MPhil Thesis

UNIVERSITY OF GHANA
COLLEGE OF BASIC AND APPLIED SCIENCES
IMPROVING INTELLIGENT TRANSPORTATION SYSTEMS (ITS) THROUGH

ANALYTICAL INVESTIGATION OF MACROSCOPIC TRAFFIC FLOW MODEL
IN VEHICULAR AD HOC NETWORKS (VANETS)
BY
ADAMS ADDISON KOBLA AZAMETI
(10551563)
THIS THESIS IS SUBMITTED TO THE UNIVERSITY OF GHANA, LEGON IN

PARTIAL FULFILLMENT OF THE REQUIREMENT FOR THE AWARD OF
MPHIL IN COMPUTER SCIENCE DEGREE
MARCH 2018
0
Declaration
I wish to declare that this thesis has not been previously accepted in substance for the award of
any degree. I state that this thesis is the result of my own independent investigation or work,
except where otherwise stated. I hereby give consent for my thesis, if accepted, to be available
for photocopying and I understand that any reference to or quotation from my thesis will
receive an acknowledgement.
Signed: ______________________________________________________________
Date: _________________________________________________________________
Adams Addison Kobla Azameti

(10551563)
I hereby certify that I have thoroughly read this thesis and, in my opinion, it is fully adequate in
scope and quality, for the degree of Master of Philosophy in Computer Science.
SUPERVISORS
Dr. Ferdinand Apietu. Katsriku Signed: --------------------------------

(Principal Supervisor)
Date: ----------------------------------
Dr. Jamal-Deen Abdulai Signed: --------------------------------

(Co-Supervisor)
Date: -----------------------------------
i
Abstract
In recent times, wireless communication has witnessed unprecedented growth due mainly to
advances in instrumentation and computer science. Substantial research efforts have been
devoted to exploring the application of Intelligent Transportation Systems (ITS) within the
context of vehicular ad hoc networks (VANETs) to ensure safety and for an efficient and
integrated transport network. Vehicle-to-vehicle (V2V) communication in VANETs poses
many challenges. In sparse as well as in dense networks. V2V communication can result in
intermittent connectivity and broadcast storm problems. Finding solutions to these challenges
in an infrastructure-less vehicular ad hoc network (VANETs) with safety and non-safety
applications is quite complex.
These limitations lead to the analytical investigations of the macroscopic traffic flow model to
understand the effect of these spatio-temporal variations in traffic density while varying the
value of the sensitivity factor for a two-vehicle interaction, λ at different time intervals during
each simulation. The result from this simulation is used to develop Congestion Control Models
in the congested network and Honeycomb Model in a sparsely connected network for improved
performance in VANETs. The variation in the value of λ affects traffic flow in vehicles. The
analytical investigation is implemented in MATLAB in which vehicular density is constant and
the value of λ varies directly from 1 to 10 within the time intervals in sparse and dense network
conditions. The simulation result in dense network indicates vehicular trajectories when λ
increases with time.
ii
The outcomes of this analytical investigation demonstrate that there is no need to increase λ
beyond the thresholds of 12. However, further increases would increase the number of vehicles
(nodes) which will result in a traffic jam. Empirical evidence suggests this range 1≤ λ ≤12 to be
used to design traffic models in dense networks. These results in principle proved that this
Congestion Control Model within the four-lane networks may be considered the most
appropriate model to be used in designing VANETs networks. In the work, the following
design parameters were used, A: 1 ≤ λ ≤ 3; B: 1≤ λ ≤ 4 and C: 1 ≤ λ ≤ 5. The level of services in
A, B, and C, have been used to perform separate simulations in sparse networks. The
simulation results also provide a framework to design sparse networks associated with frequent
disconnections within the stated design parameters of A: 1 ≤ λ ≤ 3; B: 1 ≤ λ ≤ 4 and C: 1 ≤ λ ≤
5. These findings lead to the proposed Honeycomb Models in a sparsely connected network to
improve the challenges in sparse network conditions.
Finally, the result from this analytical investigation provide a scientific framework and further
poses greater challenges to the research communities in designing vehicular traffic flow in
VANETs. This research findings, reveal the possibilities to replicate the attributes of honeybees
in designing vehicular traffic flow in VANETs. This novel approach significantly improves ITS
applications in VANETs V2V communication.
iii
Acknowledgement
This thesis would not have been made possible without the support of my supervisors. I wish to
express an endless heartfelt appreciation to Dr. Ferdinand Apietu Katsriku and Dr. Jamal-Deen
Abdulai for their unwavering support and constructive criticisms made to arrive at the final
stage of this thesis. They have been of constant sources of inspiration, motivation, and
knowledge throughout my studies in the University of Ghana. I would like also to express my
deepest gratitude to Dr. Ferdinand Apietu Katsriku for his support, contributions and guidance
towards the final revision of this thesis. I would like also to express my sincere gratitude to all
my colleagues for their insightful criticism during seminars and presentations.
I am highly indebted to GETFUND for the financial support during my MPhil program, the
University of Ghana, Computer Science Department for the opportunity offered me to make
these scholarly contributions to the body of knowledge.
I wish to thanks all the secretaries of the Computer Science Department: Madam Jemima
Oduro-Donkor, Madam Gladys Boatemaa, Sister Solange Ntimoah and the indefatigable
support staff of the department: Brother Isaac Eduful and Brother Erasmus Achianor, who have
helped in diverse ways to make my studies a memorable one.
Finally, I wish to thanks my deceased parents through whom I came to this world, my beloved
wife Mrs. Priscilla E.A. Azameti and my dear children Abigail Senam Abla Azameti, Audery
Sedem Azameti and Austen Seyram Kwesi Azameti, who endured this long process with
support and love. God bless you all and reward each one of you for your stewardship.
iv
Dedication
I wish to dedicate this thesis to the Almighty God, who is the architect and creator of every
invention and ideas. God has truly helped me with all the necessary resources to come out with
this novel approach to vehicular ad hoc networks (VANETs). This approach in VANETs, can
provide an enhanced Intelligent Transportation Systems (ITS) which will reduce the fatalities
on the road networks.
I wish also to dedicate this achievement to all my children since this will serve as a benchmark
to help them climb the academic ladder with the greatest motivation and confidence.
v
Table of Contents
Declaration……………………………………………….………………………………………i
Abstract………………………………………..……………………..…………………………iii
Acknowledgement………………………………………………………………………………iv
Dedication……………………………………………………………………………………….v
Table of contents......…………………………………………………………………………….x
List of Figures………………………………………………………………………………….xii
List of Tables…………………………………………………………………………………..xiii
List of Abbreviations…………………………………………………………………………..xiv
List of Symbols……………………………………………………………………………..….xv
Chapter 1 Introduction ……………………...………………..……………….................1

1.1 Background of the study……………………………………………………….………...3
1.2 Problem statement………………………………………………………………………..5
1.3 Objectives of the study…………………………………………….……………………..5
1.3.1 General Objective……………………………………………………………………....5
1.3.2 Specific Objectives……………………………………………………………………..6
1.4 Scope of the study………………………………………………………………………..7
1.5 Justification of the proposed approach…………………………………………………..7
1.6 Organization……………………………………………………………………………..8
1.7 Summary of chapters…………………………………………………………………….8
Chapter 2 Literature Review……………………………………………….…..…….....9

2.1 Introduction……………………………………………………………………..............10
2.2 The technology challenges………………………………………………………….......10
2.2.1 Cellular technology (2/2.5/3G/4G/5G) ……………………………………………….10
2.2.2 IEEE 802.11p technology……………………………………………………………..11
2.2.3 Issues of spectrum……………………………………………………………………..11
2.3 VANETs simulation and modelling………………………………………………..…...12
2.3.1 Wireless transmission and multiple access…………………………………………....13
vi
2.4 VANETs characteristics………………………………………………………………...13
2.4.1 One-dimensional………………………………………………………………………14
2.4.2 Mobility pattern……………………………………………………………………......14
2.4.3 Dynamic topology…………………………………………………………………......15
2.4.4 Connectivity…………………………………………………………………………...15
2.5 Intelligent Transportation Systems (ITS)………………………………………………16
2.5.1 The vehicle…………………………………………………………………………….16
2.5.2 The user………………………………………………………………………………..17
2.5.3 The infrastructure……………………………………………………………………...17
2.5.4 The communication systems…………………………………………………………..18
2.6 Research directions in VANETs…………………………………………………….....18
2.6.1 The needed technologies………………………………………………………………18
2.6.1.1 IEEE 802.11…………………………………………………………………………19
2.6.1.2 The Dedicated Short-Range Communication…………………………………….....19
2.6.1.3 Time Division Duplex of UMTS Terrestrial Radio Access…………………………20
2.6.2 MAC layer issues………………………………………………………………….......20
2.6.3 Applications……………………………………………………………………………21
2.6.3.1 Safety applications……………………………………………………………….......21
2.6.3.2 Traffic Monitoring……………………………………………………………….......22
2.6.3.3 Automated highways and cooperative driving………………………………………23
2.6.3.4 Entertainments……………………………………………………………………….23
2.7 An overview of VANETs applications…………………………………………………26
2.8 The vehicular mobility………………………………………………………………….27
2.8.1 The vehicle mobility models and traffic simulations………………………………….28
2.8.1.1 The mobility models………………………………………………………………….29
2.8.1.2 The traffic flow models………………………………………………………………30
2.8.1.3 The vehicle speed…………………………………………………………………….31
2.8.1.4 The collisions………………………………………………………………………...31
vii
2.8.1.5 The road topology……………………………………………………………………32
2.9 Performance analysis of the traffic flow theory………………………………………..33
2.9.1 Traffic flow characteristics……………………………………………………….........36
2.10 Fundamental relationship of traffic flow……………………………………………….39
2.10.1 Traffic flow models…………………………………………………………………..40
2.10.1.1 Macroscopic (continuum) traffic models…………………………………………..42
2.10.1.2 Car-following models…………………………………………………………........46
2.10.1.3 Two-fluid models…………………………………………………………………..48
2.10.1.4 Cellular automata (CA) models…………………………………………………….49
2.11 Phase transition and order parameters...……………………………………………....53
2.12 Traffic simulation models………………………………………………………….....54
2.12.1 The microscopic model……………………………………………………………...54
2.12.2 The mesoscopic model…………………………………………………………........54
2.12.3 The macroscopic model……………………………………………………………...55
2.13 Summary and conclusions………………………………………………………….....59
Chapter 3 Methodology……………………………………….………………………….60
3.1 Introduction……………………………………………………………………………61
3.2 Macroscopic traffic flow characteristics………………………………………………61
3.2.1 The macroscopic traffic flow models…………………………………………………63
3.3 Parameter selection for new traffic flow models………………………………………63
3.3.1 The mobility models…………………………………………………………………..64
3.3.2 The traffic flow models……………………………………………………………….65
3.3.3 The vehicle speed……………………………………………………………………..66
3.3.4 The collisions………………………………………………………………………....66
3.3.5 The road topology…………………………………………………………………….67
3.3.6 The vehicular mobility in sparse networks……………………………………….......68
3.4 Steps in developing the simulation models…………………………………………….69
3.4.1 Flowchart for the simulation models…………………………………………………70
viii
3.5 The traffic flow state for different densities and level of services……………………71
3.5.1 System parameters for dense network simulations……………………………………72
3.5.2 System parameters for sparse network simulations……………………………………73
3.6 Conclusions……………………………………………………………………………74
Chapter 4 Improving VANETs congestion in dense networks and

intermittent connectivity in spares networks………………………..……………...75
4.1 Introduction………………………………………………………………………….....76
4.2 Performance metrics under dense network conditions……….…………………..........76
4.2.1 The four-lane traffic flow under dense network conditions when λ =1………………78
4.2.8 The four-lane traffic flow under dense network conditions when λ =8, 9 and 10……90
4.3 The congestion control models using four-lane networks……………………………..92
4.4 Improving connectivity in sparse networks……………………………………………94
4.4.1 Performance metrics in sparse network conditions…………………………………..98
4.5 The proposed Honeycomb Models under sparse network conditions………………...100
4.5.1 The Honeycomb Models (HCM)………………………………………………........102
4.6 Conclusions…………………………………………………………………………...103
Chapter 5 Evaluation of the simulation results……………………………….......104

5.1 Introduction…………………………………………………………………………...105
5.2 Result obtained under dense network traffic flow…………………………………....105
5.2.1 Network reachability………………………………………………………………...107
5.2.2 Network scalability……………………………………………………………….....108

ix
5.3 Result obtained under sparse network traffic flow……………………………….…..109
5.3.1 Totally disconnected nodes (vehicles)……………………………………………....110
5.3.2 Partially sparsely connected nodes (vehicles)…………………………………….....111
5.4 Conclusions………………………………………………………………………......113
Chapter 6 Conclusions and Future Directions………………………….………...114

6.1 Introduction……………………………………………………………………….....115
6.2 Summary of the results……………………………………………………………...120
6.3 Conclusions and Future Work………………………………………………………121
Appendix…………………………………...………………………………………………122
Appendix A: Publications……………..……………………………………………………..122
A Publications during the MPhil course………………………………………………………122
Appendix B: Performance analysis under sparse network conditions and the value
of λ……………………………………………………..……………………………………...124
References………………………………………………………………………………….133
x
List of Figures
Figure 2.1 Overview of VANETs Connected Vehicles……….…………………………….26

Figure 2.2 Spot and instantaneous traffic observation………………………………………34
Figure 2.3 Fundamental Diagram of Road Traffic…………………………………………..37
Figure 2.4 The speed-density relationship (Pipes Equation) …………...………………… ..45
Figure 2.5 Phase Transition………………………………………………………………….50
Figure 2.6 A reverse λ shape fundamental diagram showing free-flow, congested and
mixed state traffic conditions………………...………………………………………………51
Figure 3.1 Flowchart showing steps used to develop the simulation model………………...70
Figure 4.1a The effect of traffic flow when λ = 1…………………………………………...77
Figure 4.1b Vehicle trajectories in each lane…………………………………………..........78
Figure 4.2a The effect of traffic flow when λ = 2…………………………………………..79
Figure 4.2b Vehicle trajectories in each lane…………………………………………..........79
Figure 4.3a The effect of traffic flow when λ = 3……………………………………………80
Figure 4.3b Vehicle trajectories in each lane…………………………………………...........81
Figure 4.4b Vehicle trajectories one each lane……………………………………………....82
Figure 4.5b Vehicle trajectories in each lane…………………………………………….......84
Figure 4.6b Vehicle trajectories in each lane…………………………………………...........85
Figure 4.10a The effect of traffic flow when λ = 10…………………………………………90
xi
Figure 4.10b Vehicle trajectories in each lane ………………………………………………90
Figure 4.11 Four-Lane at the right side of the road showing the direction and movement of
Vehicles.………………………………………………………………………………………91
Figure 4.12 Four-Lane at the left side of the road showing the direction and movement of
Vehicles.…………………………………………………………………………………........92
Figure 4.13 Two-way Congestion Control Model showing different threshold of λ………...92
Figure 4.14 The effect of traffic flow when λ = 1……………………………………………96
Figure 4.19 Showing the rotation of the two clusters based on the waggling dance of
the honeybees in A and B……………………………………………………………………..99
Figure 4.20 Showing the angle 1 at the feeding station………………………………………99
Figure 4.21 Showing the angle 2 at the bee hive……………………………………….........100
Figure 4.22 Showing how the angle of rotation matches the angle between the feeding
station and the bee hive……..………………………………………………………………..100
Figure 4.23 Proposed Honeycomb Models (HCM)…..…………………………………… ..102
xii
List of Tables
Table 3.1 Traffic flow state for different densities………………………………………......71

Table 3.2 Level of service for freeway under ideal conditions and 70mpr design speed........71
Table 3.3 Dense networks system parameters……………………………………………….....72
Table 3.4 Sparse networks system parameters……………………………………………….73
xiii
List of Abbreviations
3G Third Generation
4G Fourth Generation
C2CCC Car-to-Car Communications Consortium
C2CC Car-to-Car Communications
CA Cellular Automata
CDF Cumulative Distribution Function
PD Probability Density
DSRC Dedicated Short Range Communication
GPS Global Positioning System
ISM Industrial, Scientific, Medical
ITS Intelligent Transportation Systems
IVC Inter-Vehicle Communication
PHY Physical
QoS Quality of Service
V2V vehicle-to-vehicle
V2R vehicle-to-roadside
CSMA/CA Carrier Sense Multiple Access with Collision Avoidance
IEEE Institute of Electrical and Electronics Engineers
MAC Medium Access Control
MANETs Mobile Ad Hoc Networks
VANETs Vehicular Ad Hoc Networks
Wi-Fi Wireless Fidelity
WLAN Wireless Local Area Networks
WiMAX Worldwide Interoperability for Microwave Access
xiv
List of Symbols
λ = sensitivity factor of a two-vehicle interaction in the car-following model

