STRUCTURE, COORDINATION, SENSING AND
ALLOCATION IN COGNITIVE RADIO VANETS
SHAHID HUSSAIN ABBASSI
A Thesis
Submitted in Partial Fulfillment of the
Requirements for the Degree of
Doctor of Philosophy
DEPARTMENT OF ELECTRICAL ENGINEERING
AIR UNIVERSITY
2016
STRUCTURE, COORDINATION, SENSING AND
ALLOCATION IN COGNITIVE RADIO VANETS
Ph.D. Dissertation
SUBMITTED BY
SHAHID HUSSAIN ABBASSI REG. NO. Ph.D.-EE-091313
SUPERVISOR
PROF. DR. IJAZ MANSOOR QURESHI
DEPARTMENT OF ELECTRICAL ENGINEERING AIR UNIVERSITY
ISLAMABAD
February, 2016
CERTIFICATE OF APPROVAL
Department of Electrical Engineering
It is hereby certified that Shahid Hussain Abbassi (Reg # Ph.D.-EE-091313) has successfully completed his dissertation.
_____________________________
Dr. Ijaz Mansoor Qureshi
Air University Supervisor
____________________________ ____________________________ Dr. Fida Muhammad Khan Dr. Syed Ahmed Pasha Internal Examiner 1 Internal Examiner 2 Guidance and Evaluation Committee Guidance and Evaluation Committee
____________________________ ____________________________ Dr. Adnan Omer Dr. Abdul Jalil External Examiner External Examiner Guidance and Evaluation Committee Guidance and Evaluation Committee
____________________________ ___________________________
AVM Saleem Tariq Dr. Zafar Ullah Koreshi Chair Department Senior Dean
STRUCTURE, COORDINATION, SENSING AND
ALLOCATION IN COGNITIVE RADIO VANETS
Ph.D. Dissertation
SHAHID HUSSAIN ABBASSI REG. NO. Ph.D.-EE-091313
SUPERVISOR
PROF. DR. IJAZ MANSOOR QURESHI
FOREIGN RESEARCH EVALUATION EXPERTS
Dr. Amir Hussain, Divisional PhD Director, University of Stirling, Scotland, UK
Dr. Wen-Hsien Fang, Professor and Chairman, Department of Electronic and Computer Engineering, National Taiwan University of Science and
Technology, Taipei, TAIWAN
DEPARTMENT OF ELECTRICAL ENGINEERING
AIR UNIVERSITY ISLAMABAD
2016
ABSTRACT
With the increase in population, vehicle traffic has also increased on roads. This has caused an
increase in accidents, due to which thousands of people lose their lives and millions get injured
annually. Hence, a foolproof and secure Vehicular Ad-hoc Network (VANET) structure is required
to reduce the number of accidents considerably. Pre-danger information must be communicated in
real-time, in order to implement preventive measures to avoid accidents. VANETs are specially
designed in order to communicate information about hazards. The protocol
DSRC/WAVE/IEEE802.11p is proposed for VANETs, but it may not be enough to cope with
increasing network traffic, especially emergency messages. Hence the use of cognitive Radio (CR)
technology has been introduced. A lot of methods for coordination and channel allocation in the
context of VANETs are being introduced. As such, the need of a framework to reliably compare
the relative performances of different channel sensing, allocation and coordination schemes which
take into account the movement of vehicles is felt. Different techniques like Independent Spectrum
sensing and various forms of Cooperative techniques have been proposed in the near past.
In this dissertation, a VANET structure has been proposed for highways and urban environments.
In the Highway model, separate Road Side Units have been provided for the traffic on each side. In
this way group formation for localized traffic will be easy on highways. Simulation results show
that by using the proposed model, average throughput and end-to-end delay have improved
considerably, while packet loss has also been reduced.
We also propose an efficient spectrum sensing mechanism for sensing and sharing the CR
spectrum by mobile vehicles, which combines best of stand-alone sensing and cooperative sensing
techniques. The proposed mechanism not only improves the probability of correct detection, but
also almost eliminates the probability of misdetection. Then we have introduced a framework that
can be used to define and compare such schemes in a variety of scenarios. Simulation results
clearly show the robustness of our technique by almost eliminating the misdetections and reducing
to a great extent the false alarms. Afterwards we have proposed a sensing technique which prepares
a database for small road segments, time slots for the hours of the day and different frequencies of
the spectrum based on the sensing of vehicles throughout the day. Based on this database, the
future utilization of the spectrum is proposed. Simulations and results clearly indicate the success
and usefulness of our proposed technique.
In the end we have proposed a model based on fuzzy logic for the allocation of different types of
TV channels having different ON/OFF timings in different hours of the day taking time, vehicle
speed, message priority and CR channel sensing results as input. The simulations performed show
the utilization of every type of channel in speed versus time and message priority versus time.
These results can be utilized well for the allocation patterns of CR channels.
Copyright by
SHAHID HUSSAIN ABBASSI
2016
All rights reserved. No part of the material protected by this copyright notice may be reproduced or
utilized in any form or by any means, electronic or mechanical, including photocopying, recording
or by any information storage and retrieval system, without the permission from the author.
DEDICATED TO
My parents, wife and children
CERTIFICATE OF APPROVAL FROM SUPERVISOR
It is certified that the research work contained in this Ph.D. dissertation has been carried out
under my supervision in the Department of Electrical Engineering, Air University, Islamabad. It
is based on original work carried out by the student individually and has not been submitted for
any other degree anywhere else. Moreover, all the other requirements mentioned in the road map
of PhD have been completed. The thesis has also undergone plagiarism test using Turnitin. Its
similarity index is _________________.
Signature: _____________________
Supervisor:
Prof. Dr. Ijaz Mansoor Qureshi
Department of Electrical Engineering
Air University,
Islamabad.
i
LIST OF PUBLICATIONS
1. S. H. Abbassi, I. M. Qureshi, H. Abbasi and B. R. Alyaie, “History Based Spectrum
Sensing in CR-VANETs,” EURASIP Journal on Wireless Communication and
Networking 2015, 2015:163.(ISI indexed, impact factor 0.72)
2. S. H. Abbassi, I. M. Qureshi, B. R. Alyaei, H. Abbasi and K. Sultan, “An Efficient
Spectrum Sensing Mechanism for CR-VANETs,” Journal of Basic and Applied Scientific
Research, vol. 12, no. 3, pp. 365-378, 2013. (ISI indexed)
3. S. H. Abbassi, I. M. Qureshi and H. Abbasi, “Performance of Uni-directional Road Side
Units in Vehicular Adhoc Networks,” in World Congress on Computer Applications and
Information Systems, Hammamet, 2014.
4. S. H. Abbassi, “Algorithm for Topology Discovery in Resilient Packet Ring
IEEE802.17” in SETIT 2009, Hammamet, Tunisia. 22 to 26 March 2009.
5. S. H. Abbassi, “Topology Discovery in Resilient Packet Ring Technology” in INC 2004,
Plymouth, UK. June 2004.
6. S. H. Abbassi, Faheem Ahmed, Mohammad Saeed, “Induction of Survivability into
Rational Unified Process” in SERP 2003, Las Vegas, USA. June 2003.
List of Submitted Paper
1. H. Abbasi, S. H. Abbassi and I. M. Qureshi, “A framework for the simulation of CR-
VANET channel sensing, coordination and allocation,” Journal of Ad Hoc & Sensor
Wireless Networks, Submitted, 2014.(ISI indexed, impact factor 0.478)
The material presented in this dissertation is based on the published papers 1 to 3 and the
submitted paper No. 1.
ii
ACKNOWLEDGMENTS
Thanks to almighty Allah whose blessings have encouraged and provided me strength to conduct
this research and to complete this dissertation. There have been moments when I felt it
impossible to complete my research but almighty Allah has always shown me the way how to do
it.
I am extremely thankful to my supervisor Dr. Ijaz Mansoor Qureshi whose continuous guidance,
and support made it possible to complete this dissertation. His pushing attitude and
encouragement was the key factor throughout my course and research work.
I am highly thankful to Dr. Fida Mohammad Khan whose fatherly attitude provided me help and
moral support. I am also thankful to Dr. Syed Ahmed Pasha whose guidance has helped me to
correct my mistakes.
I am grateful to Mr. Bahman R. Alyaie who’s moral as well as support in problem formulation
and solutions has enabled me to complete the work. I have to give strong credit to my son Mr.
Hameer Abbasi whose help in simulating the problem was a key factor throughout my work. I
am extremely thankful to my wife Ms. Paras Abbasi and my son Janib Abbasi whose handling of
household affairs throughout my research work has enabled me to complete the dissertation.
February 17, 2016
Shahid Hussain Abbassi
iii
TABLE OF CONTENTS
LIST OF PUBLICATIONS ................................................................................................................ i
ACKNOWLEDGMENTS .................................................................................................................. ii
LIST OF TABLES ............................................................................................................................ vii
LIST OF FIGURES ........................................................................................................................ viii
LIST OF ABBREVIATIONS ........................................................................................................... xi
LIST OF SYMBOLS ........................................................................................................................ xv
Chapter 1 INTRODUCTION ......................................................................................... 1
1.1 VANETs ........................................................................................................................... 1
1.2 Cognitive Radio and VANETs ......................................................................................... 2
1.3 contributions of thesis ...................................................................................................... 4
1.4 Organization of the Thesis ............................................................................................... 5
Chapter 2 BACKGROUND ............................................................................................ 6
2.1 MANETs .......................................................................................................................... 6
2.2 VANET ............................................................................................................................ 8
2.3 Security Issues in VANETS ............................................................................................. 9
2.3.1 Communication Patterns ......................................................................................... 10
2.3.2 Overview of Attacks in VANET............................................................................. 10
2.4 IEEE 802.11p / WAVE / DSRC..................................................................................... 12
2.5 Issues in VANETs .......................................................................................................... 15
2.5.1 Vehicle Density ....................................................................................................... 15
2.5.2 High Mobility.......................................................................................................... 16
iv
2.5.3 Intermittent Connectivity ........................................................................................ 16
2.5.4 Definition of Services ............................................................................................. 16
2.5.5 Identification of Service Recipients ........................................................................ 16
2.5.6 Incremental Deployment of VANET ...................................................................... 17
2.5.7 Open Approach to VANET Architecture ............................................................... 17
2.5.8 Unreliable Components Generate Unreliable Data................................................. 17
2.5.9 Privacy .................................................................................................................... 17
2.5.10 Authentication ......................................................................................................... 17
2.5.11 Non Repudiation ..................................................................................................... 18
2.5.12 Reliability, Integrity and Scalability ....................................................................... 18
2.5.13 Real Time Guarantees ............................................................................................. 18
2.6 CR-VANETS ................................................................................................................. 19
2.6.1 Standalone CR Sensing ........................................................................................... 20
2.6.2 Centralized and Cooperative CR Sensing ............................................................... 21
2.6.3 Detection Techniques and Fading Models .............................................................. 24
2.6.4 Spectrum Management and QoS Support ............................................................... 25
2.6.5 Distance Segmentation............................................................................................ 26
2.7 Fuzzy Logic .................................................................................................................... 26
Chapter 3 STRUCTURE OF VANETs ....................................................................... 29
3.1 Message Categorization ................................................................................................. 29
3.1.1 Emergency Message ............................................................................................... 29
3.1.2 Safety Message ....................................................................................................... 30
3.1.3 GPS Message .......................................................................................................... 30
3.1.4 Probe Message ........................................................................................................ 30
v
3.1.5 Traveller Information .............................................................................................. 30
3.1.6 Location Based Service........................................................................................... 30
3.1.7 Informative message ............................................................................................... 30
3.1.8 E-mails .................................................................................................................... 31
3.2 Proposed VANET Structure ........................................................................................... 31
3.2.1 Simulation Results .................................................................................................. 33
3.3 Proposed Highway Structure .......................................................................................... 35
3.3.1 Simulation Results .................................................................................................. 36
3.4 Summary ........................................................................................................................ 38
Chapter 4 COGNITIVE RADIO AND VANETs ...................................................... 39
4.1 Proposed Spectrum Sensing Framework........................................................................ 39
4.1.1 Network Model ....................................................................................................... 39
4.1.2 Spectrum Sensing Model ........................................................................................ 40
4.1.3 Vehicle Mobility Model .......................................................................................... 41
4.1.4 Spectrum Sensing and Coordination Framework ................................................... 42
4.1.5 Coordinators Selection and Sensing Algorithms .................................................... 47
4.1.6 Simulation and Results ........................................................................................... 52
4.2 Modified Spectrum Sensing and Allocation Model ....................................................... 57
4.2.1 Models Used for the Simulation ............................................................................. 63
4.2.2 Simulation Results .................................................................................................. 65
4.3 Summary ........................................................................................................................ 71
Chapter 5 HISTORY BASED SPECTRUM SENSING AND ALLOCATION .... 73
5.1 History Updating ............................................................................................................ 74
5.2 History Preservation ....................................................................................................... 76
vi
5.3 History Utilization .......................................................................................................... 77
5.4 Simulation Results.......................................................................................................... 78
5.5 Summary ........................................................................................................................ 89
Chapter 6 CR CHANNEL ALLOCATION USING FUZZY LOGIC IN VANETs
.............................................................................................................................................................. 90
6.1 System Model ................................................................................................................. 90
6.1.1 Inputs and Membership Functions .......................................................................... 91
6.1.2 Outputs .................................................................................................................... 93
6.1.3 Fuzzy IF-THEN Rules ............................................................................................ 96
6.2 Simulation and Results ................................................................................................... 97
6.2.1 Utility Speed Vs Time ............................................................................................ 98
6.2.2 Utility Message Priority Vs Time ......................................................................... 101
6.3 Summary ...................................................................................................................... 104
Chapter 7 CONCLUSION AND FUTURE PROSPECTS ..................................... 105
7.1 Conclusion .................................................................................................................... 105
7.2 Future Prospects ........................................................................................................... 107
Appendix A ...................................................................................................................................... 108
NS2 .................................................................................................................................. 108
C# ..................................................................................................................................... 109
o Visual Studio ................................................................................................................ 109
o C# History and features ................................................................................................ 110
o .NET Framework .......................................................................................................... 111
Bibliography…………………………………………………………………………………...111
vii
LIST OF TABLES
Table 3.1: Comparison 20 to 50 nodes, RSU to DRSU 35
Table 3.2: Comparison 50 to 100 nodes, RSU to DRSU 38
Table 5.1: reply to 76
viii
LIST OF FIGURES
Figure 1-1: An example of VANET ............................................................................................... 2
Figure 2-1: Simple Mobile Ad-hoc Network .................................................................................. 7
Figure 2-2 Modes of Operation in MANETs .................................................................................. 7
Figure 2-3: IEEE 802.11p/WAVE Protocol Stack ....................................................................... 13
Figure 2-4: DSRC 75 MHz Spectrum ........................................................................................... 14
Figure 2-5: IEEE 802.11p Access Layer ...................................................................................... 14
Figure 2-6: Typical Structure of CR_VANET.............................................................................. 19
Figure 3-1: Overall Proposed Structure of VANETs .................................................................... 32
Figure 3-2: Cluster Formation for Bidirectional Traffic ............................................................... 33
Figure 3-3: Cluster Formation for Unidirectional Traffic............................................................. 33
Figure 3-4: Throughput (CASE-1, 50 Mobile nodes). .................................................................. 34
Figure 3-5: Throughput (CASE-2, 50 Mobile nodes). .................................................................. 35
Figure 3-6: Proposed Highways Structure for VANETs .............................................................. 36
Figure 3-7: Throughput (RSUs catering bi-directional traffic)..................................................... 37
Figure 3-8: Throughput (DRSUs catering unidirectional traffic). ................................................ 37
Figure 4-1: Coordinators Selection ............................................................................................... 50
Figure 4-2: Front Coordinator Selection ....................................................................................... 51
Figure 4-3: Back Coordinator Selection ....................................................................................... 51
Figure 4-4: Spectrum Sensing ....................................................................................................... 52
Figure 4-5: Probability of Correct Detection versus No. of Channels .......................................... 54
Figure 4-6: Probability of Correct Detection versus No. of Vehicles........................................... 54
Figure 4-7: Probability of Correct Detection versus Vehicle Velocity......................................... 55
Figure 4-8: Probability of Misdetection versus No. of Channels ................................................. 56
Figure 4-9: Probability of Misdetection versus No. of Vehicles .................................................. 56
Figure 4-10: Probability of Misdetection versus Vehicle Velocity .............................................. 57
Figure 4-11: The Overall Algorithm ............................................................................................. 58
Figure 4-12: Allocation, coordination and sensing ....................................................................... 59
ix
Figure 4-13: How to perform PU state changes. ........................................................................... 60
Figure 4-14: Sensing and occupation mechanism. ....................................................................... 62
Figure 4-15: Allocations Rate vs Vehicles ................................................................................... 66
Figure 4-16: False Alarms vs Vehicles ......................................................................................... 67
Figure 4-17: Misdetections vs Vehicles ........................................................................................ 67
Figure 4-18: Allocations Rate vs Channels .................................................................................. 68
Figure 4-19: False Alarms vs Channels ........................................................................................ 68
Figure 4-20: Misdetections vs Channels ....................................................................................... 69
Figure 4-21: Allocations Rate vs Speed........................................................................................ 69
Figure 4-22: False Alarms vs Speed ............................................................................................. 70
Figure 4-23: Misdetections vs Speed ............................................................................................ 70
Figure 5-1: Segment distribution in a cluster ................................................................................ 74
Figure 5-2: Clusters in a RSU based Network .............................................................................. 75
Figure 5-3: Contents of Tk time slot ............................................................................................. 76
Figure 5-4: Database maintained by the RSUs ............................................................................. 76
Figure 5-5: Data Collection and Sensing ...................................................................................... 79
Figure 5-6: Data Calculation ......................................................................................................... 80
Figure 5-7: History Records Updating .......................................................................................... 81
Figure 5-8: Allocation Rate Vs No. of Cars ................................................................................. 83
Figure 5-9: Allocation Rate Vs No. of Channels .......................................................................... 84
Figure 5-10: Allocation Rate Vs Speed ........................................................................................ 84
Figure 5-11: False Alarm Rate Vs No. of Cars ............................................................................. 85
Figure 5-12: False Alarm Rate Vs No. of Channels ..................................................................... 85
Figure 5-13: False Alarm Rate Vs Speed ..................................................................................... 86
Figure 5-14: Rejection Rate Vs No. of Cars ................................................................................. 86
Figure 5-15: Rejection Rate Vs No. of Channels ......................................................................... 87
Figure 5-16: Rejection Rate Vs Speed .......................................................................................... 87
Figure 5-17: Forced Leave Ratio Vs No. of Cars ......................................................................... 88
Figure 5-18: Forced Leave Ratio Vs No. of Channels.................................................................. 88
Figure 5-19: Forced Leave Ratio Vs Speed .................................................................................. 89
x
Figure 6-1: Overall Fuzzy Logic System ...................................................................................... 90
Figure 6-2: Membership Function of Speed ................................................................................. 91
Figure 6-3: Membership Function of Message Priority ................................................................ 92
Figure 6-4: Membership Function of Time .................................................................................. 92
Figure 6-5: Availability of Channel-1........................................................................................... 94
Figure 6-6: Availability of Channel-2........................................................................................... 94
Figure 6-7: Availability of Channel-3........................................................................................... 94
Figure 6-8: Availability of Channel-4........................................................................................... 95
Figure 6-9: Availability of Channel-5........................................................................................... 95
Figure 6-10: Availability of Channel-6......................................................................................... 95
Figure 6-11: Availability of Channel-7......................................................................................... 96
Figure 6-12: Availability of Channel-8......................................................................................... 96
Figure 6-13: Speed Vs Time for Channel 1 .................................................................................. 98
Figure 6-14: Speed Vs Time for Channel 2 .................................................................................. 98
Figure 6-15: Speed Vs Time for Channel 3 .................................................................................. 99
Figure 6-16: Speed Vs Time for Channel 4 .................................................................................. 99
Figure 6-17: Speed Vs Time for Channel 5 .................................................................................. 99
Figure 6-18: Speed Vs Time for Channel 6 ................................................................................ 100
Figure 6-19: Speed Vs Time for Channel 7 ................................................................................ 100
Figure 6-20: Speed Vs Time for Channel 8 ................................................................................ 100
Figure 6-21: Message Priority Vs Time for Channel 1............................................................... 101
Figure 6-22: Message Priority Vs Time for Channel 2............................................................... 101
Figure 6-23: Message Priority Vs Time for Channel 3............................................................... 102
Figure 6-24:Message Priority Vs Time for Channel 4................................................................ 102
Figure 6-25: Message Priority Vs Time for Channel 5............................................................... 102
Figure 6-26: Message Priority Vs Time for Channel 6............................................................... 103
Figure 6-27: Message Priority Vs Time for Channel 7............................................................... 103
Figure 6-28: Message Priority Vs Time for Channel 8............................................................... 103
xi
LIST OF ABBREVIATIONS
MANETS Mobile Ad-hoc Networks
AP Access Point
BSS Basic Service Set
VANETS Vehicular Ad-hoc Networks
NHTSA National Highways Traffic Safety Administration
RSU Road Side Unit
V-I Vehicle to Infrastructure
V-V Vehicle to Vehicle
NOW Network on Wheels Group
PATH Partners for Advanced Transit and Highways
DSR Dynamic Source Routing
AODV Ad-hoc on Demand Distance Vector
DSDV Destination Sequenced Distance Vector
DoS Denial of Service
VIN Vehicle Identification Number
WAVE Wireless Access in Vehicular Environment
DSRC Dedicated Short Range Communication
EU European Union
HTTP Hyper Text Transfer Protocol
TCP Transfer Control Protocol
UDP User Datagram Protocol
IPv6 Internet Protocol version 6
xii
LLC Logical Link Layer
FCC Federal Communication Commission
CCH Control Channel
HALL High Availability Low Latency
SCH Service Channel
QoS Quality of Support
EDCA Enhanced Distributed Coordinated Access
MAC Media Access Control
CSMA/CA Carrier Sense Multiple Access with Collision Avoidance
DCF Distributed Coordination Access
FHSS Frequency Hopping Spread Spectrum
DSSS Direct Sequence Spread Spectrum
IR Infra-Red
OFDM Orthogonal Frequency Division Multiplexing
HR High Rate
CCK Complementary Code Keying
PBCC Packet Binary Convolutional Code
DIFS Distributed Inter Frame Spacing
NAV Network Allocation Vector
CW Contention Window
AC Access Category
AIFS Arbitrary Inter Frame Space
CCI Control Channel Interval
CR Cognitive Radio
xiii
GSM Global System for Mobile Communication
CDMA Code Division Multiple Access
LTE Long Term Evolution
LAPUs Local Acquisition and Processing Units
ISM Industrial, Scientific and Medical
PU Primary User
CBS Cognitive Base Station
MAP Maximum a Posteriori
RAT Radio Access Technologies
AF Amplify and Forward
SU Secondary User
VDSA Vehicular Dynamic Spectrum Access
CCC Common Control Channel
Cog-V2V Cognitive Vehicle to Vehicle
FL Fuzzy Logic
PSO Particle Swarm Optimization
FUZZBR Fuzzy Logic based multi-hop Broadcast Protocol
NS2 Network Simulator version 2
C# See Sharp
MATLAB Matrix Laboratory
NAM Network AniMator
VINT Virtual InterNetwork Testbed
DARPA Defense Advanced Research Projects Agency
OTcl Object Tool Command Language
xiv
IDE Integrated Development Environment
GUI Graphic User Interface
CLI Command Language Infrastructure
CLR Common Language Runtime
LINPACK LINear Equations Software Package
EISPACK Eigensystem Package Subroutine Computing Facility
LCA Legal Certification Authority
RCAs Regional Certification Authorities
SPs Service Providers
VCA Verification and Controlling Authority
DRSUs Directional Road Side Units
RCUs Road Central Units
GPS Global Positioning System
VHF Very High Frequency
UHF Ultra High Frequency
CATV Community Access Television
SNR Signal to Noise Ratio
ACK Acknowledgement
NACK Non- Acknowledgement
MIMO Multi Input Multi Output
xv
LIST OF SYMBOLS
{ }
Center Frequency of individual channel
CH HALL Channel of DSRC Spectrum
Pilot frequency of primary user networks ( indicates the number of
channel and * indicate the technology used, v for VHF, u for UHF, and t
for CATV, s for secondary user)
H1 Presence of Primary user signal
H0 Absence of Primary user signal
Ci CR Channel number
( ) Detected signal
Re{.} Real part of the complex waveform
( ) Equivalent low pass representation of the detected primary or secondary
user signal.
