Broadcasting Protocols in Vehicular

Ad-Hoc Networks (VANETs)

By

Mostafa M. I. Taha
B.Sc. Electrical Engineering, Assiut University, 2004



A Thesis
Submitted in partial fulfillment of the requirements
for the degree


MASTER OF SCIENCE


Department of Electrical Engineering
Assiut University
Assiut, EGYPT.

2008
Assiut University
Faculty of Engineering







Broadcasting Protocols in Vehicular
Ad-Hoc Networks (VANETs)

By

Mostafa M. I. Taha
B.Sc. Electrical Engineering, Assiut University, 2004

A Thesis
Submitted in partial fulfillment of the requirements
for the degree


MASTER OF SCIENCE

Department of Electrical Engineering
Assiut University
Assiut, EGYPT.
2008

Supervised by:
Prof. Abdel Karim El-Wardany
(Assiut University)
Dr. Tarik K. Abdelhamid
(Assiut University)
Dr. Yassin M. Yassin
(Assiut University)

Discussion committee:
Prof. Ibrahim Elsayed Ziedan
(Zagazig University)
Prof. Hosny M. Ibrahim
(Assiut University)
Prof. Abdel Karim El-Wardany
(Assiut University)
Dr. Yassin M. Yassin
(Assiut University)
Assiut University
Faculty of Engineering



II

ABSTRACT

Wireless communications are becoming the dominant form of transferring information,
and the most active research field. In this dissertation, we will present one of the most
applicable forms of Ad-Hoc networks; the Vehicular Ad-Hoc Networks (VANETs). VANET
is the technology of building a robust Ad-Hoc network between mobile vehicles and each
other, besides, between mobile vehicles and roadside units.
The work begins with an introduction to VANET technology, its possible applications,
unique characteristics and promising challenges. It also demystifies some excerpts from the
IEEE 802.11 standard that are related to the operation in the Ad-Hoc mode and illustrates the
main points of its amendment in vehicular environments (IEEE 802.11p). Reliable
broadcasting of messages in self-organizing Ad-Hoc networks is a promising research field
with hundreds of published papers. This work presents a comprehensive study of the
significant broadcasting protocols in VANET environments.
The thesis contribution is a novel reliable broadcasting protocol that is especially designed
for an optimum performance of public-safety related applications. There are four novel ideas
presented in this thesis, namely choosing the nearest following node as the network probe
node, headway-based segmentation, non-uniform segmentation and application adaptive. The
integration of these ideas results in a protocol that possesses minimum latency, minimum
probability of collision in the acknowledgment messages and unique robustness at different
speeds and traffic volumes.
The performance of the proposed protocol has been studied using simulation programs and
it proved a superior performance over all previously published ones.



III

TABLE OF CONTENTS

Chapter Page
Chapter 1 Introduction ............................................................................................................... 1
1.1 What is VANET ............................................................................................................... 1
1.2 Why VANET .................................................................................................................... 2
1.3 What is Ad-Hoc ................................................................................................................ 4
1.4 Why Ad-Hoc .................................................................................................................... 6
1.5 Why Broadcasting ............................................................................................................ 6
1.6 Thesis Contributions ........................................................................................................ 7
1.7 Outline .............................................................................................................................. 7
Chapter 2 Background................................................................................................................ 8
2.1 VANET Applications ....................................................................................................... 8
2.2 VANET Characteristics .................................................................................................. 11
2.3 VANET Open-Research Challenges .............................................................................. 13
2.4 VANET Simulation ........................................................................................................ 14
2.5 IEEE 802.11 MAC ......................................................................................................... 16
2.5.1 Channel Access Functions ...................................................................................... 17
2.5.2 Interframe Spaces (IFS) .......................................................................................... 17
2.5.3 Random Backoff Time ............................................................................................ 20
2.5.4 RTS/CTS Handshaking ........................................................................................... 21
2.6 WAVE System Architecture .......................................................................................... 23
2.6.1 WAVE Physical Layer ............................................................................................ 25
2.6.2 WAVE Channel Coordination ................................................................................ 26
2.6.3 WAVE Basic Service Set ........................................................................................ 27
2.6.4 WAVE Communication Protocols .......................................................................... 28
2.6.4.1 Internet Protocol Version 6 (IPv6) ................................................................... 28
2.6.4.2 WAVE Short Message Protocol (WSMP) ....................................................... 28
2.6.5 WAVE Management Plane ..................................................................................... 29
2.6.6 WAVE Synchronization .......................................................................................... 29
Chapter 3 Previous Work ......................................................................................................... 31
3.1 Categories of Broadcasting Protocols ............................................................................ 31
3.2 Why not IEEE 802.11 .................................................................................................... 32
3.3 Reliable Protocols .......................................................................................................... 33
3.3.1 Rebroadcasting ........................................................................................................ 33
3.3.2 Selective Acknowledgment ..................................................................................... 35
3.3.3 Changing Parameters ............................................................................................... 35
3.4 Dissemination Protocols ................................................................................................. 36
3.4.1 Flooding .................................................................................................................. 37



IV

3.4.2 Single Relay ............................................................................................................ 38
Chapter 4 Theoretical Analysis ................................................................................................ 41
4.1 Introduction .................................................................................................................... 41
4.1.1 The Design Objective .............................................................................................. 41
4.1.2 Broadcasting Goals ................................................................................................. 42
4.1.3 Assumptions ............................................................................................................ 43
4.2 The Starting Block ......................................................................................................... 43
4.2.1 Frame Exchange Sequence ...................................................................................... 44
4.2.2 The Basic Algorithm ............................................................................................... 44
4.3 Step-1: Safety Related Applications .............................................................................. 45
4.3.1 Discussion ............................................................................................................... 46
4.4 Step-2: A Headway-Based Segmentation ...................................................................... 47
4.4.1 Discussion ............................................................................................................... 50
4.5 Step-3: Non-uniform Segmentation (Headway Model) ................................................. 52
4.5.1 Headway Model ...................................................................................................... 52
4.5.1.1 The Semi-Poisson Distribution ........................................................................ 54
4.5.2 Protocol Improvement ............................................................................................. 55
4.5.3 Analytical Results ................................................................................................... 59
4.6 Step-4: Application Adaptive (Modes of Operation) ..................................................... 60
4.6.1 Mode 0 Basic Broadcasting ................................................................................. 60
4.6.2 Mode 1 The Furthest Following Vehicle ............................................................. 61
4.6.3 Mode 2 The Nearest-in-time Following Vehicle ................................................. 61
4.6.4 Mode 3 The Furthest Leading Vehicle ................................................................ 62
4.7 The Proposed Algorithm ................................................................................................ 63
4.7.1 Algorithm of the Transmitting node ........................................................................ 63
4.7.2 Algorithm of Other Vehicles ................................................................................... 64
Chapter 5 Simulation Results ................................................................................................... 67
5.1 Performance Metrics ...................................................................................................... 67
5.2 Measurement Methodology ............................................................................................ 68
5.3 Simulation Parameters .................................................................................................... 69
5.4 Random Number Generator ........................................................................................... 69
5.5 Simulation Results .......................................................................................................... 69
5.6 Robustness at Different Traffic Volumes ....................................................................... 71
5.7 Protocol Comparison ...................................................................................................... 74
Chapter 6 Conclusion ............................................................................................................... 76
Appendix A - List of Co-authored Publications ....................................................................... 77
Appendix B - Word-Wide VANET Projects ............................................................................ 78
Appendix C - VANET Simulation Programs ........................................................................... 79
Appendix D - MATLAB Scripts .............................................................................................. 80
Appendix E - References .......................................................................................................... 94



V


LIST OF TABLES

Table Page

Table 2-1 IEEE 802.11 channel access functions .............................................................................................. 17
Table 2-2 QoS Access Categories ....................................................................................................................... 19
Table 2-3 WAVE physical characteristics ........................................................................................................ 26
Table 2-4 EDCA parameter set used in CCH ................................................................................................... 26
Table 2-5 Default EDCA parameter set used in SCH ...................................................................................... 27
Table 4-1 Best segmentation points for 330 vehicle /h (in headway sec) ........................................................ 59
Table 5-1 Matlab parameters ............................................................................................................................. 69
Table 5-2 Best segmentation points for 1300 vehicle /h (in headway sec) ...................................................... 72







VI

LIST OF FIGURES

Figure Page

Fig. 1-1. Node types in VANETs .......................................................................................................................... 2
Fig. 1-2. Uses of Ad-Hoc networks in wars and emergencies ............................................................................ 4
Fig. 1-3. Wireless Sensor Network and a sample tiny sensor ............................................................................ 4
Fig. 1-4. Wireless Mesh Network ......................................................................................................................... 5
Fig. 2-1. The GM's V2V system and a sample transceiver ................................................................................ 9
Fig. 2-2. Interframe spaces in 802.11 ................................................................................................................. 18
Fig. 2-3. Exponential increase of CW ................................................................................................................ 21
Fig. 2-4. Hidden node problem ........................................................................................................................... 21
Fig. 2-5. RTS/CTS/data/ACK timeline .............................................................................................................. 22
Fig. 2-6. WAVE system components .................................................................................................................. 23
Fig. 2-7. WAVE protocol stack .......................................................................................................................... 24
Fig. 2-8. Spectrum of WAVE Channels ............................................................................................................. 25
Fig. 2-9. WSM frame format .............................................................................................................................. 28
Fig. 2-10. WAVE Synchronization .................................................................................................................... 30
Fig. 3-1. Different categories of broadcasting protocols .................................................................................. 32
Fig. 4-1. Arrangement of segments for the basic algorithm ............................................................................ 45
Fig. 4-2. Arrangement of segments for step-1 modification ............................................................................ 45
Fig. 4-3. Collisions at far range nodes ............................................................................................................... 46
Fig. 4-4. Headway ................................................................................................................................................ 48
Fig. 4-5. Distance-based segmentation ............................................................................................................... 49
Fig. 4-6. Headway-based segmentation ............................................................................................................. 49
Fig. 4-7. Assuming a single lane highway .......................................................................................................... 50
Fig. 4-8. Sample Headway models ..................................................................................................................... 53
Fig. 4-9. Headway at different traffic volumes ................................................................................................. 53
Fig. 4-10. Semi-Poisson Headway Model .......................................................................................................... 54
Fig. 4-11. Non-uniform headway-based segmentation ..................................................................................... 55
Fig. 4-12. Study area of the analytical solution ................................................................................................. 55
Fig. 4-13. Probabilities associated with an arbitrary segment ........................................................................ 57
Fig. 4-14. Suggested Distribution of Collisions ................................................................................................. 58
Fig. 4-15. Analytical calculation of Pc for best segmentation .......................................................................... 60
Fig. 4-16. Mode 0 Basic Broadcasting ........................................................................................................... 61
Fig. 4-17. Priority arrangement of mode 1 ........................................................................................................ 61
Fig. 4-18. Priority arrangement of mode 2 ........................................................................................................ 62
Fig. 4-19. Priority arrangement of mode 3 ........................................................................................................ 62
Fig. 4-20. The suggested WSM frame format ................................................................................................... 63
Fig. 4-21. Actions of the transmitting MAC ...................................................................................................... 64
Fig. 4-22. Actions of other vehicles .................................................................................................................... 65
Fig. 5-1. RTB/CTB/data/ACK timeline ............................................................................................................. 68
Fig. 5-2. Histogram of one of the variables ....................................................................................................... 70
Fig. 5-3. Simulated calculation of Pc for best segmentation ............................................................................ 70



VII

Fig. 5-4. Simulated calculation of latency at best segmentation ...................................................................... 71
Fig. 5-5. Headway distribution at 330v/h and 1300v/h ..................................................................................... 72
Fig. 5-6. P
C
for 6-seg at 1300v/h ......................................................................................................................... 73
Fig. 5-7. Latency for 6-seg at 1300v/h ................................................................................................................ 73
Fig. 5-8. Probability of Collision (protocol comparison) .................................................................................. 75
Fig. 5-9. Latency (protocol comparison) ........................................................................................................... 75



VIII

LIST OF ABBREVIATIONS

AC Access Category
ACK Acknowledgment
AFR Asynchronous Fixed Repetition (Xu, et al. algorithm)
AFR-CS Asynchronous Fixed Repetition with Carrier Sensing (Xu, et al. algorithm)
AIFS Arbitration Interframe Space
APR Asynchronous p-persistent Repetition (Xu, et al. algorithm)
APR-CS Asynchronous p-persistent Repetition with Carrier Sensing (Xu, et al. algorithm)
BMMM The Batch Mode Multicast MAC Protocol (Huang, et al. algorithm)
BMW The Broadcast Medium Window (Tang, et al. algorithm)
BPSK Binary Phase Shift Keying
CCH WAVE Control Channel
CEN European Committee for Standardization
CSMA/CA Carrier Sense Multiple Access with Collision Avoidance
CTB Clear to Broadcast
CTS Clear to Send
CW Contention Window
DCF Distributed Coordination Function
DDB The Dynamic Delayed Broadcasting (Heissenbttel, et al. algorithm)
DHCP Dynamic Host Configuration Protocol
DIFS Distributed Coordination Function Interframe Space
DSRC Dedicated Short Range Communications
EDCA Enhanced Distributed Channel Access Function
edf empirical density function
EDR Event Data Record
EIFS Extended Interframe Space
ETC Electronic Toll Collection
GPS Global positioning systems
HCCA Hybrid Controlled Channel Access
IEEE Institute of Electrical and Electronics Engineers
IFS Interframe Space
IPv6 Internet Protocol Version 6
ITS Intelligent Transportation Systems
LAN Local Area Networks
LLC Logical Link Control
MAC Media Access Control
MANET Mobile Ad-Hoc Network
MCDS Minimum Connected Dominating Set
MLME MAC Layer Management Entity
NAV Network Allocation Vector
OBU On Board Unit
PB Probability of success broadcast
PC Probability of Collision