d = distance between two vehicles
n = number of nodes
nl = number of traffic lanes
N= number of vehicles on a highway segment
L= length of highway segments
T = observation period of spot measurements
t = time
k= traffic density
space mean speed
ui,= speed of vehicles
q= traffic flow
uT = time-mean speed
h= headway between two consecutive vehicles
η= traffic quality of service factor
fs = fraction of stopped vehicles
Τs = vehicle’s stopped time
Τt = vehicle’s average trip time
𝓍 = number of vehicles migrating to a highway
D= density measured in vehicle per mile per lane (veh/mile/lane)
V= traffic volume measured in vehicle per hour (veh/hr)
V= space mean speed measured in mile per hour (mile/hr
xv
Chapter 1 Introduction
1.1 Background of the study
Communication among vehicles has attracted the attention of the research communities in the
world in recent times according to (Forecast, 2013; Schrank, Eisele, & Lomax, 2012). A major
initiative in this direction was the European Union declaration on cooperative and ITS. As part of
this initiative some research projects were investigated into the future potential of reducing road
fatalities under the electronic safety initiative according to (Anda, LeBrun, Ghosal, Chuah, &
Zhang, 2005; Balasubramanian, Mahajan, Venkataramani, Levine, & Zahorjan, 2008; Sheng,
Mahapatra, Zhu, & Leung, 2015).
Almost simultaneously, similar initiatives were taking place in countries such as the USA and
Japan (Balasubramanian et al., 2008; Lai, 2016). The Vehicle-2-Vehicle communication in the
context of vehicular ad hoc networks (VANETs), can enable new applications for vehicular
networks. This will create numerous opportunities for safety improvements in VANETs. The
vehicle-2-vehicle communication has the potential to support drivers and allow safety
applications like collision warning, complete traffic and weather information coupled with the
navigation systems. However, these applications in VANETs, pose numerous challenges
associated with the technology, protocol designs, security and many more. This has become an
active research area that has attracted the attention of both industry and academia in the globe
according to (Lai, 2016; Lu, Cheng, Zhang, Shen, & Mark, 2014; Ramadan, Al-Khedher, & Al-
Kheder, 2012).
1
Inter-Vehicle Communication (IVC) Systems in VANETs, has attracted considerable attention
from the research community and automotive industry in recent years with several automobiles
manufacturers planning to build communication devices into their vehicles for the purposes of
safety, comfortable driving, and entertainment (Lai, 2016; Ramadan et al., 2012). Every year,
more than six million crashes occur in the U.S resulting in thousands of fatalities with many
more left with lifelong injuries (Shrank et al., 2012). The proposed Intelligent Transportation
Systems (ITS) for VANETs, that will enable vehicle-to-vehicle and vehicle-to-infrastructure
communication have the potential to prevent these collisions once developed and deployed. The
successful development of this ITS applications will eventually provide a warning to drivers in
vehicles concerning dangerous situation like accidents and many others.
VANETs characteristics are similar to mobile ad hoc networks that form multi-hop
communications (Olaverri-Monreal, Gomes, Fernandes, Vieira, & Ferreira, 2010) with the
exception that the mobile units involved are vehicles moving at high speeds. As the result of high
mobility of vehicles (nodes), the network topology is highly dynamic and subject to fast changes
with all the vehicles (nodes) utilizing the same channel. The resultant effect is traffic congestion
in dense networks (Balasubramanian et al., 2008; Forecast, 2013; Schrank et al., 2012). The
dynamic capability of VANETs, calls for a paradigm shift to new system concepts in the
development of V2V communication protocols. New wireless communication standard is
required to be used for message dissemination schemes capable of exchanging messages at
different networks conditions will improve high-speed real-time communication to enable
collision warning in ITS.
2
These mobilities in VANETs is usually considered as consisting of decentralized and self-
organizing networks made up of high-speed moving vehicle (Schrank et al., 2012). These
characteristics of VANETs is used to establish communication among vehicles and therefore
indicate pervasive application for road safety (Forecast, 2013). Meanwhile, the acceptable
communication mode in VANETs is usually through broadcast and these unique features of
VANETs are the major challenges that impede the Intelligent Transportation Systems in
VANETs. These challenges eventually render emergency traffic alert message dissemination
very difficult. This is due to high mobility and fast topology changes in vehicular environments.
This phenomenon in VANETs makes it very important to substantiate the position of the farthest
vehicles (nodes) through different methodology according to (Lai, 2016).
To achieve this in Vehicular Ad hoc networks, it is important to use decentralized controllers that
will pre-determine the exact threshold parameters variations. This will help vehicles (nodes) to
dynamically decides on the exact timelines the rebroadcast message should be sent and when to
discard received emergency traffic alert messages (Balasubramanian et al., 2008; Chitra &
Sathya, 2013; Lai, 2016). These frequent generation of alert messages in VANETs creates the
broadcast storm problems (BSPs) which are of great importance for the research community in
their quest for the development of new protocols design. It is therefore imperative to design
scalable applications that mitigate the broadcast storm problems and the intermittent connectivity
problems in sparse Vehicular Ad Hoc Networks (VANETs), this would allow free-flow and
disconnection of emergency traffic alert message (Chitra & Sathya, 2013; Lai, 2016).
3
1.2 Problem statement
The traffic congestion and the intermittent connectivity problems can be attributed to two main
factors which may be classified as predictable and unpredictable events. The predictable events
refer to the nature of the road construction (i.e. rural road, urban road, and highways) together
with the number of vehicles on the road at any moment. These unpredictable events refer to
weather, accident and the attitude of drivers on the road. It may however, be argued that in a
fully connected world of internet of things, weather could be classified as a predictable event.
When the drivers are not aware of pending congestion and traffic flow conditions ahead, it will
be difficult to pre-determine the severity of the congestion but eventually add up to the situation
without knowing. This phenomenon requires a lot of effort to mitigate the congestion and
intermittent connectivity in VANETs. The ability of the driver to pre-determine the traffic
conditions ahead on the road will aid them to take an alternate route to save time, fuel and avoid
the accident and improve traffic alert information dissemination and this is the direction of this
thesis.
The vehicular traffic congestion is a very difficult situation in VANETs according to (Schrank et
al., 2012). Unlike the current systems such as helicopter traffic reports, which are found to be
very effective. For instance, in the air, it is very easy to have a good aerial picture about the
congested area within the air-space with regards to the initial point, the final point and the speed
of the helicopter. However, this is not the case in VANETs. To equip drivers with accurate
traffic information ahead, like the helicopter traffic reports according to (Balasubramanian et al.,
2008; Chitra & Sathya, 2013; Lai, 2016), would improve V2V communication in VANETs. To
mitigate this traffic congestion in VANETs, it is very important to consider these concerns:
4
i. Ability to pre-determine the congestion in a timely manner.
ii. Ability to relay this traffic information to the driver ahead before they get to the
congested area.
iii. Ability to control intelligent traffic emergency messages (ITEMs) dissemination to avoid
the broadcast storm problems in VANETs
iv. Ability to examine traffic densities within urban and highways to ascertain the exact
phenomenon in VANETs.
v. Ability to use the above information to determine various thresholds parameters to
develop holistic congestion control systems in VANETs is the main objective of this
study.
Above information is an important parameter necessary to satisfy any congestion control
systems. The VANETs, architecture to ensure that vehicle form their own picture of congestion
ahead; depend on the need to use vehicle-to-vehicle (V2V) communication or vehicle-to-
infrastructure (V2I) communication. The understanding of these architectures leads, to the
critical examination of the selection of suitable parameters to address the technology challenges
in VANETs. To achieve these, it is important to consider the specific objectives outlined below.
1.3 Objectives of the study
1.3.1. General objective
The general objective is to conduct an analytical investigation into Intelligent Transportation
Systems (ITS) through macroscopic traffic flow model under dense and sparse networks
conditions.
5
1.3.2 Specific objectives
The specific objectives of this study are to conduct the analytical investigation in vehicular ad
hoc networks.
i. To conducts extensive performance analyses of vehicular traffic flow theory together
with the tools and methods used to address the challenges confronting the selection of
suitable parameters to achieve the objectives of the study.
ii. To propose and evaluates an alternative traffic flow model to perform the analytical
investigation into the macroscopic traffic flow model to ascertain spatio-temporal
variations in network density in VANETs to mitigate the reachability and scalability
issues.
iii. To describes the evaluation of the macroscopic traffic model to demonstrate its
importance and with evidence in dense and sparse networks conditions.
iv. To present an interpretation of the results of the analytical investigation of the
macroscopic traffic flow model in vehicular ad hoc networks to address reachability and
scalability in VANETS.
1.4 Scope of the study
This thesis focused on vehicular traffic flow theory to conduct an analytical investigation into
macroscopic traffic flow model, to analyse the spatio-temporal variations in dense networks and
sparse networks. The results from this analytical investigation will lead to the development of
congestion control model towards congestion detection and control as well as develop a
Honeycomb Model to mitigate the intermittent connectivity in sparse networks. Besides, the
6
general findings in the research through simulation and visualization tools, which can be
extended to more complex traffic models.
1.5 Justification of the proposed approach
This study is very important for the global community due to its diverse applications. In the last
two decades, wireless communication technologies have witnessed unprecedented growth,
due to developments in the field of instrumentation and in communication technology (Anda
et al., 2005; Balasubramanian et al., 2008). Within the scope of development, substantial
research efforts have been devoted to Vehicular Ad hoc Networks (VANETs) to ensure
safety (Balasubramanian et al., 2008). The thesis proposes new approach to macroscopic
traffic flow model to understand network variability in density. A successful implementation
of the concepts proposed in this study will help provide greater safety within the
transportation sector and is particularly suitable in Intelligent Transport Systems.
1.6 Organization
The rest of the study is organized as follows:
Chapter 2 of this study presents a comprehensive literature on the state-of-the-art in dense
networks, sparse networks technology challenges. It also conducts an extensive performance
analysis of vehicular traffic flow theory together with the tools and methods used to address each
of these issues. The performance merit of wide range traffic flow models and how each affects
network performance in dense and sparse networks are also discussed. The third chapter
proposes and evaluates an alternative traffic flow model. The analytical investigation into
macroscopic traffic flow model is also presented. Chapter 4 describes the evaluation of the
7
macroscopic traffic model to demonstrate its importance in dense and sparse networks scenarios.
Chapter 5 present an interpretation of the results of the analytical investigation of the
macroscopic model in VANETs. Finally, the last chapter of the study summarizes the results
obtained in this thesis and outline possible direction of future work.
1.7 Summary of chapters
This chapter provides an account of what researchers have outlined for the development of the
vehicle-to-vehicle (V2V) communication in the Vehicular Ad Hoc Networks (VANETs) in
certain countries. This includes the possible projection on how fatalities on our road could be
reduced once connected vehicle is developed. However, VANETs future applications in
Intelligent Transportation Systems pose certain challenges with the technology, protocol designs,
security and how to broadcast traffic information in a timely manner to mitigate these fatalities.
Considering this, most researchers are collaborating with industries to investigate how these
findings could be used to mitigate the numerous challenges. Due to the diverse application of
VANETs, some automobile industries have started making plans to build communication devices
in their vehicles to provide safety (i.e. prevent collision), comfortable driving and entertainment.
The main motivation of this thesis is to conduct an analytical investigation into the macroscopic
traffic flow model in both dense network and sparse network conditions.
8
Chapter 2 Literature Review
2.1 Introduction
This section is divided into two parts to provide an in-depth literature review on the general
technology challenges in VANETs protocol designs and further evaluate and discusses the main
related work on performance analysis of traffic flow theory. This chapter covers the literature on
the main research within the area of VANETs, with specific focus on congested networks and
sparse networks conditions and the limitations associated with the technical gaps according to
(Fu & Atiquzzaman, 2004; Papageorgiou, Diakaki, Dinopoulou, Kotsialos, & Wang, 2003;
Wisitpongphan, Bai, Mudalige, Sadekar, & Tonguz, 2007) . The improvement of this technical
gaps through the novel approach in ITS is the major direction in this thesis.
Current research advances in MANETs applications have become clear that such technology
would be made appropriate for Intelligent Transportation Systems applications (Al-Sultan, Al-
Doori, Al-Bayatti, & Zedan, 2014; Qian & Moayeri, 2008). VANETs applications involve
onboard safety and active systems such as V2V and V2R communications. VANETs
applications would also assist drivers to avoid collisions associated with dangerous road
conditions. For example, the non-safety systems include the real traffic congestion, adaptive
cruise control, high-speed vehicles, infotainment applications according to (H. T. Cheng, Shan,
& Zhuang, 2011).
The VANETs application domain can ensure a scalable, secure and reliable Vehicular Ad Hoc
Networks (VANETs). However, this challenge poses an extraordinary engineering challenge
9
(Liang, Li, Zhang, Wang, & Bie, 2015; Niu, Li, Liu, & Talty, 2007; Zeadally, Hunt, Chen, Irwin,
& Hassan, 2012). Besides, VANETs uses several different methodologies than other ad hoc
networks due to its unique characteristics according to (Harri, Filali, & Bonnet, 2009; Yousefi,
Mousavi, & Fathy, 2006).
2.2 The technology challenges
Several wireless access standards being used in VANETs connectivity today and this standard
provides a set of air interface protocols and certain parameters for high-speed vehicular
communication using one or two several available technologies. These core technologies are:
2.2.1 Cellular technology (2/2.5/3G/4G/5G)
The primary advantages of 2G/2.5G technology are coverage and reliable security and
3G/4G/5G, slowly but steadily taking over with improved capacity and bandwidth which is
presented (Agerholm, Jensen, & Andersen, 2015; Sheng et al., 2015). Although, several projects
are implemented using cellular technology; however, due to the relatively high cost, limited
bandwidth and latency make it impossible to use as the main communication tools in VANETs
according to (Liu, 2011), hence the use alternative technologies.
2.2.2 IEEE 802.11p technology
IEEE, currently developing different kinds of 802.11 standards capable of supporting
communication between vehicles and the roadside as well as among vehicles themselves (R. S.
Cheng & Deng, 2014; Uchimura, Nasu, & Takahashi, 2010). This technology can operate at
speed up to 200km/h, within a maximum communication range of 1000 meters are neatly
10
summarized in (N. Cheng et al., 2017; Lai, 2016; Liu, 2011). In PHY and MAC layers are based
on IEEE 802.11a, with the 5.9GHz band (5.850 – 5.925 GHz within the US) based on the
literature. The technology is widely used by car industry both in Europe (Car2Car CC) and the
US (VSCC, VII) as presented in (Agerholm et al., 2015; Liu, 2011; Sheng et al., 2015).
However, the estimated deployment cost is predicted to be relatively low due to large production
volumes according to (Agerholm et al., 2015; Liu, 2011; Sheng et al., 2015).
2.2.3 Issues of spectrum
It is estimated that the period for the V2V communication system to be commercialized would
last at least 20 years and within this time the spectrum availability has to be guaranteed
according to (Qu, Wang, & Yang, 2010). In the US the FCC has already allocated 75MHz of
spectrum at 5.9 GHz (from 5.850 to 5.925 GHz) for C2C and C2I communications according to
(Cheng et al., 2017; Liu, 2011). As have been accepted by VSC and VII Consortiums, the best
technology available for the communications systems using this spectrum would be a derivative
of IEEE 802.11. Thus the already mentioned development of the IEEE 802.11p and ISO TC204,
and also accounted for the similar approach (Cheng et al., 2017). This review is to demonstrate
the variations, the suitability and the relevance of these applications in VANETs.
However, the continuous spectrum of 75 MHz in the DSRC band is not available in most
countries (Agerholm et al., 2015; Sheng et al., 2015).
2.3 VANETs simulation and modelling
The road traffic has certain unique properties that cannot be easily modelled in a straightforward
way, using the classical MANETs approach or testbed, as presented in (Balasubramanian et al.,
11
2008; Olaverri-Monreal et al., 2010; Schrank et al., 2012). Nodes move at high relative speed
and the network density changes very dynamically and intermittently depending on the
conditions of the roads according to (Anda et al., 2005; Olaverri-Monreal et al., 2010). This
unique characteristic in VANETs need further investigation to mitigate the challenges by using
appropriate simulation tools. In addition, appropriate selection of systems parameters would aid
provide an alternative methodology to address VANETs challenges of ITS applications.
2.3.1 Wireless transmission and multiple access
Many technological challenges need to be tackled in the field of VANETs. These communication
technologies are currently discussed to be used in car-to-car communication. Conventionally
IEEE 802.11 wireless LAN (WLAN), and the dedicated short-range communication (DSRC),
together with GPRS/UMTS are just some of the selected technologies to begin with.
Due to the success story in data communication in IEEE 802.11, this technology family is the
most suitable to emerge as the main communication standard to be implemented in future cars,
specifically in 802.11p. This is currently defined by an IEEE working group. The European Car-
to-Car Communication Consortium is heavily involved in the standardization process of the
IEEE 802.11p to meet automotive communication standard which is equivalent to DSRC
technologies and this is currently being used in the US.
These standards use a communication frequency band of 5.9 GHz and this depends on the
OFDM modulation scheme. The preferred medium access method is random access, which does
not need a global scheduler. This IEEE 802.11e standard usually defines Quality of Service
mechanisms for the current WLAN technology. This concept may also be used to improve
12
message dissemination in VANETs. For enhanced channel usage, they use the combination of
IEEE 802p standard for the best result.
The WLAN-based technology has proven to be the most usable for the general task of
exchanging messages between vehicles in VANETs. However, for services with specific quality
or time constraints, as well as for very large networks (i.e.500 nodes), this technology may not be
applied in (Anda et al., 2005). This phenomenon calls for reliable and efficient wireless
transmission and multiple access technologies to provide a lasting solution in VANETs vehicle-
to-vehicle communication (V2V) which is the direction in this thesis. To have a mental picture of
VANETs environment, it is very important to understand the characteristics of VANETs in
general.
2.4 VANETs characteristics
VANETs are entirely different from the superclass of MANETs applications, such as the
mobility patterns and the constraints associated with them. These VANETs special
characteristics are:
2.4.1 One-dimensional
Most research conducted in MANETs assumes that the network consists of the nodes moving
freely in 2-dimensional manner (e.g. walking people) while vehicle mobility is restricted to the
roads only. For instance, in the cellular networks where cell covers several square kilometres, the
vehicle mobility in an urban network may be considered, 2-dimensional instead. Recent
development reveals that ad hoc standards have a limited coverage area of about 1000 meters
(i.e. 1km). It is obvious that vehicles along highways move in one dimension and therefore the
13
inter-vehicle distances are beyond the acceptable transmission range according to (M. Artimy,
2007; Jia, Lu, Wang, Zhang, & Shen, 2016). These research findings need further investigation
into VANETs mobility scenarios.
2.4.2 Mobility pattern
These include many factors like road configuration, safety limits, traffic flow conditions,
including physical limit seems to affects the mobility of vehicles (Jia et al., 2016; Singh & Singh,
2005). The drivers’ behaviours and their interactions have advert consequences on the mobility
of vehicles while in motion. These vehicles do not usually move randomly on the road rather
they also follow a probabilistic set of rules to determine their mobility patterns in terms of
directions, speed, acceleration or deceleration. The direction of this study is to simulate vehicle
traffic flow, which is very complex tasks in the intelligent transportation systems due to
VANETs characteristics in sparse and dense network scenarios.
2.4.3 Dynamic topology
The vehicular ad hoc networks may be created and re-configured in varieties of traffic situations.
For example, in dense traffic conditions, vehicles may be within proximity of each other and in
sparse traffic conditions in which vehicle distance may be about a few hundreds of meters away.
In a typical situation in which vehicle approaching a congested traffic at about 137km/hr will
change its state of proximity from 1 kilometre to 5 meters which may be approximately 26.5
seconds. In the same manner, two vehicles travelling in opposite directions at such a speed will
be within about 1-kilometer range for approximately 13 seconds. In view of these characteristics,
the communication protocols used in VANETs must be designed with the capabilities of
14
operating in a highly dynamic topology to improve ITS applications (ETSI, 2010; Forecast,
2013). To improve VANETs connectivity challenges, there is the need to understand this issue.
2.4.4 Connectivity
Current research indicates that vehicle connectivity may probably be limited to one-dimensional
networks (Forecast, 2013). This consists of short-range communication networks of finite
dimensions and infinite dimensions with the number of partitions at different time intervals
according to (M. Artimy, 2007). This section reviews the important ITS applications in
VANETs.
2.5 Intelligent Transportation Systems (ITS)
ITS applications include computers, communications, sensors, and control technologies with the
capabilities to enhance the performance of the transportation systems. These systems provide
travellers with accurate information to promote safety and the efficiency of the transportation
systems in urban and rural areas (Torrent-Moreno et al., 2009). The Intelligent Transportation
Systems (ITS) also provide essential time-critical information to systems operators in a transit
vehicle, commercial vehicles, emergency as well as the security vehicles operators. These ITS
applications combined users of the systems, vehicles as well as infrastructure into a
comprehensive system for the enhanced exchange of information to improved better traffic
management according to (Fallah, Huang, Sengupta, & Krishnan, 2011).
These applications use many exciting technologies such as cellular systems and position locating
systems (Eze, Zhang, Liu, & Eze, 2016; Vinel, 2012). As the result, there has been a steady
15
increase in the role of electronics in vehicles for the past few decades, which inspired the
creation of onboard data network to organize and coordinate the activities of different electronic
units used in a vehicle to serve these functions.
The major elements of Intelligent Transportation Systems (ITS) are categorized into seven
groups according to (Gubbi, Buyya, Marusic, & Palaniswami, 2013):
1. Travel demand management
2. Commercial vehicle operations
3. Travel and transportation management
4. Electronic payment
5. Emergency management
6. Public transportation operations
7. Advanced vehicle control and safety systems
To ensure better understanding and the importance of ITS in the future transportation systems. It
is very important to look at the potential applications of Intelligent Transportation Systems (ITS),
among the four key components of the systems. These include: the vehicle, the user, the
communication systems and the infrastructure according to (Litman & Burwell, 2006; Tayyaran
& Khan, 2003) are:
2.5.1 The vehicle
Intelligent Transportation Systems (ITS), the application includes fleet management, efficient
enforcement of regulations, in-vehicle navigation and routing, electronic toll collection, tracking
of critical cargo movements, data collection and automated control systems.
16
2.5.2 The user
The Intelligent Transportation Systems (ITS), offers inter-vehicle navigation, driver monitoring
capabilities, route guidance and dynamic route guidance in response to changing traffic
conditions.
2.5.3 The infrastructure
The Intelligent Transportation Systems (ITS), and its enabling technologies provide services
such as monitoring of weather, environmental and traffic conditions. ITS, capabilities include
responding to emergencies and management of both planned and unexpected events. These
services include other control functions that are traditionally performed by traffic signals.
2.5.4 The communications systems
Intelligent Transportation Systems (ITS) provides the ability to exchange information between
the three components in the system. This allows for the gathering of suitable data for processing
into intelligence with capabilities to determine and activate appropriate command and control
actions (Ferreira, Fernandes, Conceição, Viriyasitavat, & Tonguz, 2010).
The Intelligent Transportation Systems (ITS), projects implemented in Canada, Europe, Japan
and the United States according to (Litman & Burwell, 2006), demonstrate the potential benefits
of ITS applications and these are evident in all aspects of the transportation systems. The
important key benefits of Intelligent Transportation Systems (ITS), technologies include
improved safety, reduced congestion, and improved mobility, enhanced economic productivity,
reduced impacts on the environment. However, ITS applications in VANETs are confronted with
many technical challenges (Eichler, 2007; Zeadally et al., 2012). These challenges, associated
17
with congested networks and sparse networks in VANETs, required intensive research efforts to
improve ITS applications. The improvement of ITS applications is the research direction in this
thesis.
2.6 Research directions in VANETs
It is increasingly important to know from the research perspective that ITS applications are
becoming very successful and researchers should endeavour to come out with new research
findings to mitigate ITS challenges in these directions according to (Tayyaran & Khan, 2003;
Zang, Sories, Gehlen, & Walke, 2009). These research findings need further investigation into
certain technology applications in VANETs.
2.6.1 The needed technologies
The two approaches used in VANETs include Wireless Local Area Network (WLAN)
technologies (i.e. IEEE 802.11) and the modification of the Third Generation (3G) cellular
technologies for VANETs applications (Vinel, 2012).
2.6.1.1 IEEE 802.11
IEEE 802.11, standard outlined in (Ozer, Kahriman, Aksoy, Adiguzel, & Karadogan) indicates
the Physical (PHY) and the Medium Access Control (MAC) layers of a WLAN and offers higher
layers such as IEEE 802.x Local Area Network (LAN) standards. The physical layer can be
either infrared or spread spectrum radio transmission. Other features support power management;
the handling of the hidden nodes, license-free operation in the 2.4 GHz Industrial, Scientific,
Medical (ISM) band, and data rates 1 or 2 Mbps.
18
IEEE 802.11 standard is usually used in PHY/MAC protocol in MANETs and VANETs because
the standard’s support infrastructure or ad hoc networks. There are three basic access
mechanisms explained in IEEE 802.11. The first approach is the Carrier Sense Multiple Access
with Collision Avoidance (CSMA/CA). The second approach is Ready-To-Send (RTS)/Clear-
To-Send (CTS) handshake signals to avoid the hidden terminal problem. The third approach is
the use of the mechanism to provide time-bounded services.
2.6.1.2 The Dedicated Short-Range Communication
The Dedicated Short-Range Communication (DSRC) standard reported in (Yousefi et al., 2006),
provide descriptions of MAC and PHY standard specifications for wireless connectivity. This
standard is mainly focused on the IEEE 802.11a, according to (Altintas et al., 2011; Wang,
Song, & Han, 2013) in the 5GHz Band and which is designed for used in the high-speed
vehicular environment. In fact, the PHY layer of DSRC is adapted from IEEE 802.11a PHY
based on Orthogonal Frequency Division Multiplex (OFDM) technology. For instance, the MAC
layer of the DSRC is very similar to the IEEE 802.11 MAC based on the CSMA/CA protocol
with minor modification.
The DSRC standard utilizes the 75MHz spectrum, which is being allocated by the United States’
Federal Communication Commission to provide wireless V2R or V2V communications that
occur over line-of-sight distances of less than 1000 meters between the roadside units and which
is mostly regarded as high-speed communications but occasionally stopped and evident in slow-
moving vehicles as well as high-speed vehicles. The future expectation is that with the help of
government and industry, the DSRC systems will be the first wide-scale Vehicular Ad Hoc
Networks (VANETs) applications in North America.
19
2.6.1.3 Time Division Duplex of UMTS Terrestrial Radio Access
The Car-TALK and Fleet-Net projects have selected Terrestrial Radio Access Time Division
Duplex (UMTS) with (UTRA TDD) as a suitable alternative for the air interface that can be used
for their Vehicular Ad Hoc Networks (VANETs) applications according to (Altintas et al., 2011;
Hadaller, Keshav, Brecht, & Agarwal, 2007). The UTRA TDD provides connection-oriented,
connectionless services and control Quality of Service (QoS) through the flexible assignment of
radio resources and as well as asymmetric data flows. It also includes the availability of.
unlicensed frequency band at 2010 – 2020 MHz in Europe, with support for high transmission
range, high vehicle velocities, and high data rates of up to 2Mbps
The UTRA TDD also reserves several time slots towards high priority services. The remaining
slots can be dynamically assigned and temporarily reserved by different stations of lower priority
services using Reservation ALOHA (R-ALOHA) scheme. This reduced the potential collisions
by reserving a small transmit capacity for a circuit-switched broadcast channel that is mainly
used for signalling purposes but may be used for transmitting small amounts of user data as well.
2.6.2 MAC layer issues
VANETs are usually considered a special application of MANETs with some well-defined
applications. It also includes severe operating constraints which usually result in frequent
disconnection of the network coupled with fast topology changes according to (Torrent-Moreno
et al., 2009). The sparseness of the nodes causes the network disconnection. This study aimed at
dense network and sparse network traffic due to the challenges associated with the vehicular
network ITS applications.
20
2.6.3 Applications
A lot of applications were proposed within Intelligent Transportation Systems (ITS) projects in
recent times according to (Al-Sultan et al., 2014; Boukerche, Oliveira, Nakamura, & Loureiro,
2008). In the advent of current wireless technology research efforts from the academia and
industry in mobile ad hoc and VANETs, the VANETs are seen with the capabilities of
supporting these ITS applications in the future.
2.6.3.1 Safety applications
In VANETs applications, the vehicle foresee dangerous situation ahead can report it instantly to
neighbouring vehicles. Sensors can recognise dangerous situation by detecting events such as the
deployment of airbags due to the accident, loss of tire traction, or sudden application of brakes.
The critical requirements in these applications are latency and area of coverage as reported in
(Forecast, 2013; Zhang, 2006).
According to (M. M. Artimy, Robertson, & Phillips, 2005), triggers the need to introduce the
concept of Zone-of-Relevance. Within this zone, a warning message is delivered to drivers to
assist them to avoid dangerous situations while the vehicle outside the Zone-of-Relevance may
route the message but drivers are not alerted to ensure that only the relevant messages are
delivered to drivers and all the unnecessary reactions are avoided to prevent fatality on the roads.
2.6.3.2 Traffic monitoring
The use of existing capabilities of vehicles, such as Car Area Network (CAN) and GPS,
VANETs act as an intelligent sensor and forms powerful traffic information systems. The
21
advantages of using VANETs over other infrastructure-based systems involved the rapid
deployment, self-organization, and lower-cost applications.
According to (Kloiber, Strang, Spijker, & Heijenk, 2012), VANETs is composed of fixed and
mobile nodes (vehicles). The vehicles are organized into clusters where they exchange beacon to
maintain a neighbouring relationship. This task request is generated by the nodes or from an
external network and is transmitted to immediate neighbours to monitor some relevant conditions
of the roads. The forwarding tasks request may occur to propagate the requests through the
network upon receiving the request; the vehicle itself samples the data from the relevant sensor.
This data is then summarized and forwarded back to the originating nodes as a task response
which works within certain threshold parameters.
According to (Su & Zhang, 2007; Zeadally et al., 2012), in each segment, the vehicle has the
capability to monitor the locally observed traffic situation by receiving data packets concurrently
with detailed information from other vehicles within the contour of the road. In their studies, a
traffic situation analysis was performed in each individual vehicle and the result is transmitted
through a wireless data-link to all vehicles in the local neighbourhood. This has seen a significant
improvement but failed to address ITS alert message dissemination due to the fast topology
changes in VANETs.
2.6.3.3 Automated highways and cooperative driving
This type of VANETs applications is primarily focused on the automation of some driving
functions to increase driving safety and improve the capacity of highways. This study
investigated into platoon formation and lane merge according to (Amoozadeh, Deng, Chuah,
22
Zhang, & Ghosal, 2015; Anda et al., 2005; Nekovee & Bogason, 2007) and obstacle avoidance
and blind intersection assistance according to (Boukerche et al., 2008). This result is a serious
defect that prevails during lane change at the intersection.
2.6.3.4 Entertainments
These applications ensure the delivering of multimedia content for vehicle passengers. Among
the challenges in this area is the provision of QoS guarantee to locate the media content. The
author (Toor, Muhlethaler, & Laouiti, 2008) assess the importance of a Car-to-Car (C2C) and
Peer-to-Peer (P2P) systems for delivering media content using the cellular and VANETs
applications.
2.7 An overview of VANETs applications
The vehicular ad hoc networks (VANETs) applications are seen in a broader sense as an
indispensable aspect of modern life. This V2V (vehicle-to-vehicle), V2R (vehicle-to-road
infrastructure) V2S (vehicle-to-sensor on-board), and V2I (vehicle-to-Internet) applications
according to (Agerholm et al., 2015; Ramadan et al., 2012; Sheng et al., 2015). These applications
have come to stay and will continue to evolve despite the challenges that impede the overall
benefits. Irrespective of the challenges in VANETs, wireless communication capabilities are
expected to continue to revolutionize the automobile industries. With the help of academia and
research communities, VANETs will undoubtedly produce unprecedented results and reduced
the fatalities of accidents on our roads when efficient traffic flow models are developed. The
improvement of traffic flow model is the main reason for this thesis.
23
The new direction now would allow academia and industry to collaborate to support various
applications to improve road safety such as lane change warning, cooperative merging, and
collision detection. Smart and green transportation would also help applications such as traffic
signal control, fleet management, and intelligent traffic scheduling and location dependent
services like route optimization access. It is evident that the market of VANETs applications will
grow massively according to recent global business report, that the global market is projected to
reach USD 131.9 billion by 2019 according to (Forecast, 2013), due to the numerous benefits of
vehicular networks in ITS applications.
The importance of VANETs applications is divided into two main driving forces of bringing
together the wireless vehicular connectivity to mitigate safety alert message dissemination. The
first direction is the improvement of safety and efficiency of road transportation systems. The
increasing population growth correlates with the number of vehicles in urban and in large cities.
This driving force accounts for traffic congestion and this led to environmental problems and
economic cost. It is reported (Forecast, 2013) that the traffic congestion accounts for the cost of
extra travel time and fuel consumptions in 498, United State of America, urban areas were 121
billion USD in 2011 and carbon dioxide produced during vehicular congestion was 56 billion
pounds, compared to 24 billion USD and 10 billion pounds in 1982 respectively (Schrank et al.,
2012). The VANETs applications are promising to mitigate traffic congestion by means of
intelligent traffic controls and surveillance management according to (Anda et al., 2005). This
will subsequently improve the road safety through onboard advanced traffic warning and driving
assistance systems as presented (Olaverri-Monreal et al., 2010).
24
The second direction refers to increasing mobile data demand for vehicular users on the road. In
recent times, the high demand of mobile Internet users has increased exponentially. Many people
expect to have the same vehicular Internet connectivity while driving a car as compared to what
they have at home and at workplaces. Vehicular Internet connectivity is expected not to meet the
mobile data demands according to (Balasubramanian et al., 2008). This also support safety-
related applications of online diagnosis presented in (Lai, 2016). Intelligent anti-theft tracking
according to (Ramadan et al., 2012), whereby the servers resides in the Internet cloud. The
VANETs Internet integrated vehicles are almost on the road with us and are projected that the
percentage of the Internet integrated vehicles applications will rise from 10% to 90% by 2020 as
simplified in (Lu et al., 2014). Surprisingly the US. Department of Transportation’s (DOT),
National Highway Traffic Safety Administration (NHTSA), recently announced that it will start
taking immediate steps to enable communications between light vehicles according to (Cheng et
al., 2017). Furthermore, the government has mandated the VANETs connected vehicle
revolution on a fast track mission due to its important applications. The European Commission
has also proposed the mandatory implementation of “eCell” system in cars beginning in 2015 by
which cars may automatically establish a telephone link for emergency services in the midst of
collision and congestion according to (Agerholm et al., 2015; Sheng et al., 2015). This revelation
proved the importance and the relevance of this study.
VANETs connected vehicles refer to the wireless connectivity enabled vehicles that are equipped
to communicate with their internal and external environments. This technology enabled vehicles
presented in (Tuohy et al., 2015) This has the capabilities of supporting the interactions between
V2V (vehicle-to-vehicle), V2R (vehicle-to-road infrastructure) V2S (vehicle-to-sensor on-
board), and V2I (vehicle-to-Internet), as shown in Figure. 2.1. These applications are considered
25
as the building blocks of the emerging Internet of Vehicles (IoV). The dynamic mobile
communication systems have the features of gathering, sharing, processing, computing and the
release of security information. This would spearhead the evolution of the next generation of
Intelligent Transportation Systems (ITS) according to (Liu, 2011). The development and
deployment of the fully VANETs connected vehicles require the combination of various off-the-
shelf and emerging technologies. This study focuses on the vehicular ad hoc networks
(VANETs) technologies applications and presents an overview of the industrial and academic
advances for establishing a comprehensive VANETs connectivity architecture. The study will
also present the state-of-the-art potential challenges and identify research direction to propose the
alternative solution. The literature presented in (Faezipour, Nourani, Saeed, & Addepalli, 2012;
Qu et al., 2010), serve a similar goal, but with a different focus.
Figure 2.1 Overview of VANETs Connected Vehicles
26
2.8 The vehicular mobility
Many researchers are highly interested in studying sparse vehicular networks due to its
importance in ITS applications in VANETs. This motivation from most researchers reflects the
assumptions inherent in the inter-vehicle communication associated with sparse networks. This is
seen to be more challenging due to the difficulty in maintaining connectivity. In addition, the
low-density conditions offer more flexibility in setting the traffic parameters such as flow, speed,
and density for analytical and simulation studies. The assumption of independent parameters is
valid only in free-flow traffic. Therefore, the results of these studies cannot be used directly in
dense traffic environments. This study seeks to argue that research should focus more on dense
traffic and traffic jam environments due to their important significant influence on VANETs:
1. The vehicle density can contribute to rapid message dissemination according to (Torrent-
Moreno et al., 2009).
2. Higher node density aggravates the contention and collision problem in the
communication protocols based on shared media.
3. The traffic jams contribute to the non-homogeneous distribution of vehicles and which
increases the needed transmission range to maintain connectivity.
4. The vehicle traffic may change rapidly between free-flow to congested traffic due to
traffic controls, road constraints or accidents.
The focus of this thesis is to propose a novel alternative methodology to improve ITS
applications associated with dense and sparse networks in VANETs. The review of research
directions related to the use of vehicle mobility models and traffic simulator is provided in the
next subsection.
27
2.8.1 The vehicle mobility models and traffic simulations
The vehicular mobility models and traffic simulators have been in development according to
(Choffnes & Bustamante, 2005; Harri et al., 2009). The main reasons for these simulators are in
traffic analysis, forecasting, and planning transportation networks. Due to the important
applications in VANETs, the ITS initiative has generated an enormous number of vehicular
traffic simulators since its inceptions in the United State in 1991 according to (L. Chen, Wei, &
Shi, 2011; Ros, Martinez, & Ruiz, 2014). These simulators are intended to support ITS
applications, but not necessarily VANETs, due to their recent introduction.
In VANETs research papers, simulation of vehicle mobility is often considered a special type of
mobility model such as the RWP model with some extensions (Bettstetter, Hartenstein, & Pérez-
Costa, 2004; Bettstetter, Resta, & Santi, 2003; Harri et al., 2009). These mobility models may
not produce complex vehicle manoeuvres such as car-following, accelerating, decelerating under
traffic jam situations. For the analysis of any complex phenomenon in view of VANETs
characteristics, many studies have relied on vehicular mobility models or traffic simulators to
generate vehicle movement traces to improve ITS applications. This is the major direction of this
thesis.
2.8.1.1 The mobility models
Many studies in ad hoc networks often require keeping track of nodes positions at any given time
and vehicular networks are no exception. In recent time, many VANETs applications depend on
realistic vehicle movement patterns to obtain from traffic micro-controllers which can track the
characteristics of individual vehicles including the position, speed, and direction. The early
research in inter-vehicle communication relied on micro-simulators that were designed originally
28
for studying vehicle traffic in intelligent transportation systems. VANETs research adapts
simulators to generate traces of vehicles positions at regular time interval into network simulator
according to (Breslau et al., 2000; Fall & Floyd, 1996) to perform a simulation of
communication protocols. In recent times, the research communities have devoted substantiate
effort to the development of network traffic simulators for VANETs research according to (Harri
et al., 2009; Karnadi, Mo, & Lan, 2007; Yousefi et al., 2006). Usually, these simulators have a
dual focus on the communication and mobility sides of VANETs and this is considered in this
thesis.
2.8.1.2 The traffic flow models
The traffic simulators are extensively used in evaluating vehicular ad hoc networks. The
analytical approaches in free-flow traffic conditions are usually the preferred choice. The
primary reasons for this choice are the control environments it’s offered to select at least two or
three parameters to establish the fundamental relationship of the traffic flow under free-flow
conditions (May, 1990). This study assumes that the traffic flow and the vehicle velocity are
independent under sparse network scenarios. These fundamental relationships of traffic flow are
used to derive the pdf of the communication duration between vehicles and the pdf of the time
between consecutive topology changes. Similar assumptions are made in (May, 1990) to derive
the pdf for distance at which it is possible to establish a Multi-hop connection between two
vehicles. The effective selection of time headway (i.e. the time gap between consecutive
vehicles) as an input parameter can be control at the rate by which vehicles are injected to the
simulation environment and the density of the vehicles. The time headway values are generated
randomly from a Poisson distribution with an average value that ensures the free-flow traffic. A
29
uniform distribution with a minimum value that specifies the initial separation between vehicles
is used. The analysis of traffic flow models is the major direction of this study in which the effect
of parameters would be studied to ascertained its variations in each model. The performance
analysis of traffic flow theory will be examined critically in section 2.9, to understand the merits
and the demerits from several research directions.
2.8.1.3 The vehicle speed
In most cases, the desired speed is provided as an input to the simulations. The vehicle speed can
be a constant value for each type of vehicles (e.g. cars and trucks) or may form a random
distribution according to (Boban, Vinhoza, Ferreira, Barros, & Tonguz, 2011; Sommer, German,
& Dressler, 2011). The speed distribution is chosen to correspond to real measurements from
German highways, while in (Adolf, 1990; Donnell, Ni, Adolini, & Elefteriadou, 2001; May,
1990), the speed is assigned from a normal distribution and this scenario has been considered in
this thesis.
One possible choice of (Khoury & Hobeika, 2007) that is usually used as a guide to set the limit
on the highways (Khoury & Hobeika, 2007). The 85th percentile speed is defined as the speed
85% of drivers moving at or below. It is evident that 75% of the drivers maintain a speed within
a certain range around the average speed limit. Another research finding indicates that 15% are
slow drivers while the remaining 15% are driving at high speeds. According to (Adolf, 1990), the
vehicle speed is sampled from a Normal distribution using the standard deviation to mark the
boundaries between slow, average and high speeds. Although, the speed and time headway
changes in real traffic, they are kept constant in (Alsabaan, Alasmary, Albasir, & Naik, 2013;
30
Haerri, Filali, & Bonnet, 2006; L. Li & Chen, 2017) and this assumption is also considered in
this thesis.
In addition, the choice of the average speed varies according to the choice of the environment
under considerations. In highway simulations, the average speed is set to 100-130km/hr
according to (Esser & Schreckenberg, 1997; Van Arem, Van Driel, & Visser, 2006). In another
development, simulations of city street grid set the average speed to 40-80km/hr either randomly
according to (Ahn, Rakha, Trani, & Van Aerde, 2002) or according to the street types. The
choice of speed may reflect also the posted speed limit, road conditions, and common driving
habits.
2.8.1.4 The collisions
The collision in VANETs has usually assumed in some literature that a vehicle moves freely and
collisions are not allowed. However, the vehicle density according to (May, 1990) is specified to
create congested traffic situations and vehicles may collide if drivers fail to react in time. This
finding is very necessary to evaluate the influence of the V2V and V2R communication on safety
as indicated in (May, 1990).
2.8.1.5 The road topology
This may be described as a common choice of highway geometry of two or more lanes in each
direction. This allows the traffic to flow in both directions which may be described as
asymmetric as summarised in (Mohimani, Ashtiani, Javanmard, & Hamdi, 2009; Zeadally et al.,
2012). According to (Mohimani et al., 2009; Zeadally et al., 2012), they argue that dividing the
highways may be used to limit the geographic area affected by message broadcast to one traffic
31
direction only. According to (Jerbi, Senouci, Rasheed, & Ghamri-Doudane, 2009; Santos,
Edwards, Edwards, & Seed, 2005), explained more complex geometry to be used to evaluate
routing strategies and protocols in a city traffic flow environment to mitigate topology challenges
in VANETs.
2.9 Performance analysis of the traffic flow theory
This section provides relevant background information to the traffic flow theory that involves the
summary of the terminology as well as the parameters used to describe traffic flows is the
direction in this thesis. The fundamental concepts of the traffic flow and the tools and methods
available for vehicular traffic flow analysis.
The vehicular traffic flow is seen to be a complex phenomenon since they involve human
activities, physical infrastructure, time and vehicular movement on the other hand. Several
approaches have been adopted to investigate the problems of traffic flow according to (Adolf,
1990). This process involves the use of empirical and intuitive studies to collect traffic statistics
in an event of the accident. Standard data analysis techniques are then used to determine the
interrelated parameters between relevant quantities. According to (Nagel & Herrmann, 1993;
Ros et al., 2014), several deterministic theories have been proposed to describe traffic dynamics
at the macroscopic traffic flow model. On the other hand, the probabilities flow models and
queuing theory is usually used to represent intersections or length of road to determine quantities
such as delay, capacity or the average speed. To ensure a realistic result, traffic simulations are
used to perform analytical analysis to simplify the traffic problems to some extent.
32
Vehicular traffic flow theories have been developed to define the interrelationship of quantities
with respect to vehicle density, speed, and flow. These mathematical relationships among these
three quantities have the tendencies to predict an abrupt change in one or two quantities. These
changes usually manifest when congestion erupts by a driver.
Research indicates that several advances have emerged leading to the introduction of cellular
automata (CA) models in the past. Simulation of vehicular traffic flow using CA models has
exhibited characteristic as a phase transition phenomenon. Vehicular traffic flow exhibit gas or
liquid states phenomenon which is in a homogeneous state.
This chapter defines the main characteristics that are used to describe traffic flow in section
2.7.1, follow by the fundamental relationship of traffic flow that relates these characteristics in
section 2.8. This relationship is the fundamental basis of this thesis to perform the analytical
study using macroscopic traffic flow models according to (Adolf, 1990; May, 1990; Nagel,
1996) in relation to dense and sparse networks. The section 2.8.1 explains four of the main
models that are used in vehicle traffic analysis. In section 2.9, explain the concept of phase
transition and the effect of parameters. Section 2.10, provides the summary of traffic simulation
approaches applicable in VANETs while section 2.11, conclude the chapter.
The understanding of this model leads to methodology in chapter 3, while chapter 4 introduces
concepts to improve reachability and scalability in VANETs under dense and in sparse networks
scenarios.
33
2.9.1 Traffic flow characteristics
Road traffic stream studies is primarily focused on three characteristics of the stream. In which
the flow q, measures the number of vehicles that pass an observer per unit time. The flow k,
measures the number of vehicles per unit distance. The speed u, measures the distance a vehicle
travels per unit time.
Although, traffic streams are usually not uniform, but vary over both space and time duration. In
view of this, the quantities q, k, and µ are regarded as either an average or sample random
variables. Practically, the traffic characteristics like flow, speed, and velocity are parameters used
in statistical distributions but not absolute number according to (Adolf, 1990; May, 1990;
Yousefi et al., 2006).
Figure 2.2 Spot and instantaneous traffic observations
The vehicular movement may be counted and measured through a specific location under
observation. This observation is known as spot observation while instantaneous observation
refers to the vehicle on a designated highway when the segment is counted and the speeds are
measured instantly as shown in Figure 2.2. On the other hand, the instantaneous observation
34
occurs when the traffic density and space-mean speed are directly estimated. In addition, spot
observation occurs when the traffic volume and the time-mean speed are directly estimated
according to (Adolf, 1990; May, 1990; Schumacher, Priemer, & Slottke, 2009). This relationship
pointed in this literature is relevant in a practical implementation of any traffic flow model in
VANETs. The research findings derived from the above studies provide a theoretical background
in this thesis.
In an instantaneous observation, the number of vehicles, on highway segment in each time
measurement may be changed to traffic density :
(2.1)
Where k is measured in the vehicles per kilometre per lane (veh/km/lane), is usually the length
of the segment expressed in kilometres (km), and denotes the number of traffic lanes.
Typically, density is measured in each direction separately based on events. The speeds of
vehicles, ui, and this measurement can be averaged to calculate the space-mean speed,
(2.2)
The spot observations render the number of vehicles passing a spot during an observation period,
T. This count can be converted to traffic flow, q:
35
(2.3)
The speeds of vehicles can also be measured in a spot observation. The resultant speed, called
time-mean speed, is calculated as follows:
(2.4)
where uT refer to time-mean speed which is calculated based on N spot measurements, ui is the
speed of the ith vehicle included under observation. The time-mean as well as space-mean speeds
measured for same traffic are typically not equal since the former does not account for stopped
vehicles (Adolf, 1990; May, 1990; Schumacher et al., 2009). All measurements of density and
speed are based on instantaneous observation as given in (2.2) and (2.3). As a result, the average
(mean) speed refers to the space-mean speed, i.e. = .
2.10 Fundamental relationship of traffic flow
Traffic flow is related to the headway between vehicles. The headway is defined here as the
difference in arrival time between two consecutive vehicles as measured by a person standing at
a fixed point on the road. If all the vehicles move at the same speed of , then the headway
between two consecutive vehicles, h, is obtained by driving the distance, d, between these two
vehicles by speed, , (h = d/ ). If a person measures the headways and distance between
consecutive vehicles under observation and calculates the average values, h and d, the
36
relationship between this average value will be preserved, i.e. h=d/ . Note also that q=1/h and k=
1/d according to (Adolf, 1990; May, 1990; Schumacher et al., 2009). Thus,
(2.5)
Figure 2.3 Fundamental Diagram of Road Traffic.
Equation (2.5) is called the Fundamental Traffic Flow Relationship. For the comprehensive
derivation of this relationship according to (Adolf, 1990; May, 1990; Nagel, 1996). All the
bivariate relationships among the variable in (2.5) are studied in the literature. Some of the
approaches starts with mathematical models while others are mainly empirical according to
(Adolf, 1990; May, 1990; Nagel, 1996). The bivariate relationship between traffic flow and
density considered useful is the Fundamental Diagram of Road Traffic. The q-k relationship can
be deduced intuitively in (Adolf, 1990; May, 1990; Nagel, 1996). The flow must be zero when
density is zero. The flow also falls to zero when density is at maximum (vehicles are stationary
in a traffic jam). For values of density between these limits, flow must rise to at least one
maximum. The experimental data seem to confirm this argument, but there is some conflicting
37
evidence regarding the exact shape of the curve. Previous studied assumed the curve to take a
parabolic shape (Adolf, 1990; Nagel & Herrmann, 1993). Furthermore, empirical data such as
those reported in (Burstedde, Klauck, Schadschneider, & Zittartz, 2001; Kerner, Klenov, &
Wolf, 2002; Nagel & Schreckenberg, 1992), and simulations using CA models (Kerner et al.,
2002), including the simulations produced in Chapter 3, show that the fundamental diagram may
take the so-called reverse λ shape. The latter shape is an extension of Figure 2.3 and will be
described in more detail in section 2.5.
Figure 2.3 shows that vehicle density has a maximum value denoted kjam, the density of a traffic
jam density is regarded as the reciprocal of the average length a vehicle occupies in a traffic jam.
The maximum value of flow is denoted qmax. The maximum flow is also known as the ‘capacity’
of the road.
The left branch of the fundamental diagram describes free-flow traffic. Vehicles in this traffic
condition move free from interference from other vehicles. The average free-flow speed of
vehicles, uf, is the ratio of flow to density. Figure 2.3 suggests that the free-flow speed remains
constant as density increases up to a point where the increase in vehicle density starts to
influence drivers and force them to reduce speed. Beyond this point, the vehicle speed is
dependent on the density and this assumption has been studied in this thesis under dense network
scenario.
The right branch of the fundamental diagram describes a congested traffic condition whereby q
decrease with increasing k. The slope of this branch represents the speed of the traffic jam wave
propagating backward relative to the direction of traffic, dq/dk according to (Adolf, 1990). Given
38
the previous discussion, it is possible to define the following terms to be used in the remainder of
the thesis.
Definition 3.1 congested traffic is a vehicle traffic condition in which drivers cannot maintain
the desired vehicle speed due to the close presence of other neighbouring vehicles. The speed in
this traffic is influenced by the surrounding vehicles. Vehicles in congested traffic may
experience stop-go movement patterns due to traffic jams.
Definition 3.2 free-flow traffic is a vehicle traffic condition in which drivers can maintain a
virtually constant speed. The constant speed is determined by factors such as vehicle engine
capability and traffic laws but it is not dependent on the speed of, or the distance to other
vehicles.
Definition 3.3 traffic jam refers to several vehicles blocking one another in a queue so that they
cannot move.
2.10.1 Traffic flow models
Traffic flow characteristics such as volume, density, and speed do not vary independently from
each other. This fundamental relation of traffic flow (2.5) combines these three characteristics
altogether but give room to determine one quantity when the other two are known. In addition,
another equation is needed to calculate the two unknown flow characteristics from the known
values.
There is no single theory that can explain all the complexities of traffic flow. Instead, there are
several models that try to explain the relationships between traffic parameters. The oldest among
these models are the macroscopic model (i.e. is the direction of this thesis) and based on the
39
fluid-dynamic theory, based on the assumption of a delayed adaptation of velocity. Research
papers in physics journal (Chowdhury, Santen, & Schadschneider, 2000; Helbing & Tilch, 1998)
model traffic as a system of interacting vehicles. These traffic models are more successful in
describing physical phenomena such as the phase transition. One CA model of the macroscopic
model will be used extensively in this thesis as an analytical tool to derive various relationships
in Chapter 4 through to Chapter 5. The macroscopic models will be introduced briefly in Section
2.5 and in more detail in Chapter 3.
Unlike, other traffic models, the macroscopic model does not describe all aspects of vehicle
traffic. The following subsections include an overview of three main traffic models. Each one of
these models describes some aspect of vehicles traffic that will be employed in later chapters.
The phenomenon of these models is interrelated in many ways possible. In Chapter 4 a
relationship between the car-following model and the macroscopic model will be investigated.
These relationships will be established in Chapter 5 to allow vehicles to estimate their local
density without relying on infrastructure to exchange information with other cars.
2.10.1.1 Macroscopic (continuum) traffic models
Highway traffic can be visualized as a stream or a continuum fluid. Because of this analogy,
traffic is often associated with fluid flow and treated similarly as a one-dimensional compressible
fluid. The theory can also describe traffic flow along a multi-lane highway. This section will
introduce the basics of this theory in (Hoogendoorn & Bovy, 2001). This section is the primary
focus in this thesis with a similar model in (May, 1990), would be used extensively in chapter 4
and these results will be discussed in chapter 4 and chapter 5 with their models.
40
The continuum theory of traffic flow is based on two relations. The first is the law of
conservation of vehicles and the other is an assumed relation between the flow and density
according to (Adolf, 1990; Hoogendoorn & Bovy, 2001). In one-dimensional flow consider that
measurements of the flow, q, are taken at two points, x=a and x=b, a short distance apart,
assuming there are neither entrances nor exists between these points. Then the change in the
number of cars results from car crossing at x=a and x=b only. This is expressed as follows
(Waters & Laker, 1980):
(2.6)
This equation states that if, at a specific location, the flow coming in from the left is less than the
flow going out to the right, then density must decrease at that location over some time, t. On the
other hand, the density must increase if the incoming flow is greater than the outgoing flow. An
additional equation or assumption is needed to solve (2.6). One possible option is to state speed
as a function of density, u=u(k), which is a clear reasonable assumption, but it is only valid in
dense traffic scenario. It follows from this assumption that drivers vary their speed according to
the density (i.e. as density increases hence the speed must decrease to maintain safe headway).
Then q is assumed to be a function of density only, q= ku(k). It should be noted that u(k) can be
any function (see section 2.8.3). Then, (2.6) can be written as
(2.7)
41
Where dq/dk=u(k) + ku1(k) = 0.
The solution of (2.7) has suggested that the density k is constant along a family of curves known
as characteristics or waves. A wave represents a change in flow and the density along the
roadway. The speed of the wave is c, a constant representing the slope of the flow as a function
of density is evaluated at a constant density. According to (Adolf, 1990; Hoogendoorn & Bovy,
2001), the value of c is negative when the road is beyond a critical density and it represents the
traffic wave velocity propagating backward (see Figure 2.3).
A shockwave is created when two characteristic lines intersect. The shock represents a
mathematical discontinuity in q-k, or u-k elevations. Physically, a shockwave is an interface
between congested and free-flow traffic. It is created where vehicles slow down or accelerate
rapidly. The shock wave can travel forward or backward (the end of growing queue moves
backward while the end of a dense column of vehicles behind a truck can move forward). The
speed of the shock wave, uw, is given in (Hoogendoorn & Bovy, 2001; Meshkov, 1969):
(2.8)
Where kd, qd and ku, qu represent downstream and upstream traffic parameter, respectively. In
Figure 2.3, the shockwave speed is represented by the slope of the dotted line connecting the two
traffic conditions (free-flow and congested traffic).
42
2.10.1.2 Car-following models
The car-following models depend on a single-lane dense traffic environment with no overtaking.
These models assume that each driver reacts in certain specific fashion to a stimulus from the
vehicles ahead or behind. In general, road conditions do not affect interactions between vehicles.
A different approach is useful for low densities where interaction between vehicles disappears.
The basic equation of the car-following model is found in (Adolf, 1990; Hoogendoorn & Bovy,
2001), Response = Sensitivity x Stimulus. This response is taken as the acceleration of the
following vehicle. It has been established also that the stimulus is the relative speed between the
following vehicle and the lead vehicle within a certain range. One example of the stimulus-
response equation of the car-following model according to (Adolf, 1990; Hoogendoorn & Bovy,
2001; Saltelli & Annoni, 2010) (Kleijnen, 2005; May, 1990):
(2.9)
Where xi is the position of the ith vehicle, r is the driver’s reaction time and λ, refer to as the
sensitivity of the interaction, which is considered proportional to the headway between the two
consecutive vehicles. The sensitivity may take the value of one of the several functions (Adolf,
1990; Hoogendoorn & Bovy, 2001) and (Kleijnen, 2005; Saltelli & Annoni, 2010). This
sensitivity factor λ, is used as a function of several input variables (i.e. thus increasing the value
of λ from 1 to 10) to examine the effects of the macroscopic traffic flow model under dense and
sparse networks scenario is investigated in this thesis.
43
Equation (2.9) and similar equations (Adolf, 1990; Kleijnen, 2005; Saltelli & Annoni, 2010) are
used to derive the speed-density and the flow-density relationships for single steady-state traffic
flow. In the linear case, λ is assumed to be constant and the equation (2.9) is integrated to derive
the velocity-density relationship according to (Adolf, 1990; May, 1990).
(2.10)
This consist of a linear car following model specifies an acceleration response which is
completely independent of the inter-vehicle distance. When taking the following distance into
account, the linear model can be modified by assuming that λ is inversely proportional to the
following distance according to (Adolf, 1990; Kleijnen, 2005; Saltelli & Annoni, 2010).
Equation (2.10), which was proposed by Pipes in 1953 according to (Shercliff, 1953), has an
undesirable feature. In this case, the speed tends to be unrealistically high near low densities. The
recent event for the observations of freeway traffic and simplifications in traffic modelling led to
the modification of the equation (2.10) according to (Adolf, 1990).
(2.11)
Where the speed cannot exceed a maximum value of umax.
44
Figure 2.4 shows the speed-density relationship of (2.11). The flow can be obtained by
substituting (2.11) into the fundamental relationship in (2.5).
(2.12)
Figure 2.4 The speed-density relationship (Pipes Equation).
Equation (2.12) is represented by the q-k relationship of Figure 2.3. This equation is quite useful
for understanding the effects of driver behaviour under traffic conditions. Another possible
observation is that drivers tend to slow down when density increases. Similar observation is that
when the capacity is reached at a density value. If density increases beyond certain critical value,
kc, the loss in flow caused by the speed decrease exceeds the gain caused by the density increase
and the flow drops. The capacity is determined by the headway drivers maintain between
vehicles. Drivers select spacing perceived as safe for the speed that persists in the capacity
conditions.
45
Throughout this thesis, the relationships between speed-density and flow-density are assumed to
follow equations (2.1) and (2.3), respectively. The traffic simulations produced in Chapter 3,
results in relationships that are like those shown in figure 2.3 and figure 2.4 (with some
modifications that will be clarified in Chapter 4). The reason for this similarity is that the
macroscopic model belongs to the same class of the car-following models as the Pipes model,
which assumes a linear relation between speed and the distance between vehicles. Moreover, in
Section 3.2.1 the value of the sensitivity factor, λ, will be varied from 1 to 10 during simulations.
Later, in Section 3.2.1 the value of λ will be derived analytically with the macroscopic traffic
flow model in dense networks and sparse networks. Both results will be used in Chapter 4 and
Chapter 5 to estimate the local vehicle density variations.
2.10.1.3 Two-fluid models
Two-fluid theory on traffic flow was proposed by Herman and Prigogine according to (Herman
& Prigogine, 1979), to measure the level of quality of traffic service in an urban street network.
This model applies to congested traffic in the right branch of the q-k relationship. The vehicles in
this traffic stream are divided into two classes known as running and stopped vehicles. In
addition, those in the later classes include vehicle stopped in the traffic stream due to traffic
control signals and signs, blocking vehicles, normal congestion, but exclude those that are
outside the traffic stream (e.g., parked cars). The model is based on two assumptions and this is
reported in (Herman & Prigogine, 1979):
1. The average running speed within certain street network is proportional to the fraction of
vehicles that are moving and
2. The fractional stop time within a test vehicle circulating within a network is equal to the
46
average fraction of the vehicles that stopped during the same period.
This first assumption of the two-fluid theory usually relates to the average speed of the moving
(running) vehicle, ur, to the fraction of moving vehicles, fr, in this sequence:
(2.13)
Where umax is usually the average maximum running speed and η is an indicator of the quality of
traffic service in the networks.
This average vehicle speed can be defined as u= urfr, and fr + fs = 1, where fs is the fraction of the
stopped vehicles. Therefore, (2.13) can be rewritten as,
(2.14)
Second assumption of the two-fluid model that relates to the fraction of time when a test vehicle
is circulating a network is stopped during a trip time to the average fraction of vehicles stopped
during the same time, or
(2.15)
Where Ts is the stopping time and Tt is the trip time.
47
The relation has been proven analytically in (Ichoua, Gendreau, & Potvin, 2003; Jenelius &
Koutsopoulos, 2013), and represented the principle embedded in the model, (i.e., that the
network conditions can be represented by a single vehicle appropriately sampling the network).
The parameter, η, is considered a measure of the resistance of the network to degraded operation
with increased demand. Higher values of η indicate the networks that degrade faster as demand
increases. Field studies and simulations show that values of two-simulations, show that η=0
when vehicles are moving according to the macroscopic model.
The two-fluid theory is similarly significant to this thesis because it shows that a single vehicle
may predict the surrounding traffic condition, more specifically the fraction of vehicles stopped
in the transportation network, based on the vehicle’s own movement (and stopping) pattern. This
similar result in the macroscopic traffic flow model will be used in Chapter 5 in combination
with results from traffic models to derive a relationship that allows a vehicle to estimate the local
density by varying the value of λ from 1 to 10. This investigation would help to address the
objectives of this study with regards to ITS applications in VANETs. The outcomes of this study
would significantly improve on the congested networks and spares networks conditions in
VANETs.
2.10.1.4 Cellular automata (CA) models
Quite recently, there has been much interest in studying traffic flow problems within the context
of CA models. This cellular automaton is dynamic models in which space, time, and state are
discrete. The discrete space consists of a grid of cells where each one can be in one of a finite
number of possible states. These states of the cells in grid are updated according to a local rule.
48
That is, the state of a cell at the next time step depends only on its own current state, as well as
the current states of its nearby neighbour. This CA approach follows the macroscopic traffic flow
model where the value of λ increases. All the cells in the grid are updated in parallel, so that the
state of the entire grid advances in discrete time steps. The details of the system, including lane
changing, complex turns and intersection configurations are fully represented and each driver is
given a destination and a preferred path accordingly.
Compared with the other traffic flow models, the CA models are conceptually simpler, and can
be readily implemented on computers. These models capture the complexity of the nonlinear
character of the problem and provide clear physical pictures. For example, CA models show the
existence of a transition between a free-flow phase and a congested phase in the vehicle traffic as
the car density is varied and this in investigated in this thesis (see Chapter 4 and Chapter 5). A
well-known CA model according to (Adolf, 1990; Hoogendoorn & Bovy, 2001), will be
discussed in detail in Chapter 4 in the context of macroscopic traffic flow model.
2.11 Phase transition and order parameters
The behaviour of the vehicles in a traffic jam can be described by this example in (Sugiyama et
al., 2008). Consider a situation when a constraint causing a traffic jam has just been cleared.
Assume there is a queue of five vehicles of zero speed creating a small traffic jam. In the first-
time step, only the lead vehicle can move. The second step, the second vehicle may start, and so
on. In the meantime, it can happen that other vehicles join the queue at the upstream end. Given
the right conditions (fewer vehicles joining at the upstream end than leaving at the downstream
end), this results in a cluster of vehicles of zero speed moving against the traffic direction and so
49
on. The vehicle composition of this cluster is constantly changing from the perspective of an
observer. This represents the wave phenomenon’s that have been described by the theory of
kinematic wave’s equation (2.6).
It is believed that the scenario described above may occur without necessarily in the presence of
obvious constraints but mainly due to fluctuation (noise) in driving speed. These fluctuations
may usually be caused by bumps, curves, lapses of attention due to different engine capabilities.
The amount and types of congestion induced by erratic drivers depend on how much the leading
vehicle car’s speed fluctuates. Moderate fluctuations of an erratically driven car can trigger a
train of density waves, which moves upstream like shock waves. As the average speed decreases,
isolated moving clusters may grow quickly into a nearby continuous jam in (Sugiyama et al.,
2008). This type of traffic jam has been studied extensively using traffic simulation models
(Ichoua et al., 2003; Kerner, 2000a, 2000b). The phase transition phenomenon occurs in
composite systems and is being described an emergent feature because it is observable at the
macroscopic level but not in the microscopic level according to (Adolf, 1990; Sugiyama et al.,
2008). The research findings reported as the phase transition (see Figure 2.5) is theoretically
improved in this thesis using the macroscopic traffic flow model and have been discussed
extensively in Chapter 4 and Chapter 5 respectively.
Figure 2.5 Phase Transition
50
The breaking down of vehicle traffic into jams has properties like phase transition in physical
matter. Given the gas-liquid analogy of Figure 2.4, traffic has two phases and can be in one of
the three states. Free-flow traffic will correspond to the gas phase, jammed traffic will
correspond to the liquid phase and traffic characteristics by stop-go waves will correspond to the
coexistence of liquid and gas. A transition between the two phases occurs at a critical point
according to (Fisher, Fisher, & Huse, 1991). An alternative point of view divides traffic into
three phases where the free-flow and jammed phases are separated by a synchronized flow
phase, and there is two critical points that separate the three phases (see Figure 2.5). This
phenomenon is like the congestion control model developed in this thesis (see Figure 4.11,
Figure 4.12 and Figure 4.13) as explained in chapter 5.
Figure 2.6 reflects the speed-density relationship that results from traffic conditions similar to
ones shown in Figure 2.5 according to (Sugiyama et al., 2008). The left branch (OA) of the
relationship represents free-flow traffic at densities below k1.
Figure 2.6 A reverse λ shape fundamental diagram showing free-flow, congested, and mixed state traffic conditions
51
When the densities are above k2, traffic becomes congested and hindered by traffic jams. This q-
k relationship of the congested traffic is represented by the right branch (AC). Empirical
evidence indicated in (Adolf, 1990; Sugiyama et al., 2008) and traffic simulation suggest that
traffic would also be in a coexistence phase. At densities between k1 and k2, drivers may accept
shorter headways between vehicles. Thus, achieving the maximum flow qm, at the critical
density, k2. High fluctuations of speed in this traffic phase. The dashed line (BC) shows the
relationship between the reverse λ shape and the triangular shape of the fundamental diagram
shown in Figure 3.3.
Although the traffic models described in section 2.5 predict the occurrence of traffic jams and
shock waves, they do not show the breakdown of traffic into three states. The speed-density
relationship in Figure 2.6 can be generated in MATLAB model, by the choice of appropriate
parameters (Adolf, 1990; Fall & Floyd, 1996; Martinez, Toh, Cano, Calafate, & Manzoni, 2011),
which will be discussed in details in Chapter 3
The phase transitions are classified, in physics into first and second order. The first order
transitions are discontinuous. The system changes its state from one phase to the other in a
spontaneous way such that there is an obvious interface between phases at a critical point. The
second order transitions are continuous. There is a gradual change from one phase to another
such that there is no distinct phase at a critical point.
Phase transitions describe system going between a disordered state and an ordered state. Order is
a macro feature that is characterized by an order parameter, which is a macroscopic variable that
52
describes variations between macro-states. The order parameter takes a finite value in the
ordered state and zeroes in the disordered state as investigated in this thesis (see Chapter 4).
The previous example of the road scenario indicates vehicles in a traffic jam are not permanently
stopped, instead of driver’s experience a repeated stop-go sequence. An external observer
monitoring the traffic (say from a helicopter) will notice that there is a fraction of vehicles whose
speed is zero at any given moment. This fraction can be considered an order parameter
characterizing the traffic flow according to (Claffy, Braun, & Polyzos, 1995; Cuesta, Martínez,
Molera, & Sánchez, 1993; Zhao, Lai, Park, & Ye, 2005). In the free-flow state of traffic, all
vehicles are moving and the order parameter is zero (the system is disordered). In traffic jams,
the order parameter is nonzero (the system is ordered state).
The order parameter concept described here is not different from the fraction of stopped vehicles
that was used in (2.14) to determine the average speed. The same concept will be used to analyse
spares network as describe in Chapter 5 to distinguish between free-flow and congested
conditions.
2.12 Traffic simulation models
Traffic simulation models are effective tools for analysing complex problems that cannot be
described easily in analytical terms because of the complex, simultaneous interactions of many
system components. These traffic simulation models can be used for a wide range of applications
according to (Al-Sultan et al., 2014). Almost all traffic simulation models describe dynamic
systems where time is usually the basic independent variable. Continuous simulation models
describe how each element of system change its state continuously over time in response to
53
continuous stimuli. Discrete simulation models represent real-world systems (that are either
continuous or discrete) by updating their states abruptly at any given point in time.
Generally, we have two types of discrete models: discrete time models and discrete event
models. The discrete time models segment time into a sequence of time intervals. Within each
interval, the simulation model computes the activities which change the states of selected system
elements. Discrete event models are effective and economical when the system states change
infrequently. However, for any systems that experience a continuous change in state, such as
vehicle traffic environment and where the model objectives require very detailed descriptions,
then the discrete time model is the better choice due to the frequent change of the system states
and this phenomenon applies to VANETs applications due to its fast topology changes.
Simulation models are usually classified according to the level of detail with which they
represent the systems to be studied according to (Al-Sultan et al., 2014):
2.12.1 The microscopic model
This describes both the systems entities and their interactions at a high level of detail. For
instance, a decision to change lane are determined for each individual vehicle by considering the
distance to its current leader, then its presumed leader and follower in the target lane. The
duration of the lane-change manoeuvre can be calculated.
2.12.2 The mesoscopic model
This generally represents most entities at a high level of detail but describes their activities and
interactions at a much lower level of detail than the microscopic model. In this model, the lane-
change decision could be based on relative lane densities than detailed vehicle interactions.
54
2.12.3 The macroscopic model
This describes entities with their activities as well as their interactions at a low level of detail.
For instance, a traffic stream may usually be represented by scalar values of flow, density, and
speed. Lane changes in this model may not be represented at all in this context. Instead, the
model ensures that the traffic stream is appropriately allocated to each lane which is the main
direction of this study. This macroscopic model has been investigated in the context of a
congested network and sparse network scenario to improve ITS in VANETs applications.
Microscopic models are known to be costly to develop, execute (i.e. carry-out) and to maintain,
relative to the other models but they possess the potential to be more accurate and precise.
Macroscopic models may be less accurate in their representation of the real-world system but
they are appropriate when the intended results are not sensitive to details, when the scale of
application is too big to be accommodated by microscopic models or when resources are limited.
Another classification addresses the processes represented by the model as either deterministic or
stochastic. Deterministic models define all entities and their interactions by the exact
relationships (mathematical, statistical or logical). Stochastic models usually have processes that
include probability functions. A car-following model, for example can be formulated either by a
deterministic or stochastic relationship by defining the driver’s reaction time as a constant value
or as a random variable, respectively.
The suitable simulation model to be used in this thesis is the macroscopic models and this will be
discussed in detail in Chapter 3. MATLAB, is the traffic simulator to simulate the models in both
dense and sparse networks scenarios.
55
2.13 Summary and conclusions
In this chapter, the characteristics of VANETs has discussed along with the features of MANETs
implementation since VANETs is a subset of MANETs. This is done with emphasizes on the
physical layer and the MAC layer with regards to VANETs Characteristics in general. Further
discusses challenges associated with Intelligent Transportation Systems and vehicular mobility.
The study also focused on the overview of VANETs applications such as V2I connectivity and
V2R connectivity and finally narrow down to the research directions in VANETs to improve
safety alert information dissemination in a timely fashion under sparse and dense networks
environments.
This chapter presented a general overview of the state-of-the-art wireless solutions to vehicle-
to-sensor, vehicle-to-vehicle, vehicle-to-internet and vehicle-to-road infrastructure connectivity
in VANETs. The study discussed the potential challenges and identified the space for future
improvement. The biggest challenge for efficient and robust wireless connections i n
V A N E T s is to combat the harsh communication environment inside and/or outside the
vehicle. In addition, significant research and development efforts are required to deal with the
following issues.
i. Inter-vehicle systems have a stringent requirement on latency and reliability for traffic
alert message dissemination in VANETs. The architecture such as V2I and V2R lead
to a high cost and thereby impede the development of connected vehicles VANETs.
ii. A novel solution to provide V2X communication in VANETs with low cost is the
direction of this study. In view of the above information, the best alternative
methodology is the vehicle to vehicle (V2V) communication which is the most cost-
effective approach once suitable solutions is developed.
56
This chapter recognizes the challenges that impede the intelligent transportation systems in
VANETs and provide background information on the existing approaches and the need to use the
macroscopic traffic flow model to perform analytical investigation under dense and sparse
networks to fix the gap investigated in the literature.
The next section examines the performance analysis of vehicular traffic flow theory. This covers
a wide range of existing traffic flow model developed to ascertain their merits and demerits with
regards to vehicular ad hoc networks (VANETs). In this context, certain important factors that
inhibit vehicular traffic flow is the parameters used in developing VANETs Models. These
parameters include flow, speed and density, to perform simulation on the node mobility, and the
sensitivity factor called λ. The choice of a model depends upon how well the parameters have
been selected. It is important to know which parameter should be made constant (i.e. fixed) and
which should be made varied (i.e. varying) for a model.
The vehicle traffic is characterized by three parameters: flow, density and speed. Although, there
is a simple, ‘fundamental relationship’ that relates the three parameters show a bivariate
relationship. This relationship among them has been the subject of intensive research for the past
decades. The flow-density relationship is called the Fundamental Diagram of Road Traffic (see
Figure 2.6) and it shows that traffic flow increases as the density increases until it reaches a
maximum value, known as road capacity, then the flow starts to decline until it reaches zero at
maximum density, called jam density. The density at which flow reaches capacity separate free-
flow traffic from congested traffic in which vehicle speed is influenced by vehicle density.
Several theories are proposed to define the relationship between traffic parameters and explain
these observations. However, no single theory provides a comprehensive relationship between