( ) Additive white Gaussian noise with zero mean
Energy of the detected signal
Maximum time period
( ) Probability density function
Gamma function
I Modified Bessel function
Instantaneous signal to noise ratio (SNR)
Degree of freedom
( ) Safe speed
xvi
( ) S peed of leading vehicle at time t
( ) Gap of leading vehicle at time t
Average speed
( ) Deceleration function
Driver’s reaction time (usually 1 second)
Maximum allowable speed of the vehicle
Acceleration
( ) Desired speed
Human error factor between 0 and 1
( ) Final speed at time t
Main coordinator
Forward edge coordinator
Backward edge coordinator
Request
Request categories
Cluster member vehicle
{ ( )} Likelihood of channel availability
Probability of detection
Probability of detection given
Instantaneous SNR of different (VHF, UHF, CATV) signals on CR
spectrum
Average SNR
Result Array. * represents M for Main, F for Forward and B for Backward
Coordinators
Median path loss in urban environment
xvii
Base station antenna height in meters
Mobile station antenna height
Carrier frequency in MHz
Antenna height correction factor
Distance in kilometers
Predicted median path loss for suburban areas in decibels
Rayleigh random variable
Mode of the distribution
Distance segment
Third last segment of backward cluster
Front coordinator of backward cluster
Third segment of front cluster
Backward coordinator of front cluster
M No. of channels
Sensing interval
Vehicle identity,
X and Y Coordinates received from GPS to locate the segment
CH Channel sensed from 1 to M
RES Sensing result (1 or 0)
Segment number
K No. of time slots
Time slot number
S No. of time intervals in one
Probability of availability
xviii
Ratio of historic data.
1
Chapter 1
INTRODUCTION
1.1 VANETS
Approximately 1.24 million people die every year and nearly 20 to 50 million are affected by
fatal injuries due to road accidents around the globe according to a report published by the World
Health Organization [1]. The report further states that road accidents are the eighth leading cause
of fatal injuries and may become the fifth leading cause if proper measures are not taken to
reduce road accidents. Road traffic injuries are estimated to cost over US$ 100 billion to low and
middle income countries, which is an estimated 1-2% of their gross annual product [2]. Report
by National Highway Traffic Safety Administration (NHTSA) also finds out the fatalities in road
accidents from 1994 to 2012 [3].
Vehicular Ad-hoc Networks (VANETs) are a special form of MANETs implemented on road
infrastructure, where the mobile nodes are vehicles. Road side units (RSUs) are installed at a
particular distance along the road and road junctions for vehicle to infrastructure (V-I)
communication and for vehicle to vehicle (V-V) communication for vehicles at larger distance
apart and hence cannot communicate directly. VANETs differ from MANETs due to high speed
of vehicles, high battery power, high density, intermittent connectivity, security issues, limited
vehicles movement, static geometry of road, reliability, integrity, large scale infrastructure, real-
time guarantees.
VANETs are especially designed to decrease and then eventually eliminate the road accidents
while increasing the travel comforts. Road accidents are decreased by timely inter-vehicle
communication to inform the vehicles behind about e.g road broken, road blocked, railway
crossing, bridge broken, junction ahead, security barrier, and accident warning. Travel comfort is
2
enhanced by providing gaming services, Internet, fuel station locations, restaurant locations,
service area locations, and on-line payments [4]. Figure 1-1 provides a view of a normal VANET
[5].
Figure 1-1: An example of VANET
Besides IEEE several consortiums are working on the development of VANETs. These include
Car-to-Car consortium and Network on Wheels group (NOW) in Europe; Berkley PATH in
USA; and Fleetnet Projects in Germany [6] [7].
1.2 COGNITIVE RADIO AND VANETS
Due to increase of vehicular traffic resources provided for the vehicular communication are
becoming insufficient. Especially emergency messages require reliable and fast enough
communication resources in order to avoid accidents. Provision of communication by Cognitive
Radio (CR) technology is beneficial for vehicular communication for which accurate spectrum
sensing and coordination techniques are needed.
3
A lot of wireless technologies are in practice around the globe today, like Wi-Fi, Wi-Max,
Cellular Telephone Technologies (e.g. GSM, CDMA, and LTE), TV bands, and IR remotes. It
can be found by scanning that some of the frequency bands are over utilized, while others are
under-utilized which costs a lot to the authorities. CR is an excellent way of dealing with the
spectrum under-utilization problem. It can be observed by means of scanning that some of the
slots of wireless spectrum are occupied by the primary (Licensed) users, while some slots are
empty. Empty slots can be made available to secondary (un-licensed) users which are denied
channel access due to congestion in their own spectrum. Empty slots are referred to as holes [8]
[9]. There are three transmission modes of CR, which are interweave, spectrum overlay and
spectrum underlay. Comparison of these modes is discussed in [10].
Many signal detection techniques can be used for spectrum sensing in order to improve the
probability of detection. Some techniques are energy detection, matched filter detection, cyclo-
stationary detection, and wavelet detection. Every technique has its own pros and cons [11] [12].
Spectrum sensing is a very important aspect of cognitive radio networks as the important
decision of using a licensed spectrum is to be taken based on the results obtained by spectrum
sensing. Hence spectrum sensing must be carried out accurately and in a timely manner so that
reuse of the licensed spectrum by the secondary users can be achieved with minimum or even
negligible interference to the licensed primary users [13] [14].
In CR, unlicensed (Secondary) users are allowed to borrow unutilized bandwidth from licensed
(Primary) users. Sharing of network resources continues in this fashion until primary users need
more of the spectrum, in which case secondary users have to vacate the spectrum [15] [16].
Cognitive Radios can be built so they are smart enough to use parameters (such as carrier
frequency, bandwidth, and transmission power) of that particular spectrum band and particular
wireless technology [17].
It can be observed that some spectrum bands like TV channels have spatio-temporal patterns.
During some hours of the day and in some areas some channels are on while others are off.
These patterns can be exploited to improve the efficiency of spectrum sensing by making it more
directed [18].
4
1.3 CONTRIBUTIONS OF THESIS
This dissertation highlights several issues related to the VANETs and use of cognitive radio
technology in VANETs. Main objective of this work is to eliminate the mis-detections and
reduce the false alarms in CR-VANETs. Main contributions of this thesis are summarized as
follows:
Initially a structure of VANETs is proposed for highways and urban environment and
then emphasises is given to highway portion of the structure and more simulations are
conducted to verify its worth. It is proposed to use Directional Road Side Units (DRSUs)
instead of normal Road Side Units (RSUs).
Cognitive radio concept is used in VANETs. We have proposed multi-spectrum sensing
algorithm based on three coordinators to sense and allocate un-occupied channels to
secondary users (vehicles). Results were compared with other popular approaches.
The proposed algorithm was modified and in order to create realistic environment, Hata
model [19] for large scale fading and Rayleigh Fading model [20] for small scale fading
were used. Also modification in Gibbs Mobility model has been proposed for the
movement of vehicular traffic [21].
This algorithm was further enhanced to build and periodically update the history of
cognitive sensing results during the day time in order to utilize this history to find the
most probable holes for use by the vehicles. The database is stored at the DRSU and
given to the requesting vehicles through the coordinators. Hence bulk of sensing is
decreased by utilization of the historic sensing database built on the basis of spatio-
temporal-frequency [22].
Fuzzy Logic techniques are used for the allocation of CR channels for the vehicular
network. On/Off timings of different types of TV channels, speed of the vehicles,
message priorities and time dividing twenty four hours of the day are considered as inputs
to fuzzy engine to find out the allocation of TV channels to vehicles.
5
1.4 ORGANIZATION OF THE THESIS
This thesis has been organized as follows. Chapter 2 provide the background over which our
work is based. Chapter 3 includes our proposed ‘Structure of VANETs’. Chapter 4 provide the
details of our proposed techniques for the use of cognitive radios in VANETs. Chapter 5 includes
our work involving use of historic sensing data for allocating CR channels to vehicular nodes.
Chapter 6 includes our work to use fuzzy logic techniques for the allocation of cognitive radio
channels to the vehicular nodes. Chapter 6 includes the conclusion and future work. Appendix A
provide the details of tools used for simulations for our entire work.
6
Chapter 2
BACKGROUND
2.1 MANETS
Mobile Ad-hoc Network (MANET) is a wireless temporary network which may change
continuously with the entrance and exit of mobile nodes. In MANETs mobile nodes form an
autonomous transitory association by communicating among themselves using a wireless
medium. MANETs do not rely on the fixed infrastructure for communication instead information
is shared through wireless nodes which may be single or multi-hop. Nodes which lie in each
other’s communication range dynamically discover each other and communicate on single hop
basis and the nodes which are not in the communication range of each other rely on the
intermediate nodes to act as routers to relay the packets to the destination node and communicate
on multi-hop basis [23] [24] [25]. As these nodes are battery powered, so these are often energy
constrained. These nodes randomly move around resulting in rapid and un-predictable
topological changes. In such an un-predictable multi-hop environment nodes have to act quickly
to form a network and relay packets in the absence of access points (fixed infrastructure). Figure
2-1 shows a simple example of MANET [26].
7
Figure 2-1: Simple Mobile Ad-hoc Network
Considering IEEE 802.11 as MAC (Medium Access Control) layer MANETs have two modes of
operation. One is ‘Infrastructure mode’ and another ‘Ad-hoc mode’. In Infrastructure mode all
nodes communicate with each other via an ‘Access Point (AP)’ which is connected to the
internet through wired media. AP behaves as a base station for all nodes. Each AP has a
configurable communication range termed as ‘Basic Service Set (BSS)’. In order to extend the
range more APs can be installed. There need to be an overlap among the cells of APs to avoid
breaks in the communication. In Ad-hoc mode all the users within certain area form a personal
network among each other. Every intermediate node behaves as a router in case distance between
transmitter and receiver is greater than their communication range. Figure 2-2 gives a pictorial
view of both modes [26].
Figure 2-2 Modes of Operation in MANETs
8
Major characteristics of MANETs include; autonomous, infrastructure-less and less expensive. It
is scalable and more flexible. It involves multi-hop routing, dynamic network topology and
device heterogeneity. MANETs are self-created, self-organizing and self-administrable
networks. [27] [28]:
Some complexities and challenges of MANETs include; Lower reliability compared to wired
medium, dynamic topology, routing overhead, limited physical security, time varying channels,
hidden terminal problem, packet losses due to transmission errors, mobility induced route
changes, battery constraints, security threats, interference, energy constrained operation,
bandwidth constrained variable capacity multi-hop links [27] [28].
Some of the MANET applications include; military battlefield, collaborative work, local level,
personal area networks, commercial sectors, emergency services, education, entertainment,
coverage extension [27] [28].
2.2 VANET
VANETs are being designed specially to avoid loss of lives due to road accidents of millions of
people around the world. VANETs are also intended to provide toll services, location based
services, and infotainment. Main applications of VANETs are collision avoidance, cooperative
driving, traffic optimization, payment services, location based services, entertainment
applications [29].
Unlike MANET, most VANET nodes (Vehicles) are equipped with large batteries and due to
availability of a charging mechanism within the vehicle; the power problem is almost negligible
[30] [31] [32]. VANETs are specially designed for moving vehicles that include many
applications classified as public safety, traffic management, freight/cargo transport, transit,
traveler information/support etc. The primary goal of the public safety application is to reduce or
even eliminate accidents which result in fewer injuries and fatalities, lowering the direct or
indirect financial costs, and reducing traffic congestion. Examples are forward obstacle detection
and avoidance, lane departure warning, turn accident warning, intersection accident warning, low
9
bridge warning, roll over warning, work zone warning, stopped vehicle warning and railway
crossing warning. The goal of traffic management is to improve the flow of traffic in order to
facilitate the passengers and drivers and reduce travel time. Application examples are smart
traffic signals, variable message signs, rapid response to incidents, enhanced public transit,
emergency vehicle warning, central traffic management and electronic toll collection [33].
Enhanced transit systems include traffic signal priority, bus only lane enforcement, bus turn light
priority, automated fare collection and reporting, automated passenger counting, route
optimization and schedule tracking, rider information, on demand transit services, security
systems, fleet operations and maintenance, parking, and many other on-board systems. Freight
and cargo systems include vehicle registration/ inspection and credentials, route guidance and
tracking, vehicle monitoring and maintenance systems, cargo monitoring and tracking, and fleet
operations. Traveler information and support includes pre-trip planning, route and fare
information, access to news, weather reports and internet, navigation aids, traffic information on
routes, access to personal information during the trip, restaurant and fuelling station information,
and vehicle repair center information. Other entertainment services include audio, video, and
email services [34] [35].
Due to a different nature of VANETs as compared to ordinary MANETs, different routing
protocols are being considered. These include Reactive protocols like Dynamic Source Routing
(DSR), Ad-hoc on Demand Distance Vector (AODV) and Proactive protocols like Destination
Sequenced Distance Vector (DSDV), Wireless Routing Temporally- Ordered Routing
Algorithm, and Lightweight Mobile Routing protocols [36] [32].
2.3 SECURITY ISSUES IN VANETS
Unlike other types of MANETs, loss of human lives can be caused if any incorrect warning or
message is communicated in VANETs. Hence special security measures are needed. Basically
there are two different environments in VANETs: Infrastructure environment and Ad-hoc
environment. Legal authorities, manufacturers, trusted third parties and service providers form
the Infrastructure environment which provides the services to vehicles through RSUs. Vehicle to
10
vehicle communication through communication units (On Board Units) installed in vehicles,
form the Ad-hoc environment [29].
2.3.1 COMMUNICATION PATTERNS
Different communication patterns are found in VANETs, which are detailed below [29].
a. V2V Warning Propagation
Messages like road closure or accident warning for the vehicles behind and the messages like
'ambulance or other emergency vehicle coming' in order to get the lane cleared by the preceding
vehicles.
b. V2V Group Communications
Messages for vehicles belonging to a particular area or the vehicles of particular sort like
vehicles manufactured by single manufacturer, or all public transport vehicles or all load
carrying vehicles.
c. V2V Beaconing
Messages like current speed of vehicle, direction of vehicle, use of brakes. These messages are
mostly one hop messages delivered to one vehicle behind.
d. I2V/V2I Warning
These include messages like intersection ahead, railway crossing ahead and service area ahead,
fall in the category of Infrastructure to Vehicle (I-V) whereas the messages like road closure fall
in the category of Vehicle to Infrastructure (V-I).
2.3.2 OVERVIEW OF ATTACKS IN VANET
Different sort of attacks are found in VANETs. These are classified as under [29] :
11
a. Attacks on identification and authentication
Two types of this sort of attack are Impersonation and Sybil. In Impersonation, the attacker
pretends to be another entity by stealing the attributes of any other entity (vehicle). In Sybil one
entity poses the credentials of many other entities and behaves as different entities at different
times.
b. Attacks on Privacy
Sometimes, attackers try to obtain the identity of the driver (usually owner) and sometimes they
try to find the location of the vehicle to harm the passengers.
c. Attacks like non-repudiation
Some people try to deny the fact that they have sent any message like wrong road condition
information (road closure, speed, crossing ahead etc.). Also sometimes the people deny the
receipt of any warning message and may cause accidents.
d. Attacks on confidentiality
Sometimes attackers try to listen into others' communications, to gain access to confidential
information (eavesdropping). This type of attack is quite serious and may cause huge losses to
any person or enterprise.
e. Attacks on availability
Availability of any service or hardware like RSU is very important for services to continue.
Some attackers try to hide some services and may cause the overloading of communication
channels (Denial of Service attack or DoS attack). Some attackers overload the computation
capabilities of any vehicle to cause loss to the vehicle or passengers.
f. Attacks on data trust
Some attackers create inaccurate data to harm others. This may reduce the reliability of the
whole system.
For security purposes different mechanisms have been proposed and provided. Manufacturers
issue each vehicle a vehicle identification number (VIN), whereas legal authorities issue License
12
Plate number. VIN is assigned to uniquely identify manufactured vehicle, while License Plate
number uniquely identifies the vehicle in an administrative domain.
To deal with the issue of authentication 'Electronic License Plate' has been proposed which will
not only identify the vehicle but also authenticate it. For the purpose of privacy Public Key
Certificates have been proposed. Two techniques being used for this purpose are Identity based
Cryptography and Pseudonymous Short-lived Public Key Certificates. [37]
For the location of any vehicle by legal authorities, a Location Cloaking Technique [38] has been
provided. Aggregation technique [39] has also been used for this purpose.
2.4 IEEE 802.11P / WAVE / DSRC
A special frequency band of 5.850-5.925 GHz has been allocated for the purpose of vehicular
communication in USA [40]. Similar bands have also been allocated in Japan and Europe. Due
to high mobility normal IEEE-802.11 is not suitable for VANET applications so IEEE has
developed a special IEEE-802.11p standard for VANETs [41] [7] [42]. IEEE 802.11p/WAVE is
not a single standard, but it is a group of standards which comprises P1609.1 (Resource
Manager), P1609.2 (Security Services), P1609.3 (Networking Services), 1609.4 (Multi-channel
Operations) as shown in Figure 2-3, and 802.11p (MAC Layer) [36] [43] [44]. The European
Union (EU) is also getting close to allocating 30 MHz in 5 GHz band, especially for the
vehicular communication [45].