IX

PCF Point Coordination Function
pdf probability density function
PHY Physical Layer
PIFS Point Coordination Function Interframe Space
PLME Physical Layer Management Entity
QAM Quadrature Amplitude Modulation
QoS Quality of Service
QPSK Quadrature Phase-Shift Keying
RAK Request for Acknowledgment (Huang, et al. algorithm)
RRAR The Round-Robin Acknowledge and Retransmit (Xie, et al. algorithm)
RSU Road Side Unit
RTB Ready to Broadcast
RTS Ready to Send
SB The Smart Broadcasting Protocol (Fasolo, et al. algorithm)
SCH WAVE Service Channel
SFR Synchronous Fixed Repetition (Xu, et al. algorithm)
SIFS Short Interframe Space
SPR Synchronous p-persistent Repetition (Xu, et al. algorithm)
TCP Transmission Control Protocol
TS Time-slot
UDP User Datagram Protocol
UMB The Urban Multihop Broadcast Protocol (Korkmaz, et al. algorithm)
UMB Urban Multi-Hop
UTC Coordinated Universal Time
VANET Vehicular Ad-Hoc Network
VCWC Vehicular Collision Warning Communication protocol (Yang, et al. algorithm)
WAVE Wireless Access in Vehicular Environments
WBSS WAVE Basic Service Set
WME Wave Management Entity
WSM Wave Short Message
WSMP Wave Short Message Protocol





1

Chapter 1
Introduction
Everything is becoming wireless. The fascination of mobility, accessibility and flexibility
makes wireless technologies the dominant method of transferring all sorts of information.
Satellite televisions, cellular phones and wireless Internet are well-known applications of
wireless technologies. This work presents a promising wireless application and introduces a
tiny contribution to its research community.
Wireless research field is growing faster than any other one. It serves a wide range of
applications under different topologies every one of which comes with some new specialized
protocols. In this research, we will present an introduction to a wireless technology that is
expected to be adopted by both governments and manufacturers in the very near future. It
directly affects car accidents (which is the first cause of death in the age group 1 - 44 years
[35]) and the sales of one of the largest markets. It is the technology of building a robust
network between mobile vehicles; i.e. let vehicles talk to each other. This promising
technology is literally called Vehicular Ad-Hoc Networks (VANETs).
In this research, an introduction to the technology of VANETs will be presented as well as
a new contribution with a novel broadcasting protocol.
1.1 What is VANET
VANET is the technology of building a robust Ad-Hoc network between mobile vehicles
and each other, besides, between mobile vehicles and roadside units.
As shown in Fig. 1-1, there are two types of nodes in VANETs; mobile nodes as On Board
Units (OBUs) and static nodes as Road Side Units (RSUs). An OBU resembles the mobile



2

network module and a central processing unit for on-board sensors and warning devices. The
RSUs can be mounted in centralized locations such as intersections, parking lots or gas
stations. They can play a significant role in many applications such as a gate to the Internet.


Fig. 1-1. Node types in VANETs

VANET presents a new and promising field of research, development and standardization.
Throughout the world, there are many national and international projects in governments,
industry, and academia devoted to the development of VANET protocols (Appendix
B). These projects include consortiums like The Dedicated Short Range Communications
(DSRC) (USA) [8], the Car-to-Car Communication (Europe) [6] and the Intelligent
Transportation Systems (Japan) [27], and standardization efforts like the IEEE 802.11p
Wireless Access in Vehicular Environment (WAVE) [22]. An introduction to the WAVE
standard will be discussed in Sec 2.6.
1.2 Why VANET
The Bureau of Transportation Statistics [44] reported that, in 2004 within the USA only,
there were more than 6.4 million kilometers of highway, with more than 243 million
registered vehicles of different types running through them. During that year, there were more
than 6.18 million vehicle crashes causing approximately 2.79 million injuries and 42,000
fatalities. Car accidents are the leading cause of death in the age group of 1 to 44 years [35].
These accidents cost more than $150 billion per year [11]. With these terrific numbers,
Internet
RSU
OBU



3

considerable governmental and other related agencies' as well as investments of vehicles
manufacturers have been there trying to safety of roads.
Accordingly, vehicle manufacturers are competing in equipping their vehicles with devices
that collect data from the interior and exterior of vehicles and deliver it to a central processing
unit that can analyze this data to boost the road safety while increasing the on-board luxury.
Global positioning systems (GPS), Event Data Record (EDR) resembling the Black-Box used
in avionics, small range radars, night vision, light sensors, rain sensors and navigation
systems are well-known intelligent devices used in many newly produced vehicles, what is
rather referred to as "Computers-on-Wheels".
Communication researchers have been recently working on a prominent step; if each
vehicle has a device that can communicate with other vehicles, vehicles will have a gigantic
new source of information that extends beyond the capabilities of all previously mentioned
devices. For example, all of these devices cannot warn the driver of a stopping vehicle in the
next turn and of course cannot let travelers enjoy video chatting and file sharing at no charge.
Moreover, with this technology, vehicles can talk to each other and inform each other of any
probable danger and may even respond to that danger in a cooperative manner, i.e.,
introducing what may be rather referred to as "Computer Networks-on-Wheels".
Under heavy industrial pressure, it is obvious that VANETs are likely to become the most
relevant realization of mobile Ad-Hoc networks. Motivations of the promising VANET
technology include but are not limited to,
1. Increase traveler safety
2. Enhance traveler mobility
3. Decrease travelling time
4. Conserve energy and protect the environment
5. Magnify transportation system efficiency
6. Boost on-board luxury
Related governmental authorities (e.g. [10]) are expected to set a number of new rules and
regulations forcing all vehicle manufacturers to equip their vehicles with VANET transceivers
employing some of the required safety applications.



4

1.3 What is Ad-Hoc
Mobile Ad-Hoc Network (MANET) is a wireless technology where all nodes are one level
topology and can communicate directly with each other through a single hop or multi-hop
without the need of centralized nodes. The crucial usefulness of this technology arises when it
is required to build a network with a very fast deployment time and when is difficult to have
static centralized nodes such in cases of battlefields, forests or in natural catastrophes.


Fig. 1-2. Uses of Ad-Hoc networks in wars and emergencies

Before discussing why Ad-Hoc is the preferred topology for vehicular networks, it is
suitable to mention other respectful forms of MANET that took much research efforts with a
wide range of remarkable applications. These forms are Wireless Sensor Networks and
Wireless Mesh Networks. Distinguishable characteristics of VANETs will be highlighted
based on this brief introduction.
In wireless sensor networks [39] , a large set of sensors are thrown randomly in a large
area using an airplane or any other throwing sort. Each sensor is only of a coin size (Fig. 1-3
[31]) and equipped with a transceiver, small battery and any of temperature, vibration, light or
humidity sensors and even a microphone or camera.

Fig. 1-3. Wireless Sensor Network and a sample tiny sensor
Gateway
Sensor Node
Sensor Node



5

These sensors coordinate between each other to scan the investigated area of any required
information such as conflagrations, earthquakes, animal activities or human activities. This
information could latterly be delivered to a single terminal node acting as a gateway to a
remote server. This information is of great usefulness in the prediction of natural catastrophes,
statistical studies and spying activities.
Wireless mesh networks [24] have better properties in terms of robustness, range
extendibility and density. It consists of multiple radio nodes, on condition that, there are at
least two communication links available at each node, hence redundancy and capability of
high density. The coverage area of these nodes forms a large mesh cloud. When any node can
no longer operate, all the rest nodes can still communicate with each other directly or through
one or more intermediate nodes, hence reliability. A new access to this cloud is dependent
only on being in a connection with any node in this cloud, hence extendibility. The figure
below shows a sample wireless mesh network (Fig. 1-4).


Fig. 1-4. Wireless Mesh Network

Both wireless sensor networks and wireless mesh networks received a considerable amount
of research in the past few years and resulted in new sets of standards. As for wireless sensor
networks, researchers suggest using the new ZigBee IEEE 802.15.4 [17] standard to cover
challenging problems such as low power at low data rates. As for wireless mesh networks, the
IEEE came up with IEEE 802.11s [24] as an amendment to the IEEE 802.11 Wireless
LAN Standard to cover challenging problems such as power consumption and security.
In Sec 2.2, we will present VANET distinguishable characteristics and how it is different
from other forms of MANET.



6

1.4 Why Ad-Hoc
Although positioning static centralized infrastructure nodes will even increase the
information offered to travelers and OBUs (as they may be used as gates to the Internet),
vehicular networks should make use of but not depend on these nodes. The elephantine size of
paved roads and high mobility of nodes limit the usefulness of any static infrastructure node.
Researchers recommend this network to be in the Ad-Hoc topology where RSUs act as
regular nodes. This topology will fasten the rate of deployment as the industry will not wait
for the infrastructure to be built. Besides, it will offer the service at no charge. Literally
speaking, VANET is a special case of the general MANET to provide communications among
nearby vehicles and between vehicles and nearby fixed roadside equipments.
1.5 Why Broadcasting
Duo to the high mobility of vehicles, the distribution of nodes within the network changes
very rapidly and unexpectedly that wireless links initialize and break down frequently and
unpredictably. Taking into consideration that VANET operates in the absence of servers,
force OBUs to organize network resources distributively. Thereupon, broadcasting of
messages in VANETs plays a crucial rule in almost every application and requires novel
solutions that are different from any other form of Ad-Hoc networks. Broadcasting of
messages in VANETs is still an open research challenge and needs some efforts to reach an
optimum solution.
So, what are the problems associated with broadcasting that we devoted a master level
study for its protocols? Although we let the entire Chapter 3 to answer this question, it is
convenient to summarize the answer. Broadcasting requirements are: high reliability and high
dissemination speed with minimum latency in single-hop as well as multi-hop
communications. Problems associated with regular broadcasting algorithms are: the high
probability of collision in the broadcasted messages, the lack of feedback and the hidden node
problem. In VANETs, there are two types of collisions, collisions of wireless messages in the
network domain and the physical collisions of running vehicles. Throughout this work, the



7

default type of collision is the collision between messages in the network domain except what
is explicitly said as a vehicular collision.
1.6 Thesis Contributions
The thesis contribution is a novel reliable broadcasting protocol that is especially designed
for an optimum performance of public-safety related applications. There are four novel ideas
presented in this thesis, namely choosing the nearest following node as the network probe
node, headway-based segmentation, non-uniform segmentation and application adaptive. The
integration of these ideas results in a protocol that possesses minimum latency, minimum
probability of collision in the acknowledgment messages and unique robustness at different
speeds and traffic volumes.
The performance of the proposed protocol has been studied using simulation programs and
it proved a superior performance over all previously published ones.
1.7 Outline
This dissertation is organized as follows;
- Chapter 2 is a background on the VANET technology. This chapter presents some of the
required applications of VANETs, introducing the outcomes of this new technology. It also
introduces the unique characteristics of VANETs, VANET challenging research areas,
simulation environments and the current state of standardization process. Although it is not
directly related to the new contribution, this background is mandatory to understand the area
of research.
- Chapter 3 provides a comprehensive study of the different objectives of broadcasting in
VANETs. Accordingly, this chapter provides a brief description of the currently published
broadcasting protocols formed in a new categorization.
- Chapter 4 provides the analytical analysis of the proposed protocol with excessive
description and analysis.
- Chapter 5 presents simulation results and protocol comparison.
- Chapter 6 presents the conclusion.



8

Chapter 2
Background
Since the first invention of mobile vehicles, governments and manufacturers have
researched accidents to reduce the number of vehicle crashes in order to reduce costs, injuries
and fatalities. The promising VANET technology complements this work with a research that
focuses on preventing crashes on the first place. Accordingly, related governmental
authorities initiated new projects to the study, research, development and standardization of
VANETs. The Dedicated Short Range Communications (DSRC) [8] is a pioneer ITS
(Intelligent Transportation Systems which is a branch of the U.S. Department of
Transportation [26]) project dedicated to VANET standardization. Then, the acronym
DSRC becomes a worldwide name of any set of standards that aim to put VANET
technology into life. The DSRC concerns with communication links between vehicle-to-
vehicle and vehicle-to/from-roadside units.
2.1 VANET Applications
According to the DSRC, there are over one hundred recommended applications of
VANETs. These applications are of two categories, safety and non-safety related. Moreover,
they can be categorized into OBU-to-OBU or OBU-to-RSU applications. Here we list some
of these applications
- Co-operative Collision Warning,
Co-operative collision warning is an OBU-to-OBU safety
application, that is, in case of any abrupt change in speed or driving
direction, the vehicle is considered abnormal and broadcasts a warning



9

message to warn all of the following vehicles of the probable danger. This application
requires an efficient broadcasting algorithm with a very small latency.
- Lane Change Warning,
Lane-change warning is an OBU-to-OBU safety application, that is,
a vehicle driver can warn other vehicles of his intention to change the
traveling lane and to book an empty room in the approaching lane.
Again, this application depends on broadcasting.
- Intersection Collision Warning,
Intersection collision warning is an OBU-to-RSU safety application.
At intersections, a centralized node warns approaching vehicles of
possible accidents and assists them determining the suitable
approaching speed. This application uses only broadcast messages.
In June 2007, General Motors GM addressed the previously mentioned applications and
announced for the first wireless automated collision avoidance system using vehicle-to-
vehicle communication (Fig. 2-1, [13]), as quoted from GM, If the driver doesnt respond to
the alerts, the vehicle can bring itself to a safe stop, avoiding a collision.