57
traffic parameters in VANETs. The focus of this thesis is to conduct the analytical investigation
into this bivariate relationship in traffic flow and density as the parameter λ increases with time
and this has been investigated in this thesis.
Similarities between traffic flow and fluid gave rise to continuum models, which discusses the
relationship between traffic flow and density. The models determine the speed of traffic waves
caused by congestions as bottlenecks. The models also explain discontinuities in the flow-density
relationship because of shock waves which occur due to abrupt changes in speed. A car-
following model provides a relationship between vehicles’ speed and density based on stimulus-
response models. These models apply to single-lane, congested traffic, but not to light traffic
where vehicles tend to move freely unaffected by other vehicles. Two-fluid theory predicts the
relationship between the trip time of a test and the average speed of vehicles on the road. The
previous models predict that vehicle traffic remains in free-flow state until density reaches a
certain point and beyond which sufficiently large fluctuations will cause traffic jams.
Complexities of vehicle traffic prevent the use of mathematical models in solving all but simple
problems. Studies, that involve complex traffic scenarios must resort to traffic simulations and
this is the direction of the thesis. Traffic simulators are classified according to the details in
mimicking real-world traffic in microscopic, mesoscopic and macroscopic. Microscopic traffic
simulators, micro-simulators, are highly accurate because they reproduce detailed interactions
between vehicles. The CA models are used to build micro-simulators, due to their easy
adaptation by computer systems
In MANETs, analytical studies of connectivity in one-dimension usually assume that nodes are
dispersed according to a uniform (or Poisson) distribution to derive a relationship between