13
Figure 2-3: IEEE 802.11p/WAVE Protocol Stack
The IEEE has standardized 802.11p/WAVE (Wireless Access in Vehicular Environment) for
vehicle to vehicle communication(V-V) and vehicle to infrastructure communication (V-I). This
process has been initiated with DSRC (Dedicated Short Range Communication) spectrum
allocation. The USA Federal Communication Commission (FCC), in the year 1999, allocated 75
MHz, 5.9 GHz band DSRC especially for the V-V and V-I communication as shown in figure 2-
4. As shown in the figure, DSRC is divided into seven 10 MHz channels. Channel 178 is control
channel (CCH) and is especially reserved for public safety messages. Channel 172 is High
Availability Low Latency (HALL) channel and is reserved for critical safety V-V applications.
Channel 184 is reserved for High power public safety applications. The other four channels are
service channels (SCH) used either for safety or non-safety applications [13]. Channels 174 and
176 or 180 and 182 can be combined to get two 20 MHz channels. The range of the DSRC
standard is around 300 meters to a maximum of 1000 meters. Its data rate is from 6 to 27 Mbps.
It is half-duplex communication standard.
14
Figure 2-4: DSRC 75 MHz Spectrum
IEEE 802.11p uses physical layer of IEEE 802.11 and utilizes the advanced QoS support
provided by IEEE 802.11e as shown in Figure 2-5 [46]. It utilizes the Enhanced Distributed
Coordinated Access (EDCA) feature of 802.11e QoS in the MAC layer, which is based on
Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) also known as
Distribution Coordination Function (DCF) in IEEE 802.11.
Figure 2-5: IEEE 802.11p Access Layer
In DCF, the station will only transmit a frame if it senses the medium free for a certain time
period known as Distributed Inter Frame Spacing (DIFS). Another time period known as
Network Allocation Vector (NAV) is also defined in IEEE 802.11 which specifies that for how
long the medium is busy. Stations must wait for NAV to complete before transmitting a frame.
Another timer is Contention Window (CW). If there is a collision; station start a back-off timer
15
between CWmin to CWmax and after the timer stops; stations start sensing the medium and
sending the frames.
In EDCA MAC some improved features have been introduced which include four priority
queues for Background, Best Effort, Video, and Voice traffic, which are called Access
Categories (AC). DIFS interval in EDCA is called Arbitrary Inter Frame Space (AIFS). Each AC
has different CWmin and CWmax values based on priorities and AIFS is calculated based on
different CW value. Voice traffic has highest priority and AIFS calculated for this AC is around
34 µs as compared to 41 µs DIFS in IEEE 802.11a. In case of collision higher AC wins access to
the medium.
The nature of traffic is different on highways as compared to urban environments. Hence a
flexible structure is required to address the issues of both types of road networks. Furthermore
there are several issues that may arise while designing the structure and standards for VANET.
Categorization of messages is also needed to save the bulk of overhead due to security in
message size.
2.5 ISSUES IN VANETS
VANETs face several issues which need to be addressed while designing the standards and a
foolproof structure for VANETs [4] [47]. These issues are described as follows:
2.5.1 VEHICLE DENSITY
In urban areas road network face a high density of vehicles and hence vehicles are bound to move
with slower speed, whereas there is low density of vehicles on highways and motorways allowing
the vehicles to move with higher speed. Therefore communication network design for the urban
areas needs to be treated differently than the highways considering Transmission, Routing,
Quality of Service, Security and Location [48].
16
2.5.2 HIGH MOBILITY
In normal MANETs nodes move with slower speed as compared to the Ad-hoc network formed
by the vehicles. Hence in MANETs Ad-hoc network remain intact for a larger duration compared
to that in vehicular network. Therefore VANETs are treated differently than MANETs especially
while designing the MAC layer [49].
2.5.3 INTERMITTENT CONNECTIVITY
In the vehicular network the probability of connectivity between the vehicles moving in opposite
direction is much smaller compared to the vehicles moving in same direction. Road side
Infrastructure Units (RSUs) can be installed to avoid the intermittent connectivity among the
vehicles [50].
2.5.4 DEFINITION OF SERVICES
Network infrastructure on roads for vehicles involves different type of services when compared to
normal MANETs. These services include toll collection, emergency services like ambulances,
traffic police, location based services including restaurants, fuel stations, workshops. It also needs
the fast dissemination of safety messages which include safe distance between the vehicles,
accident warnings, road closure warnings, and road congestion warnings [51] [52].
2.5.5 IDENTIFICATION OF SERVICE RECIPIENTS
There are some services which involve online payments such as toll payments, online shopping.
In these types of services actual service recipient, who has paid for the service, needs to be
identified accurately [53].
17
2.5.6 INCREMENTAL DEPLOYMENT OF VANET
As the research on VANETs is gaining pace, new ideas are pouring in faster day by day.
Therefore network design shall be more flexible in order to accommodate new changes
facilitating the passengers and drivers [53].
2.5.7 OPEN APPROACH TO VANET ARCHITECTURE
Around the globe in different continents different approaches of VANET have been applied. This
situation is causing non-uniformity in the networks. It is an ultimate requirement to bring the
uniformity to facilitate users travelling around the globe.
2.5.8 UNRELIABLE COMPONENTS GENERATE UNRELIABLE DATA
A large population on the earth can be trusted and we may find more sincere people but there is
some minority which cannot be trusted and hence can generate unreliable data harmful for the
networks. In VANETs this harmful data may cause accidents. Hence the networks shall be
designed keeping in view the rigidity and the security to avoid the harmful activities [54].
2.5.9 PRIVACY
Personal information and identity of any person on the network is needed to be secretly placed so
that no one can know the location of the person and data being used by him except the authority
controlling the network in order to avoid misuse of data and harm to any individual [55].
2.5.10 AUTHENTICATION
Any person using a network service or generating any sort of safety message is strictly needed to
be authenticated as any sort of false safety message cause road congestion or accidents and hence
loss of lives [56].
18
2.5.11 NON REPUDIATION
Cases may occur that some people may deny receipt of any message or sending of any message.
This case may be harmful if someone denies receipt of messages from controlling authorities like
speed control warnings. Hence attention shall be given to design more reliable systems in order to
avoid such occurences [57].
2.5.12 RELIABILITY, INTEGRITY AND SCALABILITY
The communication channels used by the vehicular networks shall be scalable and reliable so that
message delivery can be guaranteed and top priority shall be given to the message integrity [58].
It has been pointed out in [59] that an increase in the control channel interval (CCI) will increase
the reliability of safety as well as non-safety applications in VANETs.
2.5.13 REAL TIME GUARANTEES
Road warnings and safety messages shall be delivered in real-time, to avoid the mishaps and
hazards on the road [60]. Authors in [59] have mentioned that in IEEE 802.11p, any
acknowledgement for broadcast messages is not sent by the vehicles, hence safety related
messages, just like collision warning, may not be received within short time by the following
vehicles which may cause chain accidents. It has been shown in some simulation based papers
that in IEEE 802.11p as the vehicle density increases, the latency of the packets also increases,
which decreases the successful packet reception rate [61] [62]. [63] [64]. Authors in [61] and
[62] have formed an algorithm to control the load for periodical status messages. [63] analyzes
the channel access delay in DSRC and it has been compared with the self-organizing time-
division multiple-access scheme. It has been proved that later scheme performs with better
results. In [64] a framework has been proposed to share the DSRC between safety and non-safety
applications. In [65] a 1-D Markov model has been proposed to calculate the delay and reception
rate in VANETs. In [66] the average delay in DSRC for each access category (AC) was analyzed
without considering back-off delay.
19
2.6 CR-VANETS
Due to the increase in the number of vehicles, it has been observed that DSRC may not be able to
cope with the large number of messages to be communicated in VANETs. Hence it is evident
that vehicles may have to utilize holes in other licensed spectrums, especially for safety critical
messages. It has been observed that VHF and UHF bands are not fully utilized, hence may be
used by un-licensed users. Considering the critical safety applications of VANETs, accuracy in
spectrum sensing is important. A lot of research work is being carried out on use of cognitive
radio in VANETs. Some have proposed stand-alone sensing schemes [67] [68], and some have
proposed cooperative spectrum sensing [11] [12]. Typical architecture of VANETs including
Cognitive Radios is shown in figure 2-6 [69].
Figure 2-6: Typical Structure of CR_VANET
There may be a gap between the methods used to produce results for such techniques, which
may cause discrepancy in the data. For example, one person may use a different fading model to
the next. Another may include small-scale fading while the first may not, which introduces
additional discrepancy in the results. One may use a different mobility model to the next. One
may use a different channel sensing technique to another. All of them are equally valid, but they
lack a common platform for impartial comparison, which takes into account variables such as
movement, fading, noise, interference [70] and other such factors which may cause variability in
20
the simulation results and at the same time allows for varying parameters, in order to perform
simulations in a variety of conditions. The historic approach in the scientific method has always
been that the data being presented should be falsifiable. [71] These differences introduce a
difficulty as to the falsifiability of the data, and even more importantly, the usefulness of such
techniques as comparison is obscured.
As VANET involves many safety related issues, hence the use of cognitive radios (CRs) in
VANETs have forced researchers around the globe to propose the solid and concrete models best
suited specifically to safety of vehicles and passengers. Many different models have been
proposed for the use of CRs in VANETs emphasizing more on the priority information
exchanged among vehicles and between RSUs and vehicles. Since life of the passengers and
pedestrians will be in danger if any delay is caused in safety critical message or the received
message is wrong; hence special attention needs to be given to real time guarantees and
reliability of data considering high mobility and density of vehicles. In IEEE 802.11p/WAVE,
control channel (CCH) is very important and the load of messages on this channel is very high,
so some researchers have proposed in [72] a structure comprising Local Acquisition and
Processing Units (LAPUs), RSUs and vehicles especially for the use of cognitive radios in TV
bands. LAPUs and RSUs are basically the decision centers, hence this architecture is centrally
controlled by these units. Responsibility of the RSUs is to assign the holes to the requesting
vehicles in their cell (Coverage area of RSUs) considering the load on CCH applied by the
member vehicles.
2.6.1 STANDALONE CR SENSING
In [73] researchers have proposed that a specific channel from TV spectrum band shall be
assigned to each vehicle to sense and use. Every vehicle shall sense and use the spectrum
independently and information gathered be exchanged between vehicles for future decisions on
the use of spectrum. In this proposed scheme there is no coordination among the vehicles and a
particular hole may be occupied by more than one vehicle at a time which may damage the data
of all vehicles using that hole. Hence this problem needs special attention. Researchers in [74]
have proposed unlicensed Wi-Fi channels in urban areas and ISM bands (2.4 GHz or 5 GHz) are
21
not fully occupied by the primary users (PUs), hence the vehicles can periodically sensed, and
use the spectrum holes and share the information between each other for future usage. Another
solution is proposed in [75] based on belief propagation. In this proposal every vehicle
broadcasts a message to its neighboring vehicles containing the information about the presence
of primary user signals in its range. All vehicles receiving this message decide on the basis of
belief messages received from all of their neighbors and their own observations; that which PU
channels to use for communication. As this scheme involves messaging among all the
neighboring vehicles, which creates a lot of communication and suffers from slow processing
due to dependence on all of its neighbor data. In [43] it is proposed that each secondary user
senses the PU spectrum and uses if the hole is detected. It also shares the information on DSRC
channels with its neighbors. In [76] authors have proposed a three state model, one describing
the empty slot (hole), other specifying the presence of primary user, and third specifying the
occupancy of channel by any secondary user. This scheme also insists on individual spectrum
sensing by the secondary users with no collaboration. The proposals presented in [16], [17], [18],
[19], and [20] are based on the decentralized structure and may suffer from congestion,
interference and collisions.
2.6.2 CENTRALIZED AND COOPERATIVE CR SENSING
Cooperative spectrum sensing can be effective in high mobility environment [77] A group of
vehicles may combine to sense and form a centralized database to be used for utilization for a
cluster. Authors have shown in [78] that as a result of spatial temporal diversity received signal
strength can be increased by using cooperative spectrum sensing in high mobility environment.
To reduce the sensing overhead they have proposed an optimal number of sensors to use for
cooperation and number of times to sense. In [79] authors insist to enhance the reliability and
reduce the delay in sensing by the use of cooperative spectrum sensing. They explain that by the
use of spatial diversity the problem of sensing limitations of a single vehicle can be tackled very
well. It is also shown that sensing time can be reduced by the use of cooperative sensing
compared to sensing by a single vehicle. Another work in [80] presents a RSU controlled central
architecture. RSUs sense the spectrum of TV bands and prepare a database to be used by
22
requesting CR-enabled vehicles. In the case centralized infrastructure is not present then
clustering technique can be used with one cluster head responsible for sensing the spectrum and
sharing it with member vehicles on demand [81]. In [82] researchers have proposed to form the
clusters of the vehicles which are moving in groups and one vehicle may be assigned the duty of
cluster head. Spectrum sensing of PU spectrum is performed by every cluster member and results
are sent to the cluster head. It is the responsibility of the cluster head to assign the holes to every
cluster member based on the information received from every cluster member. In this proposal
decision is taken on the basis of hole availability information received from all vehicles and a hole
sensed by one vehicle may be assigned to another vehicle which may be in different environment
due to various factors like shadowing and distance from PU tower. This scheme may also suffer
from delay as the holes detected may be occupied till the decision is received from the cluster
head. In [83] it is proposed to divide the wideband spectrum of CR network in small sub-bands
and assign each small narrow band to a group of nodes (vehicles) to scan and use the spectrum.
Secondary users sense the spectrum and send the results to a Cognitive Base Station (CBS) which
detects the transmitted values using MAP (Maximum a Posteriori) detector and then fuse the
results to find the occupancy of the channel. This scheme may save the time of scanning as
several groups are sensing different narrow bands at a time but lacks the correct decision making
policy as no factor is considered in assigning the narrow band to a group of vehicles. A particular
narrow band may suit one group of vehicles but may not suit another group.
In [84] authors have proposed a cooperative framework for spectrum sensing. Each vehicle senses
the PU spectrum individually and shares the information with neighbors. Each sample of
information collected from a user is assigned a weight, and then the weight is adjusted and
normalized. Afterwards, a decision is made for the availability of a hole and the decision is
forwarded to all neighbors. This paper does not propose the method of utilizing of hole after it is
found. In [85] energy detection techniques have been discussed. This paper does not mention any
cooperative mechanism that vehicles will sense the spectrum and decide about the availability of a
hole and allocation to any secondary user. In [86] multi radio technologies (Multi-RAT) have
been discussed and it has been proposed that every vehicle will sense or use any technology
according to the required class of service. For example it proposes the usage of Wi-MAX for
23
video conferencing and to use Wi-Fi for free data exchange. No cooperative sensing mechanism
or channel allocation mechanism has been proposed.
In [87], a two user model has been discussed in which one acts as a relay sending the information
regarding presence or absence of a PU in the spectrum to other users. Two cooperative schemes
have been discussed. In the first scheme both users detect the PU and first one to detect tells the
other about the presence or absence of a PU signal through a central controller. In the second
scheme, users follow Amplify and Forward (AF) protocol to reduce the detection time. Since
VANET is a multiuser scheme, and each user may have different ideas and findings related to
presence or absence of PU signals, so this proposed scheme is not suitable for a VANET
structure. In [18], the authors have highlighted the challenges and research directions in dynamic
routing emphasizing that a SU (secondary user) suffers from high interference from primary
users and other secondary users. Therefore the secondary transmission must be confined to
ensure sufficient operation of PUs, which thereby deteriorates the QoS provisioning. Due to
channel switching and users’ interference, the longer paths will be chosen for transmissions
which may create a problem of larger routing graphs and end-to-end delay.
In [88], the author has checked the performance of Vehicular Dynamic Spectrum Access
(VDSA) within a TV whitespace environment. A technique for Vehicle Cooperative
Communication and Vehicle Platooning has also been proposed. A test-bed implementing the
VSDA for performance evaluation has also been proposed. In [89], the author has mainly
focused on the spectrum sensing techniques, routing methodology, and security for cognitive
radio vehicular networks. Further, the impact of changes in the network formed by vehicles on
radio propagation channel has been discussed and finally its performance has been evaluated.
In [13] combination of standalone and cooperative spectrum sensing mechanism has been
proposed. It combines the best of both approaches. In this approach, the spectrum is divided into
several non-overlapping channels with a channel spacing of 6, 7, and 8 MHz and the secondary
network has a common channel for the exchange of control information over DSRC (5.9 GHz)
band. A coordinating node (vehicle) periodically senses the spectrum and sends a group of
channels (holes) to the requesting nodes (vehicles). The requesting node on receipt of holes,
rescans the obtained holes and after reconfirmation uses the hole as per class of demand. This
24
approach has its merits and demerits. The merit is that rescanning of holes by requesting nodes
confirms the availability of channel at that time. The demerit is that every channel in a wideband
cognitive radio spectrum may not be a hole for every vehicle due to its dynamic nature and
geographical location.
2.6.3 DETECTION TECHNIQUES AND FADING MODELS
Many researchers have put lot of efforts to analyze the sensing techniques used and fading
models used for CR networks. There are a few works related specifically to the use of CR in
VANETs scenarios. Some existing works are discussed here.
In [85] it is proposed to use energy detection technique over a Gamma-shadowed Nakagami-
composite fading channel [90] for both large and small scale fading in VANETs. Authors have
compared the results obtained in simulation with other composite fading models like Suzuki, Loo
and Rice-Lognormal [91]. In this paper, the authors have not discussed the coordination among
vehicles for sensing the CR network. Researchers in [92] have proposed a model based on
swarm-intelligence (the division of labor model in ant colonies), and claim to reduce congestion
problems to the spectrum database. This paper proposes the use of cellular networks to access the
spectrum database and to use IEEE 802.11p channels for inter-vehicular communication for
cooperative sensing. Vehicles, after sensing, send the results to their neighbors. In [93] authors
have pinpointed the drawbacks of the energy detection technique and have proposed the use of
the covariance based technique. According to the authors, the energy detection technique is
sensitive to noise uncertainty [94] [95] [96]. Furthermore as signals are correlated in most
practical applications hence energy detection should not be considered the method of choice.
The authors propose the computation of the sample covariance matrix of the received signal
based on the received signal samples; then to extract test statistics from the sample covariance
matrix. Finally, a decision is made for the presence or absence of the signal is made by
comparing the ratio of two test statistics with a threshold.
25
2.6.4 SPECTRUM MANAGEMENT AND QOS SUPPORT
Many ideas can be found in relation to spectrum management which also supports the quality of
service requirements. Some interesting ideas relating to dynamic spectrum access have been
explored in [88]. Authors have presented Vehicular Dynamic Spectrum Access (VDSA) to deal
with dynamic spectrum allocation problem being faced by VANETs. For spectrum
measurements and machine learning some VDSA techniques have been presented. As a special
case for the intelligent channel selection reinforced learning is used. (Also presented by [97]).
Secondary Users are modelled as clients and RSU as server. Queuing Theory has been used to
analyse the VDSA technique used. Authors in [98] differed from the idea presented in [88]
pointing out that this technique is over optimistic as vehicles can move out of the range of RSUs.
Hence they have presented an analysis relating to the vehicular dynamics.
Works presented in [84] [75] [99]propose that channel selection algorithm shall allocate the
channel on the basis on QoS requirement of the requesting vehicle according to the specific
location of the vehicle. Another work relating to energy efficiency has been proposed in [100]
presenting an optimization problem whether to use relay base transmission or to use direct
transmission in an objective to minimize energy consumption considering delay as Qos
constraint. These presented approaches are based on centralized architecture. Approach
presented in [81] involves the clustering strategy. It is hybrid of centralised and distributed
approaches. It involves a cluster head and other vehicles acting as cluster members in the group.
In this paper the problem of optimum channel access has been studied in detail in order to
provide QoS support for data transmission. Vehicles use one radio for the DSRC channel and
another for opportunistic spectrum access. For decision making Markov Decision Process has
been used. The cluster head takes the decision if any vehicle will be added in the cluster or not
and also reserves some bandwidth for that vehicle in the DSRC channel. In this way the size of
the cluster is also controlled. Cluster head also collects the sensing information from cluster
members and broadcasts the decision of assigning holes to cluster members. Hence the problem
relating to the opportunistic spectrum access, cluster size control and reservation of some
bandwidth in DSRC channel is solved using linear programming.
26
Some proposals on distributed spectrum management focus on the design of MAC [101] as well
as QoS requirements which may be modified for vehicular environment [102]. Proposal in [102]
insists on one common control channel (CCC) and transmits a beacon frame at regular intervals.
