decision space partitioning for distributed detection,”

IEEE Trans. Aerosp. Electron. Syst., vol. 25, no. 4, pp. 536–544, Jul. 1989.

[107] Y. Tani and T. Saba, “Quantization scheme for energy detector of soft decision cooperative

spectrum sensing in cognitive radio,” in 2010 IEEE Globecom Workshops, 2010, pp. 69–73.

[108] R. L. Haupt and S. E. Haupt, Practical genetic algorithms. John Wiley & Sons, 2004.

List of References

107

[109] E. Elbeltagi, T. Hegazy, and D. Grierson, “Comparison Among Five Evolutionary-based

Optimization Algorithms,” Adv. Eng. Inform., vol. 19, no. 1, pp. 43–53, Jan. 2005.

[110] J. Kennedy and R. Eberhart, “Particle swarm optimization,” in Neural Networks, 1995.

Proceedings., IEEE International Conference on, 1995, vol. 4, pp. 1942–1948 vol.4.

[111] J. Kennedy and R. C. Eberhart, Swarm Intelligence. San Francisco, CA, USA: Morgan Kaufmann

Publishers Inc., 2001.

[112] M. Lovbjerg and T. Krink, “Extending particle swarm optimisers with self-organized criticality,” in

Evolutionary Computation, 2002. CEC ’02. Proceedings of the 2002 Congress on, 2002, vol. 2, pp.

1588–1593.

[113] C. Mendis, S. M. Guru, S. Halgamuge, and S. Fernando, “Optimized sink node path using particle

swarm optimization,” in 20th International Conference on Advanced Information Networking and

Applications - Volume 1 (AINA’06), 2006, vol. 2, p. 5 pp.-pp.

[114] R. A. Rashid, M. Adib Sarijari, N. Fisal, S. K. S Yusof, N. Hija Mahalin, and A. C. C Lo, “Spectrum

Sensing Measurement using GNU Radio and USRP Software Radio Platform,” Seventh Int. Conf.

Wirel. Mob. Commun., no. c, pp. 237–242, 2011.

[115] R. V Rao, V. J. Savsani, and D. P. Vakharia, “Teaching-learning-based Optimization: A Novel

Method for Constrained Mechanical Design Optimization Problems,” Comput. Aided Des., vol. 43,

no. 3, pp. 303–315, Mar. 2011.

[116] R. V Rao, V. J. Savsani, and D. P. Vakharia, “Teaching-Learning-Based Optimization: An

Optimization Method for Continuous Non-linear Large Scale Problems,” Inf. Sci., vol. 183, no. 1,

pp. 1–15, Jan. 2012.

[117] M. M. Puralachetty, V. K. Pamula, L. M. Gondela, and V. N. B. Akula, “Teaching-learning-based

optimization with two-stage initialization,” in 2016 IEEE Students’ Conference on Electrical,

Electronics and Computer Science (SCEECS), 2016, pp. 1–5.

[118] B. A. Bastami and E. Saberinia, “A Practical Multibit Data Combining Strategy for Cooperative

Spectrum Sensing,” IEEE Trans. Veh. Technol., vol. 62, no. 1, pp. 384–389, Jan. 2013.

[119] H. Birkan Yilmaz, T. Tugcu, and F. Alagoz, “Novel quantization-based spectrum sensing scheme

under imperfect reporting channel and false reports,” Int. J. Commun. Syst., vol. 27, no. 10, pp.

1459–1475, 2014.

[120] J. Shen, S. Liu, L. Zeng, G. Xie, J. Gao, and Y. Liu, “Optimisation of cooperative spectrum sensing

in cognitive radio network,” IET Commun., vol. 3, no. 7, pp. 1170–1178, Jul. 2009.

[121] R. V Rao, V. J. Savsani, and J. Balic, “Teaching–learning-based optimization algorithm for

unconstrained and constrained real-parameter optimization problems,” Eng. Optim., vol. 44, no. 12,

pp. 1447–1462, 2012.

[122] V. Toğan, “Design of pin jointed structures using teaching-learning based optimization,” Struct.

Eng. Mech., vol. 47, 2013.

[123] S. O. Degertekin and M. S. Hayalioglu, “Sizing Truss Structures Using Teaching-learning-based

Optimization,” Comput. Struct., vol. 119, pp. 177–188, Apr. 2013.

[124] M. Ibnkahla and A. A. Alkheir, Cooperative Cognitive Radio Networks: The Complete Spectrum

Cycle. Boca Raton, FL, USA: CRC Press, Inc., 2014.

List of Publications

108

List of Publications

[1] D.R.Keraliya, Dr.Ashalata Kulshrestha and S.D.Patel, “Optimization of Cooperative Spectrum

Sensing in Cognitive Radio Networks” in proceeding of IEEE International conference of

communication and networks, ICCN 2015, 19-21 Nov., pp. 237-242 .

[2] D.R.Keraliya and Dr.Ashalata Kulshrestha, “Optimization of Cooperative Spectrum Sensing using

quantized data fusion scheme for cognitive radio” in International Journal of Advance Research in

Engineering Science and Technology (IJAREST), Vol. 03, issue 1, pp. 192-197, Jan. 2016.

[3] D.R.Keraliya and Dr.Ashalata Kulshrestha, “Minimizing the Detection Error in Cooperative

Spectrum Sensing Using Teaching Learning Based Optimization(TLBO)” in International Journal

of Engineering Research and Technology (IJERT), Vol. 06, issue 02, pp. 495-500, Feb. 2017.

[4] D.R.Keraliya, Dr.Ashalata Kulshrestha and Hitesh Loriya, “Performance Analysis of Cooperative

Sensing In Fading Channels for Wireless Communication” in International Journal of Advanced

Research in Computer and Communication Engineering (IJARCCE), Vol. 06, issue 02, pp. 275-

279, Feb. 2017.

[5] D.R.Keraliy, Dr.Ashalata Kulshrestha and Hitesh Loriya, “Optimization of Cooperative Spectrum

Sensing Using PSO Algorithm for Wireless Communication” in International Journal of Emerging

Technology and Advanced Engineering (IJETAE), Vol. 07, issue 02, pp. 113-118, Feb. 2017

[6] D.R.Keraliya and Dr.Ashalata Kulshrestha, “Optimization of Quantized Cooperative Spectrum

Sensing Using Teaching Learning Based Optimization (TLBO) Algorithm” in Far East Journal of Electronics and Communication (FJEC), Vol. 17, issue 05, pp. 1141-1151, Oct. 2017

optimization of cooperative spectrum sensing in … · iii declaration i declare that the thesis...

Documents