58
connectivity and node density. On several occasions, the assumption of uniform distribution is
extended to vehicular networks due to their one-dimensional nature. While this assumption may
be valid for traffic in homogenous states such as in free-flow traffic, it cannot be used to describe
non-homogenous traffic in the congested traffic that exhibit stop-go traffic wave and traffic jams.
The latter case is important considering that phase transition may occur anywhere there is a
constant or bottleneck in traffic or even because of high fluctuations in driving behaviour. These
challenges have been addressed with the novel approach in this thesis using the macroscopic
traffic flow model at varied nodes densities and perform various simulations in chapter 3.
The primary objective of this investigation is to identify and outline the performance limitations
of this vehicular traffic flow models to propose a novel alternative methodology to address these
limitations in ITS applications in VANETs. Chapters 3, proposes an efficient methodology to
overcome these limitations.
59
Chapter 3 Methodology
3.1 Introduction
As has been indicated in Chapter 2, several approaches have been adopted to investigate the
problems of traffic flow according to (Adolf, 1990; Kerner, 2000b; Korkmaz, Ekici, & Özgüner,
2010; May, 1990). However, some of these challenges remain and therefore require an intensive
research to improve ITS applications. Chapter 2 serve as a framework to understand the
challenges to help perform an analytical investigation using the macroscopic traffic flow model
established in (May, 1990). To ensure a realistic result, traffic simulations are used to perform
analytical analysis to simplify the traffic problems. There is no single theory that can explain all
the complexities of traffic flow. Instead, there are several models that try to explain the
relationships between traffic parameters but these challenges still lingered on and require further
research effort. The direction of this thesis is to investigate these challenges associated with
traffic flow to mitigate congestion and sparsely connected networks to improve ITS in VANETs.
In this Chapter, a macroscopic traffic model in (May, 1990) is used to the carry-out analytical
investigation to improve congestion control in dense networks and connectivity problems in
sparse networks. This investigation is done through simulation by making certain parameters
constant while varying other parameters to model the traffic flow phenomenon for a realistic
result. Although, several traffic flow models were developed in the past, this model is the first to
be investigated in this context to improve reachability and scalability.
60
The next chapter is organized as follows: Section 3.2 describes the proposed macroscopic traffic
flow model adopted in (Adolf, 1990; Daganzo, 2011) to perform the analytical investigation in
this thesis. Section 3.3 analyses the factors to be considered in the selection of parameters for the
model. Section 3.4 explains the step used to perform simulations in this thesis. Finally, Section
3.5 outlines the density for traffic flow conditions and system parameters while section 3.6
concludes the chapter.
3.2 Macroscopic traffic flow characteristics
The traffic flow represents the traffic density associated with the transportation system and the
interaction between these densities determines the operational performance of the system. It is
important to know the flow rates, their spatio-temporal variations and the nature of the traffic
stream. The measurement of traffic has many uses in planning, designing and the operations of
highways facilities
The flow rate (volume) is the important macroscopic flow characteristics and is defined as the
number of vehicles passing a point in each period and usually expressed as an hourly flow rate.
3.2.1The macroscopic traffic flow models
The vehicles under low traffic density and their interaction between vehicles are very low and
therefore experiences free traffic flow movement. This free flow condition results in realistic
independent movement among vehicles without any restriction on the roads but in a dense
network the traffic situation is stop-and-go movement. The computational model developed in
(Adolf, 1990; Daganzo, 2011; Kerner, 2000b; May, 1990), indicates that the arrival process of
vehicles has a highway density. This can be modelled as a Poisson process with an arrival rate t .
61
Therefore, the probability that n vehicles arriving in a time t can be computed using the
equations shown in (3.1) and (3.3).
 xt n tx