Further it sub-divides beacon in three phases: channel sensing, channel contention and data
transmission. Nodes sense the channel at the first instance, and then use CCC for reservation of
time slot for data transmission. Hence in data transmission phase contention free transmission is
performed by reserving the channel first. QoS is provided by first allowing the priority users for
channel reservation then if slots are available they are utilised for remaining nodes.
2.6.5 DISTANCE SEGMENTATION
In [73] authors have proposed a cooperative sensing Cog-V2V proposal in which every vehicle
collects the sensing data and shares with the vehicles in its range on common CCH channel.
Roads are divided into segments and spectrum horizon of each vehicle is defined. Every vehicle
based on the received information and its own observations for the related road segment and
frequency, aggregates the data and finds the similarity function. It decides the availability of any
channel based on the aggregated data and the similarity function. In [84], the authors have
further enhanced the Cog-V2V framework by considering the impact of correlated shadowing on
the sensing output. Based on correlation, weight factor has been calculated and multiplied with
the sensing binary output to get the decision of primary user’s presence or absence. These two
papers do not take into account the temporal variations in the availability of any spectrum band
in deciding the spectrum band as a hole.
2.7 FUZZY LOGIC
Fuzzy Logic (FL) provides a simpler method to arrive at a definite conclusion based upon vague,
ambiguous, imprecise, noisy, or missing input information. Real world problems are not discrete
(0 or 1). Just like sound may be very low, low, medium, high, very high, and exceptionally high.
FL converts real world problems into membership functions describing graphically the variable
states. Based on those states decisions are made by if-then rules including all possible events and
27
decision is taken based on those rules. Due to several unique features FL is considered a good
choice for many control problems. FL is robust and hence does not need noise free and precise
inputs. Output is a smooth function even in the case of variations in inputs or a wide range of
inputs. FL can be modified easily as it depends on user defined rules. Any number of inputs can
be processed by FL and generating any number of outputs. But to avoid complexity FL should be
broken up into small processes. Nonlinear systems can easily be modelled by a FL system [103].
In [104], the authors have proposed a new evolutionary approach based on a hybrid fuzzy-
particle swarm optimization (PSO) hybrid algorithm to solve the problem of the distribution
networks reconfiguration.
In [105], the authors have designed a Fuzzy-Logic based spectrum allocation algorithm, by
which the RSUs check actual CCH contention conditions, and extend dynamically the CCH
bandwidth in the case of network congestion, by utilizing the detected vacant frequencies by the
sensing module. In [106], the authors have used distance, velocity, signal strength, spectrum
efficiency as input variables to fuzzy logic. Mamdani type fuzzy rule base system has been used.
Based on the fuzzy if-then rules decision is taken for the availability of the spectrum. In [107],
the Spectrum Utilization Efficiency, Degree of Mobility and Distance of Secondary user to the
PU are used as input variables. A fuzzy logic rule base is used to avoid collision among various
contenders for spectrum access. According to authors SU with higher possibility can access the
spectrum with guarantee.
The work in [108] proposes FUZZBR a fuzzy logic based multi-hop broadcast protocol for data
dissemination in VANETS. Authors insist that FUZZBR has low overhead as it uses a subset of
neighbour vehicular nodes for relaying of data. Fuzzy logic has been used to select the relay
node considering different parameters like inter vehicle distance, node mobility and signal
strength. Also here a lightweight retransmission mechanism has been proposed to transmit a
packet in case a relay node fails. Membership functions of distance between vehicles, speed, and
signal strength have been defined to select the relay for which the output function rank defines
the best relay.
28
An algorithm for CR channel selection using fuzzy logic has been proposed in [109]. Idle time
statistics are derived based on historic data collected for the channel occupancy by PUs. In order
to learn the competition level for the available radio channels among secondary users
information exchange based method is used. Then using fuzzy logic, idle time statistics and
competition level among secondary users are integrated into hybrid decision criteria. Longest
value referred by the hybrid decision criteria indicates the most available channel.
29
Chapter 3
STRUCTURE OF VANETS
Structure of VANETs plays an important role in timely dissemination of messages. If the
structure is too complicated, then messages may suffer delays. In case message is urgent, then
delay factor may cause accidents. Moreover, the structure needs to be designed keeping security
in mind. Some of the models have been proposed in literature, like [110] [111]. So the structure
shall be simple and robust in order to reduce the propagation time and increase the security
respectively. Messages also need to be categorized to discriminate while routing immediate,
urgent and ordinary messages.
3.1 MESSAGE CATEGORIZATION
Categorization of messages is important in the sense that these may be treated differently by the
channel according to their priority, authentication requirement, or privacy requirement. This
categorization may save the bulk overhead in the size of messages, improve speed and decrease
occupancy of channel by one message for long time. These messages are categorized as under
[112]:
3.1.1 EMERGENCY MESSAGE
Examples are accident information, congestion information, bridge broken, and train crossing
etc. This type of message needs authentication, but no privacy.
30
3.1.2 SAFETY MESSAGE
Examples are inter-vehicle distance, speed, intersection collision avoidance, and location
information. This type of message also needs authentication, but no privacy.
3.1.3 GPS MESSAGE
This type of message mostly provides road map with reference to location of the vehicle. This
type of message also needs authentication, but no privacy.
3.1.4 PROBE MESSAGE
This is the periodic message for keep-alive between the road side unit and the vehicle. This type
of message needs neither authentication nor privacy.
3.1.5 TRAVELLER INFORMATION
Examples are signal status, road signs, school ahead, hospital ahead, and service ahead etc. This
type of message needs authentication but does not need privacy.
3.1.6 LOCATION BASED SERVICE
Examples are toll collection, and online payments etc. This type of message needs both
authentication and privacy.
3.1.7 INFORMATIVE MESSAGE
The messages involving ordinary internet browsing are informative type of messages. This type
of message needs neither authentication nor privacy.
31
3.1.8 E-MAILS
E-mails usually need both authentication and privacy.
3.2 PROPOSED VANET STRUCTURE
Earlier research on VANETs has not proposed any structure which may cater to the needs of
both urban (city) and Highway/Motorway environments, or the different problems related to
both. We propose a VANET structure as depicted in Figure 3-1. The proposed structure includes
the LCA (Legal Certification Authority) which is the overall controlling authority of the
country's VANET communication. LCA has sub-offices CAs (Certification Authorities) for
certification in different regions to reduce the burden on one office. These issue certificates to
RSUs (Road Side Units), SPs (Service Providers), and the vehicles belonging to individual
regions.
The structure also includes the VCA (Verification and Controlling Authority) having sub-offices
for different regions. VCAs are responsible for the verification of certificates in the individual
regions. In case a certificate is found to be illegal, VCAs have the authority to shut off the
communication of the concerned vehicle or service provider and add them in the RL (Revocation
List). VCAs can also act in case any certificate holder is violating the rules.
The structure includes the SPs which provide different services like toll collection, internet
services, entertainment services (games, audio, video), and location based services like location
of restaurant, fuel station, or workshops. Concrete security measures are required for the services
involving online payments.
32
VCA
VCA 1 VCA 2 VCA 3 VCA m...
LCA
CA 1 CA 2 CA 3 CA m...
SP 1
SP 2
SP n
.
.
.
DRSU DRSU
DRSU
RCU RCU
RSU
Figure 3-1: Overall Proposed Structure of VANETs
The structure includes highways/motorways as well as urban (city) environments. On highways
instead of RSUs, DRSUs (Directional Road Side Units) have been provided. This helps in group
formation on highways. A group of vehicles moving in one direction is easier to manage on
highways as the vehicles moving in one direction stay in the group for a longer period of time
whereas vehicles moving in opposite direction stay in communication with each other for a
shorter duration of time; hence the overhead in the message specially for group formation will be
useless. Cluster formation based on bidirectional traffic as shown in Figure 3-2 is of no use as the
vehicles moving in opposite direction live in cluster for the fraction of a second. We have
proposed cluster formation for unidirectional traffic as shown in Figure 3-3. Fast and Slow
moving vehicles leave the cluster early. Usually vehicles on highways move with specified
normal speed and these vehicles live in clusters for longer duration. Also computational cost
incurred on cluster modification and reformation will be reduced to a greater extent. Hence the
group formation of the vehicles moving in opposite directions is not recommended.
33
Figure 3-2: Cluster Formation for Bidirectional Traffic
Figure 3-3: Cluster Formation for Unidirectional Traffic
In the urban environment vehicles do not stay longer near to each other so the group formation
on the basis of direction is not possible. Instead groups are formed on the basis of categorization
of vehicles into buses, taxies, official vehicles, private vehicles registered by one regional office.
Hence in cities simple RSUs have been provided. It is also proposed to have RCUs (Road
Central Units) instead of RSUs on junctions, with more resources than RSUs to cater for the
communication needs of vehicles moving on different roads joining the junction. This may help
to reduce the infrastructure and installation costs.
3.2.1 SIMULATION RESULTS
The simulation was built using NS2 for highway scenarios. The highway patch is 2 km long,
and the traffic is moving in both directions. CASE-1 is the simulation in which four RSUs are
provided and are forwarding the traffic of both directions whereas CASE-2 is the simulation in
which four DRSUs have been provided. IEEE 802.11 has been used as the MAC layer protocol
34
whereas AODV has been used as the routing protocol. 2.4 GHz band has been used with 22 MHz
channel bandwidth and a total of 14 channels are provided. For adjacent RSUs or DRSUs non
overlapping channels are selected. Radio propagation model used is two ray ground and Omni
antenna has been used for the simulation. Maximum size of queue used for packets is 50 and
droptail type queue is used. In order to simulate real time traffic Constant Bit Rate over User
Datagram Protocol (CBR over UDP) type packet traffic is used. Bit rate for CBR traffic is 64
Kb/s, packet size is 1000 bits and window size is 20 packets. Simulation time is 250 seconds. In
both cases, the simulation has been performed for 20 mobile nodes (10 in each direction) and 50
nodes (25 in each direction). Average throughput, average end to end delay and packet delivery
ratio have been calculated for all cases.
Average Throughput for 20 mobile nodes increased from CASE-1 to CASE-2 by 3.72%, while
for 50 mobile nodes it increased by 47.44%. For CASE-1 with 50 mobile nodes Figure 3-4
shows a plot of throughput vs time. Throughput in this case goes to zero after about 225 seconds
because nodes reach the final destination and get out of range of the RSU.
Figure 3-4: Throughput (CASE-1, 50 Mobile nodes).
For CASE-2 with 50 mobile nodes throughput is shown by the plot in Figure 3-5 below. Spikes
at the end are because some of the nodes have reached the final destination and are out of the
range of DRSU.
35
Figure 3-5: Throughput (CASE-2, 50 Mobile nodes).
Average End to end delay for 20 mobile nodes decreased by 7.26%, whereas for 50 mobile nodes
it decreased by 23.8%.
Packet delivery ratio for 20 mobile nodes increased by 2.93%, whereas for 50 mobile nodes it is
increased by 5.37%. Results in tabular form which shows the percent change in Average
Throughput, End-to-End Delay and Packet Delivery Ratio are given in Table 3.1 below.
Table 3.1: Comparison 20 to 50 nodes, RSU to DRSU
No. of
Nodes
Throughput
Increase
End-to-End
Delay
Decrease
Packet Delivery Ratio
Increase
20 3.72% 7.26% 2.93%
50 47.44% 23.8% 5.37%
3.3 PROPOSED HIGHWAY STRUCTURE
We propose a VANET structure as depicted in Figure 3-6. VCA has the authority to shut off all
communications of the concerned vehicle or service provider and add them in the RL
(Revocation List). VCA can also act in case any certificate holder is violating the law.
36
Figure 3-6: Proposed Highways Structure for VANETs
On highways, instead of RSUs, DRSUs (Directional Road Side Units) have been provided. This
helps in group formation on highways. A group of vehicles moving in one direction is easier to
manage on highways because vehicles moving in one direction stay in a group for a longer
period of time whereas vehicles moving in opposite direction stay in communication with each
other over a shorter duration of time. So, group formation for vehicles moving in different
directions is impractical.
3.3.1 SIMULATION RESULTS
Simulation parameters remain the same as discussed in section 3.2.1. In both cases, the
simulation has been performed for 50 mobile nodes (25 in each direction) and 100 nodes (50 in
each direction). Average end to end delay and packet drop has been calculated for all cases.
VCA
LCA
SP
DRSU DRSU
DRSU
37
Two graphs of throughput (in kbit/s) versus time (in seconds), both concerning 100 mobile
nodes, are given below. Figure 3-7 shows the case where RSUs deal with bi-directional traffic
and Figure 3-8 shows the case where DRSUs deal with unidirectional traffic.
Figure 3-7: Throughput (RSUs catering bi-directional traffic).
Figure 3-8: Throughput (DRSUs catering unidirectional traffic).
Average end to end delay for 50 mobile nodes decreased by 23.8% whereas for 100 mobile
nodes it decreased by 29.3%.
Packets dropped for 50 mobile nodes decreased by 28.3% whereas for 100 mobile nodes they
decreased by 13.3%. We will also notice an increase in the packet delivery ratio. Results in
tabular form are given in Table 3.2 below.
38
Table 3.2: Comparison 50 to 100 nodes, RSU to DRSU
# of
Mobile
Nodes
Decrease in end-
to-end Delay
Decrease in
Dropped Packets
Packet Delivery
Ratio
Increase
50 23.8% 28.3% 5.37%
100 29.3% 13.3% 12.2%
3.4 SUMMARY
In this chapter first of all categorization of the messages is provided in terms of need of
authentication and privacy [4] [112]. Then we proposed a structure for VANETs for highways
based on Directional Road side Units (DRSUs) catering for unidirectional vehicular traffic and
for urban areas based on Road Central Units at crossings and simple RSUs at other locations.
Simulation tests were conducted for DRSUs and found increase in throughput and packet
delivery ratio and decrease in end to end delay and dropped packet ratio.
39
Chapter 4
COGNITIVE RADIO AND VANETS
4.1 PROPOSED SPECTRUM SENSING FRAMEWORK
Here under we propose a spectrum sensing framework based on three coordinators; Main
coordinator, Forward Coordinator, and Backward coordinator. This technique uses a
combination of both cooperative and standalone sensing methods.
4.1.1 NETWORK MODEL
Each vehicle shall be equipped with a CR-Radio capable of sensing and utilizing the available
resources (holes) in the spectrum, a GPS system to track its position and navigate its path to the
intended destination, in addition to the equipment utilizing DSRC IEEE 802.11p resources for
preferred normal communication among the vehicles and between vehicles and RSUs at 5.9 GHz
band. In case DSRC resources fall short of the requirements, vehicles can use CR-spectrum for
their communication.
In USA TV bands use 7-1002 MHz spectrum with VHF low band using 5.9-88 MHz, VHF high
band using 175-216 MHz, and UHF using 470-890 MHz with the channel bandwidth of 6 MHz
CATV uses 7-48 and 55-1002 MHz with 4.5 MHz channel bandwidth. Hence if vehicle uses CR-
spectrum, it may divide the spectrum into k non overlapping channels {C|Ci, i=1, 2, 3, ------, k},
not necessarily equally spaced (in order to utilize the available spectrum resources of various
technologies.) centered at{ }
. Secondary network of vehicles use channel 172 (High
Availability Low Latency Channel) on 5.9 GHz band DSRC spectrum, in order to communicate
40
information and management messages for the use of CR spectrum among vehicles and between
vehicles and RSUs. This channel is referred to as CH.
4.1.2 SPECTRUM SENSING MODEL
The CR network has primary users, which are licensed and must not face any interference from
secondary users. Hence spectrum sensing shall be carried out in such a way to avoid any
interference to primary users’ communication. Every wireless technology has its own pilot
carriers for its different channels. Let indicate the pilot frequency of primary user networks (
indicates the number of channel and * indicate the technology used; such as v for VHF, U for
UHF, and t for CATV). Parallel sensing is used for VHF, UHF, and CATV channels to reduce
the time of sensing the channels. The energy detection technique is used to effectively identify
the presence and absence of primary user signals. Binary hypothesis test can be applied where H1
is used for presence and H0 is used for the absence of primary user signals. As proposed in [113]
and [13] using these two hypothesis conditions, the band pass signal observed by a secondary
user for the channel Ci can be represented as
( ) { {[ ( ) ( )]
}
{ ( )
} (4.1)
Where real part of the complex waveform is represented by Re{.}, i=1,2,3,---,M. M is the
number of CR channels, is the carrier frequency of primary channel, if we use the secondary
user pilot carrier frequency for the detection of secondary user signal occupying the primary
network we use , ( ) is the equivalent low pass representation of the detected primary or
secondary user signal. Additive white Gaussian noise with zero mean is represented by ( ).
Using the energy detection technique and bandpass filter, the energy of detected signal ( ) for
the period can be represented as
∫
( )
(4.2)
41
is a random variable and it has chi-square distribution. Its probability density function can be
expressed as [113] and [13].
( ) {
⁄ ( ⁄ ) ( ⁄ )
⁄
(
⁄ ) (
) ⁄
( ⁄ ) (√ )
(4.3)
Here is the Gamma function and defined as ( ) ∫
, I is the modified Bessel
function, is the instantaneous signal to noise ratio (SNR), and is the degree of freedom.
Situations may arise due to fading if the detected signals have low SNR confusing it with the
noise. In this case caused by misdetection meaning signal is present but it may be considered as
absent. In this case interference may be caused by the secondary users with the communication
of the primary user. There may also be a case of false alarm meaning the signal is absent but is
considered present. But this case causes no harm as far as interference is considered. User
requests are considered random to the main coordinator. Main coordinator furnishes the requests
on the first come first serve basis.
4.1.3 VEHICLE MOBILITY MODEL
Mobility model used for the vehicle mobility is mainly Gipps model [114] with a slight
modification that if the distance between the current vehicle and the leading vehicle is less than
safe distance, then the current vehicle shall overtake the leading vehicle. Other parameters follow
the Gipps model as given below.
( ) ( ) ( ) ( )
( )
(4.4)
( ) [ ( ) ( )] (4.5)
( ) [ [ ( ) ( )]] (4.6)
42
In equations (4), (5), and (6) above, ( ) is the initial speed, ( ) is the speed of leading vehicle
at time t, ( ) is the gap of leading vehicle at time t, is the average speed, ( ) is the
deceleration function, is the driver’s reaction time (usually 1 second), is the maximum
allowable speed of the vehicle, is the acceleration, ( ) is the desired speed, is the
human error factor between 0 and 1, and ( ) is the final speed at time .
4.1.4 SPECTRUM SENSING AND COORDINATION FRAMEWORK
Different ideas have been proposed for the sensing of the CR spectrum. Some are standalone,
where sensing results are faster but they cause unnecessary interference to the primary user
signals and also among secondary users using CR spectrum. Some approaches are cooperation
based which create a master/slave relationship among coordinating node or RSU and secondary
users; also sensing results may not be accurate. Further scalability and intractability problems are
also noticed in these types of approaches. One approach proposed in [13] claim to combine best
of both approaches but the results sensed by a coordinator may not be accurate in high mobility
VANET environment. Our proposed coordinating sensing idea has more than one localized
coordinating node and sensing results are based on a majority decision, which causes more
accuracy, scalability, intractability, and reduces interference.
In the proposed coordination sensing framework, a main coordinator coordinates the sensing
activities and forwards the sensing results to the requesting secondary users based on its sensing
results and the results received from , forward edge coordinator, and , backward edge
coordinator. Sensing results received by from and are finalized using best two out of
three decisions. For example if a channel sensed is considered available (hole) by two or more
coordinators, it is stored as a hole in the database. Requesting secondary users after receiving a
group of channels (holes) from re-sense the channels, but for a slightly longer time compared
to coordinators, confirming these as holes, and then pick a channel suitable for the class of
service requested. Secondary users receive information based on three differently located
coordinators which can be considered more accurate thereby causing negligible interference to
primary user networks.
43
The proposed sensing coordination framework works as follows. The main coordinator and
the two edge coordinators and sense periodically the CR spectrum channels . and
send the sensed results to periodically on channel of DSRC spectrum. Sensing is
performed using the energy detection technique based on hypothesis as given in (4.1). , after
receiving the sensed results makes a decision based on majority for declaring a channel as
occupied or empty (hole) and stores the calculated results in the database. Upon arrival of request
(received from a secondary user on channel of a particular class of service, the
main coordinator sends the sensed results of that class of service to . The requesting
secondary user (after receiving the results), re-senses the received channels, picks a channel
and sends an ACK (acknowledgement) message mentioning the number of channel picked for
use. after receiving an ACK, deletes the channel picked by the from its available channels
database. If the requesting secondary user after re-sensing the channels finds the channels
occupied, it sends a NACK (Non-acknowledgement) message to the , after receipt of which
sends the freshly sensed results to .
a. Coordinators Selection Phase
Coordinators selection is performed dynamically whenever a secondary user wants to access the
CR spectrum. In case of RSU based system, a RSU will act as the coordinator for its coverage
range hence RSUs shall be installed in such a way to fully utilize the CR resources. In case when
RSUs are not available, separate set of coordinators shall be selected for the both sides of
highways. This will ensure that the groups will remain intact for a longer period of time on
highways.