Fig. 2-1. The GM's V2V system and a sample transceiver

- Approaching Emergency vehicle,
Approaching emergency vehicle is an OBU-to-OBU public-safety
application, that is, high-speed emergency vehicles (ambulance or
police car) can warn other vehicles to clear their lane. Again, this
application depends on broadcasting.





10

- Rollover Warning,
Rollover warning is an OBU-to-RSU safety application. A RSU
localized at critical curves can broadcast information about curve angle
and road condition, so that, approaching vehicles can determine the
maximum possible approaching speed before rollover.
- Work Zone Warning,
Work zone warning is an OBU-to-RSU safety application. A RSU is
mounted in work zones to warn incoming vehicles of the probable danger
and warn them to decrease the speed and change the driving lane.
- Coupling/Decoupling,
Coupling/decoupling system is an OBU-to-OBU non-safety
application that is designed to link multiple buses or trucks into a train
to minimize the headway distance and traveling time and to decrease
rear-end crashes. In August 2003, California PATH project practically
tested this application on a three-bus platoon [5].
- Inter-Vehicle Communications,
Inter-vehicle communication is an OBU-to-OBU non-safety
application that enables travelers to communicate with each other using
instant file transfer, voice chatting or even video chatting.
- Electronic Toll Collection (ETC),
Electronic toll collection is an OBU-to-RSU non-safety application
that supports the collection of payment at toll plazas using automated
systems to increase the operational efficiency. Systems typically
consist of OBUs that are chargeable with prepaid smart cards. These
OBUs are identified by RSUs located in dedicated lanes at toll plazas.
ETC was the first widely accepted DSRC application and it is
practically implemented in many toll collection sites. As an example, it has
been used for the congestion charge region in London downtown since
2003 [43].




11

- Parking Lot Payment,
Parking lot payment is an OBU-to-RSU non-safety application that
provides benefits to parking lot operators, simplify payment for
customers, and reduce congestion at entrances and exits of parking lots.
- Traffic Management,
In-vehicle navigation is a non-safety application that is designed to
reduce driving time and fuel consumption by exchanging real-time
information about traffic conditions in the driving route.
2.2 VANET Characteristics
Although VANETs, Wireless Sensor Networks and Wireless Mesh Networks are special
cases of the general MANETs, VANETs possess some distinguishable characteristics that
make its nature a unique one. These properties present considerable challenges and require a
set of new especially designed protocols.
- Due to the high mobility of vehicles, that can be up to one hundred fifty kilometers per
hour, the topology of any VANET changes frequently and unexpectedly. Hence, the time that
a communication link exists between two vehicles is very short especially when the vehicles
are traveling in opposite directions. A one solution to increase the lifetime of links is to
increase the transmission power, but increasing a vehicles transmission range will increase
the collision probability and degrade the overall throughput of the system. The other solution
is to have a set of new protocols employing a very low latency.
- Yet another effect of the high mobility of nodes is that the usefulness of the broadcasted
messages is very critical to latency. Assuming for example that a vehicle is suddenly
stopping, it should send a broadcast message to warn other vehicles of the probable danger.
Considering that the driver needs at least 0.70 to 0.75 sec to initiate his response [14], the
warning message should be delivered at virtually zero sec latency.
- In VANETs, location of nodes changes very quickly and unpredictably, so that, building
an efficient routing table or a list of neighbor nodes will exhaust the wireless channel and



12

decrease the network efficiency. Protocols that rely on prior information about location of
nodes are likely to have a poor performance.
- Nevertheless, the topology of a VANET can be a benefit because vehicles are not
expected to leave the paved road, hence, the running direction of vehicles is predictable to
some extent.
- Although, the design challenge of protocols in wireless sensor networks is to minimize
the power consumption, this is not a problem in VANETs. Nodes in VANETs depend on a
good power supply (e.g. vehicle battery and the dynamo) and the required transmission power
is small compared with power consumption of on-board facilities (e.g. air-condition).
- It is expected that, as VANET is initially deployed, only a small percentage of vehicles
will be equipped with transceivers. Thus, the benefits of the new technology, especially OBU-
to-OBU applications, will not rise until many years. Moreover, the limited number of vehicles
with transceivers will lead to a frequent fragmentation of the network. Even when VANET is
fully deployed, fragmentation may still exist in rural areas, thereupon, any VANET protocol
should expect a fragmented network.
- Privacy and security are of crucial effect on the public acceptance of this technology. In
VANETs, every node represents a specific person and its location tells about his location.
Any lack of privacy can ease a third party monitoring persons daily activities. However, from
the other point of view, higher authorities should gain access to identity information to ensure
punishment of illegal actions, where, there is a fear of a possible misuse of this feature. The
tampering with messages could increase false alarms and accidents in some situations
defeating the whole purpose of this technology.
Finally, the key difference between VANET protocols and any other form of Ad-Hoc
networks is the design requirement. In VANETs, the key design requirement is to minimize
latency with no prior topology information. However, the key design requirement of Wireless
Sensor Network is to maintain network connectivity with the minimum power consumption
and the key design requirement of Wireless Mesh Network is reliability.
Concluding, the main characteristics of VANETs can be summarized as follows [28];
- High mobility of nodes
- No prior information about the exact location of neighbor nodes



13

- Predictable topology (to some extent)
- Critical latency requirement especially in cases of safety related applications
- No problem with power
- Slow migration rate
- High possibility to be fragmented
- Crucial effect of security and privacy
2.3 VANET Open-Research Challenges
VANET is still a virgin research area. This section walks through some of the currently
open-research challenging areas.
- Security
Authentication versus privacy [4] is considered the most intuitively confusing challenge in
the area of VANET security. Authentication of each message is a must to ensure that
messages are originated from actual vehicles suffering from actual situations. Consider what
may happen if a normal vehicle can transmit a warning beacon message of an ambulance just
to clear its travelling lane. Moreover, higher authorities (e.g. police officers) should be able to
determine causes of accidents by investigating the pre-accident transmitted messages.
However, a third party can use this information to track vehicles of important persons
remotely.
Vehicular networks, especially in cases of public-safety applications, have a very low
tolerance to errors, i.e. tampering with these messages can increase accidents.
The critical latency requirement of VANET messages prohibits the use of complicated
time-consuming cryptographic algorithms. The expected sheer scale of the network, assuming
full deployment, rules out protocols that require pre-stored information about participating
parities.
Concluding, VANET technology requires a completely new bundle of security protocols.




14

- Broadcasting
In a self-organizing Ad-Hoc network, the challenge is how we can design a protocol that is
capable of implementing a reliable broadcasting with a minimum probability of message
collision and minimum latency.
The deployed protocol should be highly distributed and does not need any prior control
messaging. Moreover, it should take into account that vehicles are expected to be travelling at
different speeds and different environments (urban and rural). Finally, as indicated in Sec 2.1,
broadcasting supports a vast range of applications that the implemented protocol should cope
with application differences efficiently.
2.4 VANET Simulation
The problem discussed in this section is how VANET researchers are going to evaluate
their proposed protocols? The ultimate evaluation tool is by doing outdoor experiments, but
this solution has many drawbacks:
- Neither easy nor cheap to have a high number of vehicles in real scenarios especially in
case of public safety related protocols.
- Difficult to analyze the performance in highly distributed environments like the case of
VANETs.
- Impossible to compare between two protocols in the exactly same situation.
Therefore, the only appropriate evaluation tool is by using simulation programs. Any
simulation program consists of two complementary parts; network model and mobility model.
The network model is responsible for identifying the communication stack; i.e. wireless
channel model, antenna model, MAC layer, network layer, application layer and similar
issues. The network model for VANET simulation programs is the same as that of MANET
programs.
The mobility model is responsible for identifying different aspects of vehicle movement. It
is the only new issue in VANET simulation programs. Vehicular mobility models are usually
classified as being either microscopic or macroscopic models. When focusing on the
macroscopic point of view, motion constraints such as roads, streets, crossroads and traffic



15

lights are considered and the generation of vehicular traffic such as traffic density, traffic
flows, and initial distribution of vehicles are defined. The microscopic point of view, instead,
focuses on the movement of each individual vehicle and on the vehicle behavior with respect
to neighbors such as lane changing and car following models. A realistic mobility model
should include [29]:
- Accurate and realistic topological maps: Such maps should include different types of
roads that consist of different number of lanes.
- Intersections with traffic lights: Maps should contain intersection where vehicles should
slow-down. Vehicles are expected to react with traffic lights appropriately.
- Lane changing models: Drivers are not expected to still in their lanes for the entire
journey. Hence, lane-changing maneuvers should be included in the simulation.
- Smooth deceleration and acceleration: Since vehicles do not breakdown and accelerate
abruptly, deceleration and acceleration models should be included.
- Obstacles: The simulation should include obstacles in the vehicular mobility and the
wireless channel.
- Intelligent driving patterns: Drivers interact with their environments, not only with
respect to static obstacles, but also to dynamic obstacles, such as neighboring cars and
pedestrians.
- Human behaviors: Drivers are humans not machines. All driving models should be
probabilistic with a tolerance of errors which results in simulated accidents.
- Non-random distribution of vehicles: As it can be observed in real life, initial positions of
vehicles are not uniformly distributed in the simulation area.
- Different types of vehicles: The VANET technology is not addressed to sedan cars only
buses, vans, trucks, trains and motorcycles are also involved. Each type should have its
own models.
- Effect of the implemented protocol: Almost all mobility models are used to generate a
predefined traffic prior to the simulation itself, without any effect of the implemented
protocol. If the researcher wants to measure the net improvement of his protocol on the
traffic flow, he must have a simulation program that allows changing of future
movements according to events from the network model.



16

All of these features are recommended for a mobility model to be as realistic as possible,
but the researcher may not use such very complicated models because this means many
variables and a lot of time. Such complicated models may be useful only in the final
evaluation of the protocol but not during the development cycle itself where the researcher
wants to study the effect of his protocol in specific situations. Note that, the network model
used in the simulation program should also be adequate to his needs with the possibility of
developing new protocols.
Although many simulation programs are available to VANET research community, it is
expected that choosing, and getting used to, an appropriate simulation tool is the most time-
consuming problem in the protocol development cycle.
Some of the popular network simulators are NS-2, GloMoSim, QualNet, OPNet, NCTUns
and MATLAB.
Some of the popular mobility generators are VanetMobiSim and CanuMobiSim.
Some of joint mobility and network simulators are TraNS and MOVE.
Web addresses for these simulators are listed in (Appendix C).
2.5 IEEE 802.11 MAC
This section provides an overview of some concepts from the IEEE 802.11 MAC standard
[23]. The IEEE 802.11 standard defines medium access control (MAC) and physical layer
(PHY) specifications for the wireless connectivity of fixed, portable, or moving stations
within a local area network. It defines a single set of MAC procedures to support packet
delivery services and several physical signaling techniques. The IEEE 802.11 includes a long
list of amendments [38] to make the standard more suitable for specific purposes. Each one of
these amendments shares the common MAC while defining some parameters of the physical
technique. Wireless Access in Vehicular Environments (WAVE) has got its own amendment
(802.11p). The first draft of which was just in Nov 2004 and it is still a draft [40]. In this
section, only general MAC concepts related to this work will be covered, based on the IEEE
802.11-REVma/D7.0 [23]; however, WAVE specific concepts will be discussed later in
Sec 2.6.