P( X  n)  e
n! (3. 1)
 , represents vehicles migrating to a highway in a time interval x. The time headway
distribution is described in (Adolf, 1990; Cowan, 1975), as an exponential distribution of this
1
parameter   with the probability density function (PDF) fT (t )  tet  , t  0 . The PDF
 t 
allows the time headway to be zero and this gives it the highest relative likelihood of occurrence
which is not possible to be zero by the set conditions. For a realistic result to be achieved, it is
important to increment the minimum allowable time headway. This PDF is given by
fT (t )   te t ( t  ), t 
0, t  (3. 2)
V
D
 (3. 3)
where V in equation (3.3) is the traffic volume measured in t h e vehicle per hour
(veh/hr) and D , density is measured in the vehicle per mile per lane in (veh/mile/lane),
62
and  represents the space mean speed in mph. The traffic flow conditions corresponding
to each density level are classified in Table 3.1. in (May, 1990)
3.3 Parameter selection for new traffic flow models
Vehicular mobility models and traffic simulators have become popular for use in VANETs
studies according to (Greenshields, Channing, & Miller, 1935). The main reasons for these
simulators are for traffic analysis, forecasting and planning transportation networks. Due to the
important applications in VANETs, the ITS initiative has generated an enormous number of
vehicular traffic simulators since its inceptions in the United State in 1991 according to (Boxill
& Yu, 2000). These simulators are intended to support ITS applications, but not necessarily
VANETs due to their recent introduction.
In VANETs research papers, simulation for vehicle movement is usually considered a special
type mobility model according to (Tian, Hahner, Becker, Stepanov, & Rothermel, 2002; Tian,
Han, & Rothermel, 2003). These mobility models cannot produce complex vehicle manoeuvres
while car-following, decelerating, and accelerating, during traffic jam situations. For the analysis
of any complex phenomenon like VANETs, many studies have relied on vehicular mobility
models to generate vehicle movement traces.
3.3.1 The mobility models
Many studies in ad hoc networks often require keeping track of node positions at any given time
and vehicular networks are no exception. In recent time, many VANETs applications depend on
63
realistic vehicle movement patterns and this can be obtained from traffic micro-controllers which
can track the characteristics of individual vehicles including the position, speed and direction.
The early research in inter-vehicle communication relied on micro-simulators that were designed
for the purpose of investigating vehicle traffic flow in transportation systems according to (Eze et
al., 2016). VANETs research adapts simulator to generate traces of vehicles position at regular
time interval into network simulator such as ns-2, ns-3, MATLAB according to (Eze et al., 2016;
Härri, Filali, Bonnet, & Fiore, 2006; Hartenstein & Laberteaux, 2008; Naumov, Baumann, &
Gross, 2006; Zeadally et al., 2012) to perform simulation of communication protocols. In recent
times, researchers have devoted substantiate effort to the development of network traffic
simulators for VANETs research investigated in (Eze et al., 2016; Härri et al., 2006). Usually,
these simulators have a dual focus on the communication and mobility of VANETs.
3.3.2 The traffic flow models
The traffic simulators are extensively used in evaluating vehicular ad hoc networks. The
analytical approaches in free-flow traffic conditions are usually the preferred choice. The
primary reason for this choice is the control environments it offers, ability to select at least two
or three parameters to establish the fundamental relationship of the traffic flow. Under free-flow
conditions, according to (Harri et al., 2009; Michael Rudack, Meincke, Jobmann, & Lott, 2003;
M Rudack, Meincke, & Lott, 2002; Weil, Wootton, & Garcia-Ortiz, 1998), it is assumed that
the traffic flow and the vehicle velocity are independent. This fundamental relationship of traffic
flow is used to derive probability density function (pdf) of the communication duration between
vehicles and the pdf of the time between consecutive topology changes. Similar assumptions are
made in (Ebner & Rohling, 2001) to derive the pdf for distance at which to establish a multi-hop
64
connection between two vehicles. The effective selection of time headway (i.e. the time gap
between consecutive vehicles) as an input parameter control both the rate by which vehicles are
injected to the simulation environment and the density of the vehicles. According to (Adachi,
Morita, Fujimura, Takatori, & Hasegawa, 2002), the time headway values are generated
randomly from a Poisson distribution with an average value that ensures the free-flow traffic. A
uniform distribution with a minimum value that specifies the initial separation between vehicles
is used in (Ebner & Rohling, 2001).
3.3.3 The vehicle speed
In most cases, the desired speed is provided as an input to the simulations. The vehicle speed can
be a constant value for each type of vehicles (e.g. cars and trucks) or may be sampled from a
random distribution. In (Füßler, Mauve, Hartenstein, Käsemann, & Vollmer, 2002), the speed
distribution is chosen to correspond to real measurements from German highways, while in
(Korkmaz et al., 2010; Korkmaz, Ekici, Özgüner, & Özgüner, 2004), the speed is assigned from
a normal distribution and this is the direction of this thesis.
One possible choice of random vehicle speed is based on the theory of the “85th percentile
speed” that is usually used as a guide to set limit on the highways as reported in (Haglund, 2001;
Johnson & Pawar, 2005; Milazzo II, Rouphail, Hummer, & Allen, 1999; Setting & Limits,
1998). The 85th percentile speed is defined as the speed 85% of drivers moving at or below. It is
evident that 75% of the drivers maintain speed within a certain range around the average speed
limit. Another research finding indicates that 15% are slow drivers while the remaining 15% are
driving at high speeds. According to (Z. D. Chen, Kung, & Vlah, 2001; M Rudack et al., 2002),
vehicle speed is sampled from a Normal distribution using the standard deviation to mark the
65
boundaries between slow, average and high speeds. Although, the speed and time headway
changes in real traffic, they are kept constant in (Cowan, 1975; Duan, Li, & Salvendy, 2013;
Kanagaraj, Asaithambi, Kumar, Srinivasan, & Sivanandan, 2013; Naus et al., 2008).
Furthermore, the choice of the average speed varies according to the choice of the environment
under considerations. For instance, in highway simulations environment the average speed is set
to 100-130km/hr according to (Füßler et al., 2002; Michael Rudack et al., 2003). Similarly, the
simulations of city street grid set the average speed to 40-80km/hr either randomly according to
(Korkmaz et al., 2004) or according to the street types (Mangharam, 2008). The choice of speed
may reflect also the posted speed limit, road conditions and common driving habits.
3.3.4 The collisions
The collision in VANETs has usually assumed in some literature that vehicles move freely and
collisions are not allowed. However, the vehicle density according to (Adachi et al., 2002) is
specified to create congested traffic situations and vehicles may collide if drivers fail to react in
time. This finding is very necessary to evaluate the influence of the V2V and V2R
communication on safety according to (Adachi et al., 2002).
3.3.5 The road topology
This may be described as a common choice of highway geometry of two or more lanes in each
direction. This allows the traffic to flow in both directions may be described as asymmetric as
indicated in (Michael Rudack et al., 2003) . According to (Al-Sultan et al., 2014), they argue
that dividing the highways may be used to limit the geographic area affected by message
66
broadcast to one traffic direction. According to (Choffnes & Bustamante, 2005; Lochert et al.,
2003; Mangharam, 2008), explained more complex geometry to be used to evaluate routing
strategies and protocols in a city traffic flow environment.
3.3.6 The vehicular mobility in sparse networks
Many researchers are highly interested in studying sparse vehicular networks due to its
importance in ITS applications in VANETs. This motivation from most researchers reflects the
assumptions inherent in the inter-vehicle communication in sparse networks as more challenging
due to the difficulty in maintaining connectivity. In addition, the low-density conditions offer
more flexibility in setting the traffic parameters such as flow, speed and density for analytical
and simulation studies. The assumption of independent parameter is valid only in free-flow
traffic. Therefore, the results of these studies cannot be used directly in dense traffic
environments. This study seeks to argue that research should focus more on dense traffic and
traffic jam environments due to their important significant influence on VANETs, since:
i. The vehicle density can contribute to rapid message dissemination
(Ferreira et al., 2010; Härri et al., 2006; Raya, Aziz, & Hubaux, 2006;
Tonguz, Wisitpongphan, Bait, Mudaliget, & Sadekart, 2007) .
ii. Higher node density aggravates the contention and collision problem in
the communication protocols based on shared media.
iii. The traffic jams contribute to certain non-homogeneous distribution of
vehicles and therefore increasing the range of transmission needed to
maintain connectivity.
67
iv. The vehicle traffic may change rapidly between free-flow in congested
traffic as the result of the traffic controls, road constraints or accidents.
The research directions related to the use of vehicle mobility models and traffic simulator in this
thesis is provided in the next section.
3.4 Steps in developing the simulation models
The first step is the problem identification and how to formulate the objectives explicitly with
alternative analytical techniques to evaluate the subsequent steps. At this stage, a lot of questions
need to be answered well for the problem to be solved. What are the desired inputs to arrive at
the outputs? What is the boundary, timelines and limitation of the problem? Are the parameters
needed to be modelled individually? In this context, when the question is answered successfully
then the next step needs to continue. The steps are based on (see Figure 3.1).
The second step need to answer the question explicitly: Is simulation the appropriate analytical
techniques for the problem? Several more questions need to be addressed such as: How could the
problem be solved without simulation? Why is simulation a better solution to the problem? Are
the needed resources available to perform the experiment? Can the problem really be solved? In
this thesis simulation is the best choice to carry-out the macroscopic traffic model according to
(Eze et al., 2016; Härri et al., 2006; Hartenstein & Laberteaux, 2008; Naumov et al., 2006;
Zeadally et al., 2012). One major reason for selecting MATLAB in this thesis is to perform
simulation on the macroscopic traffic flow models, due to easy-to-use environment in which
problems and solutions are expressed in familiar mathematical notation. Besides, MATLAB is
68
one of the network simulators that have a significant number of users in both academia and
industry.
The third step is the formulation of the problem. This is done by developing a flowchart of the
model beginning with an input, process and an output variable with emphasize on an input and
an output requirement. This thesis is to perform the analytical investigation to gain understanding
of the macroscopic traffic flow model. The input requirements include design elements, traffic
demand pattern, operational rules and control conditions.
The fourth steps are the formulation of the macroscopic traffic model. This is done by
developing an intermediate flowchart to understand the logic behind the model. This stage is
iterative in nature until the desired model is accomplished.
The fifth step consists of estimation all the required parameters in the model. To accomplish this,
some of the parameters need to be deterministic while other need to probabilistic (stochastic).
The deterministic parameters may be constant for all situations and may take on one of a set of
constant values depending on the problem at hand or this may vary over the range of continuous
manner when necessary. The probabilistic parameters require the variations of variables to meet
the set objectives.
The six stages explain the evaluation of the current state of the models. In this process judgment
and decisions are required by adding, changing and deleting variables in the model. This stage is
done until the desired solution to the problem is achieved before the process can continue to the
next stage to formulate the computer code using MATLAB to perform various simulations to
ascertain the desire solution through validation of codes.
69
3.4.1 Flowchart for the simulation models
Figure 3.1 Flowchart showing steps used to develop the simulation model
70
3.5 The traffic flow state for different densities and level of services
These traffic flow parameters were generally used to investigate traffic flow in civil engineering
according to (May, 1990). This table is used to investigate the flow conditions of uncongested
flow conditions and the congested flow conditions using the level of services with regards to the
densities (see Table 3.1).
Density Level of
Service Flow Conditions
(Veh/Mile/Lan
e)
0-12 A Free-flow operations
12-20 B Reasonable free-flow operations Uncongested
flow conditions
20-30 C Stable operations
30-42 D Borders on unstable operations
42-67 E Extremely unstable flow Near-capacity

operations flow conditions
67-100 F Forced or breakdown operations Congested flow
conditions
>100 Incident situation operations
Table 3.1 Traffic flow state for different densities
Level Service Speed Density

of Flow v/c limit Volume(veh/hr) (mph) (veh/mile/lane)
service condition
A Free 0.35 700 60 12

B Stable 0.54 1100 57 20
C Stable 0.77 1550 54 30
D High Density 0.93 1850 46 40
E Near Capacity 1.00 2000 30 67
F Breakdown Unstable Unstable <30 >67
Table 3.2 Levels of service for freeway under ideal conditions and 70mpr design speed
71
3.5.1 System parameters for dense networks simulations
This study modifies the known civil engineering traffic flow models in Four-Lane Mobility in
VANETs within a congested network. This study demonstrates how the network behaves under
varying traffic densities which involve 10Miles veh/mile/lane to 100 vehicles per each lane
(veh/mile/lane) within 36Miles road section to investigate network behaviour within four lanes
analytically.
To really understand the interactions among two vehicles in terms of traffic flow density. This
study seeks to vary the value of λ from 1 to 10 against 30 veh/mile /lane to investigate the
network behaves under the various λ in a congested network as shown (see Table 3.3).
Item Parameters Descriptions