When a secondary user wants to use the CR network for transmission; it firstly checks if there
is any main coordinator available in its vicinity. It confirms this confirming if it is periodically
receiving messages from any . If no main coordinator is available, the requesting node itself
starts behaving as main coordinator . After being selected as , it sends message on
channel to all nodes (vehicles) in its coverage range to send the GPS coordinates and current
speed. When every node replies with the GPS coordinates and speed, selects farthest node in
front with the speed equal or less than (within 5%) the speed of as the front edge coordinator
44
; and farthest node in back with the speed equal to or greater than (with in 5%) the speed of
, as the backward edge coordinator . collects the GPS coordinates and speed of the
cluster (moving cell) nodes periodically and reassign the duty of and to the nodes
matching the criteria as discussed above as due to difference in speed coordinating nodes may be
at different locations than expected.
b. Spectrum Sensing Phase
CR-spectrum sensing responsibility is on the coordinators , and . Each coordinator will
sense the channels { } in parallel, periodically, and in a proactive manner. Each coordinator
will sense the channels independently of each other. Each coordinator shall be able to detect the
presence or absence of the primary or secondary user signals while remaining efficient to reduce
the likelihood of interference. At each iteration of energy detection on channels { } , the
and will send the results on channel to and which will compile the results in terms of
likelihood of channel availability { ( )} according to the following rule.
“The channel will be declared available if and only if at least two of the coordinators will
decide the availability of the channel on the basis of hypothesis (signal absence) or (signal
presence)”.
The probability of detecting the user (primary or secondary) is defined as:
[ ]
[
] (4.7)
In order to obtain , we need to express or depending on whether the PU is present or
absent respectively in the place of , and also we need to derive joint pdf as expressed
later in this section. It is important that during the sensing phase all the coordinators shall be
synchronized. i.e they all shall sense the channel at the same time.
45
At , and the received bandpass waveform on as per equation (4.1) is given by:
( ) {
{[ ( ) ( )]
}
{ ( )
} (4.8)
Here X represents M for main coordinator, F for front coordinator and B for backward
coordinator. As per equation (4.2) the output of energy detection at each coordinator is given by
equation (4.9). In this step it is assumed that mean of noise component is zero and variance is
.
∫
( )
(4.9)
The pdf of individual ( ) is given by equation (4.10).
( )
{
⁄ ( ⁄ ) ( ⁄ )
⁄
(
⁄ )
(
) ⁄
( ⁄ ) (√
)
(4.10)
Since all the coordinators perform the energy detection independently, hence the joint pdf and
third order pdf are given by equations (4.11) to (4.14).
(
) ( ) (
) (4.11)
(
) ( ) (
) (4.12)
(
) ( ) (
) (4.13)
(
) (
) ( ) (
) (4.14)
Therefore probability of detection at can be evaluated as:
46
[
]
[
]
[
]
2 [
] (4.15)
∬ (
)
∬ (
)
∬ (
)
∬ (
)
(4.16)
∫ ( )
∫ ( )
∫ ( )
∫ ( )
∫ ( )
∫ ( )
2∫ ( )
∫ (
)
∫ (
)
(4.17)
47
The detection threshold can be obtained from equation 6 in [13]. Average detection probability
can be obtained by averaging out of over SNR of different signals on CR spectrum as
given by (4.18) below.
∫
⁄
(4.18)
Where is the instantaneous SNR of different (VHF, UHF, CATV) signals on CR spectrum
and is the average SNR.
Upon arrival of the requests from the vehicles, main coordinator send the list of detected
holes to the requesting vehicles on first come first serve basis. Requesting vehicle upon receipt of
list of holes re-senses the received holes multiple times to re-confirm the availability of the
channels and picks the first available channels for the transmission. It sends the channel
acquiring message to the so that it may delete the channel from the list of available channels.
Re-sensing by the requesting vehicle reduces the chances of misdetection and hence avoids
interference with the primary user signals.
4.1.5 COORDINATORS SELECTION AND SENSING ALGORITHMS
Coordinators selection algorithm is given below:
1:
2:
(Front Coordinator selection)
3:
4: (p front nodes)
5: ( ) ( )
48
6:
7: end
(Back coordinator selection)
8:
9: (q back nodes)
10: ( ) ( )
11:
12: end
Algorithm for sensing of the spectrum by three coordinators and compiling the results is given
below:
1: ( )
2:
3:
4:
5:
6:
7:
8: ( ( ) ( ) )
49
( ( ) ( ) )
( ( ) ( ) )
9: ( )
10:
11: ( )
12: end
Flow charts of every activity are given below in figures 4-1 to 4-4.
50
Figure 4-1: Coordinators Selection
Start
Did Nj detect NM?
No
Set NM = Nj
Perform Front Coordinator
Selection
Perform Back Coordinator
Selection
End
Yes End
51
Figure 4-2: Front Coordinator Selection
Figure 4-3: Back Coordinator Selection
Start
Set NF = N1
Set i = 2
i > p? NoD(Ni to Nj) > D(NF to Nj)
Increment iSet NF = NiYes
No
Yes
End
52
Figure 4-4: Spectrum Sensing
4.1.6 SIMULATION AND RESULTS
Simulations have been built using Microsoft C# 5.0. Simulations have been carried out by
varying number of channels, number of vehicles and by changing vehicle speed. The results were
evaluated over a single run, but varied insignificantly over multiple runs. Different scenarios
have been developed for standalone sensing case, cooperative sensing case, and our proposed
spectrum sensing case. A 2 km piece of highway is considered for the simulation, cognitive radio
towers are placed at a perpendicular distance of 91 km from the highway and transmission power
taken is 100 mW. Channels are varied from 10 to 100, number of vehicles from 10 to 100, and
speed from 40 to 110 km/h. Inter vehicle communication range is 240 m and sensing range is
400 m, deceleration factor is 2.12 s-2
, which is , a constant in ( ̅) ̅, and maximum
53
acceleration fraction is 0.01, which is the fraction of average speed a car can accelerate in a
second, Noise power density considered is . This was far too high to be
realistic but was kept high to compare performance of systems under extreme conditions. Power
density threshold is , sensing interval taken is 20 ms, vehicle communication
interval is 20 to 30 ms and vehicle silent time is taken as 0.4 to 0.5 s.
a. Probability of Correct detection
Correctness of the detection technique to identify the presence or absence of primary user signal
is an important factor in determining the validity of the algorithm and robustness of the
equipment used. We have plotted the probability of correct detection versus number of channels,
number of vehicles and vehicle velocity as shown in figure 4-5, figure 4-6, and figure 4-7
respectively. Figure 4-5 shows that as we increase the number of channels on the primary user
network, the probability of correct detection of CR network in the case of standalone and
cooperative sensing methods decreases but in the case of our proposed scheme it remains
constant and almost equal to one. Figure 4-6 shows that if we increase the number of vehicles
probability of correct detection improves with standalone sensing and it decreases with
cooperative sensing (due to load of traffic on a single coordinator and coordination time
required) whereas with our proposed scheme it remains constant at almost one. Figure 4-7 shows
that if we increase the vehicle velocity probability of correct detection with standalone and
cooperative sensing techniques is low as compared to our proposed scheme. From these three
graphs it is evident that our proposed sensing technique outclasses the standalone sensing and
cooperative sensing techniques.
54
Figure 4-5: Probability of Correct Detection versus No. of Channels
Figure 4-6: Probability of Correct Detection versus No. of Vehicles
0.997
0.9975
0.998
0.9985
0.999
0.9995
1
1.0005
0 20 40 60 80 100 120
Co
rrec
t D
etec
tio
n R
ate
# of Channels
Independent Sensing
Proposed Sensing
Cooperative Sensing
0.997
0.9975
0.998
0.9985
0.999
0.9995
1
1.0005
0 20 40 60 80 100 120
Co
rrec
t D
etec
tio
n R
ate
# of Vehicles
Independent Sensing
Proposed Sensing
Cooperative Sensing
55
Figure 4-7: Probability of Correct Detection versus Vehicle Velocity
b. Probability of Misdetection
If any vehicle wrongly identifies the presence of primary user as absence, it can start using that
channel and cause interference with the primary user signals. Hence algorithms and techniques
shall correctly detect the presence of primary user so that secondary user may not harm the
primary user network. Graphs given in figure 4-8, figure 4-9, and figure 4-10 show the
probability of misdetection versus number of channels, number of vehicles, and vehicle velocity.
These graphs show that misdetection caused by our proposed technique is almost eliminated.
Some misdetection is observed in the case of standalone and cooperative sensing techniques
which may cause interference with the primary user network by the unlicensed secondary users
hence misdetection must be eliminated completely. These graphs show that our proposed scheme
have completely eliminated the misdetection and hence does not interfere with the primary user
network.
0.997
0.9975
0.998
0.9985
0.999
0.9995
1
1.0005
60 70 80 90 100 110 120
Co
rrec
t D
etec
tio
n R
ate
Speed (Km/h)
Independent Sensing
Proposed Sensing
Cooperative Sensing
56
Figure 4-8: Probability of Misdetection versus No. of Channels
Figure 4-9: Probability of Misdetection versus No. of Vehicles
-7
-6
-5
-4
-3
-2
-1
0
0 20 40 60 80 100 120lo
g(M
isd
etec
tio
n R
ate)
# of Channels
Independent Sensing
Proposed Sensing
Cooperative Sensing
-7
-6
-5
-4
-3
-2
-1
0
0 20 40 60 80 100 120
log(
Mis
det
ecti
on
Rat
e)
# of Vehicles
Independent Sensing
Proposed Sensing
Cooperative Sensing
57
Figure 4-10: Probability of Misdetection versus Vehicle Velocity
4.2 MODIFIED SPECTRUM SENSING AND ALLOCATION
MODEL
We first lay out a consistent framework for simulating different scenarios. In this framework, we
will vary one factor and keep others constant to achieve consistency in the comparisons. We start
with a “bird’s eye” view of the entire algorithm, and then examine each process in finer detail,
until we are down to the bare essentials. We do this by describing the mobility, sensing,
coordination and scheduling algorithms individually, as well as when they are needed and what
ties them in with the rest of the process.
The proposed framework starts as any time-based simulation model should, by dividing the
entire time over which the simulation is to be performed into a number of time slices, and
simulating over each of those slices.
-7
-6
-5
-4
-3
-2
-1
0
70 80 90 100 110 120 130 140lo
g(M
isd
etec
tio
n R
ate)
Speed (Km/h)
Independent Sensing
Proposed Sensing
Cooperative Sensing
58
Figure 4-11: The Overall Algorithm
The psuedocode corresponding to the flowchart in Figure 4-11 is given below.
nTimeSlices := runningTime/timeStep
for i := 0 upto nTimeSlices - 1
[Perform vehicle movements]
[Perform sensing/coordination/allocation]
next
The procedure described in Figure 4-11 begins by calculating the number of time steps over
which the simulation is to be performed, and then running repetitive procedures over each time
slice. First, we move the vehicles as desired, and as dictated by the mobility model of choice.
Second, we perform channel sensing, coordination and allocation as required by the different
models used.
From first glance, the flexibility of this approach is immediately visible. We can choose the
mobility model while performing movements and we can also choose what methods to use for
59
the other tasks. From here, the first sub-process is completely defined by the choice of mobility
model to be used. The second one needs to be expanded, and it is shown below.
Figure 4-12: Allocation, coordination and sensing
The psuedocode corresponding to the flowchart in Figure 4-12 is given below.
[Perform PU state changes]
if i*timeStep >= nextCoordinationTime
[Perform coordinator selection]
[Perform coordinated sensing]
end if
[Perform sensing and occupation for applicable vehicles]
60
The process shown in Figure 4-12 begins by performing the state changes for PU. This is defined
as the PU (licensed) occupying or vacating the channel in the given time slice. After this, we
check if coordination needs to be performed. If it does, we perform coordination and store results
with the coordinators in a given range. We then perform sensing and occupation for individual
vehicles that need to transmit over the CR spectrum.
Here, the coordinator selection sub-process is defined completely by the choice of coordination
algorithm used. The other two need to be expanded, as they are below. Note that these processes
are for individual channels or vehicles.
Start
Is the PU scheduled to
occupy or vacate
channel?
Vacate
Tear down transmission channel as scheduled.
Occupy
Schedule next occupation of
channel
End
None
Is there SU on the channel?
Tear down SU communication
Yes
Start transmission.
No
Schedule vacation of channel
Figure 4-13: How to perform PU state changes.
The psuedocode corresponding to the flowchart in Figure 4-13 is given below.
if i*timeStep >= vacationTime && occupying
[Tear down transmission channel]
[Set occupationTime in accordance with scheduling algorithm]
61
occupying := false
else if i*timeStep >= occupationTime && !occupying
if occupied by SU
[Tear down SU transmission]
end if
[Start transmission]
[Set vacationTime in accordance with scheduling algorithm]
occupying := true
end if
Figure 4-13 defines the basic scheduling algorithm used for PU state changes. It defines how the
simulation will decide when a PU is to occupy a channel, and when it is to vacate a given
channel.
Again, a lot of flexibility as to the method of scheduling transmissions is allowed. A message
must be sent to SUs occupying this channel, possibly over the WAVE control channel.
62
Start
Is the car scheduled to
occupy or vacate
channel?
Vacate
Tear down transmission channel as scheduled.
OccupyGet channel(s)
from coordinator (if applicable)
Sense all applicable
channels and attempt to occupy
first
Schedule next occupation of
channel
Was a channel occupied?
No
Back off for a certain time period
YesSchedule next
vacation of channel
End
None
End
Figure 4-14: Sensing and occupation mechanism.
The psuedocode corresponding to the flowchart in Figure 4-14 is given below.
if i*timeStep >= vacationTime && occupying
[Tear down transmission channel]
[Set occupationTime in accordance with scheduling algorithm]
occupying := false
else if i*timeStep >= occupationTime && !occupying
if coordinated sensing scheme
[Get channel(s) from coordinator]
end if
[Attempt to sense/Occupy as necessary]
if occupation was successful
[Set vacationTime in accordance with scheduling
algorithm]
occupying := true
63
else
[Extend occupationTime by back-off time]
end if
end if
These procedures, shown in Figure 4-14, define the (more complicated) scheduling and
allocation scheme for SUs of the CR spectrum. It takes into account the fact that a channel may
not always be allocated to a SU when needed, and it also lays the groundwork for what sensing is
to be performed before a SU will attempt to occupy a channel.
Please note that this last flowchart may vary slightly depending on the needs of the coordination
and allocation scheme. For example, coordination may not be required at all; multiple channels
may be sensed before a back-off is triggered, and so on. The flowchart has been made for the
requirements of [12], but the pseudo-code is more generalized. This also takes into account the
possibility of a misdetection.
4.2.1 MODELS USED FOR THE SIMULATION
a. Hata Model
The Hata Model [19] for suburban areas was used as the large-scale fading model in this
simulation. This depends on the Hata model for urban areas. This gives the predicted median
path loss over a given path. Equation (4.19) describes the median path loss in urban
environments, in decibels.
[ ] (4.19)
Equations (4.20) and (4.21) describe the calculation of in cities with low to medium
populations and cities with a high population, respectively.
( ) (4.20)
64
{ ( )
( )
(4.21)
In equations (4.19), (4.20) and (4.21); is the median path loss for urban areas in decibels,
is the base station antenna height in meters, is the mobile station antenna height, is the
carrier frequency in MHz, is the antenna height correction factor and is the distance in
kilometers.
Equation (22) defines how to calculate the path loss in a suburban area.
(
)
(22)
In Equation (22), is the predicted median path loss for suburban areas in decibels. This
second formula is the one that was used.
The Hata model is a simple model that simulates large scale urban fading. It takes into account
obstacles that may be encountered, and gives the median path loss that a given signal will
experience under given conditions. The simulation for a particular point, which will account for
instantaneous variations in power, is given by Rayleigh fading, which is fed the median power
flux density calculated by the Hata model.
b. Rayleigh Fading
Whereas the Hata model is a large-scale model that predicts averages over a given distance,
Rayleigh fading predicts sudden changes due to rapidly varying channel conditions. Equation
(4.23) gives the probability density function for a Rayleigh random variable.
( )
⁄ (4.23)
In Equation (4.23), is the mode of the distribution. Since the median of this distribution is
given by √ , we can easily use this model for a given median as well, as required by
the Hata model [20].
65
Rayleigh fading is a long-standing model used in wireless communication systems for power
prediction, where a line-of-sight component is absent. Considering the scenario, which is a
suburban environment, it is highly unlikely that one will be present. Therefore, we have chosen
the Rayleigh model for our simulation.
c. Gipps Mobility Model
The Gipps mobility model describes how traffic behaves in a typical vehicular scenario.
Equations (4.4), (4.5) and (4.6) describe how to calculate the new velocity of a vehicle given the
old velocity in a given time step [114].
The Gipps model is one of the classical models used to simulate urban mobility. We choose it for
its simplicity and effectiveness in predicting vehicle behavior. We made one modification to this
model for our purposes: That the vehicle will not go below a certain minimum speed threshold.
Given we are simulating a highway scenario where vehicles may overtake each other, this
assumption is reasonable. Equation (4.24) is the replacement we use for equation (4.6) in our
simulations.
( ) [ [ ( ) ( )]] (4.24)
4.2.2 SIMULATION RESULTS
Here, we give simulation results for three different sensing and allocation schemes, for three
different kinds of data. We vary the number of available channels, the number of vehicles in the
simulation and the average speed of the vehicles. The schemes used are standalone sensing,
cooperative sensing, and proposed sensing, as used in [115]. The other parameters were: Vehicle
sensing range was 400 m and communication range was 240 m, the length of road used was 2
km, the noise power was -133.16 dBW (500 K system noise temperature at a bandwidth of 7
MHz). Two PU transmission towers were used, each 10 km away from the closest point on the
road. The mobility model used was the Gipps model [114] and the wave propagation model used
was the Hata model [19] for large scale fading and the Rayleigh model [20] for small scale
66
fading. The frequency used was 150 MHz, the base station height was 50 m and the mobile user
height was 1.5 m. The average speed for vehicles was set to 100 km/h ± 20% per vehicles (when
not varied), the number of channels to 100 (50 per PU tower) and number of vehicles to 50 (25
per side). A vehicle could deviate to within 10% of its average speed. The reaction time was set
to 1 second, and the deceleration function was 2.12 times velocity. The human error factor could
vary between 0.25 and 0.4 for each vehicle. The vehicle back-off time was 10 ms and the
coordination time was 20 ms.
The recorded data in the simulations:
a) The number of successful allocations of a channel.
b) The number of false alarms (a channel was vacant but was sensed as occupied).
c) The number of misdetections (a channel was occupied but was sensed as vacant).
Misdetections are dangerous because they can interfere with PU or other SU
communications.
The following figures 4-15 to 4-23 show the results of simulations run under a number of
scenarios.
Figure 4-15: Allocations Rate vs Vehicles
-0.05
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0 20 40 60 80 100 120
Allo
cati
on
Rat
e
# of Vehicles
Independent Sensing
Proposed Sensing
Cooperative Sensing
67
Figure 4-16: False Alarms vs Vehicles
Figure 4-17: Misdetections vs Vehicles
-4
-3.5
-3
-2.5
-2
-1.5
-1
-0.5
0
0 20 40 60 80 100 120lo
g(Fa
lse
Ala
rm R
ate)
# of Vehicles
Independent Sensing
Proposed Sensing
Cooperative Sensing
-4
-3.5
-3
-2.5
-2
-1.5
-1
-0.5
0
0 20 40 60 80 100 120
log(
Mis
det
ecti
on
Rat
e)
# of Vehicles
Independent Sensing
Proposed Sensing
Cooperative Sensing
68
Figure 4-18: Allocations Rate vs Channels
Figure 4-19: False Alarms vs Channels
-0.05
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0 20 40 60 80 100 120
Allo
cati
on
Rat
e
# of Channels
Independent Sensing
Proposed Sensing
Cooperative Sensing
-4.5
-4
-3.5
-3
-2.5
-2
-1.5
-1
-0.5
0
0 20 40 60 80 100 120
log(
Fals
e A
larm
Rat
e)
# of Channels
Independent Sensing
Proposed Sensing
Cooperative Sensing
69
Figure 4-20: Misdetections vs Channels
Figure 4-21: Allocations Rate vs Speed
-4
-3.5
-3
-2.5
-2
-1.5
-1
-0.5
0
0 20 40 60 80 100 120lo
g(M
isd
etec
tio
n R
ate)
# of Channels
Independent Sensing
Proposed Sensing
Cooperative Sensing
0
0.05
0.1
0.15
0.2
0.25
0.3
70 80 90 100 110 120 130 140
Allo
cati
on
Rat
e
Speed (Km/h)
Independent Sensing
Proposed Sensing
Cooperative Sensing
70
Figure 4-22: False Alarms vs Speed
Figure 4-23: Misdetections vs Speed
The differences between the proposed model and the other schemes were:
a) Cooperative sensing had cluster formation on both sides, and just one coordinator per
cluster. Every vehicle senses the spectrum and sends the result to coordinator. In this
-3
-2.5
-2
-1.5
-1
-0.5
0
70 80 90 100 110 120 130 140lo
g(Fa
lse
Ala
rm R
ate)
Speed (Km/h)
Independent Sensing
Proposed Sensing
Cooperative Sensing
-4
-3.5
-3
-2.5
-2
-1.5
-1
-0.5
0
70 80 90 100 110 120 130 140
log(
Mis
det
ecti
on
Rat
e)
Speed (Km/h)
Independent Sensing
Proposed Sensing
Cooperative Sensing
71
model, the coordinator compiles the result and allocate the channel, but cars blindly
followed the coordinator instead of sensing it themselves before occupation.
b) Independent sensing had no clusters at all and each car sensed the channel independently
before occupying it.