17

2.5.1 Channel Access Functions
The IEEE 802.11 MAC defines four access functions (as shown in Table 2-1)
- DCF The Distributed Coordination Function
- PCF The Point Coordination Function
- EDCA The Enhanced Distributed Channel Access Function
- HCCA The Hybrid Controlled Channel Access

Table 2-1 IEEE 802.11 channel access functions
Ad-Hoc Coordinator Point
non-QoS DCF PCF
QoS EDCA HCCA

The DCF is the fundamental access function and the one that must be implemented by all
stations, whether the network was Ad-Hoc or server-based. The DCF is a distributed protocol
where all nodes, must first contend for access on the channel. The DCF access protocol
reduces collision probability by using carrier sense multiple access with collision avoidance
(CSMA/CA) and a random backoff time. The EDCA is similar to DCF but it is used when a
certain quality of service (QoS) is required. It provides four access priorities by assigning
each node one out of four access categories according to the running application.
Contrarily, the PCF is an optional access method, and is used in server-based networks
only. The PCF is a contention-free protocol where the coordinator point passes the channel
control to network nodes in a round robin fashion. Finally, the HCCA is just similar to PCF in
cases of QoS server-based networks.
The EDCA is the recommended access function in VANETs because the communications
in VANET environments does not depend on centralized infrastructure nodes and the
deployed applications should have different access priorities (from life-safety to file-sharing).
2.5.2 Interframe Spaces (IFS)
The Interframe space (IFS) is the time interval between transmission of two consecutive
frames from different nodes, whether it was a new session or just a handshaking packet in the



18

same session. Each station should wait for a different IFS according to its priority. There are
five different IFSs listed here from the shortest to the longest (Fig. 2-2)
- SIFS Short Interframe Space
- PIFS Point Coordination Function (PCF) Interframe Space
- DIFS Distributed Coordination Function (DCF) Interframe Space
- AIFS Arbitration Interframe Space (used by the QoS facility)
- EIFS Extended Interframe Space

Fig. 2-2. Interframe spaces in 802.11

The timing unit of the IEEE 802.11 is the Time-Slot, which is defined as the minimum
time that is required by nodes to sense the channel as idle and start a new transmission.
The SIFS should be used before transmission of frames that belong to the same session like
ACK frames, CTS frames, and the second or subsequent fragments of data. The SIFS is the
time interval from the end of a frame to the beginning of the next frame as seen at the air
interface assuming that the node responds directly without sensing the channel. the SIFS is
the shortest interframe space. It gives nodes involved in the current session the control over
the wireless medium until the end of the frame exchange sequence.
In case of server-based networks, the coordinator point should control access to the
wireless medium. Although all nodes in the network shall wait for DIFS before starting a new
session, the coordinator point gives a single node the permission to start after PIFS only. This
gives it a higher priority over other nodes. The PIFS is the tool used by the coordinator point
to maintain a contention-free medium.
PIFS = SIFS + Time-Slot
The DIFS is the default waiting time of nodes before starting a new session in both Ad-
Hoc and server-based networks. DIFS is longer than both SIFS and PIFS, which inhibits all
SIFS
Busy media
PIFS
AIFS
DIFS
EIFS
time



19

nodes from interrupting a running session they are not involved in, or taking a time-slot that
they are not allowed to.
DIFS = SIFS + 2 Time-Slot
If all nodes start transmission after the same DIFS, an unavoidable collision will happen,
hence, the IEEE 802.11 utilizes a contention algorithm that depends on assigning a random
back-off time to each node (will be discussed in details in the next section). If a node wants to
start a new session, it must sense the channel as idle for the duration of DIFS and an extra
random time.
All nodes should use the AIFS instead of DIFS whenever it is required to employ a
quality of service (QoS). The AIFS is used by nodes deploying EDCA access function. The
EDCA provides differential access to the channel by assigning to each node one out of four
access categories. These access categories are labeled according to Table 2-2, where the
Voice gets the highest priority. The AIFS is a different value for each category with a
minimum value for the Voice (highest priority).
AIFS[AC] = SIFS + AIFSN[AC] Time-Slot
where AC is the access category and AIFSN[AC] is a number associated with AIFS[AC].
Table 2-2 QoS Access Categories
Priority AC Designation
Lowest


Highest
AC_BK Background
AC_BE Best Effort
AC_VI Video
AC_VO Voice

Unlike other IFSs, EIFS is not used to control access onto the radio link, but it is only used
when there has been an error in the last transmitted frame. If the present session ends
correctly, nodes wait for DIFS and a random backoff before starting a new transmission.
However, if the present session ends erroneously, all other nodes should use the EIFS waiting
time to provide enough time for session involved nodes to correct this error.
EIFS = SIFS + DIFS + ACK transmission time



20

2.5.3 Random Backoff Time
In contention-based access functions (DCF and EDCA), channel access protocol should be
efficient while being distributed, that network nodes should achieve low collision probability
without the help of coordinator points. Recalling that, if a node wants to start a new session, it
must sense the channel as idle for the duration of DIFS (or AIFS[AC]) and an extra random
backoff time. This section discusses specifications of the random backoff time. The pool of
random numbers that is used should be big enough for minimizing collision probability in
cases of high-density networks and small enough for shorter useless waiting time in cases of
low-density networks. The IEEE 802.11 employs an adaptive size of random pool by defining
the contention window size (CW) which increases in high-density cases and decreases in low-
density ones.
Backoff Time = Random Time-Slot
where Random is a uniformly distributed random integer in the interval (0, CW), and CW is
an integer of (CW
min
CW CW
max
).
The procedure is as follows,
1- The node must first sense the channel as idle for the DIFS (or AIFS[AC]) time.
2- Choose a random backoff counter in the interval (0) to (CW
min
).
3- Sense the channel on every Time-Slot (TS).
4- If the channel was idle, decrement the backoff counter by one. If not (a busy medium),
hold the backoff counter.
5- If it reached zero, start the transmission.
If it received an ACK from the destination as an indication of a correct transmission, then it
should move on to the next fragment. However, if there was no ACK as an indication of a
collision in the transmitted message (there are two or more nodes got the same random
number and the network is denser than thought), it should increase the CW to a higher value
and redo the procedure from the beginning.
Summarizing, the CW should take a higher value if a collision happens until reaching
CW
max
and it should be reset to CW
min
after every successful transmission.
Note that, values of CW of nodes deploying DCF should be
CW = 2
(i)
- 1



21

where i equals 3 to 8 as shown in Fig. 2-3. In EDCA (VANET case), the CW
min
and CW
max
are different for each AC as will be shown in Sec 2.6.2.


Fig. 2-3. Exponential increase of CW
2.5.4 RTS/CTS Handshaking
So far, we have studied how the 802.11 minimizes collision probability by using carrier
sense mechanism and different channel-access waiting times (different IFSs and random
backoff times). However, there is still another source of collision that cannot be avoided by
the CSMA/CA, which is the hidden node problem.
Consider the case that there are four nodes arranged as shown in Fig. 2-4. N
2
is in the
communication range of both N
1
and N
3
, while N
3
is out of range of N
1
. If there is a
concurrent transmission between N
1
N
2
and between N
3
N
4
, there will be a collision at N
2

because it can hear the transmission of both N
1
and N
3
simultaneously. Note that, the
CSMA/CA has nothing to do with this type of collision as when N
3
is willing to initiate its
transmission, it cannot hear N
1
, hence it senses the channel as idle, and proceeds with the
transmission after the associated IFS.

Fig. 2-4. Hidden node problem
N
1
N
2
N
3
N
4
Busy medium
DIFS/AIFS
7 TS (CW
min
)
1
st
trial
Busy medium
DIFS/AIFS
15 TS
2
nd
trial
Busy medium
DIFS/AIFS
255 TS / CW
max

6
th
and all following trials
A new session can start at
any of these time-slots



22

The 802.11 standard addressed this problem and suggested that the transmitter should,
prior to any transmission, reserve his communication range as well as the receiver range (N
1

and N
2
in the example) by using ready to transmit / clear to transmit (RTS/CTS) handshaking.
In case that N
1
(transmitter) has a long message to send to N
2
(receiver), the procedure will be
as follows:
1- It sends an unencrypted broadcast with the RTS message indicating the transmitter address
(N
1
), intended receiver address (N
2
) and the expected time required.
2- The receiver (N
2
) should reply with an unencrypted broadcast with the CTS message
indicating the CTS-transmitter address (N
2
), CTS-receiver address (N
1
) and the expected
time required.
The RTS reserves the transmitter communication range, while the CTS reserves the receiver
communication range. The hidden node (N3) will hear the CTS message, know about the
medium reservation and wait for the time reservation before resuming contention for the
channel.
Each node should maintain a network allocation vector (NAV) as an indicator of time periods
when transmission is not allowed. Data in the NAV is updated by time requirements in the
RTS and CTS messages.
The timeline of the sequence [RTS/CTS/DATA/ACK] is shown in Fig. 2-5.


Fig. 2-5. RTS/CTS/data/ACK timeline

Note that, the RTS message itself may still suffer from unexpected collisions due to hidden
node problem and should only be used prior to long messages, however, for short messages,
the RTS/CTS handshaking will just increase the overhead.

SIFS
DIFS
SIFS
SIFS
RTS DATA Transmitter
Receiver
CTS ACK



23

2.6 WAVE System Architecture
Worldwide, hundreds of projects, laps, and consortiums are competing in developing a
robust set of standards for VANET environments (Appendix B). In USA, the Dedicated Short
Range Communication (DSRC) [8] Committee of the IEEE Transportation Technology
Council is preparing the new Wireless Access in Vehicular Environments (WAVE)
standard, which will be illustrated in this section. In Europe, the European Committee for
Standardization (CEN) [7] (CEN stands for Comit Europen de Normalisation) has got its
own standard namely General Specifications for Medium-Range Pre-Information Via
Dedicated Short-Range Communication (CEN ISO/TS 14822-1:2006). In Japan, the
Association of Radio Industries and Businesses [1] issued the standard Dedicated Short-
Range Communication System (ARIB STD-T75) in 2001 with an updated version in 2007.
This section presents a brief overview of the IEEE WAVE system architecture as an
indication of the current state of standardization process. WAVE system Architecture is a set
of standards that describes the communication stack of vehicular nodes and the physical
airlink between them (Fig. 2-6). Any RSU may have two interfaces, one for the wireless
WAVE stack and the other for external interfaces like wireline Ethernet that may be used to
enable connectivity to the Internet. Similarly, each OBU may have two interfaces, one for the
wireless WAVE stack and the other for sensor-connections and human interaction.

Fig. 2-6. WAVE system components
On-Board Unit Road Side Unit
Applications
Applications
W
A
V
E

s
t
a
c
k

W
A
V
E

s
t
a
c
k

W
i
r
e
l
i
n
e

s
t
a
c
k

W
i
r
e
l
i
n
e

s
t
a
c
k

Airlink
Optional
External interface
On-Board
Human Interfaces
Intra-Vehicle
systems
External
systems
Covered by
WAVE standards



24

WAVE standard consists of five complementary parts,
- 802.11p Wireless Access in Vehicular Environments (WAVE) [22], which is an
amendment to the well-known IEEE 802.11 Wireless LAN Standard and covers the
physical layer of the system.
- 1609.1 Resource Manager [18] that covers optional recommendations for the
application layer.
- 1609.2 Security Services for Applications and Management Messages [19] that
covers security, secure message formatting, processing, and exchange.
- 1609.3 "Networking Services [20] that covers the WAVE communication stack.
- 1609.4 Multi-Channel Operation [21] that covers the arrangement of multiple
channels and how they should be used.
The WAVE communication stack and the coordination between standards are shown in
Fig. 2-7. Definition and operation of each layer of the stack will be demystified in the
following sections.


Fig. 2-7. WAVE protocol stack

Applications
1609.1,
et al.
1609.3
1609.4
802.11p
802.11p
LLC
Multi-Channel
Operation
IPv6
UDP / TCP
Management Plane Data Plane
WME
MLME
WSMP
PLME
A
i
r
l
i
n
k

WAVE PHY
WAVE MAC



25

2.6.1 WAVE Physical Layer
In October 1999, the Federal Communication Commission (FCC) allocated a 75 MHz of
bandwidth in the 5.9 GHz band (5.850 5.925 GHz) for applications of the DSRC [36]. The
WAVE spectrum is composed of seven channels of 10 MHz each, as shown in Fig. 2-8, with
an option of grouping two adjacent channels to have a spectrum of 20 MHz. Channel 178 is
the only control channel (CCH), and other channels are service channels (SCH). Channels
175 and 181 are the 20 MHz channels. Note that channel numbering are defined according to
the relation,
Channel center frequency = 5 GHz + (5 channel number) MHz
The modulation scheme used by WAVE is the Orthogonal Frequency Division
Multiplexing (OFDM) using 52 orthogonal subcarriers. The OFDM is a multi-carrier
modulation scheme where data is split into multiple lower rate streams. Each stream is used to
modulate one of the closely spaced orthogonal subcarriers. The primary advantage of OFDM
is its ability to cope with frequency-selective fading due to multipath channels without
complex equalization filters. This modulation scheme enables data rates of 3, 4.5, 6, 9, 12, 18,
24, and 27 Mbit/s in the 10 MHz channels and up to 54 Mbit/s in the 20 MHz channels. The
orthogonal subcarriers should be modulated using BPSK (Binary Phase Shift Keying), QPSK
(Quadrature Phase-Shift Keying), 16-QAM (Quadrature Amplitude Modulation), or 64-QAM
depending on the data rate required.


Fig. 2-8. Spectrum of WAVE Channels

Frequency 5.850 5.860 5.870 5.880 5.890 5.900 5.910 5.920 5.925 GHz
Channel number 172 174 176 178 180 182 184
175 181
10 MHz 5 MHz



26

Before leaving the physical layer, Table 2-3 summarizes some of the physical-dependant
parameters related to 802.11 MAC [22].

Table 2-3 WAVE physical characteristics
Characteristic Value for WAVE
Time-slot 16 s
SIFS 32 s
DIFS 64 s

2.6.2 WAVE Channel Coordination
The WAVE spectrum is composed of only one control channel (CCH) and six service
channels (SCHs). The control channel is considered as the public room for all WAVE devices
and its critical resource. Efficient organization and minimization of traffic on the CCH is a
challenging problem. The CCH should only be used for service advertisement frames and
broadcast messages (i.e. when the transmitter has not negotiated with a specific receiver yet);
however, no active connections between two or more devices are allowed to exchange data
over the CCH (i.e. after handshaking, the transmitter and receiver must pursue talking in
another channel). The channel access function used to organize contention over the CCH (and
SCHs as well) is the EDCA. Table 2-4 summarizes CW and AIFSN parameters for different
access categories over the CCH. Note that, CW
min
=15 and CW
max
=1023

Table 2-4 EDCA parameter set used in CCH
ACI AC CW
min
CW
max
AIFSN
0 Background CW
min
CW
max
9
1 Best Effort (CW
min
+1)/2 1 CW
min
6
2 Video (CW
min
+1)/4 1 (CW
min
+1)/2 1 3
3 Voice (CW
min
+1)/4 1 (CW
min
+1)/2 1 2

The other six SCHs are considered as private rooms for any connection to exchange long
streams of data. Before initiating a connection over a SCH, a node must first join an active
logical private network (namely, the WAVE Basic Service Set WBSS). Advertisement of
new services should be transmitted over the CCH, however, actual data exchange of the



27

service is done over any SCH. Table 2-5 summarizes CW and AIFSN parameters for different
access categories over SCHs.

