Upload
others
View
4
Download
0
Embed Size (px)
Citation preview
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
Name of Research Scholar: Keraliya Divyesh Rudabhai
Place: Ahmedabad
Signature of Supervisor:…………………………… Date: 10/08/2018
Name of Supervisor: Dr. Ashalata Kulshrestha
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:………………………………
Name of Research Scholar: Keraliya Divyesh Rudabhai
Date: 10/08/2018 Place: Ahmedabad
Signature of Supervisor:……………………………...
Name of Supervisor: Dr. Ashalata Kulshrestha
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
FIGURE 5.20: Convergence performance on over 15 Iteration ................................... 74
xxii
FIGURE 5.21: Convergence performance on over 20 Iteration ................................... 75
FIGURE 5.22: Convergence performance on over 30 Iteration ................................... 75
FIGURE 5.23: Convergence performance on over 40 Iteration ................................... 76
FIGURE 5.24: Convergence performance on over 50 Iteration ................................... 76
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 6.11: Convergence performance on for .............................................. 93
FIGURE 6.12: Convergence performance on for .............................................. 93
FIGURE 6.13: Convergence performance on for .............................................. 94
FIGURE 6.14: Convergence performance on for ............................................ 94
FIGURE 6.15: Convergence performance on for ............................................ 95
FIGURE 6.16: Convergence performance on for ............................................ 95
FIGURE 6.17: Convergence performance on for ............................................ 96
FIGURE 6.18: Convergence performance on for ............................................ 96
FIGURE 6.19: Convergence performance on for ............................................ 97
FIGURE 6.20: Convergence performance on for ............................................ 97
FIGURE 6.21: Convergence performance on for ............................................ 98
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
Background of Spectrum Sensing for CR
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.
Background of Spectrum Sensing for CR
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)
Background of Spectrum Sensing for CR
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.
Background of Spectrum Sensing for CR
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.
Background of Spectrum Sensing for CR
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.
Background of Spectrum Sensing for CR
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)
Background of Spectrum Sensing for CR
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)
Background of Spectrum Sensing for CR
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).
Background of Spectrum Sensing for CR
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)
Optimization Techniques for Cooperative Spectrum Sensing
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
Optimization Techniques for Cooperative Spectrum Sensing
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
Optimization Techniques for Cooperative Spectrum Sensing
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.
Classification of Optimization Technique for CSS
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
Optimization Techniques for Cooperative Spectrum Sensing
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]
Classification of Optimization Technique for CSS
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)
Optimization Techniques for Cooperative Spectrum Sensing
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
Optimization Techniques for Cooperative Spectrum Sensing
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)
Teaching–Learning Based Optimization
49
FIGURE 4.8: Teaching–Learning-Based Optimization (TLBO) flow chart
Optimization Techniques for Cooperative Spectrum Sensing
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
Teaching–Learning Based Optimization
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.
Optimization Techniques for Cooperative Spectrum Sensing
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
Teaching–Learning Based Optimization
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
4 – Level Quantizer
4 – Level Quantizer
Sensing Channel
∑
Reporting Channel
𝐿 𝐿𝑛 𝐿
Global Decision
(H0/H1)
FC
PU
TLBO Assisted CSS for Maximizing the Probability of Detection
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
Proposed TLBO Based System Model
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.
TLBO Assisted CSS for Maximizing the Probability of Detection
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)
Proposed TLBO Based System Model
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.
TLBO Assisted CSS for Maximizing the Probability of Detection
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:
TLBO Assisted CSS for Maximizing the Probability of Detection
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
TLBO Assisted CSS for Maximizing the Probability of Detection
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
TLBO Assisted CSS for Maximizing the Probability of Detection
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
Simulation Results and Discussion
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
TLBO Assisted CSS for Maximizing the Probability of Detection
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
Simulation Results and Discussion
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
TLBO Assisted CSS for Maximizing the Probability of Detection
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
Simulation Results and Discussion
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
TLBO Assisted CSS for Maximizing the Probability of Detection
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
Simulation Results and Discussion
73
FIGURE 5.17: Convergence performance on over 2 Iteration
FIGURE 5.18: Convergence performance on over 5 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
TLBO Assisted CSS for Maximizing the Probability of Detection
74
FIGURE 5.19: Convergence performance on over 10 Iteration
FIGURE 5.20: Convergence performance on over 15 Iteration
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
Simulation Results and Discussion
75
FIGURE 5.21: Convergence performance on over 20 Iteration
FIGURE 5.22: Convergence performance on over 30 Iteration
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
TLBO Assisted CSS for Maximizing the Probability of Detection
76
FIGURE 5.23: Convergence performance on over 40 Iteration
FIGURE 5.24: Convergence performance on over 50 Iteration
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
Simulation Results and Discussion
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
TLBO Assisted CSS for Maximizing the Probability of Detection
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
Simulation Results and Discussion
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
TLBO Assisted CSS for Maximizing the Probability of Detection
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
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 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.
Proposed TLBO Based System Model
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
TLBO Assisted CSS for Minimizing the Cooperative Sensing Error
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
TLBO Assisted CSS for Minimizing the Cooperative 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
TLBO Assisted CSS for Minimizing the Cooperative Sensing Error
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
TLBO Assisted CSS for Minimizing the Cooperative Sensing Error
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
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 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
Simulation Result and Discussion
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
TLBO Assisted CSS for Minimizing the Cooperative Sensing Error
92
FIGURE 6.9: Convergence performance on for
FIGURE 6.10: 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
Simulation Result and Discussion
93
FIGURE 6.11: Convergence performance on for
FIGURE 6.12: Convergence performance on for
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
TLBO Assisted CSS for Minimizing the Cooperative Sensing Error
94
FIGURE 6.13: Convergence performance on for
FIGURE 6.14: Convergence performance on for
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
Simulation Result and Discussion
95
FIGURE 6.15: Convergence performance on for
FIGURE 6.16: Convergence performance on for
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
TLBO Assisted CSS for Minimizing the Cooperative Sensing Error
96
FIGURE 6.17: Convergence performance on for
FIGURE 6.18: Convergence performance on for
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
Simulation Result and Discussion
97
FIGURE 6.19: Convergence performance on for
FIGURE 6.20: Convergence performance on for
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
TLBO Assisted CSS for Minimizing the Cooperative Sensing Error
98
FIGURE 6.21: Convergence performance on for
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