Number
1 Platform MATLAB
2 Speed of Vehicles 60km/hr
3 No. of Vehicles on the Road 1800
4 Type of Road (Capacity) Highways, Urban, Semi-Urban
5 Number of Lanes in Each Direction 4
6 Average Percentage of Vehicles 30%
7 Expected Threshold Changes 15%
8 Speed Limit 36 mile/hr
Table 3.3 Dense Networks System Parameters
3.5.2 System parameters for sparse networks simulations
This explains the traffic flow models in Four-Lane Mobility in VANETs within a sparse
network. This study demonstrates how the network behaves under varying traffic densities which
72
involve 10Miles veh/mile/lane to 30 vehicles per each lane (veh/mile/lane) within 60Miles road
section to investigate network behaviour with four lanes analytically.
This explains the interactions among two vehicles in terms of traffic flow density. This study
seeks to vary the value of λ from 1 to 10 against 30 veh/mile /lane to investigate the network
behaves under the various λ in a sparse network as shown in (see Table 3.4).
Item Number Parameters Descriptions
1 Platform MATLAB
2 Speed of Vehicles 54 - 60m/hr
3 No. of Vehicles on the Road 700 -1550
4 Type of Road (Capacity) Highways, Urban, Semi-
Urban
5 Number of Lanes in Each 4
Direction
6 Average Percentage of Vehicles 15%
7 Expected Threshold Changes 10%
8 Speed Limit 60 mile/hr
Table 3.4 Sparse Networks System Parameters
3.6 Conclusions
This chapter used a well-known macroscopic traffic flow model developed in (May, 1990) and
presented a novel approach to model traffic flow in vehicular ad hoc networks (VANETs)
through simulation. This approach utilizes the challenges in traffic flow models to improve
reachability and scalability through analytical investigation to mitigate congestion in dense
networks and mitigate the challenges associated with sparsely connected networks under four-
lane road network (i.e. off-peak and peak periods).
73
This study performs several simulations with the macroscopic traffic flow models to investigate
the spatio-temporal variations in the dense network and sparse network associated in VANETs.
The analytical investigation and performance analysis have been conducted under different
traffic densities. In the first place, the study outlines the simulation set-up with rigor in a
flowchart to examine each step iteratively until the model is fully developed (see Figure 3.1). In
the second place, the impact of traffic density on the performance of the traffic flow model is
assessed by varying the number of vehicles (nodes) under fixed traffic flow conditions (see Table
3.1). In the third place, the sensitivity factor of two-vehicle interaction in the car-following
model λ, is varied from 1 to 10 under dense network and sparse network conditions to assess the
performance of the traffic flow model to measure reachability and scalability. In the fourth place,
the simulation results were assessed to developed congestion control model in a dense network
and Honeycomb model in sparsely connected networks to be used to mitigate congestion in the
dense network and sparsely connected network to improve (ITS) in VANETs.
74
Chapter 4 Improving VANETs congestion in dense networks & intermittent
connectivity in sparse networks
4.1 Introduction
Due to inter-vehicle communication, vehicle speed, driver behaviour and connectivity have
become a very important performance indicator in VANETs. The challenges associated with
regards to reliable connectivity, fast communication among transmitting and the receiving nodes
have become an important research challenge in recent times. Two nodes are connected if they
can establish a reliable communication between each node either through a single hop or
multiple hops. In recent times, several research studies have investigated into VANETs
connectivity problems both in sparse and dense networks according to (L. Li & Chen, 2017;
Reis, Sargento, & Tonguz, 2011). Connectivity in VANETs is primarily depended on the inter-
vehicle time headway being the time interval between the lead vehicles and the successive
vehicles on the road.
The time headway depends on vehicle speed, traffic flow, driver behaviour and so on. Under
normal traffic flow condition, three states of classification are associated with vehicular mobility:
free flow, synchronized flow, and wide moving jam according to (Daganzo, 2011; Nagel &
Herrmann, 1993). During free flow state, traffic density is low and vehicles on the road have the
tendency to move at a high speed. This is associated with intermittent connectivity since some of
the vehicles move out of the acceptable transmission range in VANETs. During the synchronized
state, traffic flow reaches its maximum and the vehicles in different lanes travel with a
synchronized speed where stop-and-go movement of vehicles may occur steadily. In
75
synchronized and wide moving jam occur under a congested traffic flow (Adolf, 1990; May,
1990; Nagel & Herrmann, 1993), where vehicles experience heavy congestions due to a high
density of vehicles on the roads. In vehicular traffic flow condition, there exists a correlation
between traffic flow and traffic density during a free flow condition and up to critical density. In
congested traffic flow, the headway is very low and vehicles are within the connected
transmission rage.
The rest of the chapters are organized as follows. Section 4.2 describes the performance metrics
in dense networks and performs the analytical investigation to simulate the macroscopic traffic
flow model in detail. Section 4.3 proposed a congestion control model based on the simulation
results in dense networks. Section 4.4 presents how to improve connectivity in sparse networks
through analytical investigation. Section 4.5 proposed a Honeycomb Model in sparsely
connected networks to improve connectivity. Finally, section 4.6 concludes this chapter.
4.2 Performance metrics under dense network conditions
This λ, is an important parameter used to model the behaviour and the interaction of vehicles at
different traffic densities. The study used 0veh/mile/lane to 100veh/mile/lane and varies the λ from
1 to 10 to investigate the effects of λ on the interaction of vehicles within the four-lane according to
(Azameti, Katsriku, Chong, & Abdulai, 2018)
4.2.1 The four-lane traffic flow under dense network conditions when λ =1
The network behaviour with respect to the lane changes is shown in the figure below If λ = 1, the
traffic flow trajectories within the four-lane exhibits exponential distribution where lane 1: has the
76
least congestion with less priority in terms of packet reception. The next less priority occurs in lane
2 but has an increased number of vehicles than in lane 1. Lane 3, experiences a high number of
vehicles while lane 4 shows a higher number of vehicles and therefore receives the highest priority
reception rate to control the congestion. Lane 3, is the second higher lane with the next higher
priority reception rate when lane 4, is reduced to the acceptable threshold as shown in Figure 4.1a
and 4.1b. The effect also indicates the correlation between traffic density and speed (see Figure 2.4)
and (see Figure 2.5), showing the fundamental diagram on free flow, congested and mixed state
traffic conditions as explicitly indicated in chapter 2. These relationships are the results in this thesis
and have been explained in this section as the value of the λ increases. These effects are further
explained under vehicle trajectories when the value of λ increases the density also increases.
Figure 4.1a the effect of traffic flow when λ = 1
77
Figure 4.1b Vehicle trajectories in each lane
λ = 2, the traffic flow within the lanes also indicate an exponential distribution at lane 1 and lane 2
and therefore have the least priority while lane 2 having a fewer number of vehicles than in lane 1.
Lane 3 and lane 4 begin to show the partial normal distribution in Figure 4.2a and 4.2b. Lane 4, has
the highest priority in terms of packet reception and lane 3 has the next higher priority after lane 4
has successfully received the needed reception. It also explains the relationship between traffic
density and speed.
78
Figure 4.2a The effect of traffic flow when λ = 2
79
Again, if λ = 3, the traffic flow within the four-lane shows an exponential distribution at lane 1 and
indicates a partial normal distribution from lane 3 and lane 4 in Figure 4.3a and 4.3b. Since lane 3
and lane 4 with partial normal distribution indicate that there is an increased number of vehicles but
lane 4 having the highest number of vehicles than in lane 3. Therefore, lane 4 is congested with the
highest priority reception rate and the next priority is lane 3. When λ = 3 shows the initial threshold
in which lanes continue to experience an increased number of vehicles anytime the value of λ is
increased.
80
When λ = 4, the traffic flow shows a complete normal distribution at lane 3 and lane 4 with an
increased number of vehicles as indicated in Figure 4.4a and 4.4b. At this stage, lane 3 and lane 4
are congested but lane 4 has the highest number of vehicles and therefore need to receive the
highest priority of reception rate before lane 3. Lane 1, is still the least but with an increased
number of vehicles once the value λ, increases. Lane 1, has high mobility than in lane 2 since the
number of vehicles in lane 2 is higher than in lane 1. Therefore, lane 1 has the highest mobility rate.
81
82
Subsequently, when λ = 5, the traffic flow exhibits a complete normal distribution at lane 2, lane 3
and lane 4. Lane 1, has the least number of vehicles and therefore has the highest mobility rate than
in lane 2. Lane 3 and lane 4 still have the highest number of vehicles, since both lanes become
congested with lane 4 having the highest number of vehicles indicating a high priority in terms of
packet reception and move subsequently to lane 3. This phenomenon also demonstrates the
relationship between traffic density and speed as shown in figure 4.5a and 4.5b.
83
In addition, when λ = 6, the traffic flow also shows a complete normal distribution at lane 2, lane 3
and lane 4. At this stage lane, 3 and lane 4 have an increased number of vehicles and therefore
congested but lane 4 is congested than lane 3. Lane 4, seems to experience higher congestion than
lane 3. Lane 1, still has the least number of vehicles with a reasonable amount of mobility rate than
in lane 2, lane 3 and lane 4. Therefore, establishes a relationship between density and speed in
figure 4.6a and 4.6b.
84
85
If λ = 7, the traffic flow shows a complete normal distribution at lane 2, lane 3 and lane 4. Lane 4
has the highest priority in terms of packet reception and move subsequently to next lane 3. At this
instance, is worth to note that anytime the value of λ increases then all the four-lane also seen an
appreciable increase indicating a direct proportion in each lane. This shows a correlation between
traffic density and speed as shown in figure 4.7a and 4.7b.
86
4.2.8 The four-lane traffic flow under dense network conditions when λ = 8, 9 and 10
When λ = 8, 9 and 10, the traffic flow shows a complete normal distribution for the entire four-lane.
Lane 4 has the highest priority in terms of packet reception and move subsequently to next lane 3
but with an increased number of vehicles. This trend shows that once all the lanes assumed a
complete normal distribution it would correlate with an increased number of vehicles but with lane
3 and lane 4 term a congested network. In this instance, lane 4 has the highest packet reception next
to lane 3, lane 2 and lane 1 in that order of magnitude in terms of traffic density as shown in figure
4.8a 5.8b, 4.9a and 4.9b and 4.10a and 4.10b. In these cases, the graph shows a definite relationship
between density and speed.
87
88
89
90
4.3 The congestion control models using four-lane networks
This study used Four-Lane Mobility in VANETs, within a congested network which is similar (see
Figure 2.5) with increased traffic density as the value of λ from 1 to 10. These effects are shown in
(see Figure 4.11 and Figure 4.12) in this thesis and the combined effects of the congestion control
model (see Figure 4.13) which is like Phase transition (see Figure 2.5), showing the level of
transitions on each lane according to (Adolf, 1990; Sugiyama et al., 2008). This transition is
demonstrated in the congestion control models with the effects of the value of parameter λ (Figure
4.11, 4.12 and figure 4.13) as outlined below.
Figure 4.11 Four-Lane at the right side of the road showing the direction and movement of vehicle
91
Figure 4.12 Four-Lane at the left side of the road showing the opposite direction and movement of vehicles
Figure 4.13 Two-way Congestion Control Model showing different threshold of λ
92
4.4 Improving connectivity in sparse networks
This thesis analyse the connectivity challenges in sparse networks under free-flow condition
according to (May, 1990) and other related studies challenges have been discussed in chapter 2
extensively. Chapter 3 discussed how the challenges are addressed to improve VANETs
problems in sparse network and dense networks.
In free flow condition, extensive studies have been reported concerning the connectivity
challenges of mobile ad-hoc networks in different context in (Kloiber et al., 2012). The
connectivity of mobile ad-hoc networks primarily depends on the pattern of mobility and the
channel conditions. In most studies, on ad-hoc networks were investigated with predefined node
placement and static transmission range to examine the network phenomenon associated with
connectivity. This phenomenon is then analysed according to the node density, speed,
transmission range etc.
According to (Adolf, 1990; Bettstetter et al., 2004) the connectivity probability density of one
dimensional network is analysed with uniform node distribution and is extended to multiple
dimensions to find the bounds of connectivity through extensive simulation. The authors
(Kleijnen, 2005; Ros et al., 2014), derived a critical range of nodes for connectivity with
probability of one and considered as infinity. This type of node distribution is not applicable in
vehicular network due to high dynamic fast topology changes. However, vehicular movement of
nodes are considered in (Zeadally et al., 2012). In (Adolf, 1990; May, 1990), the author relies on
some road statistics and numerical solution for the connectivity distance, platoon size derived
based on queuing theoretical views. According to author (Adolf, 1990; Schumacher et al.,
2009), real road scenario is created through simulation to determine the minimum transmission
93
range to arrive at full network connectivity and the effect of vehicle density on the transmission
range is investigated in thesis.
According to (Choffnes & Bustamante, 2005), the enhancement of connectivity is done by
considering both direct connectivity from a source node as well as the indirect connectivity
through multi hop forwarding to ensure the available connectivity. According to (May, 1990)
(Adolf, 1990) , the link lifetime from the PDF of relative velocity of consecutive vehicles and
PDF of vehicle headway distance were derived. The link lifetime corresponding to multi-hop
connectivity in two-lane road with opposing vehicle direction is also analysed. The rest of the
chapter is organized as follows: Section 4.2 describes in detail the simulation outcome
associated with sparse networks. Section 4.3 analyse the connectivity problems associated with
sparse network in detail. Finally, section 4.4 concludes the chapter.
4.4.1 Performance metrics in sparse network conditions
In lane 1,2,3 and 4, when λ=1, 2, 3, 4 and 8; the smaller standard deviation than the mean
indicates that certain number of the vehicles on the roads are closed or clustered around the mean
value. This indicates reasonable distribution among few numbers of vehicle on the road, with
respect to high speed and distances. This separates vehicles beyond the acceptable transmission
range (i.e. 1000m). The vehicle connectivity range for effective communication, When λ=1; lane
1: 79.5%.; lane 2: 79.9%; lane 3: 81.1% and lane 4: 82.3%, when λ=2; lane 1: 70.8%.; lane 2:
78.4%; lane 3: 81.9% and lane 4: 83.9%, when λ=3; lane 1: 75.5%.; lane 2: 81.8%; lane 3: 84.7%
and lane 4: 86.5%, when λ=4; lane 1: 78.4%.; lane 2: 83.9%; lane 3: 86.5% and lane 4: 87.9%
and finally when λ=8; lane 1: 83.9%.; lane 2: 87.9%; lane 3: 89.8% and lane 4: 90.9%. When
λ=1: all the four lanes, shows an exponential distribution when λ=2: lane 1 and 2, show an
94
exponential distribution while lane 3 and 4 begin to show normal distribution but skewed to the
right. When λ=3 in lane 1, shows an exponential distribution while lane 2,3 and 4 indicate a
normal distribution but skewed to the right at varying degree of skewness. When λ=4: lane 1,
shows an exponential distribution, in lane 2, 3 and 4; indicates a normal distribution but skewed
to the right at varying degree of skewness.
In lane 1,2 and 3, when λ=5, the smaller standard deviation than the mean indicates that certain
number of the vehicles on the roads are closed or clustered around the mean value. This indicates
fair distribution among few numbers of vehicle on the road, with respect to high speed and
distances. In lane 4; the vehicles are spread-out at varying degree when the standard deviation is
greater than the mean value. On the other hand, vehicles in lane 4; are much further away from
each other due to fewer vehicles on the road with very high speed and high distances within each
neighbour. The vehicles in lane 1, 2 and 3, are at reasonable distances from each vehicle but
beyond the acceptable transmission range (i.e. 1000m). The vehicle connectivity range for
effective communication, When λ=5, lane 1: 80.3%.; lane 2: 85.4%; lane 3: 97.3% and lane 4:
88.0%. When λ=5: lane 1 begin to show a normal distribution but skewed to the right while lane
2, indicates a normal distribution but also skewed to right. Lane 3 and 4, indicate a complete
normal distribution but lane 3 slightly skewed to right while lane 4 also slightly skewed to left.
95
Figure 4.14 The effect of traffic flow when λ =1
96
97
4.5 The proposed Honeycomb Models under sparse network conditions
This study introduces a Honeycomb Model (HCM) to improve sparsely connected network using
the idea on how honeybees form honeycomb and in search of food. This idea is based on the
orientation of the waggling dance of the honeybees. This study has found out that the angel of
rotation precisely matches the angle between the feeding station and the hive (honeycomb). This
clue demonstrates how the bees can share information about the location of food. The honey bee
form clusters and one cluster of A and B may accurately share important information by
communicating among the two clusters. This phenomenon can be replicated in VANETs traffic
information dissemination in sparsely connected networks to enable vehicles to share traffic
information in a cluster within the acceptable transmission range.
98
This idea follows a macroscopic traffic flow model which is often used in civil engineering to
describe the traffic aggregation behaviour of traffic flow on the road for both free-flow and
congested traffic conditions. This level is very important for deciding the traffic flow state
coupled with density disruptions according to (M. Artimy, 2007; Mohimani et al., 2009). This
proposed model is shown in Figure 4.19, 4.20 and 4.21, 4.22 and the complete Honeycomb
Model is shown in Figure 4.23.
Angle 1 at the feeding station Angle 2 at the Hive
Figure 4.19 Showing the rotation of the two clusters based on the waggling dance of the
Honeybees in A and B
Figure 4.20 Showing the angle 1 at the feeding station
99
Figure 4.21 Showing the angle 2 at the bee hive
Figure 4.22 Showing how the angle of rotation matches the angle between the feeding stations
and the bee hive.
4.5.1 The Honeycomb Model (HCM)
The traffic flow models developed in (May, 1990) are summarized into subsection for simplicity.
The result in (Adolf, 1990; Mohimani et al., 2009), show that the traffic flow characteristics
depends to a large extent on the traffic flow conditions on the roads and this is usually based on
100
the condition imposed by the level of interaction between the vehicles on the roads. For instance,
under a low traffic density, the interaction between the vehicles is negligible. Similarly, the
vehicles moving independently enjoy the maximum level of services in A or B as indicated in
Table 3.1. When the traffic density increases, it corresponds with the interaction between
vehicles increase and this eliminates the free flow conditions. For instance, when the traffic
density approaches the road capacity, the interaction that exists between vehicles further
increases. This subsequently, further eliminates randomness in the vehicle movement. These
occurrences lead to the steady movement with deterministic traffic characteristic values
proceeded by the car following models.
Both circles represent cluster P and Q. Cluster P is where the source vehicle, segment vehicle
and the neighbour vehicle are located while the destination vehicle is in cluster Q. Cluster P
contains two right-angled triangles: namely angle ASC and angle BSC. Angle ASC is the source
vehicle while angle BSC is the destination vehicle. The angle formed in ASC is the same as the
angle formed in BSC. Therefore, angle ASC = angle BSC (see Figure 4.19, Figure 4.20 and
Figure 4.21) for these assumptions. The angle ASC = angle BSC is explained in the Model (see
Figure 4.23) to replicate the attributes of Honeybees and how this phenomenon is implemented
in VANETs to mitigate the sparse network problem.
The source vehicle within cluster P has two directions to transmit messages to the destination
vehicle in cluster Q. The angle at which to transmit messages within cluster P to cluster Q is
1800, since the angle of rotation from cluster P matches cluster Q precisely. Each triangle within
cluster P behaves differently when transmitting messages to the destination vehicles.
101
In this model, any vehicle that has the least distance will serve as the cluster head to communicate
with destination vehicles within the acceptable transmission range R.
Figure 4.23 Proposed Honeycomb Models (HCM)
4.6 Conclusions
This chapter has suggested a new congestion control model in dense networks and Honeycomb
Model in sparsely connected networks based on the macroscopic analytical investigations
through simulations. The simulations result of the analytical investigations in dense networks
and sparse networks were represented graphically to demonstrate the performance analysis under
traffic density while varying the value of λ. The simulation result reveals the vehicle trajectories
as the value of λ increases with time. The outcome of this study indicates that there is no need
increasing the value of λ beyond 12, since further increases would only increase the number of
vehicles leading to traffic jam. In principle, it is very important to use 1≤ λ ≤12 in designing
102
traffic models in VANETs. The analytical result proved that the Congestion Control Model for
the four-lane may be the most appropriate model to be adopted and use in designing VANETs
networks. The analytical investigation also provides a model to mitigate the frequent
disconnections in sparse vehicular networks (VANETs). This model provides a scientific basis in
designing vehicular networks for different level of services. For service A: 1 ≤ λ ≤ 3; B: 1 ≤ λ ≤
4 and C: 1 ≤ λ ≤ 5, the study proposed a Honeycomb Models in sparsely connected model to
mitigate the challenges in sparse networks. The first model is the congestion control model that
has been used to describe four-lane road networks and indicate the various thresholds with
respect to the value of λ. The second model is the Honeycomb model in sparsely connected
networks based on the science behind how honey bees move from the feeding station to the hive.
These models were developed to help VANETs researchers to develop scalable algorithm to
enhance traffic alert message dissemination in VANETs.
103
Chapter 5 Evaluation of the simulation results
5.1 Introduction
In Chapter 2, the Cellular Automata (CA) model was discussed in the context of macroscopic
traffic flow model as the discrete model. In this approach, parameter such as density, speed and
traffic flow were investigated in which the level of service in (Adolf, 1990; May, 1990; Rickert,
Nagel, Schreckenberg, & Latour, 1996), were selected based on traffic flow conditions under
congested and free-flow scenario to ascertained the spatio temporal variations. Chapter 3,
adopted the macroscopic traffic flow model according to (May, 1990), using the novel approach
in this thesis to perform simulation in MATLAB and proposed a congestion control model along
with the Honeycomb Model to mitigate ITS problems in VANETs for enhanced future
applications. Chapter 4 demonstrates the performance of the adopted macroscopic traffic flow
model and used equations (3.1) and (3.3) to perform simulation in terms of density, speed and
traffic flow.
This chapter evaluates the result of the simulations in dense and sparse networks as reported in
chapter 3 while chapter 4 conduct an experiment through simulation to address the specific
objectives of this thesis and briefly show the results graphically.
The rest of the chapter is organized as follows: Section 5.2 describes in detail the result
associated with dense network traffic flow. Section 5.3 also describes in detail the result
concerning the sparse network and demonstrates the percentage link intermittent connectivity in
104
each lane with respect to the acceptable transmission range. Finally, section 5.4 concludes the
chapter.
5.2 Result obtained under dense network traffic flow
This evaluation is based on the simulation using MATLAB and depends on the node’s density of
each lane. This demonstrates the spatio temporal variations in vehicular trajectories with respect
to the node density. The nodes in each lane indicate the vehicular (nodes) trajectories as the
value of λ increases (see Figure 4.1b, 4.2b, 4.3b, 4.4b, 4.5b, 4.6b, 4.7b, 4.8b, 4.9b and 4.10b).
5.2.1 Network reachability
Figure 4.1a, 4.1b shows the average packet penetration distance as a function of λ. The faster the
packet delivery the more efficient the Congestion Control Model (CCM). The model is
considered efficient if it can send a message to the target destination within an acceptable time
frame. Under an ideal condition, it is expected as the time increases, the packet penetration
distance also increases up to the end of the Region of Interest (ROI). In contrast, if there is no
vehicle (node) within the transmission range at the time of broadcast, then the disconnected
vehicle moves out of (ROI) (i.e.1000 m). In this case, lane 1, lane 2 has seen a few vehicles than
in lane 3 and lane 4. Therefore, lane 3 and lane 4 experiences reasonable penetration distance
while lane 1 and lane 2 are within (ROI). The results are like Figure 4.2a and 4.2b but the
average packet penetration distance is better as the value of λ increases the faster the packet
delivery until the normal distribution occurs.
Similarly, Figure 4.3a, 4.3b shows the average packet penetration distance as a function of λ,
when lane 3 and lane 4 have seen normal distribution around nodes 10 (veh/mile/lane) and
105
therefore experiences faster packet delivery and more efficient as indicated in the Congestion
Control Model (CCM). As the value of λ increases the model is considered efficient since it can
send a message to the target destination within an acceptable time frame. Lane 1 and lane 2 also
seen fewer packet penetration under an ideal condition as the time increases, the packet
penetration distance also increases up to the end of the Region of Interest (ROI). In the same
way, figure 4.4a and 4.4b have seen a complete normal distribution indicating how the vehicles
spread across the road when the value of λ increases with time. The complete normal distribution
between 5 nodes and 15 nodes indicate faster packet delivery than in lane 1. As the value of λ
increases the model is considered efficient since it can send a message to the target destination
within an acceptable time frame than in lane 1.
In addition, Figure 4.5a and 4.5b are like Figure 4.4a and 4.4b, have also seen a complete normal
distribution indicating how the vehicles spread across the road as the value of λ increases with
time. The complete normal distribution between 8 nodes and 20 nodes indicate faster packet
delivery than in lane 1. As the value of λ increases the model is considered efficient since the
target destination is within an acceptable time frame than in lane 1. In Figure, 4.6a and 4.6b have
seen an increased node between 10 and 26 nodes as the value of λ increases with time than in
lane 1. The packet penetration distance as a function of λ is greatly enhanced since lane 2, lane 3
and lane 4 have seen increased packet penetration rate under the normal distribution while lane 1
experiences fewer packet penetration rate.
Figure 4.7a and 4.7b have also seen an improvement over Figure 4.6a and 4.6b, since almost all
the lanes nearly assumed a normal distribution. In this case, the nodes are distributed between 10
and 35 nodes within lane 2, lane 3 and lane 4 as the value of λ increases with time than in lane 1.
106
This is evident when the packet penetration distance is expressed as a function of λ and therefore
have increased packet penetration rate in lane 2, lane 3 and lane 4 while lane1 experiences an
average packet penetration rate with increased nodes.
Similarly, Figure 4.8a and 4.8b, have seen an increased in nodes distribution between 10 and 40
nodes and lane 2, lane 3 and lane 4 have also demonstrated a complete normal distribution than
lane 1 as the λ increases. Lane 2, lane 3 and lane 4 have experienced high packet penetration rate
as a function of λ.
5.2.2 Network scalability
Figure 4.9a and 4.9b, almost assumed a complete normal distribution in all the lanes with nodes
density between 10 and 50 nodes as the λ increases with time. In this case, network scalability
emerges as the nodes density increases and the average network overhead for a single broadcast
is observed in each vehicle (node) with respect to each lane. This network overhead seems to
scale very well in lane 2, lane 3 and lane 4; especially in heavy traffic conditions in which
broadcast storm is likely to occur. The result from the analytical investigation indicates that the
amount of overhead is about four times less than in lane 2 and lane 1 and therefore no broadcast
storm occurred.
Similarly, Figure 4.10a and 4.10b depicted a complete normal distribution within all the lanes.
The nodes are distributed between 10 and 55 nodes as the value of λ increases to 10. This
phenomenon shows the average network overhead for a single broadcast as observed by each
vehicle (node) within lanes. This network overhead seems to scale very well as the value of λ
increases with time; especially in heavy traffic conditions in which broadcast storm is likely to
107
occur. The result from the analytical investigation also indicates that the amount of overhead is
about four times less than in lane 2 and lane 1.
This network overhead seems to scale very well; especially in heavy traffic conditions in which
broadcast storm is likely to occur. The result from the analytical investigation indicates that the
amount of overhead is about four times less than in lane 2 and lane 1. In this scenario, further
increases in the value of λ beyond 10 would result in a jam situation where reachability will
suffer. Therefore, in designing traffic flow models it is advisable to use 1≤ λ ≤12 to avoid a
traffic jam.
In conclusion, the analytical investigation of the macroscopic traffic flow model presents the
performance in terms of reachability and scalability within the congested network in highways
scenario under multiple traffic conditions. The Four-lane Congestion Control Model is developed
to mitigate the extreme conditions of heavy traffic. This phenomenon, during simulation, was
evident according to the network behaviour under lane 1, lane 2, lane 3 and lane 4.
5.3 Result obtained under sparse network traffic flow
The result in (Adolf, 1990; Daganzo, 2011; Kloiber et al., 2012; May, 1990), show that the
traffic flow characteristics depends to a large extent on the traffic flow conditions on the roads
and this is usually based on the condition imposed by the level of interaction between the
vehicles on the roads. For instance, under a low traffic density, the interaction between the
vehicles is negligible.
Similarly, the vehicles moving independently enjoy the maximum level of services in A or B as
indicated in Table 3.1. When the traffic density increases, it corresponds to the interaction
108
between vehicles and this eliminates the free flow conditions. For instance, when the traffic
density approaches the road capacity, the interaction that exists between vehicles further
increases. This subsequently, eliminates randomness in the vehicle movement. These
occurrences lead to the steady movement with deterministic traffic characteristic values
proceeded by the car following models.
As described in chapter 2 and chapter 3, the simulation consists of a traffic flow model. The
mobility pattern generated among the four-lane using the traffic model to describe in Chapter 3.
This mobility pattern is used under sparsely connected networks to perform each simulation
under different densities scenario to gain deep insight into network performance metrics such as
reachability and scalability. The simulation parameter used are presented in Chapter 3, where
vehicle density of (0 – 30 veh/mile/lane) are also selected for sparsely connected networks to
perform various simulation while varying the value of λ from 1 to 10 at a different time interval.
When the simulation result indicates non-normal distribution, it may lead to drop and peaks in
graph demonstrating unexpected increases in packet penetration rate when the density is reduced.
On the other hand, when the simulation results also indicate normal distribution, this may lead to
dense traffic in the graph with expected increases in the packet penetration rate. To mitigate the
challenges in sparsely connected networks where packet penetration rate is reduced due to low
traffic density, a Honeycomb model is developed with respect to the simulation results under
sparse networks. This result indicates the percentage disconnection within each lane and with
regards to the level of service volume as shown below:
109
5.3.1 Totally disconnected nodes (vehicles)
A vehicle operating in a sparse traffic condition is regarded as totally disconnected nodes if it has
no neighbour in the message forwarding direction and is not connected to any node in the
opposite direction. The simulation result, reveals the traffic situation in a typical disconnected
network. To mitigate these challenges, a Honeycomb model is proposed to improve the packet
penetration rate.
5.3.2 Partially sparsely connected nodes (Vehicles)
A vehicle is operating in a sparse traffic condition if it is the last node in a cluster. Furthermore, a
vehicle in this regime is said to be in a sparsely connected neighbourhood if there exists at least
one neighbour in the opposite direction as indicated in the Figure below
In Figure 4.14, 4.15 and 4.16 with lane 1,2,3 and 4, when λ=1, 2, 3 and 4; this explains that the
number of vehicles on the roads are closed or clustered. This indicates reasonable distribution
among few numbers of a vehicle on the road, with respect to high speed and distances. This
separate vehicle beyond the acceptable transmission range (i.e. 1000m) and therefore
experiences low packet penetration rate with respect to the lanes. The vehicle connectivity range
for effective communication. When λ=1; lane 1: 79.5%.; lane 2: 79.9%; lane 3: 81.1% and lane
4: 82.3%, when λ=2; lane 1: 70.8%.; lane 2: 78.4%; lane 3: 81.9% and lane 4: 83.9%, when λ=3;
lane 1: 75.5%.; lane 2: 81.8%; lane 3: 84.7% and lane 4: 86.5%, finally when λ=4; lane 1:
78.4%.; lane 2: 83.9%; lane 3: 86.5% and lane 4: 87.9% and When λ=1: all the four lanes, shows
an exponential distribution when λ=2: lane 1 and 2, show an exponential distribution while lane
3 and 4 begin to show normal distribution but skewed to the right. When λ=3 in lane 1, shows an
exponential distribution while lane 2,3 and 4 indicate a normal distribution but skewed to the
110
right at varying degree of skewness. When λ=4: lane 1, shows an exponential distribution, in lane
2, 3 and 4; also indicates a normal distribution but skewed to the right at varying degree of
skewness.
In lane 1,2 and 3, when λ=5, this indicates that a certain number of vehicles on the roads are
closed or clustered. This indicates fair distribution among few numbers of the vehicle on the
road, with respect to high speed and distances. In lane 4; the vehicles spread-out at varying
degree.
On the other hand, vehicles in lane 4; are much further away from each other due to fewer
vehicles on the road moving at a very high speed and high distances within each neighbour. The
vehicles in lane 1, 2 and 3, are at reasonable distances from each vehicle but beyond the
acceptable transmission range (i.e. 1000m). The vehicle connectivity range for effective
communication. In Figure 4.17, when λ=5, lane 1: 80.3%.; lane 2: 85.4%; lane 3: 97.3% and
lane 4: 88.0%. When λ=5: lane 1 begin to show a normal distribution but skewed to the right
while lane 2, indicates a normal distribution but also skewed to right. Lane 3 and 4, indicate a
complete normal distribution but lane 3 slightly skewed to right while lane 4 also slightly skewed
to left explains the variations in node density. These intermittent percentage disconnections of
nodes (vehicles) show the severity in the sparse network with regards to information
dissemination. The model ensures that the nodes are cluster using Honeycomb model to mitigate
the frequent disconnection associated with the sparse network in VANETs.
111
5.4 Conclusions
This chapter presented a new macroscopic traffic flow model and run various simulations under
dense network and sparse network to ascertain traffic flow variability in VANETs. This model is
implemented in MATLAB.
The performance of the macroscopic traffic flow model is measured in terms of these
performance metrics: reachability and scalability. Performance analysis has been done using
various system parameters for dense networks (See Table 3.3) and for sparse networks (See
Table 3.4). In the first place, the impact of the network density concerning the performance of
the macroscopic traffic flow model is assessed by varying the number of nodes placed in a fixed
level of service (See Table 3.1). In the second place, the impact of varying the value of λ as a
function of average packet penetration distance. Finally, the performance analysis of the
macroscopic traffic flow model has been conducted under varying node densities by varying the
value of the nodes in the network.
The simulation results of the performance analysis demonstrate that the performance of the four-
lane road of the macroscopic traffic flow model in terms of scalability as the value of λ increases.
It performs well in the dense network than in sparse network. The simulation results of the
second performance analysis also indicate that the performance of the four-lane road
macroscopic traffic model in terms of reachability is achieved by varying the value of λ to ensure
the connectivity of nodes to determine the packet penetration rate within dense and sparse
network and the acceptable threshold in which the nodes reaches its destination under both
scenarios. The results in dense networks have seen high packet penetration rate than in sparse
network.
112
Finally, in the dense network the simulation result reveals the vehicle trajectories as the value of
λ increases with time. The performance analysis indicates that there is no need for increasing the
value of λ beyond 12 since further increases would only increase the number of vehicles leading
to the traffic jam. It is very important to use 1≤ λ ≤12 in designing traffic models in VANETs to
achieve good reachability and scalability. In the sparse network, the performance analysis
provides the basis to mitigate the frequent disconnections in sparse vehicular networks
(VANETs). This range provides the basis in designing vehicular networks for the different level
of services. For service A: 1 ≤ λ ≤ 3; B: 1 ≤ λ ≤ 4 and C: 1 ≤ λ ≤ 5 to improve reachability and
scalability.
113
Chapter 6 Conclusions and Future Directions
6.1 Introduction
Vehicular Ad Hoc Networks (VANETs) have attracted the attention of academia and industry
recently according to (Dressler, Kargl, Ott, Tonguz, & Wischhof, 2011; Hartenstein &
Laberteaux, 2009). Due to current advances in Intelligent Transportation Systems (ITS)
applications cannot be overemphasized. This research effort cut across international boundaries
due to its important applications to improve vehicular ad hoc networks (VANETs) according to
(Hartenstein & Laberteaux, 2009). Although the nodes in VANETs have similar characteristics
of MANETs, they present certain unique challenges inherent in dense networks and sparse
networks due to the wireless communication medium and the distribution functions of their
medium access mechanism (Bianchi, 2000; Bianchi, Fratta, & Oliveri, 1996). This result in
frequent topology changes in vehicular mobility pattern (Fiore, Harri, Filali, & Bonnet, 2007;
Zeadally et al., 2012). In view of the important (ITS) applications in VANETs, a lot of research
efforts have been devoted to finding immediate solutions to these challenges according to
(Hartenstein & Laberteaux, 2009).
The provision of congestions control models and sparsely connected models would improve
information dissemination in ITS applications in VANETs. When this is achieved, it will greatly
mitigate the frequent topology changes, and improve reachability and scalability being the most
significant challenges in VANETs according to (F. Li & Wang, 2007). To accomplish this, quite
several traffic flow models have been proposed as indicated in (Choffnes & Bustamante, 2005;
Hoogendoorn & Bovy, 2001; Kerner et al., 2002). In these articles, the models have been
114
grouped into congestion control models and sparsely connected model. The models developed
for both congestion control model and sparsely connected model are considered not to be
scalable and reachable because of the excessive overhead and intermittent connectivity
associated with traffic alert message disseminations under critical life-changing situations.
To mitigate these challenges in VANETs, several traffic flow models were proposed according
to (Adolf, 1990; Hoogendoorn & Bovy, 2001; May, 1990). However, no single traffic flow
models have improved these limitations and therefore require a new methodology to mitigate the
challenges for an improved VANETs application in general. This, therefore requires a paradigm
shift and system concepts approach for enhanced collaboration across the globe to mitigate the
challenges.
6.2 Summary of the results
i. Figure 4.1a, 4.1b shows the average packet penetration distance as a function of λ. The
faster the packet delivery the more efficient the Congestion Control Model (CCM). The
model is considered efficient if it can send the message to the target destination within an
acceptable time frame. Under an ideal condition, it is expected as the time increases, the
packet penetration distance also increases up to the end of the Region of Interest (ROI).
In contrast, if there is no vehicle (node) within the transmission range at the time of
broadcast, then the disconnected vehicle moves out of ROI (1000 m). In this case, lane 1,
lane 2 has seen a few vehicles than in lane 3 and lane 4. Therefore, lane 3 and lane 4
experiences reasonable penetration distance while lane 1 and lane 2 are within (ROI).
These results are seen in Figure 4.2a and 4.2b but the average packet penetration distance
115
is better as the value of λ increases the faster the packet delivery until the normal
distribution occurs.
ii. Similarly, Figure 4.3a, 4.3b shows the average packet penetration distance as a function
of λ, when lane 3 and lane 4 have seen normal distribution around nodes 10
(veh/mile/lane) and therefore experiences faster packet delivery and more efficient as
indicated on the Congestion Control Model (CCM). As the value of λ increases the model
is considered efficient since it can send the message to the target destination within an
acceptable time frame. Lane 1 and lane 2 also seen fewer packet penetration under an
ideal condition as the time increases, the packet penetration distance also increases up to
the end of the Region of Interest (ROI). In the same way, Figure 4.4a and 4.4b have seen
a complete normal distribution indicating how the vehicles spread across the road when
the value of λ increases with time. The complete normal distribution between 5 nodes and
15 nodes indicate faster packet delivery than in lane 1. As the value of λ increases the
model is considered efficient since it can send the message to the target destination within
an acceptable time frame than in lane 1.
iii. In addition, Figure 4.5a and 4.5b are like Figure 4.4a and 4.4b have also seen a complete
normal distribution indicating how the vehicles are spread across the road as the value of
λ increases with time. The complete normal distribution between 8 nodes and 20 nodes
indicate faster packet delivery than in lane 1. As the value of λ increases the model is
considered efficient since the target destination is within an acceptable time frame than in
lane 1. In Figure 4.6a and 4.6b, have seen an increased node between 10 and 26 nodes as
the value of λ increases with time than in lane 1. The packet penetration distance as a
function of λ is greatly enhanced since lane 2, lane 3 and lane 4 have seen increased
116
packet penetration rate under the normal distribution while lane 1 experiences fewer
packet penetration rate.
iv. Figure 4.7a and 4.7b have also seen an improvement over Figure 4.6a and 4.6b since
almost all the lanes nearly assumed a normal distribution. In this case, the nodes are
distributed between 10 and 35 nodes within lane 2, lane 3 and lane 4 as the value of λ
increases with time than in lane 1. This is evident when the packet penetration distance is
expressed as a function of λ and therefore have increased packet penetration rate in lane
2, lane 3 and lane 4 while lane1 experiences an average packet penetration rate with
increased nodes.
v. Similarly, Figure 4.8a and 4.8b have seen an increased in nodes distribution between 10
and 40 nodes and lane 2, lane 3 and lane 4 have also demonstrated a complete normal
distribution than lane 1 as the value of λ increases. Lane 2, lane 3 and lane 4 have
experienced high packet penetration rate as a function of λ.
vi. In Figure 4.9a and 4.9b, almost assumed a complete normal distribution in all the lanes
with nodes density between 10 and 50 nodes as the value of λ increases with time. In this
case, network scalability emerges as the nodes density increases where the average
network overhead for a single broadcast is observed by each vehicle (node) with respect
to each lane. This network overhead seems to scale very well in lane 2, lane 3 and lane 4,
especially in heavy traffic conditions in which broadcast storm is likely to occur. The
result from the analytical investigation indicates that the amount of overhead is about
four times less than in lane 2 and lane 1 and therefore no broadcast storm occurred.
vii. Similarly, Figure 4.10a and 4.10b depicted a complete normal distribution within all the
lanes. The nodes are distributed between 10 and 55 nodes as the value of λ increases to
117
10. This phenomenon shows the average network overhead for a single broadcast as
observed by each vehicle (node) within lanes. This network overhead seems to scale very
well as the value of λ increases with time; especially in heavy traffic conditions in which
broadcast storm is likely to occur. The result from the analytical investigation also
indicates that the amount of overhead is also about four times less than in lane 2 and lane
1.
viii. In the sparse network, Figure 4.14, 4.15 and 4.16 with lane 1,2,3 and 4, when λ=1,
2, 3 and 4; indicates that a certain number of the vehicles on the roads are closed or
clustered. This indicates reasonable distribution among few numbers of a vehicle on the
road, with respect to high speed and distances. This separate vehicle beyond the
acceptable transmission range (i.e. 1000m) and therefore experiences low packet
penetration rate within a certain lane. The vehicle connectivity range for effective
communication is affected due to the position of the vehicles on the road.
ix. This network overhead seems to scale very well; especially in heavy traffic conditions in
which broadcast storm is likely to occur. The result from the analytical investigation
indicates that the amount of overhead is subsequently about four times less than in lane 2
and lane 1. In this scenario, further increases in the value of λ beyond 10 would result in a
jam situation where reachability will suffer. Therefore, in designing traffic flow models it
is advisable to use 1≤ λ ≤12 in designing traffic models in VANETs to improve ITS
applications.
x. The performance of the macroscopic traffic flow model is measured in terms of these
performance metrics: reachability and scalability. Performance analysis has been done
using various system parameters. In the first place, the impact of the network density on
118
performance of the macroscopic traffic flow model is assessed by varying the number of
nodes placed in a fixed level of service in Table 3.1. In the second place, the impact of
varying the value of λ as a function of average packet penetration distance. Finally, the
performance analysis of the macroscopic traffic flow model has been conducted under
varying node mobility by varying the maximum node speed in the network.
xi. In conclusion, the analytical investigation of the macroscopic traffic flow model presents
the performance in terms of reliability and scalability within the congested network on
highways under multiple traffic conditions. The Four-lane Congestion Control Model is
developed to mitigate the extreme conditions of heavy traffic. This phenomenon during
simulation was evident according to the network behaviour under lane 1, lane 2, lane 3
and lane 4.
xii. The simulation results of the performance analysis demonstrate that the performance of
the four-lane road of the macroscopic traffic flow model in terms of scalability as the
value of λ increases. It performs well in the dense network than in sparse network.
xiii. The simulation results of the second performance analysis also indicate that the
performance of the four-lane road macroscopic traffic model in terms of reachability is
achieved by varying the value of λ to ensure the connectivity of nodes to determine the
packet penetration rate within dense and sparse network and the acceptable threshold in
which the nodes reaches its destination under both scenarios. The results in dense
networks have seen high packet penetration rate than in sparse network.
xiv. Finally, in the dense network, the simulation result reveals the vehicle trajectories
as the value of λ increases with time. The performance analysis indicates that there is no
need for increasing the value of λ beyond 12 since further increases would only increase
119
the number of vehicles leading to a traffic jam. It is very important to use 1≤ λ ≤12 in
designing traffic models in VANETs to achieve good reachability and scalability. In the
sparse network, the performance analysis provides the basis to mitigate the frequent
disconnections in sparse vehicular networks (VANETs). This provides basis for in
designing vehicular networks for a different level of services. For service A: 1 ≤ λ ≤ 3;
B: 1 ≤ λ ≤ 4 and C: 1 ≤ λ ≤ 5 to improve reachability and scalability.
6.3. Conclusions and Future Work
This thesis has revealed very interesting challenges that require further investigation have
emerged to promote further research in the future.
One important novel articles in congestion control model have been published from this thesis
according to (Azameti et al., 2018). This research article has been widely read, indicating the
novel contributions that would be used by both academia and industry for the development of
new protocols and ITS applications in VANETs. These are clearly outlined below:
i. This thesis has performed an analytical investigation into macroscopic traffic
flow model to gain deep insight into the traffic density to ascertain the spatio-
temporal variation as the value of λ increases under dense network and sparse
network according to (May, 1990)
ii. This study used traffic flow models (Civil Engineering) and concentrated on the
macroscopic model stated in (May, 1990) and has presented a scalable solution
to a congestion control in highly populated city environment. This proposed
approach is a Congestion Control model designed for a four-lane road network.
The lanes are numbered from 1 to 4 and in either direction (see Figure 4.14a).
120
iii. The simulation in MATLAB provides a practical implementation of V2V and
V2R Communication where each lane shows the vehicle trajectories and the lane
with the highest congestion and the spatial variation in the successive lanes.
Therefore, vehicular movements are not entirely random but also sequential and
depend largely on the drivers’ perception and the interactions with other road
users. This Congestion Control Model (CCM) would help in designing vehicular
traffic flow. Therefore, there is no need for increasing the value of λ beyond this
range 1≤ λ ≤12 in congested networks.
iv. This study provides a fast, reliable and scalable model to congestion control.
Although there exist many interventions to provide a lasting solution to
congestion problem in VANETs, this proposed Congestion Control Model
(CCM) provides an alternative methodology and framework for the research
community to develop a lasting protocol for safety alert message dissemination
in VANETs for ITS applications.
v. Finally, the research findings reveal the possibilities to replicate the attributes of
honeybees in designing vehicular traffic flow in VANETs. This novel approach
would significantly improve ITS applications in VANETs V2V communication
and therefore requires further investigation in VANETs protocol design.
121
Appendix
Appendix A: Publications
A Publications during the MPhil course