Table 2-5 Default EDCA parameter set used in SCH
ACI AC CW
min
CW
max
AIFSN
0 Background CW
min
CW
max
7
1 Best Effort CW
min
CW
max
3
2 Video (CW
min
+1)/2 - 1 CW
min
2
3 Voice (CW
min
+1)/4 - 1 (CW
min
+1)/2 - 1 2

2.6.3 WAVE Basic Service Set
The WAVE Basic Service Set (WBSS) is a concept that should be clear before discussing
the deployed communication protocols. Duo to the distributed manner of WAVE protocols,
applications that want to establish a new connection with remote devices must first announce
for the new service on the CCH within a WBSS advertisement frame. The WBSS
advertisement frame contains the originating application, intended recipient devices (which
could be a broadcast), data rate and the intended SCH to be used.
On receiving of the WBSS advertisement frame, the provider node as well as user nodes
should switch to the indicated SCH to proceed with data exchange. Hence, the WBSS is a
logical private network of two or more WAVE devices having same active application(s) and
participating in data exchange over any of the SCHs (no WBSS is allowed on the CCH).
Any node can announce for a new WBSS while other nodes, on receiving of the
advertisement frame, have the right to join it according to their currently active applications.
A device can join only one WBSS at any time. A WBSS can support services for multiple
applications and can be joined by many users.
There are two types of WBSS, persistent WBSS and non-persistent WBSS. A persistent
WBSS is announced periodically in each CCH interval (the time interval when all WAVE
nodes listen to the CCH). This type could be used to support services of indefinite lifetime
(e.g. a RSU offering Internet access) so that they can be joined by nodes that newly come into
range. A non-persistent WBSS is announced only once on its initiation, and could be used to
support WBSS with limited lifetime.



28

2.6.4 WAVE Communication Protocols
WAVE supports two protocol stacks, the standard Internet Protocol Version 6 (IPv6) and a
new specially designed WAVE Short Message Protocol (WSMP).
2.6.4.1 Internet Protocol Version 6 (IPv6)
WAVE networking services support data exchange using the Internet Protocol version 6
(IPv6) [25] with both TCP and UDP at the transport layer. The existence of IPv6 protocol in
the wireless device within vehicles opens the Internet access with a tremendous variety of
possible applications. Connection using IPv6 is permitted only on SCHs after joining a
WBSS.
2.6.4.2 WAVE Short Message Protocol (WSMP)
The WAVE Short Message Protocol (WSMP) is a new protocol designed especially for an
optimized operation in WAVE environments. If any node prefers not to join a WBSS (for
example, a transmitter has a short data to broadcast) it will have to use only WSMP over the
CCH. WSMP is used for direct transmission of short messages without joining WBSS.
Messages of this protocol are designed to consume minimal channel capacity. Hence, it is the
only protocol allowed over the CCH (and may be used on any SCH as well). The suggested
frame format of a WAVE Short Message (WSM) is shown in Fig. 2-9 (lengths are in octets of
bits).

1 1 1 1 1 4 2 variable
WSM
Version
Security
Type
Channel
Number
Data
Rate
Tx Power
Level
Provider
Service
Identifier
WSM
Length
WSM
Data
Fig. 2-9. WSM frame format

The WSM Version is used version of WSMP (currently, its value is zero). The Security
Type indicates the security processing of the WSM Data i.e. the transmitter application can
sign or encrypt the message with an indication in security field. The Channel Number is
used to identify the radio channel used for the WSM. The Data Rate indicates the data rate
used for the WSM. The Tx Power Level indicates the transmit power used for the WSM.



29

The Provider Service Identifier identifies the application that originated the WSM (each
application will have a unique number). The WSM Length indicates the length in octets of
the following WSM Data field (limited to 1400 in its default value). The WSM Data
contains the application data being transferred.
2.6.5 WAVE Management Plane
The WAVE management plane is considered a logical low-level database of the system
and performs system configuration and maintenance functions. It consists of the WAVE
management entity (WME) with a special part to serve the MAC layer namely MAC layer
management entity (MLME) and another one to serve the physical layer namely Physical
layer management entity (PLME). Examples of its use include:
- Prior to the first operation of the transceiver (i.e. network configuration phase) different
system parameters are loaded into the devices WME. This field is known as Local
Information.
- Active applications register their parameters with the WME. Therefore, MAC layer can
determine whether a received WBSS advertisement is of interest to any of its applications or
not. This field is known as User Service Information.
- The WME is responsible for generating the WAVE service advertisement frame on an
application request. This field is known as Provider Service Information.
- On the initiation or joining of a WBSS, network parameters are registered in the WME.
2.6.6 WAVE Synchronization
During data exchange within a WBSS over a SCH, critical events (e.g. public safety
related messages) and new service advertisements with higher priorities may take place over
the CCH. Thereupon, the WAVE system requires that all participating devices should monitor
the CCH during a small common time interval (CCH interval) on a regular basis.
WAVE depends on GPS devices to acquire synchronization with reference to the
Coordinated Universal Time (UTC). Each UTC second is divided into ten sync intervals,
which in turn divided into a CCH interval followed by a SCH interval separated by a guard
interval, as shown in Fig. 2-10 .




30



CCH interval SCH interval CCH interval SCH interval CCH interval


Fig. 2-10. WAVE Synchronization

Devices without access to a precise timing signal (e.g. GPS) may acquire synchronization
from other WAVE devices upon receiving of WAVE advertisement frames, as the time will
be included within the frame.
This concludes the introduction of the field of study. Next chapter will briefly cover
published broadcasting protocols in VANET environments.


Guard interval
Sync Interval (0.1 sec)
Start of every
UTC second



31

Chapter 3
Previous Work
Although broadcasting has a limited usage in Ethernet and MANET (e.g. a DHCP
Dynamic Host Configuration Protocol request), it has got a wider range of implementation
in VANET applications. Almost all applications discussed in Sec 2.1 depend on sending
messages to intended vehicles without explicitly determining their identity, which is a
broadcast in its nature. Note that, all signaling techniques that are currently deployed in
vehicles (e.g. brake lights and turning right / left lights) are considered a broadcast. With
VANET technology, these signals will be exchanged directly between vehicles themselves.
This will increase the driver awareness of the road and the traveling luxury as well.
In this chapter, we will discuss previous promising contributions in broadcasting protocols
in VANET environments. Within the discussion of each protocol, we will clarify the work
objective, the new algorithm proposed and the key strengths / weaknesses regarding VANET
environments.
3.1 Categories of Broadcasting Protocols
All of these contributions try to solve just two questions; the first one is "How to deliver
the broadcast message to nodes within a single communication range with the highest
possible reliability?" which will be designated as reliable protocols. The second one is "How
to deliver the broadcast message to the entire network?" which will be designated as
dissemination protocols. Although both questions look similar to each other, the first one is
used with applications related to direct neighbors (e.g. collision avoidance) and the second is
used with applications related to the entire network (e.g. traffic management).



32

Fig. 3-1 shows different categories of broadcasting protocols along with some sample
protocols that will be discussed in this chapter.
Broadcasting Protocols
Reliable Protocols Dissemination Protocols
Rebroadcasting Selective ACK
Changing
Parameters
Flooding Single relay
Xu [46]
(2004)
Tang [42]
(2001)
Balon [2]
(2006)
Ni [37]
(1999)
Zanella [48]
(2004)
Yang [47]
(2004)
Huang [16]
(2002)
Heissenbttel [33]
(2006)
Korkmaz [30]
(2004)
Alshaer [15]
(2005)
Xie [45]
(2005)


Fasolo [12]
(2006)

Fig. 3-1. Different categories of broadcasting protocols

Published reliable protocols use three methods: Rebroadcasting where the transmitter
node retransmits the same message for many times, Selective ACK where the transmitter
requires ACK from a small set of the neighbors, and Changing parameters where the
transmitter changes transmission parameters according to the expected state of the network.
Published dissemination protocols use two methods: Flooding where each node is
responsible for determining whether it will rebroadcast the message or not, and Single relay
where the transmitter is responsible for determining the next hop node.
3.2Why not IEEE 802.11
As quoted from the IEEE 802.11 standard [23], There is no MAC-level recovery on
broadcast or multicast frames. As a result, the reliability of this traffic is reduced, relative to
the reliability of directed traffic, due to the increased probability of lost frames from
interference, collisions, or time-varying channel properties.
Although the probability of collisions may be dropped down using the RTS/CTS
mechanism, the 802.11 standard says that, The RTS/CTS mechanism cannot be used for



33

messages with broadcast and multicast immediate destination because there are multiple
recipients for the RTS, and thus potentially multiple concurrent senders of the CTS in
response. As a result, the area of reliable broadcasting is still an open research challenge and
needs some new innovations.
3.3 Reliable Protocols
Broadcasting in wireless networks can serve numerous applications where reliability is not
necessary and time is not a critical requirement. The emergence of VANETs opened a new
research challenge of time-critical reliable broadcasting that intended to serve a bunch of
public safety related applications. The problem statement for reliable protocols is to design a
protocol that can deliver a message from a single source to every node in his transmission
range with the highest possible reliability and minimum latency.
The key performance metrics for reliable protocols are:
Success rate: the number of nodes that have successfully received the broadcast, divided by,
the number of nodes in the transmitter communication range.
Latency: the total time required in a single broadcast phase.
Researchers used three methods to increase the broadcast reliability: Rebroadcasting,
Selective Acknowledgment and Changing Parameters.
3.3.1 Rebroadcasting
The first method of increasing broadcast reliability is by retransmitting the same message
for many times. The problem discussed in this situation is, how many times are considered
practically enough?
Xu, et al. (2004) [46] explored the effect of retransmission on increasing the reliability and
developed six MAC protocols:
- Asynchronous Fixed Repetition (AFR): where the message is repeated in each time-slot for
a fixed number of times.
- Asynchronous p-persistent Repetition (APR): where the transmitter node transmits the
message in each time-slot with probability P, where P is a configurable parameter.



34

- Synchronous Fixed Repetition (SFR): is the same as AFR except that all nodes in the
network are synchronized to a global clock.
- Synchronous p-persistent Repetition (SPR): is the same as APR except that all nodes in the
network are synchronized to a global clock.
- Asynchronous Fixed Repetition with Carrier Sensing (AFR-CS): is the same as AFR except
sensing the channel before transmission.
- Asynchronous p-persistent Repetition with Carrier Sensing (APR-CS): is the same as APR
except sensing the channel before transmission.
Although both SFR and AFR-CS protocols gave the best success rate, the author suggests
using the AFR-CS as it does not require a global synchronization and it uses the minimum
overhead.
Key strengths: He was the first to address retransmission as a method of increasing
reliability.
Key weaknesses: He did not solve the hidden node problem, and the AFR-CS protocol
requires the same number of repetitions neglecting the effect of network condition and traffic
volume.
Vehicular Collision Warning Communication protocol (VCWC) (Yang, et al. 2004) [47]
proposed two new concepts. The first one shows that, the same degree of reliability can be
achieved by retransmitting with a decreasing rate, and hence the protocol saves some
unnecessary transmissions. The second one is that a single communication range (10-sec
traveling time with a minimum of 110 meters and a maximum of 300 meters as suggested by
the DSRC [8] consortium) is sufficient for an easy slowing down. i.e. vehicles that are
running away from that range should not be interrupted by that event. In case of the following
vehicles react aggressively, they will be considered abnormal and send new warning
messages by their own. In such situation, we will have a danger area that is covered by a
cloud of warning messages initiated by the newly affected vehicles.
Key weaknesses: there are still many unnecessary transmissions considering that the
minimum retransmission rate suggested by the author is 10 messages/sec!
Alshaer, et al. (2005) [15] proposed an adaptive rebroadcasting algorithm where each
vehicle determines its own probability of retransmission according to an estimate of the



35

density of vehicles around it within two-hops. The density information is obtained from the
periodical packets that are involved in the operation of the Ad-Hoc routing protocols.
Key weaknesses: the operation of this protocol depends on the routing protocol used.
Besides, it ignored the effect of hidden node problem.
3.3.2 Selective Acknowledgment
There is no doubt that, acknowledging is the ultimate method of reliable communication
and it is widely implemented in unicast messages. However, in broadcasting, the destination
node is unidentified. The problem discussed in this section is, how acknowledging can be
used in improving the reliability of the broadcast.
The Broadcast Medium Window (BMW) (Tang, et al. 2001) [42] protocol treats
broadcasting as multiple unicast operations. For every broadcast message, the transmitter
unicasts it to every node of the neighborhood using the RTS/CTS/DATA/ACK scenario.
Key weaknesses: the protocol possesses very high latency, especially in cases of high-
density nodes.
The Batch Mode Multicast MAC Protocol (BMMM) (Huang, et al. 2002) [16] protocol
requires the transmitter to use RTS/CTS frames with every node sequentially. Then the
transmitter should broadcast the message once, and use a new control frame RAK (Request
for ACK), to collect ACKs from every node.
Key weaknesses: if an ACK from a receiver is lost, the transmitter will restart the whole
process again, which is unnecessary.
The Round-Robin Acknowledge and Retransmit (RRAR) (Xie, et al. 2005) [45] protocol
suggests that every broadcast message should contain a request of ACK to only one of the
neighbors (network-probe node). For each new packet to be broadcasted, the transmitter
selects a different node in a round-robin style.
Key weaknesses: the operation of this protocol assumes that each node has an updated list
of the neighborhood nodes, which is impractical in VANET environments (Sec. 2.2).
3.3.3 Changing Parameters
Balon, et al. (2006) [2] proposed a protocol that minimizes the collision rate and hence
increases broadcast reliability of a single communication range. The author used an ordinary