Independent sensing had a lack of coordination, due to which errors were high in most cases. In
cooperative sensing, clusters were plagued by vehicles leaving and entering the clusters at high
rates, because of which coordination was rendered difficult.
In our model, cluster formation was concrete and a three coordinator system established high
certainty of the result of sensing the spectrum. The number of allocations was slightly lower due
to a small delay coming from coordination, but it was offset by a smaller number of false alarms
and (more importantly) misdetections.
As one can clearly see, in all three cases, the number of misdetections is almost eliminated. In
this limited simulation, they actually came out to be nearing zero in all cases. This is shown in
Figure 4-17, 4-20 and 4-23. The number of false alarms is also greatly reduced, as shown in
Figure 4-16, 4-19 and 4-22. One can see a fall in the allocation rate as compared to standalone
sensing and cooperative sensing in Figure 4-15, 4-18 and 4-21. This is because coordination may
have an adverse effect on the allocations, due to the time taken to communicate channels to and
from the coordinator. Further as the false alarms are greatly reduced and misdetections are
almost eliminated the reduction in allocation rate is eminent. In any case, the benefits outweigh
this slight disadvantage, particularly that of the elimination of misdetections. It is very important
that a SU should not interfere with a PU’s communication. This is achieved in the proposed
sensing model. Graphs show the standard deviation and error bars not visible on certain graphs
are too small to be displayed.
4.3 SUMMARY
In this chapter we have proposed a cognitive radio sensing model for VANETs based on three
coordinators (Main, Front, and Back). Sensing is performed by the three coordinators and results
72
are prepared on best two out of three basis. Gibbs Mobility model has been used for vehicle
mobility. In simulations probability of correct detection and probability of misdetection have
been calculated and compared with standalone and cooperative approaches and found
considerable improvement. Then we used Hata model for large scale fading, Rayleigh Fading
model for small scale fading and suggested a modification in Gibbs Mobility model. Simulation
results show a considerable improvement in allocation rate and decrease in false alarms and
misdetections.
73
Chapter 5
HISTORY BASED SPECTRUM
SENSING AND ALLOCATION
Here, we propose a system which relates to suburban areas and highways. It is understood that RSUs are
installed at proper distances on the sides of highways. For efficient and systematic communication among
vehicles and between vehicles and infrastructure Directional RSUs (DRSUs) shall be installed as
proposed in Chapter 3. As DSRUs have separate processors and storage locations for each direction,
cluster management can be achieved easily because clusters remain intact for longer times on separate
directions of any highway. Vehicles are managed in clusters on each direction and each cluster has three
coordinators: (Main Coordinator), (Front Coordinator), and (Back Coordinator) for CR
spectrum sensing and coordination activities as proposed in section 4.1. The Cognitive Radio spectrum,
especially the TV spectrum, is divided into M channels. Communication range of is divided into
fifteen equal distance segments denoted by [73], [84] as shown in Figure 5-1. is moving in distance
segment , in , and in ; which are progressively changing as vehicles move forward and
jump into the next distance segment. is the third last segment of the backward cluster and carries the
front coordinator of that cluster. is the third segment of front cluster and carries the back
coordinator of that cluster. The segments and are overlapping segments between the
backward and current cluster; whereas the segment and are overlapping segments between the
front and current cluster. It is important to mention here that highway considered here is busy, but as
traffic conditions and patterns change dynamically and may be some of the segments contain no vehicles
74
(in shallow traffic conditions) then positions of back and front coordinators may shift one or two
segments front or backwards. The selection of the coordinators was defined in section 4.1.
Figure 5-1: Segment distribution in a cluster
Every vehicle sends and receives the control information on the control channel (CCH) of 5.9 GHz
IEEE 802.11p/DSRC spectrum. Vehicles in segments to send and receive control messages to
; vehicles in segments to to ; and vehicles in segments to to . Vehicles in
segments , also send and receive control messages to and vehicles in segments ,
also send and receive control messages to to receive the information on next segments from front
cluster or to provide information to the backward cluster. Further every vehicle is equipped with a GPS
(Global Positioning System) chip in order to locate its position and track the roads and intended
destination.
5.1 HISTORY UPDATING
Every vehicle (Secondary User) senses the M channels of TV spectrum every second using the
energy detection technique in order to identify the presence or absence of licensed users (Primary Users)
in the spectrum. Binary hypothesis test is conducted; indicating for the presence of primary user and
for the absence of primary user as given by equation 1.
Every after sensing sends a sense message containing the results (Binary 1 for and binary 0 for )
for every channel it sensed along with its coordinates indicating the distance segment and also its identity
75
in order to avoid duplication of messages on CCH channel to , , or depending on its location in
the cluster. Each entry in the list contained in the message is formatted as follows.
< X Y CH RES Time>
Where is the vehicle identity, X and Y are the coordinates received from GPS to locate the segment,
CH is the channel sensed from 1 to M, RES is the sensing result (1 or 0), and Time is the sensing time. It
shall be noted that in to and to also send the sense message to and
respectively in order to be aware of the future locations on the path. Figure 5-2 shows the cluster
formation in a RSU based network. Instead of RSUs we have provided Directional RSUs (DRSUs) as
proposed in Chapter 3.
Figure 5-2: Clusters in a RSU based Network
, , or on receipt of sensed messages from the vehicles for one interval compile the results
based on majority decision if more than one result is obtained from one segment for the same channel
f. Results for the space, time, and frequency are finalized and sent to the main coordinator for onward
submission to the RSU. A sample entry in the list contained in message sent to RSU is of the format as
given below. Output indicates the binary 1 for primary or secondary user presence and binary 0 for
absence. RSUs on receipt of update message updates in spatial-time-frequency database.
76
< f Output>
5.2 HISTORY PRESERVATION
RSU divides the twenty four hours of the day into K time slots. Each time slot is further subdivided
into S time intervals. For each data entries are collected from the clusters for every distance segment
and every channel. One time slot containing S time intervals is represented in Figure 5-3.
Figure 5-3: Contents of Tk time slot
Probability for the availability of the channel is computed by the RSU based on number of zeros in the
time slot (5.1) and a database for K time slots, each channel, and each segment is stored. Database entries
computed and stored by the RSUs will look like as shown in Figure 5-4.
(5.1)
Figure 5-4: Database maintained by the RSUs
77
RSUs exchange the database with each other and each RSU stores the database for the segments covered
by its basic service set (BSS) and also for the segments covered by the BSS of next RSU. This will enable
the vehicles to utilize the spectrum proactively on their future path to the destination.
5.3 HISTORY UTILIZATION
Each RSU sorts the database in order of high probability for the next time slots and shares the information
with main coordinators of the clusters for the segments covered by that main coordinator and its future
path. The main coordinator shares the information with and . All three coordinators sense the
channels as per procedure discussed in [115], compile the results and send back to main coordinator. ,
on receipt of results from itself and the other two coordinators, compiles the results based on current
availability and high availability probability received from RSU for the segments in its coverage range.
On receipt of CR spectrum utilization request from any moving in segment and time slot ,
sends back the channel numbers in order of higher probability of availability to the lower one for the next
distance segments and time slot . Sample format of reply sent by to the requesting will look like
as shown in Table 5-1.
Table 5-1: reply to
Segment Channels
12 5 3 14 8
7 5 1 2 9
13 11 3 7 4
3 12 15 8 6
78
on receipt of reply from the as per format in Table 5-1, rescans the channels in order of higher to
lower probability and utilizes the first available channel for its communication and continuously scans the
channels for the next segments and switches to other channels if required on its way to the destination.
5.4 SIMULATION RESULTS
Simulations have been performed using Visual C#. Spectrum sensing mechanism used is as presented in
section 4.2 which is the modified version of section 4.1. The mechanism for the movement of vehicles as
well as other parameters has been lifted from sections 4.1 and 4.2. The only thing added on is spatio-
temporal history, databases and predictive sensing as discussed below. As shown in Figure 5-5, whenever
a vehicle wants to use the CR-spectrum, it checks the sensing data it has available in its own database,
corresponding to the current time and space segment. If unavailable, it requests new history data for its
time segment and future segments, and similarly for space, from the coordinators.
79
Figure 5-5: Data Collection and Sensing
Similarly, if the coordinator finds that its own data are inadequate, it requests data from the RSU
in batches, for its entire communication range and then some more, so it won’t have to request
new data too much.
Sensing is performed as detailed in [21], but with a minor difference: Channels with a higher
probability of being empty (also for a longer amount of time) in the given spatio-temporal slot
are sensed first; with the first available channel being occupied.
80
This process is then reversed, with simple (non-coordinating) vehicles sending data to the
coordinators and the coordinators sending it to the RSUs, with the RSUs updating daily records.
Figures 5-6 and 5-7 detail the process.
Figure 5-6: Data Calculation
81
Figure 5-7: History Records Updating
There were a number of different parameters used during the simulation, which are listed as
follows: the vehicle communication range was 240 meters, the average car speed varied from 80
km/h to 120 km/h, with each individual car varying its speed between 5% to 10%. Noise power
was watts (measured by the popular formula for Additive White Gaussian
Noise , where B is the bandwidth (7 MHz for a typical TV channel), k is the Boltzmann
constant and T is the noise temperature (we used 500 Kelvin). The threshold used to measure if
another user was present was watts. The Hata model for suburban areas was used for
fading, with height of transmitting antenna = 50 m, height of receiving antenna = 1.5 m, and the
center frequency = 1.5 GHz. Cars would communicate for 20 to 30 ms, and stay “silent” for 0.4
to 0.5 seconds. The patch of road was 2 km long.
A number of different metrics have been observed in these simulations. The most important of these
metrics are Allocation Rate, False Alarms, Misdetections, Rejection Rate, and Forced Leaves. Allocation
82
Rate is the ratio of number of successful allocations to the number of total attempts hence it provides a
clear picture of the success of the algorithm. False Alarm indicates an empty channel detected busy and
the ratio is number of false alarms to the number of total attempts. Lower false alarms are better for
efficient resource utilization but are not harmful for the spectrum and primary user. Misdetection
indicates a busy channel detected empty. Misdetections are very harmful as it leads to interference with
the primary user signals and are needed to be eliminated. Our algorithm as discussed in section 4.1 and
section 4.2 has successfully eliminated Misdetections. Rejection Rate tells the number of un-successful
attempts to the total number of attempts. Lower Rejection Rate is a measure of success of the algorithm.
Forced Leave ratio indicates the number of times once the vehicles acquire the channels is forced to leave
the channels due to the return of primary user. Lower Forced Leave ratio indicates the fewer disturbances
to the communication of the secondary users.
These metrics have been measured in a number of different scenarios: varying numbers of cars, number
of CR-spectrum channels and varying average speed of the vehicles, keeping all other factors constant.
Results observed for Allocation Rate are given by graphs in Figure 5-8, Figure 5-9 and Figure 5-10. It is
clear from the graphs that Allocation Rate for our proposed Algorithm is better as compared to the
Independent sensing and Cooperative sensing mechanisms. In Figure 5-8, a higher number of cars cause
congestion in the spectrum, resulting in a lower allocation rate. In Figure 5-9, the proposed model has
lower results in a low number of channels due to it being heavily reliant on PU timings. The lower the
channels, the lower the reliance on the PU timings. In Figure 5-10, it was noted that the number of
requests were higher when there were drops in the graph, resulting in a lower allocation rate. We may see
an increase in the allocation rate compared to the technique presented in section 4.2. Results obtained for
False Alarms Ratio are given by Figure 5-11, Figure 5-12 and Figure 5-13. Results show that False Alarm
ratio of our proposed algorithm is better than Independent sensing mechanism but Cooperative sensing is
better in this respect. This is due to time slots being large: it was possible that in the time slot, the PU had
83
vacated the channel, but it wasn’t added to the empty channels list due to a low probability in the rest of
the time slot. As False Alarms are not harmful so it is not a problem if we obtain Allocation Rate higher,
Rejection Rate lower and Forced Leave Ratio lower. Figure 5-14, Figure 5-15 and Figure 5-16 show the
Rejection Rate graphs. It is evident from the graphs that our proposed algorithm has a lower rejection rate
compared to the other two mechanisms. Figure 5-17, Figure 5-18 and Figure 5-19 indicate the Forced
Leave ratio. The forced Leave Ratio of our proposed mechanism is zero. This is due to the fact that our
proposed algorithm monitors the channels and calculates the higher probability of availability for the
duration of the intended duration of the transmission. Hence it provides the list of those channels which
will be available for a longer duration calculated based on the history. All the graphs show the error bars
specifying standard deviation. Error bars not visible on some graphs are too small to be displayed.
Figure 5-8: Allocation Rate Vs No. of Cars
84
Figure 5-9: Allocation Rate Vs No. of Channels
Figure 5-10: Allocation Rate Vs Speed
85
Figure 5-11: False Alarm Rate Vs No. of Cars
Figure 5-12: False Alarm Rate Vs No. of Channels
86
Figure 5-13: False Alarm Rate Vs Speed
Figure 5-14: Rejection Rate Vs No. of Cars
87
Figure 5-15: Rejection Rate Vs No. of Channels
Figure 5-16: Rejection Rate Vs Speed
88
Figure 5-17: Forced Leave Ratio Vs No. of Cars
Figure 5-18: Forced Leave Ratio Vs No. of Channels
-6
-5
-4
-3
-2
-1
0
10 20 30 40 50 60 70 80 90 100Lo
g(Fo
rce
d L
eav
e R
ate
)
No. of Cars
Proposed Sensing
Independent Sensing
Cooperative Sensing
-6
-5
-4
-3
-2
-1
0
10 20 30 40 50 60 70 80 90 100
Log(
Forc
ed
Le
ave
Rat
e)
No. of Channels
Proposed Sensing
Independent Sensing
Cooperative Sensing
89
Figure 5-19: Forced Leave Ratio Vs Speed
5.5 SUMMARY
In this chapter we have proposed to build a database based on spatial-temporal frequency slots
for the road segments in different time and different frequency channels. Results are obtained
from the moving vehicles and stored in a database at DRSUs and three coordinators discussed
earlier and are continuously updated giving a proper weightage to history and current results.
Allocation is performed based on better probability of availability of channels during that
particular time and for that road segment. Results obtained are compared with other popular
approaches and have found considerable improvement.
-6
-5
-4
-3
-2
-1
0
70 80 90 100 110 120 130 140Lo
g(Fo
rce
d L
eav
e R
ate
)
Speed Km/h
Proposed Sensing
Independent Sensing
Cooperative Sensing
90
Chapter 6
CR CHANNEL ALLOCATION USING
FUZZY LOGIC IN VANETS
It has been observed that some wireless bands especially TV and radio channels have a specific
on/off criteria in a day. Based on this observation a system can be designed to utilize the
spectrum available due to off timings of these channels. We have modeled this system using
fuzzy logic [116] as detailed in this chapter.
6.1 SYSTEM MODEL
Our fuzzy logic system is multi-input multi-output (MIMO). The main purpose of the system is
the selection of the channels from TV channels available on Cognitive Radio (CR) network
based on different input scenarios. The block diagram of our system is given in Figure 6-1
below.
Figure 6-1: Overall Fuzzy Logic System
91
6.1.1 INPUTS AND MEMBERSHIP FUNCTIONS
Our system is a six input eight output system. The first input is the speed of the vehicle. The set
of linguistic values for speeds are very slow, slow, medium, fast, and very fast. Vehicles moving
specifically on highways have a speed from zero to 150 km/h. This range is however different
from models used in chapters 4 and 5 but we wanted to check the system at higher range. Speed
plays an important role in channel assignment. As vehicles move in clusters and sensing is based
on control by the coordinators of the cluster [115] [117], so the vehicles moving with very fast or
very slow speed leave the cluster early whereas vehicles moving with medium speed stay in the
cluster for a longer duration. Thus, vehicles moving with very fast or very slow speed shall be
allocated those channels which are available for a shorter duration of time and the vehicles
moving with medium speed shall be allocated the channels which are available for a longer
duration of time. The speed of the vehicles is assumed to follow the truncated Gaussian
distribution [13]. The membership function of speed is given in Figure 6-2.
Figure 6-2: Membership Function of Speed
The second input linguistic variable is message priority. The set of linguistic values for message
priority are immediate, urgent, normal and ordinary. Immediate messages include emergency
messages like accident information, congestion information, bridge broken, train crossing, and
safety messages like inter-vehicle distance, speed, intersection collision avoidance, and location
information. Urgent messages include traveler’s information like signal status, road signs, school
ahead, hospital ahead, service area ahead and location based services like toll collection, and
online payments. GPS and probe messages fall in the category of normal messages. While net-
surfing and emails fall in the category of ordinary messages [4]. Message category plays an
92
important role in the channel allocation mechanism. Higher priority messages need to have stable
and longer duration channels, whereas lower priority messages can have channels with lower
stability or duration. The membership function in trapezoid shape is shown in Figure 6-3 below.
Figure 6-3: Membership Function of Message Priority
The third input linguistic variable is time in hours of the day. Linguistic values for the time are
mid-night, morning, noon, after-noon, evening and night. Different TV channels have different
on/off timings. Some channels are on late night, some during evening and some during mid-day.
So time is an important factor for the determination of channel allocation of TV channels. The
membership function in trapezoid shape of time is shown in Figure 6-4.
Figure 6-4: Membership Function of Time
Channel sensing is crucial in the channel allocation scheme. We use the spectrum sensing results
obtained by using techniques proposed in our previous work [115] [117] [118]. In this chapter
we have used the sensing result based on the categorization of the different types of TV channels
93
as discussed in next paragraph. We have combined the sensing results of channels 1 to 3 in one
input variable, 4 to 6 in another input variable and 7 to 8 in another input variable.
6.1.2 OUTPUTS
Output variables are eight different types of channels. We have categorized these channels as;
channel-1 is entertainment channel, channel-2 is news channel, channel-3 is infotainment
channel, channel-4 is kids channel, channel-5 is informative channel, channel-6 is educational
channel, channel-7 is food channel and channel-8 is midnight programs channel. These different
categorized channels have different on/off timings. We have formed the membership function for
the availability (off timing) of each type of channel as shown in Figures 6-5 to 6-12.
Figure 6-5: Availability of Channel-1
Figure 6-6: Availability of Channel-2
Figure 6-7: Availability of Channel-3
95
Figure 6-8: Availability of Channel-4
Figure 6-9: Availability of Channel-5
Figure 6-10: Availability of Channel-6
96
Figure 6-11: Availability of Channel-7
Figure 6-12: Availability of Channel-8
We have taken on/off timings of above categorized channels as arbitrary as different countries
have different timings for TV channels. Anyone willing to use this technique has to alter the
timings according to their own available timing slots.
6.1.3 FUZZY IF-THEN RULES
By using vehicle speed, message priority, hours of the day and channel sensing results we can
model the fuzzy IF-THEN rules based the criteria discussed in above paragraphs. Our system
presented in this paper has been modeled using 320 rules. Some rules are given below for better
understanding.