Adams A.K. Azameti et al. “The Effect of Congestion Control Model on Congested Traffic Flow
in Vehicular Ad Hoc Networks (VANETs)” EAI Endorsed Transactions on Mobile
Communications and Applications, vol.3. no.10, March 2018.doi={10.4108/eai.22-3-
2018.154371}, http://eudl.eu/doi/10.4108/eai.22-3-2018.154371
Appendix B: Performance analysis under sparse network conditions and the value of λ
Figure B.1 The effect of traffic flow when λ =6
122
123
124
References
Adachi, M., Morita, Y., Fujimura, K., Takatori, Y., & Hasegawa, T. (2002). On an autonomous
cruising traffic flow simulator including inter-vehicle and road-to-vehicle
communication networks. Paper presented at the Intelligent Transportation Systems,
2002. Proceedings. The IEEE 5th International Conference on.
Adolf, D. (1990). May. Traffic flow fundamentals. University of California, Berkeley.
Agerholm, N., Jensen, R., & Andersen, C. S. (2015). Identification of Hazardous Road Locations
on the basis of jerks. Paper presented at the ICTCT.
Ahn, K., Rakha, H., Trani, A., & Van Aerde, M. (2002). Estimating vehicle fuel consumption
and emissions based on instantaneous speed and acceleration levels. Journal of
transportation engineering, 128(2), 182-190.
Al-Sultan, S., Al-Doori, M. M., Al-Bayatti, A. H., & Zedan, H. (2014). A comprehensive survey
on vehicular ad hoc network. Journal of network and computer applications, 37, 380-
392.
Alasmary, W., & Zhuang, W. (2012). Mobility impact in IEEE 802.11 p infrastructureless
vehicular networks. Ad Hoc Networks, 10(2), 222-230.
Alsabaan, M., Alasmary, W., Albasir, A., & Naik, K. (2013). Vehicular Networks for a Greener
Environment: A Survey. IEEE Communications Surveys and Tutorials, 15(3), 1372-
1388.
Altintas, O., Ihara, Y., Kremo, H., Tanaka, H., Ohtake, M., Fujii, T., . . . Tsuru, M. (2012). Field
tests and indoor emulation of distributed autonomous multi-hop vehicle-to-vehicle
communications over TV white space. Paper presented at the Proceedings of the 18th
annual international conference on Mobile computing and networking.
Altintas, O., Nishibori, M., Oshida, T., Yoshimura, C., Fujii, Y., Nishida, K., . . . Tsuru, M.
(2011). Demonstration of vehicle to vehicle communications over TV white space.
Paper presented at the Vehicular Technology Conference (VTC Fall), 2011 IEEE.
Amadeo, M., Campolo, C., & Molinaro, A. (2012). Enhancing IEEE 802.11 p/WAVE to provide
infotainment applications in VANETs. Ad Hoc Networks, 10(2), 253-269.
Amoozadeh, M., Deng, H., Chuah, C.-N., Zhang, H. M., & Ghosal, D. (2015). Platoon
management with cooperative adaptive cruise control enabled by VANET. Vehicular
communications, 2(2), 110-123.
Anda, J., LeBrun, J., Ghosal, D., Chuah, C.-N., & Zhang, M. (2005). VGrid: vehicular adhoc
networking and computing grid for intelligent traffic control. Paper presented at the
Vehicular Technology Conference, 2005. VTC 2005-Spring. 2005 IEEE 61st.
Artimy, M. (2007). Local density estimation and dynamic transmission-range assignment in
vehicular ad hoc networks. IEEE Transactions on intelligent transportation systems,
8(3), 400-412.
Artimy, M. M., Robertson, W., & Phillips, W. J. (2005). Assignment of dynamic transmission
range based on estimation of vehicle density. Paper presented at the Proceedings of the
2nd ACM international workshop on Vehicular ad hoc networks.
Azameti, A. A., Katsriku, F. A., Chong, P., & Abdulai, J. D. (2018). EAI Endorsed Transactions.
Balasubramanian, A., Mahajan, R., Venkataramani, A., Levine, B. N., & Zahorjan, J. (2008).
Interactive wifi connectivity for moving vehicles. ACM SIGCOMM Computer
Communication Review, 38(4), 427-438.
125
Bettstetter, C., Hartenstein, H., & Pérez-Costa, X. (2004). Stochastic properties of the random
waypoint mobility model. Wireless Networks, 10(5), 555-567.
Bettstetter, C., Resta, G., & Santi, P. (2003). The node distribution of the random waypoint
mobility model for wireless ad hoc networks. IEEE Transactions on mobile computing,
2(3), 257-269.
Bianchi, G. (2000). Performance analysis of the IEEE 802.11 distributed coordination function.
IEEE journal on Selected Areas in Communications, 18(3), 535-547.
Bianchi, G., Fratta, L., & Oliveri, M. (1996). Performance evaluation and enhancement of the
CSMA/CA MAC protocol for 802.11 wireless LANs. Paper presented at the Personal,
Indoor and Mobile Radio Communications, 1996. PIMRC'96., Seventh IEEE
International Symposium on.
Boban, M., Vinhoza, T. T., Ferreira, M., Barros, J., & Tonguz, O. K. (2011). Impact of vehicles
as obstacles in vehicular ad hoc networks. IEEE journal on Selected Areas in
Communications, 29(1), 15-28.
Boukerche, A., Oliveira, H. A., Nakamura, E. F., & Loureiro, A. A. (2008). Vehicular ad hoc
networks: A new challenge for localization-based systems. Computer communications,
31(12), 2838-2849.
Boxill, S. A., & Yu, L. (2000). An evaluation of traffic simulation models for supporting its.
Houston, TX: Development Centre for Transportation Training and Research, Texas
Southern University.
Breslau, L., Estrin, D., Fall, K., Floyd, S., Heidemann, J., Helmy, A., . . . Xu, Y. (2000).
Advances in network simulation. Computer, 33(5), 59-67.
Burstedde, C., Klauck, K., Schadschneider, A., & Zittartz, J. (2001). Simulation of pedestrian
dynamics using a two-dimensional cellular automaton. Physica A: Statistical
Mechanics and its Applications, 295(3-4), 507-525.
Chen, L., Wei, S., & Shi, L. (2011). Simulation of Vehicular Ad Hoc Networks with real city
scenes. Paper presented at the Communication Software and Networks (ICCSN), 2011
IEEE 3rd International Conference on.
Chen, Z. D., Kung, H., & Vlah, D. (2001). Ad hoc relay wireless networks over moving vehicles
on highways. Paper presented at the Proceedings of the 2nd ACM international
symposium on Mobile ad hoc networking & computing.
Cheng, H. T., Shan, H., & Zhuang, W. (2011). Infotainment and road safety service support in
vehicular networking: From a communication perspective. Mechanical Systems and
Signal Processing, 25(6), 2020-2038.
Cheng, N., Zhou, H., Lei, L., Zhang, N., Zhou, Y., Shen, X., & Bai, F. (2017). Performance
analysis of vehicular device-to-device underlay communication. IEEE Transactions on
Vehicular Technology, 66(6), 5409-5421.
Cheng, R. S., & Deng, D. J. (2014). Congestion control with dynamic threshold adaptation and
cross‐layer response for TCP Vegas over IEEE 802.11 wireless networks. International
Journal of Communication Systems, 27(11), 2918-2930.
Chitra, M., & Sathya, S. S. (2013). Efficient broadcasting mechanisms for data dissemination in
vehicular ad hoc networks. International Journal of Mobile Network Communications
& Telematics (IJMNCT), 3(3), 47-63.
Choffnes, D. R., & Bustamante, F. E. (2005). An integrated mobility and traffic model for
vehicular wireless networks. Paper presented at the Proceedings of the 2nd ACM
international workshop on Vehicular ad hoc networks.
126
Chowdhury, D., Santen, L., & Schadschneider, A. (2000). Statistical physics of vehicular traffic
and some related systems. Physics Reports, 329(4-6), 199-329.
Claffy, K. C., Braun, H.-W., & Polyzos, G. C. (1995). A parameterizable methodology for
Internet traffic flow profiling. IEEE journal on Selected Areas in Communications,
13(8), 1481-1494.
Cowan, R. J. (1975). Useful headway models. Transportation Research, 9(6), 371-375.
Cuesta, J. A., Martínez, F. C., Molera, J. M., & Sánchez, A. (1993). Phase transitions in two-
dimensional traffic-flow models. Physical review E, 48(6), R4175.
Daganzo, C. F. (2011). On the macroscopic stability of freeway traffic. Transportation Research
Part B: Methodological, 45(5), 782-788.
Donnell, E., Ni, Y., Adolini, M., & Elefteriadou, L. (2001). Speed prediction models for trucks
on two-lane rural highways. Transportation Research Record: Journal of the
Transportation Research Board(1751), 44-55.
Dressler, F., Kargl, F., Ott, J., Tonguz, O. K., & Wischhof, L. (2011). Research challenges in
intervehicular communication: lessons of the 2010 Dagstuhl Seminar. IEEE
Communications magazine, 49(5).
Duan, J., Li, Z., & Salvendy, G. (2013). Risk illusions in car following: Is a smaller headway
always perceived as more dangerous? Safety science, 53, 25-33.
Ebner, A., & Rohling, H. (2001). A self-organized radio network for automotive applications.
Paper presented at the Conference Proceedings ITS 2001, 8th World Congress on
Intelligent Transportation Systems.
Eichler, S. (2007). Performance evaluation of the IEEE 802.11 p WAVE communication
standard. Paper presented at the Vehicular Technology Conference, 2007. VTC-2007
Fall. 2007 IEEE 66th.
Esser, J., & Schreckenberg, M. (1997). Microscopic simulation of urban traffic based on cellular
automata. International Journal of Modern Physics C, 8(05), 1025-1036.
ETSI, T. (2010). 102 637-2, Intelligent Transport Systems (ITS); Vehicular Communications;
Basic Set of Applications; Part 2: Specification of Co-operative Awareness Basic
Service. ETSI, Sophia Antipolis Cedex, France.
Eze, E. C., Zhang, S.-J., Liu, E.-J., & Eze, J. C. (2016). Advances in vehicular ad-hoc networks
(VANETs): Challenges and road-map for future development. International Journal of
Automation and Computing, 13(1), 1-18.
Faezipour, M., Nourani, M., Saeed, A., & Addepalli, S. (2012). Progress and challenges in
intelligent vehicle area networks. Communications of the ACM, 55(2), 90-100.
Fall, K., & Floyd, S. (1996). Simulation-based comparisons of Tahoe, Reno and SACK TCP.
ACM SIGCOMM Computer Communication Review, 26(3), 5-21.
Fallah, Y. P., Huang, C.-L., Sengupta, R., & Krishnan, H. (2011). Analysis of information
dissemination in vehicular ad-hoc networks with application to cooperative vehicle
safety systems. IEEE Transactions on Vehicular Technology, 60(1), 233-247.
Ferreira, M., Fernandes, R., Conceição, H., Viriyasitavat, W., & Tonguz, O. K. (2010). Self-
organized traffic control. Paper presented at the Proceedings of the seventh ACM
international workshop on VehiculAr InterNETworking.
Fiore, M., Harri, J., Filali, F., & Bonnet, C. (2007). Vehicular mobility simulation for VANETs.
Paper presented at the Simulation Symposium, 2007. ANSS'07. 40th Annual.
127
Fisher, D. S., Fisher, M. P., & Huse, D. A. (1991). Thermal fluctuations, quenched disorder,
phase transitions, and transport in type-II superconductors. Physical Review B, 43(1),
130.
Forecast, C. C. (2013). Global Connected Car Market to Grow Threefold Within Five Years.
GSMA Connected Living Programme: mAutomotive.
Fu, S., & Atiquzzaman, M. (2004). SCTP: State of the art in research, products, and technical
challenges. IEEE Communications magazine, 42(4), 64-76.
Füßler, H., Mauve, M., Hartenstein, H., Käsemann, M., & Vollmer, D. (2002). A comparison of
routing strategies for vehicular ad hoc networks. Technical reports, 2.
Greenshields, B., Channing, W., & Miller, H. (1935). A study of traffic capacity. Paper presented
at the Highway research board proceedings.
Gubbi, J., Buyya, R., Marusic, S., & Palaniswami, M. (2013). Internet of Things (IoT): A vision,
architectural elements, and future directions. Future generation computer systems,
29(7), 1645-1660.
Gupta, P., & Kumar, P. R. (2000). The capacity of wireless networks. IEEE Transactions on
information theory, 46(2), 388-404.
Hadaller, D., Keshav, S., Brecht, T., & Agarwal, S. (2007). Vehicular opportunistic
communication under the microscope. Paper presented at the Proceedings of the 5th
international conference on Mobile systems, applications and services.
Haerri, J., Filali, F., & Bonnet, C. (2006). Performance comparison of AODV and OLSR in
VANETs urban environments under realistic mobility patterns. Paper presented at the
Proceedings of the 5th IFIP mediterranean ad-hoc networking workshop.
Haglund, M. (2001). Speed Choice: The Driver, the Road and Speed Limits. Acta Universitatis
Upsaliensis.
Harri, J., Filali, F., & Bonnet, C. (2009). Mobility models for vehicular ad hoc networks: a
survey and taxonomy. IEEE Communications Surveys & Tutorials, 11(4).
Härri, J., Filali, F., Bonnet, C., & Fiore, M. (2006). VanetMobiSim: generating realistic mobility
patterns for VANETs. Paper presented at the Proceedings of the 3rd international
workshop on Vehicular ad hoc networks.
Hartenstein, H., & Laberteaux, K. (2009). VANET: vehicular applications and inter-networking
technologies (Vol. 1): John Wiley & Sons.
Hartenstein, H., & Laberteaux, L. (2008). A tutorial survey on vehicular ad hoc networks. IEEE
Communications magazine, 46(6).
Helbing, D., & Tilch, B. (1998). Generalized force model of traffic dynamics. Physical review E,
58(1), 133.
Herman, R., & Prigogine, I. (1979). A two-fluid approach to town traffic. Science, 204(4389),
148-151.
Hoogendoorn, S. P., & Bovy, P. H. (2001). State-of-the-art of vehicular traffic flow modelling.
Proceedings of the Institution of Mechanical Engineers, Part I: Journal of Systems and
Control Engineering, 215(4), 283-303.
Ichoua, S., Gendreau, M., & Potvin, J.-Y. (2003). Vehicle dispatching with time-dependent
travel times. European Journal of Operational Research, 144(2), 379-396.
Jenelius, E., & Koutsopoulos, H. N. (2013). Travel time estimation for urban road networks
using low frequency probe vehicle data. Transportation Research Part B:
Methodological, 53, 64-81.
128
Jerbi, M., Senouci, S.-M., Rasheed, T., & Ghamri-Doudane, Y. (2009). Towards efficient
geographic routing in urban vehicular networks. IEEE Transactions on Vehicular
Technology, 58(9), 5048-5059.
Jia, D., Lu, K., Wang, J., Zhang, X., & Shen, X. (2016). A survey on platoon-based vehicular
cyber-physical systems. IEEE Communications Surveys & Tutorials, 18(1), 263-284.
Johnson, S. L., & Pawar, N. (2005). Cost-benefit evaluation of large truck-automobile speed
limit differentials on rural interstate highways. Retrieved from
Kanagaraj, V., Asaithambi, G., Kumar, C. N., Srinivasan, K. K., & Sivanandan, R. (2013).
Evaluation of different vehicle following models under mixed traffic conditions.
Procedia-Social and Behavioral Sciences, 104, 390-401.
Karnadi, F. K., Mo, Z. H., & Lan, K.-c. (2007). Rapid generation of realistic mobility models for
VANET. Paper presented at the Wireless Communications and Networking Conference,
2007. WCNC 2007. IEEE.
Kerner, B. S. (2000a). Experimental features of the emergence of moving jams in free traffic
flow. Journal of Physics A: Mathematical and General, 33(26), L221.
Kerner, B. S. (2000b). Phase transitions in traffic flow Traffic and Granular Flow’99 (pp. 253-
283): Springer.
Kerner, B. S., Klenov, S. L., & Wolf, D. E. (2002). Cellular automata approach to three-phase
traffic theory. Journal of Physics A: Mathematical and General, 35(47), 9971.
Khoury, J. E., & Hobeika, A. (2007). Incorporating uncertainty into the estimation of the passing
sight distance requirements. Computer‐Aided Civil and Infrastructure Engineering,
22(5), 347-357.
Kleijnen, J. P. (2005). An overview of the design and analysis of simulation experiments for
sensitivity analysis. European Journal of Operational Research, 164(2), 287-300.
Kloiber, B., Strang, T., Spijker, H., & Heijenk, G. (2012). Improving information dissemination
in sparse vehicular networks by adding satellite communication. Paper presented at the
Intelligent Vehicles Symposium (IV), 2012 IEEE.
Korkmaz, G., Ekici, E., & Özgüner, F. (2010). Supporting real-time traffic in multihop vehicle-
to-infrastructure networks. Transportation Research Part C: Emerging Technologies,
18(3), 376-392.
Korkmaz, G., Ekici, E., Özgüner, F., & Özgüner, Ü. (2004). Urban multi-hop broadcast protocol
for inter-vehicle communication systems. Paper presented at the Proceedings of the 1st
ACM international workshop on Vehicular ad hoc networks.
Lai, J.-Y. (2016). A Study on Remote Wireless Diagnostic System for Vehicles by Integrating
the Technology of OBD firmware adapter, Smartphone, and Laptop. (12), 11-25.
Li, F., & Wang, Y. (2007). Routing in vehicular ad hoc networks: A survey. IEEE Vehicular
technology magazine, 2(2).
Li, L., & Chen, X. M. (2017). Vehicle headway modeling and its inferences in
macroscopic/microscopic traffic flow theory: A survey. Transportation Research Part
C: Emerging Technologies, 76, 170-188.
Liang, W., Li, Z., Zhang, H., Wang, S., & Bie, R. (2015). Vehicular ad hoc networks:
architectures, research issues, methodologies, challenges, and trends. International
Journal of Distributed Sensor Networks, 11(8), 745303.
Litman, T., & Burwell, D. (2006). Issues in sustainable transportation. International Journal of
Global Environmental Issues, 6(4), 331-347.
Liu, N. (2011). Internet of Vehicles: Your next connection. Huawei WinWin, 11, 23-28.
129
Lochert, C., Hartenstein, H., Tian, J., Fussler, H., Hermann, D., & Mauve, M. (2003). A routing
strategy for vehicular ad hoc networks in city environments. Paper presented at the
Intelligent Vehicles Symposium, 2003. Proceedings. IEEE.
Lu, N., Cheng, N., Zhang, N., Shen, X., & Mark, J. W. (2014). Connected vehicles: Solutions
and challenges. IEEE internet of things journal, 1(4), 289-299.
Mangharam, R. (2008). Real-time embedded wireless networks: Algorithms and experiences.
Carnegie Mellon University.
Martinez, F. J., Toh, C. K., Cano, J. C., Calafate, C. T., & Manzoni, P. (2011). A survey and
comparative study of simulators for vehicular ad hoc networks (VANETs). Wireless
Communications and Mobile Computing, 11(7), 813-828.
May, A. D. (1990). Traffic flow fundamentals.
Meshkov, E. (1969). Instability of the interface of two gases accelerated by a shock wave. Fluid
Dynamics, 4(5), 101-104.
Milazzo II, J., Rouphail, N., Hummer, J., & Allen, D. (1999). Quality of service for interrupted-
flow pedestrian facilities in highway capacity manual 2000. Transportation Research
Record: Journal of the Transportation Research Board(1678), 25-31.
Mohimani, G. H., Ashtiani, F., Javanmard, A., & Hamdi, M. (2009). Mobility modeling, spatial
traffic distribution, and probability of connectivity for sparse and dense vehicular ad
hoc networks. IEEE Transactions on Vehicular Technology, 58(4), 1998-2007.
Nagel, K. (1996). Particle hopping models and traffic flow theory. Physical review E, 53(5),
4655.
Nagel, K., & Herrmann, H. J. (1993). Deterministic models for traffic jams. Physica A:
Statistical Mechanics and its Applications, 199(2), 254-269.
Nagel, K., & Schreckenberg, M. (1992). A cellular automaton model for freeway traffic. Journal
de physique I, 2(12), 2221-2229.
Naumov, V., Baumann, R., & Gross, T. (2006). An evaluation of inter-vehicle ad hoc networks
based on realistic vehicular traces. Paper presented at the Proceedings of the 7th ACM
international symposium on Mobile ad hoc networking and computing.
Naus, G., Van Den Bleek, R., Ploeg, J., Scheepers, B., van de Molengraft, R., & Steinbuch, M.
(2008). Explicit MPC design and performance evaluation of an ACC Stop-&-Go. Paper
presented at the American Control Conference, 2008.
Nekovee, M., & Bogason, B. B. (2007). Reliable and effcient information dissemination in
intermittently connected vehicular adhoc networks. Paper presented at the Vehicular
Technology Conference, 2007. VTC2007-Spring. IEEE 65th.
Niu, W., Li, J., Liu, S., & Talty, T. (2007). Intra-vehicle ultra-wideband communication testbed.
Paper presented at the Military Communications Conference, 2007. MILCOM 2007.
IEEE.
Olaverri-Monreal, C., Gomes, P., Fernandes, R., Vieira, F., & Ferreira, M. (2010). The See-
Through System: A VANET-enabled assistant for overtaking maneuvers. Paper
presented at the Intelligent Vehicles Symposium (IV), 2010 IEEE.
Omar, H. A., Zhuang, W., & Li, L. (2013). VeMAC: A TDMA-based MAC protocol for reliable
broadcast in VANETs. IEEE Transactions on Mobile Computing, 12(9), 1724-1736.
Ozer, U., Kahriman, A., Aksoy, M., Adiguzel, D., & Karadogan, A. (2008). The analysis of
ground vibrations induced by bench blasting at Akyol quarry and practical blasting
charts. Environmental geology, 54(4), 737-743.
130
Papageorgiou, M., Diakaki, C., Dinopoulou, V., Kotsialos, A., & Wang, Y. (2003). Review of
road traffic control strategies. Proceedings of the IEEE, 91(12), 2043-2067.
Qian, Y., & Moayeri, N. (2008). Design of secure and application-oriented VANETs. Paper
presented at the Vehicular Technology Conference, 2008. VTC Spring 2008. IEEE.
Qu, F., Wang, F.-Y., & Yang, L. (2010). Intelligent transportation spaces: vehicles, traffic,
communications, and beyond. IEEE Communications Magazine, 48(11).
Ramadan, M. N., Al-Khedher, M. A., & Al-Kheder, S. A. (2012). Intelligent anti-theft and
tracking system for automobiles. International Journal of Machine Learning and
Computing, 2(1), 83.
Raya, M., Aziz, A., & Hubaux, J.-P. (2006). Efficient secure aggregation in VANETs. Paper
presented at the Proceedings of the 3rd international workshop on Vehicular ad hoc
networks.
Reis, A. B., Sargento, S., & Tonguz, O. K. (2011). On the performance of sparse vehicular
networks with road side units. Paper presented at the Vehicular Technology Conference
(VTC Spring), 2011 IEEE 73rd.
Rickert, M., Nagel, K., Schreckenberg, M., & Latour, A. (1996). Two lane traffic simulations
using cellular automata. Physica A: Statistical Mechanics and its Applications, 231(4),
534-550.
Ros, F. J., Martinez, J. A., & Ruiz, P. M. (2014). A survey on modeling and simulation of
vehicular networks: Communications, mobility, and tools. Computer communications,
43, 1-15.
Rudack, M., Meincke, M., Jobmann, K., & Lott, M. (2003). On traffic dynamical aspects of inter
vehicle communications (IVC). Paper presented at the Vehicular Technology
Conference, 2003. VTC 2003-Fall. 2003 IEEE 58th.
Rudack, M., Meincke, M., & Lott, M. (2002). On the dynamics of ad hoc networks for inter
vehicle communications (IVC). Paper presented at the proc. ICWN.
Saltelli, A., & Annoni, P. (2010). How to avoid a perfunctory sensitivity analysis. Environmental
Modelling & Software, 25(12), 1508-1517.
Santi, P. (2005). Topology control in wireless ad hoc and sensor networks. ACM computing
surveys (CSUR), 37(2), 164-194.
Santos, R. A., Edwards, A., Edwards, R., & Seed, N. L. (2005). Performance evaluation of
routing protocols in vehicular ad-hoc networks. International Journal of Ad Hoc and
Ubiquitous Computing, 1(1-2), 80-91.
Schrank, D., Eisele, B., & Lomax, T. (2012). TTI’s 2012 urban mobility report. Texas A&M
Transportation Institute. The Texas A&M University System, 4.
Schumacher, H., Priemer, C., & Slottke, E. N. (2009). A simulation study of traffic efficiency
improvement based on Car-to-X communication. Paper presented at the Proceedings of
the sixth ACM international workshop on VehiculAr InterNETworking.
Setting, N. R. C. T. R. B. C. f. G. o., & Limits, E. S. (1998). Managing speed: review of current
practice for setting and enforcing speed limits (Vol. 254): Transportation Research
Board.
Sheng, Z., Mahapatra, C., Zhu, C., & Leung, V. C. (2015). Recent advances in industrial wireless
sensor networks toward efficient management in IoT. IEEE access, 3, 622-637.
Shercliff, J. (1953). Steady motion of conducting fluids in pipes under transverse magnetic fields.
Paper presented at the Mathematical Proceedings of the Cambridge Philosophical
Society.
131
Singh, T., & Singh, V. (2005). An intelligent approach to prediction and control ground vibration
in mines. Geotechnical & Geological Engineering, 23(3), 249-262.
Sommer, C., German, R., & Dressler, F. (2011). Bidirectionally coupled network and road traffic
simulation for improved IVC analysis. IEEE Transactions on mobile computing, 10(1),
3-15.
Su, H., & Zhang, X. (2007). Clustering-based multichannel MAC protocols for QoS
provisionings over vehicular ad hoc networks. IEEE Transactions on Vehicular
Technology, 56(6), 3309-3323.
Sugiyama, Y., Fukui, M., Kikuchi, M., Hasebe, K., Nakayama, A., Nishinari, K., . . . Yukawa, S.
(2008). Traffic jams without bottlenecks—experimental evidence for the physical
mechanism of the formation of a jam. New journal of physics, 10(3), 033001.
Tayyaran, M. R., & Khan, A. M. (2003). The effects of telecommuting and intelligent
transportation systems on urban development. Journal of Urban Technology, 10(2), 87-
100.
Tian, J., Hahner, J., Becker, C., Stepanov, I., & Rothermel, K. (2002). Graph-based mobility
model for mobile ad hoc network simulation. Paper presented at the Simulation
Symposium, 2002. Proceedings. 35th Annual.
Tian, J., Han, L., & Rothermel, K. (2003). Spatially aware packet routing for mobile ad hoc
inter-vehicle radio networks. Paper presented at the Intelligent Transportation Systems,
2003. Proceedings. 2003 IEEE.
Tonguz, O., Wisitpongphan, N., Bait, F., Mudaliget, P., & Sadekart, V. (2007). Broadcasting in
VANET. Paper presented at the 2007 mobile networking for vehicular environments.
Toor, Y., Muhlethaler, P., & Laouiti, A. (2008). Vehicle ad hoc networks: Applications and
related technical issues. IEEE Communications Surveys & Tutorials, 10(3).
Torrent-Moreno, M., Mittag, J., Santi, P., & Hartenstein, H. (2009). Vehicle-to-vehicle
communication: fair transmit power control for safety-critical information. IEEE
Transactions on Vehicular Technology, 58(7), 3684-3703.
Tuohy, S., Glavin, M., Hughes, C., Jones, E., Trivedi, M., & Kilmartin, L. (2015). Intra-vehicle
networks: A review. IEEE Transactions on Intelligent Transportation Systems, 16(2),
534-545.
Uchimura, Y., Nasu, T., & Takahashi, M. (2010). IEEE 802.11-based wireless sensor system for
vibration measurement. Advances in Civil Engineering, 2010.
Van Arem, B., Van Driel, C. J., & Visser, R. (2006). The impact of cooperative adaptive cruise
control on traffic-flow characteristics. IEEE Transactions on intelligent transportation
systems, 7(4), 429-436.
Vinel, A. (2012). 3GPP LTE versus IEEE 802.11 p/WAVE: Which technology is able to support
cooperative vehicular safety applications? IEEE Wireless Communications Letters,
1(2), 125-128.
Wang, T., Song, L., & Han, Z. (2013). Coalitional graph games for popular content distribution
in cognitive radio VANETs. IEEE Transactions on Vehicular Technology, 62(8), 4010-
4019.
Waters, M., & Laker, I. (1980). Research on fuel conservation for cars. Retrieved from
Weil, R., Wootton, J., & Garcia-Ortiz, A. (1998). Traffic incident detection: Sensors and
algorithms. Mathematical and computer modelling, 27(9-11), 257-291.
132
Wisitpongphan, N., Bai, F., Mudalige, P., Sadekar, V., & Tonguz, O. (2007). Routing in sparse
vehicular ad hoc wireless networks. IEEE journal on Selected Areas in
Communications, 25(8).
Yousefi, S., Mousavi, M. S., & Fathy, M. (2006). Vehicular ad hoc networks (VANETs):
challenges and perspectives. Paper presented at the ITS Telecommunications
Proceedings, 2006 6th International Conference on.
Zang, Y., Sories, S., Gehlen, G., & Walke, B. (2009). Towards a European Solution for
Networked Cars-Integration of Car-to-Car technology into cellular systems for
vehicular communication in Europe. Speech, ITU, Geneva, Switzerland, 14.
Zeadally, S., Hunt, R., Chen, Y.-S., Irwin, A., & Hassan, A. (2012). Vehicular ad hoc networks
(VANETS): status, results, and challenges. Telecommunication Systems, 50(4), 217-
241.
Zhang, Z. (2006). Routing in intermittently connected mobile ad hoc networks and delay tolerant
networks: overview and challenges. IEEE Communications Surveys & Tutorials, 8(1),
24-37.
Zhao, L., Lai, Y.-C., Park, K., & Ye, N. (2005). Onset of traffic congestion in complex networks.
Physical review E, 71(2), 026125.
133