36

DSRC required application (that each node should broadcast a status message every 100ms)
as a source of information about the network condition. In this protocol, each node should
include its own MAC address and a sequence number within the status message. Hence,
nodes can estimate the count of lost messages and change the contention window size
accordingly. In case of low loss rate, as an indication of good network condition and small
number of vehicles, the protocol attempts to decrease the contention window size and hence
decrease latency. On the contrary, if the loss rate is high, the protocol increases the contention
window size and limits collisions.
Key strengths: the author used an ordinary DSRC service as a source of information about
the network condition. The proposed protocol can be implemented in the application layer of
the currently working devices without any modification.
Key weaknesses: although the protocol minimizes collision probability, it ignores other
sources of degrading reliability like hidden node problem and channel fading.
3.4 Dissemination Protocols
Broadcasting messages to the entire network in Ad-Hoc mode is not an easy job especially
in case of highly mobile nodes. Building a routing table will consume a heavy messaging load
and is useful for only a couple of seconds before every node changes its location. The
problem statement for dissemination protocols is to design a protocol that can coordinate
between network nodes to deliver the message to the largest number of nodes in the network
within the shortest time duration. The key performance metrics for dissemination protocols
are:
Success rate: the number of nodes that have successfully received the broadcast divided by
the total number of nodes in the network.
Redundancy: the total number of useless transmissions i.e. when all nodes within the
broadcast range have already received the message.
Dissemination speed: speed of the message along the propagation direction; i.e. do all
broadcasting hops are allied with the direction of propagation, which is an indication of the
cumulative time that a message will take to reach all nodes.



37

Researchers used two methods to achieve better dissemination: Flooding and Single
Relay.
3.4.1 Flooding
Flooding protocols are highly distributive, since it is each nodes responsibility to
determine whether it will participate in the rebroadcasting or not. This is based on the number
of messages it has already overheard and the current locations of their sources.
Ni, et al. (1999) [37] was the first to use flooding techniques in mobile Ad-Hoc networks,
and introduced the term broadcast storm problem. That problem happens when attempting
to send the intended message to all nodes by forcing each node to rebroadcast the message
(simple flooding). Simple flooding will result in a serious redundancy (all neighbors have
already received the message), contention (nodes severely contend on the channel), and
collision (concurrent transmissions and the lack of RTS/CTS). He presented different schemes
to reduce the redundancy by inhibiting some nodes from rebroadcasting. These schemes are;
- Probabilistic Scheme: where a node rebroadcasts the message with a probability p where
0p1. Note that when p=1, this scheme will be identical to the simple flooding.
- Counter-Based Scheme, where a node rebroadcasts the message only if it overheard the
message for c < C times, where C is a constant (equals 3 or 4 as recommended by the author).
- Distance-Based scheme: where a node rebroadcasts the message only if its distance from
the transmitter is d > D, where D is a constant. i.e. when the expected increase in coverage
range is greater than the threshold.
- Location-Based scheme, each node compares its location with the transmitter location
and calculates the additional coverage that can be provided assuming that all nodes have
omnidirectional coverage. A node rebroadcasts the message only if the additional
coverage > A, where A is a constant.
- Cluster-Based scheme, in this scheme, the author suggests dividing the network into
circular clusters each cluster has a small set of nodes acting as a gateway to the neighbor
clusters. Here, only gateway nodes have the right to rebroadcast the message.
Finally, the author concludes that the location-based scheme resulted in the minimum
redundancy.



38

Key strengths: although it was the first time to clarify the problem associated with multi-
hop broadcasting in Ad-Hoc networks, the author presented a good analysis and some elegant
solutions.
Key weaknesses: the algorithm is not effective at high packet loads and it suffers from
hidden node problem.
The Dynamic Delayed Broadcasting (DDB) (Heissenbttel, et al. 2006) [33] protocol is
just an updated version of the Location-Based scheme, where nodes that receive the
broadcast packet calculate the additional coverage that can be provided. Depending on the
size of this additionally area, each node introduces a delay before relaying the packet, where
the delay is longer for smaller additional areas. In this way, nodes that have a higher
probability of reaching additional nodes will broadcast the packet first.
Key strengths: the best nodes for rebroadcasting are chosen in a completely distributed
way at the receiving nodes without any prior topology information.
Key weaknesses: the protocol reduces redundancy efficiently, but nothing improved in the
collision probability i.e. hidden node problem. Besides, the lack of explicit ACK degrades the
overall reliability.
3.4.2 Single Relay
We can mind single relay protocols as sequential ones, where the transmitter node handles
the responsibility of the broadcast to a another following node. The best node to handle such a
job in dissemination protocols is the furthest one. The problem discussed with these protocols
is, how to choose the furthest node without any prior information.
The Minimum Connected Dominating Set (MCDS) (Zanella, et al. 2004) [48] is defined as
the minimum set of connected nodes that, every other node in the network is one-hop
connected with a node in this set. In the MCDS-based broadcast protocols, the message is
forwarded only by the nodes of the MCDS. This protocol achieves the largest progress along
the propagation line, while guaranteeing the coverage of the entire network. The MCDS gives
the theoretical optimum performance. However, this protocol needs extensive real-time
information about the exact location of every node in the network, which is not practical in



39

VANETs. The same problem affects any solution that relies on a complete or partial
knowledge of the network topology.
Korkmaz, et al. (2004), the authors of The Urban Multihop Broadcast Protocol (UMB)
[30], defined the term RTB/CTB (Ready to Broadcast / Clear to Broadcast), which they used
as an equivalent to the IEEE RTS/CTS in cases of broadcasting. In this protocol, the
transmitter sends an RTB message containing its geographical location and the intended
direction of message propagation. The protocol logically divides the transmission area into
adjacent and non-overlapping segments of equal lengths. The node located in the furthest non-
empty segment should reply the transmitter with a CTB message containing its identity and
prepare itself to be the relay node for the incoming broadcast.
The mechanism of electing the furthest node is: upon reception of an RTB message, each
node transmits a channel-jamming signal (black-burst) of a length proportional to its current
distance from the transmitter. Then, it checks the status of the wireless channel. If the channel
is busy with another black-burst signal i.e. another node is still transmitting, the node exits the
contention phase and listens to the incoming broadcast. On the contrary, if the node senses the
channel as idle i.e. it is the furthest node, the node sends a CTB message as indicated above.
In case that, there was more than one node in the furthest non-empty segment, their CTBs
may collide. In such situation, the transmitter will have to resend the RTB again requiring a
farther division of the intended segment into sub-segments and redo the same procedure.
Key strengths: the proposed algorithm does not require any prior topology information and
is robust at any traffic volume.
Key weaknesses: selection of the furthest node is based on transmitting of the longest
black-burst signal. This involves high latency and limits its usefulness in emergencies;
moreover, black-burst signal is not a candidate of the present version of the 802.11 standard.
The Smart Broadcasting Protocol (Fasolo, et al. 2006) [12] addressed the same objective
as UMB using a different methodology. Upon reception of an RTB message, each node
should determine its segment and accordingly, set a random back-off time. Segments will be
associated with non-overlapping contention windows ordered from outermost to innermost.



40

For example, assume that the contention window size is (4) TS (time-slot); Nodes in the
furthest segment should randomly choose a back-off time between (0) to (3) TS, nodes in the
next nearer segment choose a value between (4) to (7) TS, and so on.
Nodes will decrement their backoff timers by one in each time-slot while listening to the
physical channel. While waiting, if any node receives a valid CTB message, it will exit the
contention phase and listen to the incoming broadcast. On the contrary, if any node finishes
its backoff timer, it will send the CTB containing its identity and relay any incoming
broadcast.
Key strengths: this protocol performs the same operation as UMB, but depending on the
minimum waiting time.
Key weaknesses: the protocol depends on the size of the contention window. There are
different optimal values for different traffic volumes. The protocol did not provide a method
for estimating the current traffic volume; hence, the contention window size will be static
with a predetermined value. Thereupon, the implementation will never reach the optimal
values.

There is a point that needs a re-comment, what is the difference between Single Relay
protocols (e.g. the Smart protocol) and Selective ACK protocols using single network-probe
(e.g. RRAR protocol), that makes them in two different categories?
Simply, the difference is in the objective of the design. Selective ACK protocols are
designed for a better reliability through acknowledging with every neighbor node, one by one.
However, Single Relay protocols are designed for a better dissemination of the broadcast
sacrificing some of the reliability (there could be a near miss-synchronized node that is
completely overwhelmed by the single relay protocol).



41

Chapter 4
Theoretical Analysis
This chapter presents the theoretical analysis of the new contribution. The thesis
contribution is a novel VANET broadcasting protocol. This protocol is designed especially
for safety-related applications. It also integrates other non-safety applications with a new
multi-mode feature. The new protocol is presented as a series of four successive steps starting
from The Smart Broadcasting Protocol [12]. These steps are: 1- reversing the order of
priority, 2- headway-based segmentation, 3- non-uniform segmentation based on naturalistic
model of drivers reactions, and 4- application adaptive multi-mode scheme. The necessity of
each step will be clear after evaluating the design objective. The performance of the proposed
protocol will be evaluated logically, analytically and with simulation.
4.1 Introduction
This section formulates the objective of the new design, the broadcasting goals and the
assumptions.
4.1.1 The Design Objective
The design objective is the key behind any new innovation. The proposed protocol is
designed for only emergencies. Assuming that the driver is suddenly stopping or changing his
driving lane, the vehicle will be considered abnormal. Here, it is required to open an instant
communication channel with the vehicle in the most dangerous situation. It is a case of
unicast information packed in a broadcast protocol where suggested actions and maneuvers
are intended only to a specific vehicle. However this information is packed in a broadcast
protocol for two reasons: the first one is, there isnt enough time for handshaking and moving



42

to a service channel. The second one is that all surrounding vehicles should also process the
message to take actions in their turn.
The problem addressed in this protocol is, how to identify the vehicle in the most
dangerous situation in a robust and low latency protocol.
According to the categorization of Sec 3.1, the proposed protocol can be considered in a
third category namely Reliable Protocols for a Specific Purpose. It addresses a single
transmitter with a single transmission range to achieve reliability using a network probe node.
However, the network probe node is chosen as being the most affected by the running
application (not a random one as in Selective ACK, nor the furthest one as in Single Relay).
4.1.2 Broadcasting Goals
Any broadcasting protocol should satisfy all of the following goals;
1 - High reliability: the transmitter should be acknowledged of delivering the message to
the intended vehicle(s).
2 - Low latency: the time duration from the first attempt of transmission to the end of the
broadcasting phase, should be as small as possible.
3 - Low probability of collision: the protocol should suffer from as minimum collisions as
possible, hence, higher reliability and lower latency.
4 - Hidden node problem: collisions at receiving nodes duo to hidden node problem should
be avoided.
5 - No prior control messaging: in VANET environment, the broadcasting protocol should
be highly distributive and does not need any prior information. It is assumed that each node in
the network knows nothing about type or exact location of the other nodes.
6 - Human factors: mobile nodes in VANETs are controlled by human drivers, which
means that node movement is not completely random and it can be expected to some extent.
Fortunately, driver naturalistic behaviors are studied extensively in hundreds of papers. Any
VANET protocol should take advantage of these studies.
7 - High robustness: location of nodes changes very rapidly and unpredictably (to some
extent). Thus, any VANET protocol should be robust at different speeds, different traffic
volumes and different environments (rural or urban).



43

8 - Different applications: broadcasting in VANETs services many categories of
applications. Thus, the deployed protocol should cope with the differences in these
applications.
4.1.3 Assumptions
There are some reasonable assumptions that are considered throughout the development
cycle. These assumptions are:
Each vehicle participating in this protocol should be equipped with:
- High accuracy positioning device: that each vehicle can obtain its own geographical
location (GPS for example).
- One wireless transceiver: that can communicate in the DSRC range (5.9 GHz).
- Speed sensor: that the DSRC device gets the vehicles current running speed.
It is also assumed that the broadcasted message (RTB or the message itself depending on the
message length, as indicated in Sec 2.5.4) contains the following:
- The transmitting node MAC address
- The geographical location of the transmitting node
- The current traveling speed of the transmitting node
- The message propagation direction
- The broadcast mode (BM), which is a number to determine the mode of operation, (as
discussed later in Sec 4.6)
Following the work of UMB protocol [30], the proposed protocol uses the notation RTB/CTB
instead of the IEEE RTS/CTS.
4.2 The Starting Block
This section presents a more in-depth analysis of The Smart Broadcasting Protocol (2006)
[12]. The four modification steps proposed in this thesis start from this algorithm, hence, a
good understanding of this protocol is a mandatory before discussing the proposed
modifications.