97
If (Speed is Very-Fast) and (Message-Priority is Normal) and (Time is Noon) then (CH1
is not T1)(CH2 is not T8)(CH3 is not T12)(CH4 is T18)(CH5 is not T21)(CH6 is not
T24)(CH7 is not T29)
If (Speed is Very-Slow) and (Message-Priority is Immediate) and (Time is Noon) then
(CH1 is not T1)(CH2 is not T6)(CH3 is not T12)(CH5 is not T21)(CH6 is not T24)(CH8
is T35)
If (Speed is Fast) and (Message-Priority is Ordinary) and (Time is Evening) and (Ch-1-3-
Sensing-Results is not T9) and (Ch-4-6-Sensing-Results is T26) then (CH1 is not
T3)(CH2 is not T9)(CH3 is not T14)(CH4 is not T18)(CH5 is not T23)(CH6 is T26)(CH7
is not T31)(CH8 is not T35)
If (Speed is Slow) and (Message-Priority is Normal) and (Time is Evening) and (Ch-4-6-
Sensing-Results is T26) then (CH1 is not T3)(CH3 is not T14)(CH4 is not T20)(CH5 is
not T23)(CH6 is T26)(CH7 is not T31)(CH8 is not T35)
If (Speed is Medium) and (Message-Priority is Immediate) and (Time is Night) and (Ch-
7-8-Sensing-Results is T35) then (CH1 is not T3)(CH2 is not T9)(CH3 is not T14)(CH4
is not T20)(CH5 is not T23)(CH7 is not T33)(CH8 is T35)
A widely used implication in fuzzy systems and controls is Mamdani Implications. The argument
in the favor of Mamdani Implications is that it considers fuzzy IF-THEN rules as local. For
global implications other type of implications are available.
6.2 SIMULATION AND RESULTS
Simulations have been carried out using MATLAB version R2012a, Fuzzy Logic Toolbox.
Mamdani Implications have been used to model the Channel Selection Fuzzy model. Input,
output linguistic variables used and membership functions modeled are described in detail in
System Model section above. Three hundred and twenty IF-THEN rules have been used to
model the system. Results have been obtained checking the utility of the proposed system by
98
looking at the utility of the channels in speed-time domain and in message priority-time domain.
These utility results actually show the allocation of that particular channel. Figures 6-13 to 6-20
show the utility of channel-1 to 8 in speed-time domain and Figures 6-21 to 6-28 show the utility
of the same channels in message priority-time domain.
6.2.1 UTILITY SPEED VS TIME
Figure 6-13: Speed Vs Time for Channel 1
Figure 6-14: Speed Vs Time for Channel 2
99
Figure 6-15: Speed Vs Time for Channel 3
Figure 6-16: Speed Vs Time for Channel 4
Figure 6-17: Speed Vs Time for Channel 5
100
Figure 6-18: Speed Vs Time for Channel 6
Figure 6-19: Speed Vs Time for Channel 7
Figure 6-20: Speed Vs Time for Channel 8
From the above figures it is evident that channel-1 is well utilized for very fast and very slow
speeds. Utility of channel-2 is better in early hours of the day. Channel-3 is best utilized for all
the speeds during its availability timings. Channel-4 is better for slow and fast speeds in the early
hours of the day. Channel-5 is better for all speeds during morning hours. Channel-6 is better for
very slow and very fast speeds in night hours. Channel-7 is better in early hours for very slow,
101
slow, fast and very fast speeds. Channel-8 is better during available hours for very slow, very
fast and medium speeds but its utility for slow and fast speeds drops during noon and after-noon
hours. Utility results indicate the allocation of these channels for different speed and time.
6.2.2 UTILITY MESSAGE PRIORITY VS TIME
Figure 6-21: Message Priority Vs Time for Channel 1
Figure 6-22: Message Priority Vs Time for Channel 2
102
Figure 6-23: Message Priority Vs Time for Channel 3
Figure 6-24:Message Priority Vs Time for Channel 4
Figure 6-25: Message Priority Vs Time for Channel 5
103
Figure 6-26: Message Priority Vs Time for Channel 6
Figure 6-27: Message Priority Vs Time for Channel 7
Figure 6-28: Message Priority Vs Time for Channel 8
From the above figures it is clear that channel-1 is well utilized during morning hours for all
priority messages. Channel-2 is better during first half of the day for all priorities except
immediate. Channel-3 is better in early hours for all priorities and in late hours for immediate
and urgent messages. Channel-4 is better for lower priority messages and channel-5 for all
104
priorities during morning hours. Channel-6 is better for higher priority messages during late
hours. Channel-7 is better during morning hours for all sort of messages and channel-8 very well
utilized especially for immediate and urgent messages. These results indicate the allocations for
different message priority and time.
6.3 SUMMARY
In this chapter we have proposed a fuzzy logic based cognitive radio channel allocation scheme
for the vehicles. Channel sensing model used is based on as proposed in chapter 5. These
channels sensing results along with speed, time in hours of the day and message priority are
taken as inputs and allocation possibilities are obtained. Better allocation possibilities are
achieved by using this model as shown by results.
105
Chapter 7
CONCLUSION AND FUTURE
PROSPECTS
7.1 CONCLUSION
As the vehicles move with high speed on highways they switch from one RSU to another very
quickly. Hence it will be difficult for bidirectional Road Side Units to bear the load of the traffic
nodes on both sides. Furthermore as on highways speed of most of the vehicles remains in the
similar range with the minor difference of 5% to 10 % hence the formation of groups for
management and security purposes is easier for the vehicles moving in one direction. But with
bidirectional RSUs it will be difficult to form the groups/clusters for vehicles moving in both
directions. With the introduction of directional Road Side Units and from simulation results it is
observed that average throughput is increased considerably when the traffic on the road increases.
Also the end to end delay is reduced by a better margin with the increase in traffic. Packet
delivery ratio is also increased. Also the groups/cluster formation will be easier and manageable.
As discussed in the introduction section many different techniques have been proposed for
spectrum sensing of cognitive radio networks. Since primary users are licensed users and are
authorized to utilize that network hence their communication shall not be disturbed in any case
by the unlicensed secondary users. So the technique and algorithm used shall correctly identify
the presence or absence of primary user and it shall force the secondary users to leave the
primary network channel in case primary user comes back again. As proposed in section 4.1 and
shown by the simulation results our technique has a better probability of correct detection as
compared to standalone and cooperative sensing techniques further our technique has almost
106
eliminated misdetection resulting in negligible interference being caused with the primary user
signals. Since vehicular communication is needed to be error free and quick so our technique
provides the opportunity to utilize CR networks in case the resources on the DSRC spectrum are
overburdened.
In section 4.2, we introduced a framework for the impartial comparison of CR-VANET sensing
and allocation schemes. We described the framework and all the models that could be used to fill
in various parts of the algorithm. We then described the schemes we used to produce a complete
simulation. We then ran simulations in line with the framework for three different channel
sensing and coordination schemes. The proposed coordination scheme with three coordinators
per cluster showed marked benefits over independent and cooperative sensing schemes.
Day by day the vehicular traffic is increasing on the roads; so it is the need of the hour to search
for the vacant frequency slots in CR-spectrum and then utilize until these are occupied again by
the primary users. Our proposed algorithm discussed in section 4.3 is based on the historic data
of sensing the CR-spectrum so it provides a clear picture of spatio-temporal and frequency slots
for its future activity. It observes the primary user’s activity and timing related to acquiring and
leaving the channel for a particular distance slot and frequency. Based on these computations and
observations a list is prepared giving priority to the channels which are most likely to be
available for the duration of the intended transmission. The simulation results obviously show
the success of our algorithm.
The model proposed in chapter 5 provides a clear picture while allocation of CR channels to the
moving vehicles depending upon speed, message priority, hours of the day, and sensing results.
Considering the utility of channels, priority in allocation can be achieved easily. There may be
more than one channels of one category available at a time with same instants availability or
different. This model can be utilized well for all sorts of channels depending on sensing results.
107
7.2 FUTURE PROSPECTS
In future it is proposed that opposite side RSUs may be integrated into one RSU to reduce the
cost of the hardware to be installed. Further work may also be carried out to check formation of
groups for localized traffic and security issues. Further work may also be carried out on
formation of groups categorized in buses, trams, taxies, cars, vans etc. for urban traffic
environments and improvements in security. In urban areas Road Central Units (RCUs) may be
installed at crossings and simulations may be performed. Vehicular traffic in urban areas is a
major concern for researchers. Hence a concrete structure is needed for growing need of
VANETs in urban areas.
Day by day traffic in urban areas as well as on highways is increasing leaving researchers to
think about providing sufficient resources to cope with the increasing vehicle to vehicle and
vehicle to infrastructure communication. Our technique may help to distribute the burden of
increased communication and avoid unwanted incidents.
As the concept of the application of CRs is relatively new to VANETs, there is a lot of room for
improvement. Further work needs to be done on the application of effective spectrum
management and allocation in VANETs. This is particularly true for CR-VANETs, as there is a
possibility of PU involvement. Furthermore, work needs to be done on the possibility of using
bandwidth-efficient techniques. As power generation isn’t a huge issue, work needs to be done
on immersive entertainment systems while inside a vehicle, while improving the safety. Vehicle
to vehicle communication in real-time may open up the door for maintaining a safe distance from
other vehicles, with facilities for auto-braking in the case of risks. Other similar features for lane
guidance and optimal speed may also help improve the utility of vehicles in general.
We are also working to add channel bandwidth, fading, end to end delay, packet delivery ratio,
packet drop ratio and distance segments as further input variables in the model proposed in
chapter 6 to enhance the capability of the proposed system. This picture will provide a realistic
environment in vehicular networks and will help to build a stable system for VANETs.
108
Appendix A
TOOLS USED
During my research work I have used following different tools for the simulations of my
proposed models.
1 NS2
2 C#
Some discussion on these tools is given below.
NS2
Network Simulator version 2 is open source event driven simulation tool specially designed for
simulating different scenarios in computer communication networks. NS2 was started in 1989
and since then various researchers from education, industry, and government sectors have used it
for their valuable proposals. It is continuously under enhancements and contains modules for
application, transport, network and data link layer simulations. Various network performance
parameters can be integrated in NS2 in order to measure the objects of researchers using easy to
use scripting language. There may be some modules needed by researchers which are beyond the
scope of NS2 but such modules can be integrated in NS2 by programmers using back end object
oriented language. For this purpose profound knowledge of NS2 architecture is required. Some
of the widely used modules are Network AniMator (NAM) for visualizing animations and
Xgraph for generating graphs of the results. As the researchers are developing new modules for
their specific purposes; NS2 developers are integrating new modules enhancing its performance.
Due to its flexibility and modular built up NS2 has gained too much popularity. University of
109
California and Cornell University have played an important role in the development of real NS2
simulator. Virtual InterNetwork Testbed (VINT) project with the support of Defense Advanced
Research Projects Agency (DARPA) was main cause for the development of Network Simulator
[119].
NS2 consists of mainly two key languages. C++ and OTcl (Object Tool Command Language).
OTcl sets up simulation by the assembly and configuration of the objects whereas C++ works as
the back end. Both languages are linked together by TclCL (TCL with classes). The variables in
OTcl objects are called handles which are mapped to C++ objects. Handle acts as a front end for
interaction with users and other OTcl objects. There is one to one correspondence between C++
and OTcl hierarchies. C++ has compiled and OTcl has interpreted Hierarchies. In compiled
hierarchy member or class variables and member or class functions are referred to as variables
and functions respectively. In interpreted hierarchy these are referred to as instance variables and
instant procedures [119].
NS2 is supported by FreeBDS, Linux, Solaris, Windows and Mac. NS2 supports Wired as well
as Wireless networks. In wired networks the features supported are Routing, Transportation,
Traffic sources, Queuing, disciplines, and QoS. In Wireless netowrks Ad hoc routing, mobile IP,
and sensor-MAC are supported. NS2 creates trace file for graphs and data manipulations and for
visualizing NAM file is generated. NS2 supports applications and traffic models such as FTP,
CBR, Telnet, Web and real audio. It supports unicast transport protocols such as TCP and UDP
and multicast such as SRM. It supports RED and Drop-tail queuing protocols [120] [121].
C#
O VISUAL STUDIO
Visual Studio developed by Microsoft is an Integrated Development Environment (IDE) which
can be used to develop console applications and graphic user interface (GUI) applications. It can
also be used to develop window form applications, web pages, web applications in both managed
110
as well as native code for all supported platforms of Microsoft windows, window CE, .NET
framework, windows phone, Microsoft silvernight and .NET compact framework [122].
O C# HISTORY AND FEATURES
C# pronounced as ‘see sharp’ was developed by Microsoft in the year 2000. Its main designer
and lead architect was Anders Hejlsberg. He wanted to design a new programming language
which could create class libraries in .NET framework. It was understood that C# is imitation of
Java. However in spite of strong object oriented approaches both have distinct features. C# has
support for automatic garbage collection; array bound checking, and software engineering
principles. It has software components which can be integrated for development in distributed
environment. It has source code similarity features as in C or C++.
C# understands everything as object. It reflects directly the CLI (Common Language
Infrastructure). Its intrinsic types are similar to value types in CLI framework. It is more type
safe than C++. It strongly supports operator overloading like C++ and unlike Java. Due to
garbage collection property memory leak problems are avoided. C# has try…finally construct in
addition to try…catch construct which guarantees (in case an exception occurs or not) execution
of code in finally block. C# unlike Java does not has checked exceptions. In order to simplify
architectural requirements throughout CLI, it does not support multiple inheritance. Structs are
different in C# than classes. These are not reference types but value types so they are passed by
value. These are not derived from any base class so they are sealed. Unlike C++ inheritance is
always public in C#. A class in C# can only be derived from one base class. In case base class is
not specified then class will automatically be derived from System Object. C# permits delegates,
events and properties in addition to C++ class members such as variables, functions,
constructors, destructors and operator overloads. In C# two more access modifiers are
introduced. Those are a) Internal, b) Protected Internal [123]
111
O .NET FRAMEWORK
This framework primarily runs on Microsoft Windows. It has large library and supports many
programming languages. This feature supports language inter-operability. Its library supports all
the programming languages supported by .NET framework. Programmes written with the
support of .NET framework can execute in a unique software environment called CLR (Common
Language Runtime). Services provided by CLR are security, exception handling and memory
management. So .NET framework is the combination of CLR an application virtual machine and
class libraries [122].
112
BIBLIOGRAPHY
[1] WHO, "Road Traffic Injuries," World Health Organization, 2015. [Online]. Available:
http://www.who.int/mediacentre/factsheets/fs358/en/.
[2] "Global Status Report on Road Safety," World Health Organization, 2013.
[3] "Fata Analysis Reporting System," National Highway Traffic Safety Administration,
[Online]. Available: http://www-fars.nhtsa.dot.gov.
[4] S. H. Abbassi, I. M. Qureshi, O. Khalid and H. Abbasi, "Basic Structural Change in
Vehicular Adhoc Networks," Journal of Networking Technology, vol. 4, no. 3, pp. 149-
154, Dec.2013.
[5] "Sensor and Vehicle Networks," 24 February 2012. [Online]. Available:
http://sensorandvehiclenetworks.blogspot.com/2012/02/primera-toma-de-contacto-
hsvn.html.
[6] L. Figueiredo, I. Jesus, J. Machado, J. Ferreira and J. Martins de Carvalho, "Towards the
development of intelligent transportation systems," in IEEE Intelligent Transportation
Systems, Hong Kong, 2001.
[7] A.-S. K. Pathan, "Security of Self-Organizing Networks," CRC Press, Taylor and Francis
Group, 2011.
[8] S. Haykin, "Fundamental Issues in Cognitive Radio," in Cognitive Wireless
113
Communication Networks, Springer US, 2007, pp. 1-43.
[9] P. D. Barjoei, G. Dadashzadeh, F. Razzazi and S. M. Razavizadeh, "Power and Time Slot
Allocation in Cognitive Relay Networks Using Particle Swarm Optimization," The
Scientific World Journal, vol. 2013, no. 2013, 2013.
[10] L. Giupponi and C. Ibars, "Distributed cooperation in cognitive radio networks: Overlay
versus underlay paradigm," in 69th IEEE Vehicular Technology Conference, Barcelona,
2009.
[11] K. B. Letaief and W. Zhang, "Cooperative Spectrum Sensing, Cognitive Wireless
Communication Networks," Springer Science + Business Media LLC., pp. 115-138, 2007.
[12] J. Chen, X. Zheng and M. Zhu, "Cooperative Spectrum Sensing Algorithm based on the
Relay," Procedia Engineering, vol. 15, no. 2011, pp. 2349-2353, 2011.
[13] X. Y. Wang and P.-H. Ho, "A Novel Sensing Coordination Framework for CR-VANETs,"
IEEE Transaction on Vehicular Technology, vol. 59, no. 4, pp. 1936-1948, 2010.
[14] P. Mankar, S. S. Pathak and R. V. Rajakumar, "A Cooperative Secondary User
Localization based Primary User Localization Method for Cognitive Radio Networks," in
National Conference on Communications, Kharagpur, 2012.
[15] J. Mitola and G. Q. Maguire, "Cognitive radio: Making software radios more personal,"
IEEE Personal Communications, vol. 6, pp. 13-18, 1999.
[16] S. Haykin, "Cognitive radio: brain-empowered wireless communications," IEEE J. Sel.
Areas Commun, vol. 23, pp. 201-220, 2005.
[17] W. Zhang, R. K. Mallik and K. Letaief, "Cooperative Spectrum Sensing Optimization in
Cognitive Radio Networks," in IEEE International Conference on Communications, 2008.
114
[18] S. Tauqeer and B. H. Halabi, "Interference-Aware Routing in Cognitive Radio Ad-hoc
Networks: Chanlenges and Research Directions," MAGNT Research Report, vol. 2, no. 3,
2014.
[19] M. Hata, "Empirical Formula for Propagation Loss in Land Mobile Radio Service," IEEE
Transactions Vehicular Technology, vol. 29, no. 3, pp. 317-325, August 1980.
[20] B. Sklar, "Rayleigh Fading Channels in Mobile Digital Communication Systems Part I:
Characterization," IEEE Communications Magazine, vol. 35, no. 7, pp. 90-100, July 1997.
[21] H. Abbasi, S. H. Abbassi and I. M. Qureshi, "A framework for the simulation of CR-
VANET channel sensing, coordination and allocation," Journal of Chinese Institute of
Engineers, vol. Submitted, 2014.
[22] S. H. Abbassi, I. M. Qureshi, H. Abbasi and B. R. Alyaie, "History Based Spectrum
Sensing in CR-VANETs," vol. 2015:163, 2015.
[23] U. Universitet, "Introduction to Mobile Ad hoc Networks (MANETs)," [Online].
Available: http://user.it.uu.se/~erikn/files/DK2-adhoc.pdf.
[24] M. Zitterbart and K. Weniger, "Mobile ad hoc networks (MANET)," TeleMatics, 2005.
[25] C. Imrich, C. Marco and J.-N. L. Jennifer, "Mobile ad hoc networking: imperatives and
challenges," Adhoc Networks, vol. 1, no. 2003, pp. 13-64, 2003.
[26] A. Tønnesen, "Mobile Ad-Hoc Networks," [Online]. Available: www.olsr.org/docs/wos3-
olsr.pdf.
[27] J. Hoebeke, I. Moerman, B. Dhoedt and P. Demeester, "An Overview of Mobile Ad Hoc
Networks: Applications and Challenges," Journal of the Communications Network, vol. 3,
no. July 2004, pp. 60-66, 2004.
115
[28] Aarti and S. S. Tyagi, "Study of MANET: Characteristics, Challenges, Application and
Security Attacks," International Journal of Advanced Research in Computer Science and
Software Engineering, vol. 3, no. 5, pp. 252-257, 2013.
[29] R. K. Sakib and B. Reza, "Security Issues in VANET," BRAC University, Dhaka,
Bangladesh, 2010.
[30] T. Taleb and E. Sakhaee, "A Stable Routing Protocol to Support ITS Services in VANET
Networks," IEEE Transactions on Vehicular Technology, vol. 56, no. 6, pp. 3337-3347,
2007.
[31] S. Wasiq, W. Arshad, N. Javaid and A. Bibi, "Performance Evaluation of DSDV, OLSR
and DYMO using 802.11 and 802.11p MAC-Protocols," in IEEE 14th International
Multitopic Conference (INMIC), Karachi, 2011.
[32] S. Sagar, J. Saqib, A. Bibi and N. Javaid, "Evaluating and Comparing the Performance of
DYMO and OLSR in MANETs and in VANETs," in IEEE 14th International Multitopic
Conference (INMIC), Karachi, 2011.
[33] P. I. Offor, "Vehicle Ad Hoc Network (VANET): Safety Benefits and Security
Challenges," Social Science Research Network, Nova Southeastern University, 3 Dec.
2012.
[34] D. V. Jamthe and S. S. Dorle, "Collision Avoidance in Ivan to maintain Inter-Vehicular
Distance on Highways," International Journal of Engineering Science and Advanced
Technology, vol. 2, no. 3, pp. 672-678, 2012.
[35] J. Antonio, G. Ibanez, C. Flores-Cortes and J. M. Ramirez-Alcaraz, "Intelligent Signaling
for Prevention of Intersection Collision in Urban Zones," in 8th International Conference
on Intelligent Environment, 2012.
116
[36] Irshadullah and S. u. Rehman, "Analysis of Black Hole Attack on MANETs Using
Different MANET Routing Protocols," School of Computing, Blekinge Institute of
Technology, Sweden, 2010.