MPhil Thesis

Uploaded by

Copyright:

Available Formats

MPhil Thesis

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

MPhil Thesis

Uploaded by

Copyright:

Available Formats

UNIVERSITY OF GHANA

COLLEGE OF BASIC AND APPLIED SCIENCES

IMPROVING INTELLIGENT TRANSPORTATION SYSTEMS (ITS) THROUGH

THIS THESIS IS SUBMITTED TO THE UNIVERSITY OF GHANA, LEGON IN

Adams Addison Kobla Azameti

Dr. Ferdinand Apietu. Katsriku Signed: --------------------------------

Dr. Jamal-Deen Abdulai Signed: --------------------------------

integrated transport network. Vehicle-to-vehicle (V2V) communication in VANETs poses

in an infrastructure-less vehicular ad hoc network (VANETs) with safety and non-safety

applications is quite complex.

analytical investigation is implemented in MATLAB in which vehicular density is constant and

increases with time.

design parameters were used, A: 1 ≤ λ ≤ 3; B: 1≤ λ ≤ 4 and C: 1 ≤ λ ≤ 5. The level of services in

disconnections within the stated design parameters of A: 1 ≤ λ ≤ 3; B: 1 ≤ λ ≤ 4 and C: 1 ≤ λ ≤

improve the challenges in sparse network conditions.

applications in VANETs V2V communication.

my colleagues for their insightful criticism during seminars and presentations.

these scholarly contributions to the body of knowledge.

helped in diverse ways to make my studies a memorable one.

on the road networks.

Chapter 1 Introduction ……………………...………………..……………….................1

Chapter 2 Literature Review……………………………………………….…..…….....9

Chapter 4 Improving VANETs congestion in dense networks and

Chapter 5 Evaluation of the simulation results……………………………….......104

5.2.2 Network scalability……………………………………………………………….....108

Chapter 6 Conclusions and Future Directions………………………….………...114

Figure 2.1 Overview of VANETs Connected Vehicles……….…………………………….26

Table 3.1 Traffic flow state for different densities………………………………………......71

λ = sensitivity factor of a two-vehicle interaction in the car-following model

1.1 Background of the study

Mahapatra, Zhu, & Leung, 2015).

navigation systems. However, these applications in VANETs, pose numerous challenges

vehicles concerning dangerous situation like accidents and many others.

development of V2V communication protocols. New wireless communication standard is

required to be used for message dissemination schemes capable of exchanging messages at

different networks conditions will improve high-speed real-time communication to enable

collision warning in ITS.

characteristics of VANETs is used to establish communication among vehicles and therefore

vehicles (nodes) through different methodology according to (Lai, 2016).

the broadcast storm problems in VANETs

v. Ability to use the above information to determine various thresholds parameters to

Above information is an important parameter necessary to satisfy any congestion control

ahead; depend on the need to use vehicle-to-vehicle (V2V) communication or vehicle-to-

infrastructure (V2I) communication. The understanding of these architectures leads, to the

1.3 Objectives of the study

1.3.1. General objective

The general objective is to conduct an analytical investigation into Intelligent Transportation

i. To conducts extensive performance analyses of vehicular traffic flow theory together

suitable parameters to achieve the objectives of the study.

investigation into the macroscopic traffic flow model to ascertain spatio-temporal

variations in network density in VANETs to mitigate the reachability and scalability

importance and with evidence in dense and sparse networks conditions.

iv. To present an interpretation of the results of the analytical investigation of the

1.4 Scope of the study

extended to more complex traffic models.

1.5 Justification of the proposed approach

two decades, wireless communication technologies have witnessed unprecedented growth,

due to developments in the field of instrumentation and in communication technology (Anda

traffic flow model to understand network variability in density. A successful implementation