44

4.2.1 Frame Exchange Sequence
In case of a transmitter has a message to broadcast over the air, it should use the sequence
[RTB/CTB/DATA/ACK] with a network-probe node. However, only if the message length is
smaller than a predefined threshold, the source shall use the sequence [DATA/ACK] directly.
Throughout this work, we assumed that the message is long, and we are using the sequence
[RTB/CTB/DATA/ACK]. Hence, the problem statement becomes, which node will be the
first to reply with the CTB message?
With this frame exchange sequence, we cope with best efforts in reducing the effect of
hidden node problem.
4.2.2 The Basic Algorithm
The Smart Broadcasting Protocol seeks the best performance as a dissemination protocol.
It elects the furthest node to relay the broadcast it its followers. The election methodology is
by logically dividing the transmission range into a number of adjacent and non-overlapping
equal segments. The node located in the furthest non-empty segment should reply with a CTB
message containing its identity and prepare itself to be the relay node for the incoming
broadcast.
On receiving of an RTB message, every node in the message propagation direction should
perform these steps (Fig. 4-1):
- Find the segment number (based on its distance from the transmitting node).
- Choose a random backoff period within the contention window assigned to its segment
(assuming a contention window size of (4)).
- On receiving of a valid CTB, exit the contention phase.
- On receiving of a colliding CTB messages, hold its countdown timer until the end of
collision.
- On the end of its countdown timer, send a CTB message.
In this algorithm, decisions of the receiving node depend solely on information from the
RTB message and GPS device without using any prior information.
Note that, The Smart Broadcasting Protocol assumes dividing the transmission area into
ten segments, but Fig. 4-1 shows only four segments for simplicity. The contention windows



45

indicated in Fig. 4-1 reveals that, this protocol chooses the furthest node with a plain uniform
distance-based segmentation algorithm.


Fig. 4-1. Arrangement of segments for the basic algorithm
4.3 Step-1: Safety Related Applications
As indicated earlier, location of network-probe node determines the purpose of protocol
design; one by one for better reliability in Selective ACK and the furthest for better
dissemination in Single Relay protocols. However, our primary design purpose is reliability
for safety related applications, that we choose the network-probe node as the one in the most
dangerous situation. This enables the abnormal vehicle of sending suggested actions and
maneuvers to the vehicle in the most dangerous situation.
In almost all emergency situations (e.g. co-operative collision warning), the most
threatened vehicle is the nearest one running behind the source vehicle. Hence, the first
modification is simply, reversing the order of priority as shown in Fig. 4-2.


Fig. 4-2. Arrangement of segments for step-1 modification

The contention windows indicated in Fig. 4-2 reveals that, the first modification chooses
the nearest following node, still with a plain uniform distance-based segmentation method.
Distance (meter)
Contention Window [0-3] [4-7] [8-11] [9-12]
Segments S1 S2 S3 S4
Distance (meter)
Contention Window [12-15] [8-11] [4-7] [0-3]
Segments S4 S3 S2 S1



46

4.3.1 Discussion
The following are some points that need more comments, to clarify the effect of this
proposed step:
- Collisions at far range nodes:
While communicating between the abnormal vehicle and the nearest following one, there
could be collisions at far range nodes due to hidden node problem as shown in Fig. 4-3.


Fig. 4-3. Collisions at far range nodes

Although this looks like a defect in the methodology, it is actually the core improvement.
If there is a collision at far range nodes, this means that there was a communication session
taking place. In critical life-safety situations, the abnormal vehicle should not wait for all
nodes within two-hop distance to finish their sessions as long as they will not collide with its
urgent message to the most threatened vehicle. Yes, there could be collisions at far range
nodes and in such situations, step-1 modification gives the protocol a logically minimum
latency.
As a compensation for this type of collision, we recommend that the ACK message should
be the same as the broadcast. Hence, we include an ACK field in the broadcast; which
should be set in the ACK message. From now on, we assume that the ACK message is of the
same length as the original broadcast.
- Broadcasting beyond a single transmission range:
As recommended by the DSRC [8], the communication range of the abnormal vehicle is
10 sec travel time. Vehicles beyond this range is expected to have a sufficient distance and
time for an easy slowing down ( [14] and [41]). The proposed protocol considers that this
Urgent message
S D
Regular message
D S
Expected collision
at this vehicle



47

range is very efficient for one abnormal vehicle. In case that a following vehicle reacts
aggressively, it will become abnormal and issue a new warning message itself. Consequently,
there will be a transmission range surrounding all abnormal vehicles with minimum
interruption to the rest of the network.
- Malicious messages:
The objective of a malicious attack is to harm members or the functionality of the network,
hence its means and targets may not be clearly identified ( [3] and [4]). The proposed protocol
can logically prevent malicious messages from broadcasting and swamping the network,
because its broadcasting beyond the communication range depends on reactions of the
following drivers who would not react aggressively to messages while not seeing anything in
the near range or getting clear inconsistencies with the real-time scene.

The improvement of step-1 is logically clear, but there isnt any change in the average
latency.
In this step, we stated that location of the best probe node is an application dependant and
it is the nearest following vehicle in safety related applications which represent the core
usefulness of VANET technology.
4.4 Step-2: A Headway-Based Segmentation
This modification continues with figuring out where is the most threatened vehicle. In the
previous step, we stated that it is the nearest vehicle running behind the abnormal one
neglecting the effect of different speeds. Of course, vehicles running at high speed are in more
danger than those running at low speed even if they were located further from the abnormal
vehicle. Accordingly, this section studies the effect of using a novel headway-based
segmentation instead of the regular distance-based segmentation.
Using a headway-based segmentation leads us to a new definition:
Time headway or headway for short (Fig. 4-4) is the time interval between two vehicles
passing a point as measured from the front bumper to the front bumper. The headway is the
in-between distance divided by the following vehicles speed. It may be of different meter



48

lengths corresponding to different speeds, with a minimum length of 4m, which is the average
length of a sedan car.


Fig. 4-4. Headway

Although headway has never been used as a basis of segmentation, it looks more suitable
to DSRC requirements [8]. According to DSRC consortium, vehicle-to-vehicle
communications should have a transmission range of 10 seconds travel time, thus the
transmission range will vary with vehicle speed with a minimum range of 110 meters and a
maximum range of 300 meters. For example, vehicles traveling at 100 kilometer per hour
should transmit at a power level appropriate to reach approximately 278 m and vehicles
traveling at 40 kilometer per hour or lower should transmit at a power level appropriate to
reach approximately 110 m.
The only change we will introduce in this step is how the following vehicles will calculate
the segment number; assuming that the communication range is divided into (10) segments,
each is only of one second.
1 - Get the source vehicle location (l
s
) from the RTB message
2 - With the receiving vehicle current location (l

) and speed (s), calculate the Headway

(E) with this very simple equation
E =
l
s
-l

s

3- The segment number is the headway rounded to +.
Fig. 4-5 and Fig. 4-6 show the difference between headway-based and distance-based
segmentation with vehicles running at different speeds.

Headway (sec)
Headway (sec)



49


Fig. 4-5. Distance-based segmentation


Fig. 4-6. Headway-based segmentation

Fig. 4-5 shows a 3-lane highway with three vehicles running at different speeds,
(30,60,120 Km/h) with reference to distance (meter). This figure is a real snap-shot image.
Fig. 4-6 shows the same situation after calculating the headway for each vehicle to produce an
imaginary calculated image. This image reveals that headway-based segmentation mimics
dangerous situations better than distance-based, that it puts the 120Km/h-vehicle into the first
segment, which is identical to the intuitive analysis of the situation.
From now on, figures of the highway will be of two types: the first one is figures with
reference to distance (meter) (e.g Fig. 4-5) which is a real snap-shot image. The second one is
that with reference to headway (sec) (e.g. Fig. 4-6) which is an imaginary calculated image
based on the location and speed of each vehicle.


Headway (sec)
CW [0-3] [4-7] [8-11] [9-12]
Segment S1 S2 S3 S4
Distance (meter)
120 Km/h
30 Km/h
CW [0-3] [4-7] [8-11] [9-12]
Segment S1 S2 S3 S4



50

4.4.1 Discussion
- Assuming multiple lanes highway:
A question that may arise if the analysis is for multiple lanes is do vehicles in different
lanes are prone to the same danger as vehicles running in the same lane following the
abnormal one? Studies found that some drivers avoid obstacles by steering rather than by
braking or even perform the both [14]. It is found that the response time for steering is about
0.3 sec faster than that for breaking. This is the cause that we should consider that the
abnormal vehicle may often use steering in conjunction with breaking and that the danger area
is not only the same lane of the abnormal vehicle, but also adjacent lanes.
- Assuming a single lane highway:
A question that may arise if the analysis is for a single lane is will a far fast vehicle
overtakes a near slow one?
Let us study this problem quantitatively; assume that there are two vehicles following the
abnormal one in a single lane highway as indicated in the figure.


Fig. 4-7. Assuming a single lane highway

The first vehicle is running in a speed S1, and the distance to the abnormal vehicle is d1
meters and H1 secs, where
E1 =
J1
S1

The second vehicle is running in a speed S2, and the distance to the first vehicle is d2 meters
and H2 secs, where
H1 & d1 H2 & d2
Ht & dt
Headway (sec)
S1 S2



51

E2 =
J2
S2

and the distance to the abnormal vehicle is dt meters and Ht secs, where
Et =
Jt
S2

The condition stated in the question is
Et < E1
J1 +J2
S2
<
J1
S1

J1
S2
+E2 <
J1
S1

J1 _
1
S1
-
1
S2
] > E2
J1 _
S2 -S1
S1 - S2
] > E2
J1 > E2_
S1 - S2
S2 -S1
]
E1 =
d1
S1
E1 > E2[
S2
S2-S1

For example, assume that S1 = 80km/h (22.2m/s), S2 = 120km/h (33.3m/s) and H2 is the
average minimum headway (1 sec) [41]
E1 > S sccs
Thus, if the two vehicles are running at 80, 120km/h (in developed countries, such a speed
difference is not expected to happen in the same highway lane), this situation may happen at a
minimum of three seconds away from the abnormal vehicle; i.e. a relaxed situation. Then,
either the fast vehicle will slow down (not to hit the slower one) or try to pass it and truly
become more threatened than the slower one and the one that should logically reply with the
CTB (or ACK) message.
For short, the stated situation may happen only if
- The speed difference is very large
- and the two vehicles are still far away from the abnormal vehicle




52

With step-2 modification, the algorithm elects the nearest-in-time vehicle with a plain
uniform headway-based segmentation method.
In this step, we emphasized that any VANET protocol should expect that vehicles are
running at different speeds and it should cope with these variations efficiently.
4.5 Step-3: Non-uniform Segmentation (Headway Model)
What modification that can be applied to the proposed protocol if we highlighted that, "the
distribution of vehicle-headway along highways has never been uniform"?
The analysis of the distribution of vehicle-headways is classified under the Headway
Model. The concept of Headway Model will be discussed in Sec 4.5.1, before continuing
with its impact on the protocol development with Sec 4.5.2.
4.5.1 Headway Model
The Headway Model is a mathematical equation that describes the average naturalistic
headway that drivers tend to leave apart. This model is fundamental in any traffic engineering
application because it provides a laboratorial method of generating vehicles in any traffic flow
simulator.
Traffic engineering researchers introduced many headway models trying to mimic realistic
situations. Some of these models are: the negative exponential distribution, the shifted
exponential distribution, the gamma distribution, the lognormal distribution and the semi-
poisson distribution.
Luttinen (1996) [32] ,in his PhD thesis, gave a good study comparing each model with
empirical data. Fig. 4-8 shows the comparison between the probability density function (pdf)
of these distributions and the empirical density function (edf) with reference to 10-sec
headway. The headway distribution shows that drives cannot hold a zero sec headway and the
average headway expected is about 1.5 sec. The headway model of Fig. 4-8 is measured at a
traffic density of 1439 vehicle/hour.
He stated that the model that gives the best goodness-of-fit results when compared with
the edf is the Semi-Poisson distribution, and that it is the recommended distribution for use in



53

simulators with adequate computational facilities. The Semi-Poisson distribution is the one
used throughout this work.



Fig. 4-8. Sample Headway models

Studying the headway model shows that, broadcasting protocols that depend on the time
headway (only this contribution at this time) have a high robustness at different traffic
volumes. This claim is clear comparing the headway model at different traffic volumes (Fig.
4-9 [32]) and with the simulation in Sec 5.5.


Fig. 4-9. Headway at different traffic volumes
a. 1439 veh/h b. 795 veh/h c. 328 veh/h
d. Lognormal Distribution e. Semi-Poisson
a. Negative Exponential b. Shifted Exponential c. Gamma Distribution



54

Although nothing has been changed in the developing protocol, this section satisfies a new
goal, namely High robustness at different traffic volumes.
4.5.1.1 The Semi-Poisson Distribution
The probability density function (pdf) of the Semi-Poisson distribution is recalled here;
t t
e
t
p e
t
p p t f

) 1 (
) (
) , (
) 1 (
) (
) (
) , , , | (
1

1 0 ; 0 , , ; 0 > p t Eq. 4-1
Where, is the Gamma Function and is the Incomplete Gamma Function. Parameters p
and changes according to the desired traffic volume; however, parameters and do
not correlate with the traffic volume. Parameters chosen in (Fig. 4-10) are considered at a
traffic volume of approximately 330 veh/h [32].