[37] Cruz-Cunha and M. Manuela, Handbook of Research on Mobility and Computing:
Evolving Technologies and Ubiquitous Impacts, IGI Global, 2011.
[38] G. Xu, "Location cloaking for location privacy protection and location safety protection,"
Iowa State University, 2010.
[39] M. Rezvani, A. Ignjatovic, E. Bertino and S. Jha, "Secure Data Aggregation Technique for
Wireless Sensor Networks in the Presence of Collusion Attacks," IEEE Transactions on
Dependable and Secure Computing, vol. 12, no. 1, pp. 98-110, 2014.
[40] "DSRC: The Future of Safer Driving," Intelligent Transportation Systems, [Online].
Available: http://www.its.dot.gov/factsheets/dsrc_factsheet.htm.
[41] IEEE, "P802.11p/D5.0, IEEE 802.11 Amendment 7: Wireless Access in Vehicular
Environments," Nov. 2008.
[42] J. B. Kenney, "Dedicated Short-Range Communications (DSRC) Standards in the United
States," Proceedings of the IEEE, vol. 99, no. 7, pp. 1162-1182, July 2011.
[43] E. R. Nyanhete, M. Mzyece and K. Djouani, "Operation and Performance of Cognitive
Radios in Vehicular Environments," Tshwane University of Technology, Aug. 2011.
[44] IEEE, "Std 1609 family, IEEE Trial-Use Standard for Wireless Access in Vehicular
Environments(WAVE)," Nov. 2006.
[45] D. Jiang and D. Luca, "IEEE 802.11p: Towards an International Standard for Wireless
Access in Vehicular Environments," in IEEE Vehicular Technology Conference,
117
Singapore, 2008.
[46] K. Bilstrup, E. Uhlemann and E. G. Ström , "Medium access control in vehicular networks
based on the upcoming IEEE 802.11p standard," in 15th World Congress on Intelligent
Transport Systems, 2008.
[47] J. Kakarla, S. S. Sathya and B. G. Laxmi, "A Survey on Routing Protocols and its Issues in
VANET," International Journal of Computer Applications, vol. 28, no. 4, Aug. 2011.
[48] A. Reyes, C. Barrado and M. López, "Vehicle density in VANET applications," Journal of
Ambient Intelligence and Smart Environments, vol. 6, no. 4, pp. 469-481, 2014.
[49] W. Alasmary and W. Zhuang, "Mobility impact in IEEE 802.11p infrastructureless
vehicular networks," Ad Hoc Networks, vol. 10, no. 2, pp. 222-230, 2012.
[50] N. Loulloudes, G. Pallis and M. D. Dika, "The Dynamics of Vehicular Networks in Urban
Environments," Cornell University Library, 2011.
[51] Y. Toor, P. Muhlethaler and A. Laouiti, "Vehicle ad hoc networks: applications and related
technical issues.," IEEE Commun. Surv. Tutor., vol. 10, no. 3, pp. 74-88, 2008.
[52] H. Hartenstein and K. Laberteaux, "A tutorial survey on vehicular ad hoc networks," IEEE
Communication Magzine, vol. 46, no. 6, pp. 164-171, 2008.
[53] S. Kukliński and G. Wolny , "CARAVAN: Context-AwaRe Architecture for VANET," in
Mobile Ad-Hoc Networks: Applications, Rijeka, Croatia, Intech, 2011.
[54] K. Sha, S. Wang and W. Shi, "RD4: Role-Differentiated Cooperative Deceptive Data
Detection and Filtering in VANETs," IEEE Transactions on Vehicular Technology, vol.
59, no. 3, pp. 1183-1190, 2010.
118
[55] S. Mahajan and A. Jindal, "Security and Privacy in VANET to reduce Authentication
Overhead for Rapid Roaming Networks," International Journal of Computer Applications,
vol. 1, no. 20, pp. 17-21, 2010.
[56] A. Studer, E. Shi, F. Bai and A. Perrig, "TACKing Together Efficient Authentication,
Revocation, and Privacy in VANETs," in 6th Annual IEEE Communications Society
Conference on Sensor, Mesh and Ad Hoc Communications and Networks, Rome, 2009.
[57] A. Frederik, F. Andreas, W. Dirk and Z. Ke, "Cross-layer Privacy Enhancement and Non-
repudiation in Vehicular Communication," in ITG-GI Conference on Communication in
Distributed Systems, Bern, Switzerlad, 2007.
[58] N. Balon and J. Guo, "Increasing broadcast reliability in vehicular ad hoc networks," in
3rd international workshop on Vehicular ad hoc networks, Los Angeles, 2006.
[59] K. A. Hafeez, L. Zhao, B. Ma and J. W. Mark, "Performance Analysis and Enhancement
of the DSRC for VANET’s Safety Applications," IEEE Transactions on Vehicular
Technology, vol. 62, no. 7, pp. 3069-3083, 2013.
[60] B. Hughes, R. Meier, R. Cunningham and V. Cahill, "Towards real-time middleware for
vehicular ad hoc networks," in First ACM International Workshop on Vehicular Ad Hoc
Networks, Philadelphia, 2004.
[61] M. Torrent-Moreno, J. Mittag, P. Santi and H. Hartenstein, "Vehicle-to-vehicle
communication: Fair transmit power control for safety-critical information," IEEE
Transactions on Vehicular Technology, vol. 58, no. 7, pp. 3684-3703, 2009.
[62] E. M. Vaneenennaam, W. Kleinwolterink, G. Karagiannis and G. J. Heijenk, "Exploring
the solution space of beaconing in VANETs," in 1st IEEE Vehicular Networking
Conference, Tokyo, Japan, 2009.
119
[63] K. Bilstrup, E. Uhlemann, E. G. Strom and U. Bilstrup, "Evaluation of the IEEE 802.11p
MAC method for vehicle-to-vehicle communication," in IEEE 68th Vehicular Technology
Conference, Calgary, BC , 2008.
[64] Z. Wang and M. Hassan, "How much of DSRC is available for non-safety use?," in 5th
ACM Int. Workshop on Vehicular Inter-Networking, San Francisco, California, USA,
2008.
[65] M. X. and X. B. Chen, "Delay and broadcast reception rates of highway safety applications
in vehicular ad hoc networks," in Mobile Networking Vehicular Environment, Anchorage,
AK, 2007.
[66] S. Eichler, "Performance evaluation of the IEEE 802.11p WAVE communication
standard," in IEEE Vehicular Technology Conference, Baltimore, MD, 2007.
[67] W. Ejaz, N. U. Hasan and H. S. Kim, "Stand-Alone Architecture for Spectrum Sensing in
Cognitive Radio Networks for Smart Buildings," in 2nd International Conference on
Networking and Information Technology, 2011.
[68] X. Y. Wang, A. Wang and P. H. Ho, "Prioritized spectrum sensing in cognitive radio based
on spatiotemporal statistical fusion," in IEEE Wireless Communications and Networking
Conference, Budapest, 2009.
[69] H. T. Cheng, H. Shan and W. Zhuang, "Infotainment and road safety service support in
vehicular networking: From a communication perspective," Journal of Mechanical
Systems and Signal Processing, vol. 25, no. 6, pp. 2020-2038, 2011.
[70] W. Chen, "Effect of adjacent channel interference on uplink capacity in a macrocell‐
microcell‐overlapping WCDMA system," Journal of the Chinese Institute of Engineers,
vol. 31, no. 1, pp. 157-161, 2008.
120
[71] K. Popper, The Logic of Scientific Discovery, Routledge, 1959.
[72] K. Fawaz and A. Ghandour, "Improving Reliability of Safety Applications in Vehicle Ad
hoc Networks through the Implementation of a Cognitive Network," in 17th International
Conference on Telecommunications, Doha, 2010.
[73] M. Di Felice, K. R. Chowdhury and L. Bononi, "Analyzing the Potential of Cooperative
Cognitive Radio Technology on Inter-Vehicle Communication," in Proceedings of the
2010 IFIP Wireless Days, Venice, 20-22 Oct. 2010.
[74] W. Kim and S. Y. Oh, "CoRoute: A New Cognitive Anypath Vehicular Routing Protocol,"
in Proceedings of the 2011 7th International Wireless Communications and Mobile
Computing Conference, Istanbul, 4-8 July 2011.
[75] L. Husheng and D. K. Irick, "Collaborative Spectrum sensing in Cognitive Radio
Vehicular Ad hoc Networks: Belief Propagation on Highways," in Proceedings of the 2010
71st IEEE Vehicular Technology Conference, Taipei, 16-19 May 2010.
[76] D. B. Rawat, Y. Zhao and G. Yan, "CRAVE: Cognitive Radio Enabled Vehicular
Communications in Heterogeneous Networks," in Proceedings of the 2013 IEEE Radio
and Wireless Symposium, Austin, 20-23 Jan. 2013.
[77] I. Akyildiz, B. Lo and R. Balakrishnan, "Cooperative spectrum sensing in cognitive radio
networks: a survey," Physics Communications, vol. 4, no. 1, pp. 40-62, 2011.
[78] A. Min and K. Shin, "Impact of mobility on spectrum sensing in cognitive radio
networks," in ACMWorkshop on Cognitive Radio Networks, Beiging, China, 2009.
[79] D. Borota, G. Ivkovic, R. Vuyyuru, O. Altintas, I. Seskar and P. Spasojevic, "On the delay
to reliably detect channel availability in cooperative vehicular environments," in IEEE
121
73rd Vehicular Technology Conference, Yokohama, 2011.
[80] R. Doost-Mohammady and K. Chowdhury, "Design of spectrum database assisted
cognitive radio vehicular networks," in 7th International ICST Conference on Cognitive
Radio Oriented Wireless Networks and Communications, Stockholm, 2012.
[81] D. Niyato, E. Hossain and P. Wang, "Optimal channel access management with QoS
support for cognitive vehicular networks," IEEE Transactions on Mobile Computing, vol.
10, no. 4, pp. 573-591, 2011.
[82] I. H. Brahmi, S. Djahel and Y. Ghamri-Doudane, "A Hidden Markov Model based Scheme
for Efficient and Fast Dissemination of Safety Messages in VANETs," in Proceedings of
the 2012 IEEE Global Communication Conference, California, 3-7 Dec. 2012.
[83] A. R. Biswas, T. C. Aysal, S. Kandeepan, D. Kliazovich and R. Piesiewicz, "Cooperative
Shared Spectrum Sensing for Dynamic Cognitive Radio Networks," in Proceedings of the
2009 IEEE International Conference on Communications, Dresden, 14-18 June 2009.
[84] M. Di Felice and K. R. Chowdhury, "Cooperative Spectrum Management in Cognitive
Vehicular Ad Hoc Networks," in Proceedings of the 2011 IEEE Vehicular Networking
Conference, Amsterdam, 14-16 Nov. 2011.
[85] H. Rasheed and N. Rajatheva, "Spectrum Sensing for Cognitive Vehicular Networks over
Composite Fading," International Journal of Vehicular Technology, vol. 2011, pp. 1-9,
2011.
[86] F. Riaz, I. Shafi, S. F. Hasan, W. Younas and Y. Mehmood, "Vehicle-to-Vehicle
Communication Enhanced by Cognitive Approach and Multi-Radio Technologies," in
Proceedings of the 2012 IEEE International Conference on Emerging Technologies,
Islamabad, 8-9 Oct. 2012.
122
[87] G. Ganesan and L. Ye, "Cooperative Spectrum Sensing in Cognitive Radio, Part I: Two
User Networks," IEEE Transactions on Wireless Communications, vol. 6, no. 6, pp. 2204-
2213, 2006.
[88] S. Chen, "Vehicular Dynamic Spectrum Access: Using Cognitive Radio for Automobile
Networks," Worcester Polytechnic Institute, 2012.
[89] K. Komathy, "Application of Cognitive Radio in VANET," Anna University, India.
[90] I. Kostic, "Composite Phase PDF in Gamma Shadowed Nakagami Fading Channel,"
Wireless Personal Communications, vol. 41, no. 4, p. 465, 2007.
[91] C. Press, Signal Processing for Mobile Communications, Taylor & Francis Group, 2004.
[92] M. Di Felice, A. J. Ghandour, H. A. Hartail and L. Bononi, "Integrating Spectrum
Database and Cooperative Sensing for Cognitive Vehicular Networks," in 78th IEEE
Vehicular Technology Conference, Las Vegas, Nevada, USA, 2-5 Sept. 2013.
[93] Y. Zeng and Y.-C. Liang, "Spectrum-Sensing Algorithms for Cognitive Radio Based on
Statistical Covariances," IEEE Transactions on Vehicular Technology, vol. 58, no. 4, pp.
1804-1815, May 2009.
[94] A. Sonnenschein and P. M. Fishman, "Radiometric Detection of Spread-Spectrum Signals
in Noise of Uncertainty Power," IEEE Trans. Aerosp. Electron. Syst., vol. 28, no. 3, pp.
654-660, July 1992.
[95] R. Tandra and A. Sahai, "Fundamental Limits on Detection in Low SNR Under Noise
Uncertainty," in International Conference on Wireless Networks, Communications and
Mobile Computing, Wuhan, 13-16 June 2005.
[96] D. Cabric, A. Tkachenko and R. W. Brodersen, "Spectrum Sensing Measurements of Pilot,
123
Energy, and Collaborative Detection," in IEEE Military Communications Conference.
MILCOM 2006, Washington DC, 23-25 Oct. 2006.
[97] S. Chen, R. Vuyyuru, O. Altintas and A. Wyglinski, "On optimizing vehicular dynamic
spectrum access networks: automation and learning in mobile wireless environments," in
IEEE Vehicular Networking Conference, Amsterdam, 2011.
[98] M. Khabbaz, C. Assi and A. Ghrayeb, "Modeling and analysis of DSA-based vehicle-to-
infrastructure communication systems," IEEE Transactions on Intelligent Transport
Systems, vol. 14, no. 3, pp. 1186-1196, 2013.
[99] I. F. Akyildiz, W. Y. Lee, M. C. Vuran and S. Mohanty, "A Survey on Spectrum
Management in Cognitive Radio Networks," IEEE Communications Magazine, vol. 46, no.
4, pp. 40-48, April 2008.
[100] C. Yang, Y. Fu, Y. Zhang, S. Xie and R. Yu, "Energy-efficient hybrid spectrum access
scheme in cognitive vehicular ad hoc networks," IEEE Communications Letters, vol. 32,
no. 18, pp. 1983-1997, 2009.
[101] S. Jha, M. Rashid, V. Bhargava and C. Despins, "Medium access control in distributed
cognitive radio networks," IEEE Transactions on Wireless Communications, vol. 18, no. 4,
pp. 41-45, 2011.
[102] S. Jha, U. Phuyal, M. Rashid and V. Bhargava, "Design of OMC-MAC: an opportunistic
multi-channel MAC with QoS provisioning for distributed cognitive radio networks,"
IEEE Transactions on Wireless Communications, vol. 10, no. 10, pp. 3414-3425, 2011.
[103] S. D. Kaehler, "Fuzzy Logic Tutorial," Encoder. [Online].
[104] R. Mohadeseh, T. Hamid and R. Anise, "Reconfiguration Optimization for Loss Reduction
in Distribution Networks Using Hybrid PSO Algorithm and Fuzzy Logic," MAGNT
124
Research Report, vol. 2, no. 5, pp. 903-911, 2014.
[105] A. J. Ghandour, K. Fawaz, H. Artail, M. D. Felice and L. Bononi, "Improving vehicular
safety message delivery through the implementation of a cognitive vehicular network,"
Journal of Adhoc Networks, vol. 11, no. 8, pp. 2408-2422, Nov. 2013.
[106] P. P. Bhattacharya, R. Khandelwal, R. Gera and A. Agarwal, "Smart Radio Spectrum
Management for Cognitive Radio," International Journal of Distributed and Parallel
Systems, vol. 2, no. 4, pp. 12-24, july 2011.
[107] K. R, D. N. Kumar C. and P. A., "Fuzzy Logic System for Opportunistic Spectrum Access
using Cognitive Radios," International Journal of Computer Science, vol. 10, no. 1, pp.
703-709, Jan. 2013.
[108] C. Wu, S. Ohzahata and T. Kato, "VANET Broadcast Protocol based on Fuzzy Logic and
Lightweight Retransmission Mechanism," IEEE Transactions on Communications, Vols.
E95-B, no. 2, pp. 415-425, 2012.
[109] Y. Yong, P. Alexandra and P. Adrian, "On Fuzzy Logic Based Channel Selection in
Cognitive Radio Networks," in Handbook of Research on Software-Defined and Cognitive
Radio Technologies for Dynamic Spectrum Management, University of North Dakota,
USA, Information Science Reference, 2014, pp. 255-274.
[110] S. Khan and N. Chauhan, "Region Authority (RA) Collaborated Certificate Organization
and Management in VANET," in World Congress in Computer Science Computer
Engineering and Applied Computing, Las Vegas, Nevada, USA, 2014.
[111] L. Mohan, "Security in VANETs," 1 December 2014. [Online]. Available:
http://www.cse.wustl.edu/~jain/cse571-14/index.html .
125
[112] J. SAE, "Dedicated Short Range Communications (DSRC) message set dictionary".
[113] F. F. Digham, M. S. Alouini and M. K. Simon, "On the energy detection of unknown
signals over fading channels," IEEE Transactions on Communication, vol. 55, no. 1, pp.
21-24, 2007.
[114] P. G. Gipps, "A Behavioral Car Following Model for Computer Simulation," Journal of
Transportation Research, vol. 15, no. 2, pp. 105-111, 1981.
[115] S. H. Abbassi, I. M. Qureshi, B. R. Alyaei, H. Abbasi and K. Sultan, "An Efficient
Spectrum Sensing Mechanism for CR-VANETs," Journal of Basic and Applied Scientific
Research, vol. 12, no. 3, pp. 365-378, 2013.
[116] S. H. Abbassi, I. M. Qureshi, H. Abbasi and B. R. Alyaie, "CR Channel Allocation Using
Fuzzy Logic in VANETs," MAGNT Research Report (ISSN. 1444-8939), vol. 3, no. 3, pp.
898-909, 2015.
[117] H. Abbasi, S. H. Abbassi and I. M. Qureshi, "A framework for the simulation of CR-
VANET channel sensing, coordination and allocation," Ad-Hoc and Sensor Wireless
Networks, vol. Submitted, 2014.
[118] S. H. Abbassi, I. M. Qureshi, H. Abbasi and B. R. Alyaie, "History Based Spectrum
Sensing in CR-VANETs," EURASIP Journal on Wireless Communications and
Networking, vol. 2015:163, 2015.
[119] T. Issariyakul and E. Hossain, Introduction to Network Simulator NS2, New York:
Springer Science+Business Media, LLC, 2009.
[120] U. T. Malaysia, "Network Simulator 2," 26 8 2003. [Online]. Available:
http://sps.utm.my/wp-content/uploads/2013/12/NS2-Manual2.pdf.
126
[121] School of E. Engineering and C. Science, "CDA6530: Performance Models of Computers
and Networks, Chapter 10: Introduction to NS2," [Online]. Available:
http://www.cs.ucf.edu/~czou/CDA6530-11/NS2-tutorial.pdf.
[122] H. P. HALVORSEN, Introduction to Visual Studio and C#, Telemark: Telemark
University College, 2014.
[123] S. Dhall and P. Sundararaman, "C# and .NET Framework," [Online]. Available:
http://www.cs.colorado.edu/~kena/classes/5448/f12/presentation-materials/dhall.pdf.
[124] D.-G. Wang, W.-Y. Song and L. Zou, "Continuity of Fuzzy Similarity Inference Method,"
Journal of Multiple-Valued Logic and Soft Computing, vol. 18, no. 5-6, pp. 525-539, 2012.
[125] K. D. Singh, P. Rawat and J.-M. Bonnin, "Cognitive radio for vehicular ad hoc Networks
(CR-VANETs): approaches and challenges," EURASIP Journal on Wireles
Communications and Networking, vol. 1, no. 49, 2014.
[126] S. H. Abbassi, I. M. Qureshi and H. Abbasi, "Performance of Uni-directional Road Side
Units in Vehicular Adhoc Networks," in World Congress on Computer Applications and
Information Systems, Hammamet, 2014.
[127] S. H. Abbassi, I. M. Qureshi, H. Abbasi and B. R. Alyaie, "CR Channel Allocation Using
Fuzzy Logic in VANETs," MAGNT RESEARCH REPORT, vol. 3, no. 3, pp. 898-909,
2015.
[128] S. N. Sivanandam, S. Sumathi and S. N. Deepa, Introduction to Fuzzy Logic using
MATLAB, Heidelberg: Springer, 2007.
[129] S. Pandey and P. Agrawal, "A Survey on Localization Techniques for Wireless Networks,"
Journal of the Chinese Institute of Engineers, vol. 29, no. 7, pp. 1125-1148, 2006.