Fig. 4-10. Semi-Poisson Headway Model
Notwithstanding that there is a common verbal instruction used in driving manuals around
the world to maintain a minimum headway of 2 seconds [9], Authors in [41] found that the
0 2 4 6 8 10
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
Headway (sec)
D
e
n
s
i
t
y
Semi-Poisson Distribution at 330 veh/h


P = 0.14
[ = 2.929
= 5.488
0 = 0.06



55

naturalistic average minimum headway for drivers is about 0.66 second as indicated in Fig.
4-10, and it is insensitive to the driving speeds. This 0.66 sec is a very strict timing taking into
account that the best driver response time is about 0.70 to 0.75 sec [14].
4.5.2 Protocol Improvement
The headway model can dramatically change the segmentation algorithm. It can be a basis
for a non-uniform segmentation where the width of each segment is adapted to give any
required distribution of collision probability where a minimum probability of collision leads
directly to a minimum latency (Fig. 4-11).

Fig. 4-11. Non-uniform headway-based segmentation
Without loss of generality, assume that there are only 2 vehicles in the transmission range
of the abnormal vehicle. The headway between the abnormal vehicle and the first following
one is
1
X sec, and the headway between the two following vehicles is
2
X sec as shown in
Fig. 4-12. It is clear that both and are random variables with a Semi-Poisson
probability distribution function identical to Fig. 4-10. According to the probability theory,
both variables can be replaced with a common variable X , having the same pdf.

Fig. 4-12. Study area of the analytical solution
1
X
2
X
i
l
f
l
Study area
Headway (sec)
1
X
2
X
Headway (sec)
Segment S1 S2 S3 S4



56

Assuming that, the highway is only one lane and the CW is static ( = 1 ) i.e. there is no
contention or backoff, all vehicles within one segment will wait for exactly one TS before
initiating the CTB transmission i.e. for a successful broadcast, segments should include
exactly one vehicle.
To study the collision probability in one of the segments, we assume that the segment
under consideration is in-between any arbitrary headways l

and l
]
sec (as shown in Fig.
4-12). Although the headway is a contentious variable and the probabilities should have been
in the integral form, we list here the discrete forms (i.e. after discritization) to be identical to
the ones used in the optimization program that will be discussed in Sec 4.5.3.
There will be a collision in the CTB message only if there were more than one vehicle in
the segment, that the probability of collision (P
C
) is the probability that X
1
lies inbetween
l

& l
]
and X
2
is smaller than ( l
]
-X
1
)

=
< = =
f
i
l
l x
f C
x l X P x X P P ) ( ) ( Eq. 4-2
For completeness and check of correctness, probabilities of the other conditions that may
happen in the study area are:
The probability of success broadcast (P
B
) (i.e. only one node in the segment) is the probability
that X
1
lies inbetween l

& l
]
and X
2
is larger than ( l
]
-X
1
)

=
> = =
f
i
l
l x
f B
x l X P x X P P ) ( ) ( Eq. 4-3
The probability of idle (P
I
) (i.e. there is no nodes in the segment) is the probability that X
1
is
larger than l
]
.
) (
f I
l X P P > = Eq. 4-4
The probability (P
0
) (one of the prior nodes has captured the broadcast phase) is the
probability that X
1
is smaller than l

.
) (
0 i
l X P P < = Eq. 4-5
As a check of correctness, we note that, these probabilities sum to one as shown in Fig. 4-13.



57


Fig. 4-13. Probabilities associated with an arbitrary segment

In the analysis, we are only concerned with the probability of collision and latency (i.e. P
C

and P
B
). Hence, P
C
and P
B
are rewritten below with a normalization factor so that their
summation equals one i.e. the P
C
and P
B
given that the first vehicle lies within the segment.

=
<
< <
=
=
f
i
l
l x
f
f i
C
x l X P
l X l P
x X P
P ) (
) (
) (

Eq. 4-6

=
>
< <
=
=
f
i
l
l x
f
f i
B
x l X P
l X l P
x X P
P ) (
) (
) (

Eq. 4-7
Eq. 4-6 is the one that will be used in the analytical analysis of the proposed protocol.
Although we assumed only two vehicles for simplification, the stated equations are
applicable to any number of following vehicles because only the fastest two ones are the main
cause of collision. Following the same procedures, the analysis is expandable to as multiple
lanes as practically needed.
0 2 4 6 8 10
0
0.05
0.
0.15
0.
0.25
0.
0.35
Headway (sec)
D
e
n
s
i
t
y
pdf of
1
X
I
P
0
P
B C
P P +



58

Now, we can state the design problem as; within a single communication range (10 sec as
recommended by DSRC [8]), find the best points of segmentation that result in P
C
s that are
linearly increasing with a minimum slop as shown in Fig. 4-14.


Fig. 4-14. Suggested Distribution of Collisions

There are two reasons behind minimizing the slop instead of the absolute minimum:
- Intuitively, vehicles in the first segments are more threatened to danger than those in the last
segments. Hence, they should take a higher weight in the minimization process. Each vehicle
is exposed to a danger that is inversely proportional to the time before collision (a vehicle
collision) which is the same as the headway time. Fortunately, segmentation points are in
headway time and we will get a linearly increasing latency through a linearly increasing
C
P
.
- The other reason is a traffic concept that if there were no vehicles in the first segment, we
can expect that the traffic is moderate or low, and let later segments be of wider width.

0 2 4 6 8 10 12
Headway (sec)
P
r
o
b
a
b
i
l
i
t
y

o
f

C
o
l
l
i
s
i
o
n
s

(
r
e
q
u
i
r
e
d
)
The required probability of collision
Suggested
distribution
of P
C
Find the
minimum



59

4.5.3 Analytical Results
Using Matlab commercial program, a simple program was developed to compute the Semi-
Poisson equation (Sec 4.5.1.1) and to perform the minimization process as indicated above for
different number of segments ranging from (4) to (10) segments. The best points of
segmentation are listed in Table 4-1.

Table 4-1 Best segmentation points for 330 vehicle /h (in headway sec)
10-seg = [1.01 1.74 2.51 3.34 4.26 5.27 6.36 7.52 8.75 10];
9-seg = [1.12 1.93 2.79 3.73 4.79 5.96 7.23 8.58 10];
8-seg = [1.25 2.15 3.12 4.22 5.47 6.85 8.35 10];
7-seg = [1.42 2.45 3.59 4.93 6.47 8.18 10];
6-seg = [1.62 2.81 4.18 5.86 7.81 10];
5-seg = [1.86 3.27 5.03 7.28 10];
4-seg = [2.15 3.88 6.38 10];

We note that the width of each segment is monotonically increasing as indicated earlier;
with an upper bound to 10 sec. For example, for (6) segments, the width of each segment is
{1.19 - 1.37 - 1.68 - 1.95 - 2.19}. The width of the first segment is {1.62 sec} as an indication
of the effect of the minimum headway (as stated in 4.5.1.1).
Using the above results, the probability of collision are computed as in Eq. 4-6 and the
results are as shown in Fig. 4-15.
There is a note in this figure and all subsequent figures; values (here, P
C
) are measured as
an average on the segment, and are plotted on the last point of it. For example, the value of P
C

of the last segment of 4-seg case is an average over the range {6.38 10} and is plotted at
headway of 10 sec.
It is clear that the program did its job correctly, and it indicates that increasing the number
of segments leads directly to decreasing the overall average probability of collisions.



60


Fig. 4-15. Analytical calculation of Pc for best segmentation
4.6 Step-4: Application Adaptive (Modes of Operation)
Although the majority of VANET applications require message broadcasting, each
application has its unique flavor and requires a special treatment. The difference between
them is which of the following vehicles should have the highest priority to respond first,
either with a reply to the source vehicle or a relay to the following vehicles. This modification
enables changing the order of priority according to the application requirement. The RTB
message should contain a field for the Mode which will inform other vehicles of the category
of the broadcasting application so that the following nodes arrange priority accordingly.
Without loss of generality, we propose only four modes covering major applications.
4.6.1 Mode 0 Basic Broadcasting
The zero mode is the original basic mode, where broadcasting is omnidirectional with no
intended vehicle nor acknowledgment. This mode is still useful in VANET environment
especially in case of the status message, where, as recommended by DSRC [8], every
0 2 4 6 8 10 12
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
Headway (sec)
P
r
o
b
a
b
i
l
i
t
y

o
f

C
o
l
l
i
s
i
o
n
Pc for Non-uniform seg at 330 v/h (analytical)


10-seg
9-seg
8-seg
7-seg
6-seg
5-seg
4-seg



61

vehicle should broadcast its position, speed, direction of travel, and acceleration every
300 ms, and this transmission is intended for all vehicles within 10-sec travel time.


Fig. 4-16. Mode 0 Basic Broadcasting
4.6.2 Mode 1 The Furthest Following Vehicle
The intended vehicle in this mode is the furthest one following the transmitting vehicle
(Fig. 4-17). This mode is suitable to be used with dissemination protocol for applications like
Traffic Information, and Work Zone Warning. With these applications, the broadcast is
required to be delivered to the physically furthest node; that we recommend the regular
uniform distance-based protocols (e.g. The Smart Broadcasting Protocol) to be used in this
mode. Acknowledgment is recommended to be with a basic ACK packet.


Fig. 4-17. Priority arrangement of mode 1
4.6.3 Mode 2 The Nearest-in-time Following Vehicle
The intended vehicle, in this mode, is the nearest-in-time one running behind the
transmitting vehicle (Fig. 4-18). This mode is suitable to be used with reliable protocols for
Distance (meter)
Segment S4 S3 S2 S1



62

all public-safety related applications like Cooperative Collision Warning and Stop Light
Assistant. As indicated above, the proposed non-uniform headway-based protocol is the
superior in this mode. Acknowledgment is recommended to be with the same message setting
the ACK flag. This same message ACK is to compensate collisions at far range nodes due to
hidden node problem (Sec 4.3.1).


Fig. 4-18. Priority arrangement of mode 2
4.6.4 Mode 3 The Furthest Leading Vehicle
The intended vehicle, in this mode, is the furthest one leading the transmitting vehicle.
This mode is suitable for emergency applications like Approaching Emergency Vehicle
either it was an ambulance or a police car. In this case, the headway is identical to distance
because the speed is constant (headway is measure with reference to the speed of the vehicle
that comes later). However, with headway-based protocols, we can implement a non-uniform
segmentation based on headway studies. Acknowledgment is recommended to be with a basic
ACK packet.

Fig. 4-19. Priority arrangement of mode 3
Headway (sec)
Segment S1 S2 S3 S4
Headway (sec)
Segment S1 S2 S3 S4



63

Although these four modes cover major application categories, any other mode can be
added according to the application requirements.
4.7 The Proposed Algorithm
Although we already presented every modification and its impact on performance, this
section is devoted for implementation aspects.
The suggested Wave Short Message (WSM) frame format is the same as that indicated in
Fig. 2-9 with a new 8-bit field for Mode and ACK. This field can be a part of the Provider
Service Identifier, but as indicated in Fig. 4-20, it is recommended to be a part of the WSM
data, not to confuse with the P1609.3 standard [20].

1 1 1 1 1 4 2 1 variable
WSM
Version
Security
Type
Channel
Number
Data
Rate
Tx
Power
Level
Provider
Service
Identifier
WSM
Length
Mode
&
ACK
WSM
Data
Fig. 4-20. The suggested WSM frame format
4.7.1 Algorithm of the Transmitting node
Actions of the transmitting vehicle:
In case of an OBU has a message to broadcast, the MAC layer of the system has to proceed
with the following (Fig. 4-21),
1- It sends an RTB message including its MAC address, current location, current speed,
message propagation direction and the mode of operation.
2- It waits for a valid CTB message within SIFS+N+1 time-slots (assuming N segments).
If it received a valid CTB, then it should send the unencrypted broadcast to with intended
receiver as that indicated in the CTB message. Otherwise (if not), repeat the procedure from
the beginning (as long as the application requires).





64


Fig. 4-21. Actions of the transmitting MAC

4.7.2 Algorithm of Other Vehicles
Actions of the other vehicles/nodes MAC:
Upon receiving of an RTB message, other nodes proceed with the following algorithm
(Fig. 4-22),
1- Set the NAV to be SIFS+N+2 time-slots so that nodes will not start a new session until
the end of the current broadcast.
2- Check the broadcasting mode field.
3- Compare the geographical coordinates of the transmitting vehicle with their own, and
obtain its relative position. If the receiving vehicle is in the opposite driving direction or not
in the message propagation direction, ignore the message and go to end. Otherwise, if the
receiving vehicle is in the message propagation direction, continue to Step 4.
4- Compute the headway in seconds (or distance in meter for mode 2) then determines its
segment number with reference to the operating mode. Widths of each segment are
implemented according to Table 4-1.

send RTB
received CTB
within
SIFS+N+1 Ts?
again?
start
end
yes
no
no
yes

continue
session



65


Fig. 4-22. Actions of other vehicles

5- Assuming that the segment number equals S
i
where ( i <= N ) and ( i ) is the segment
number. Set the back-off counter to be equal to ( i-1 ).
So that, nodes wait for the SIFS then decrement the back-off counter by one in each time slot
while listening to the channel for any valid CTB message, if locked with a valid CTB
message then the node should exit the contention phase and listen for the incoming broadcast.

Set NAV to
SIFS+N+2 TS
start
end
yes
Is an affected
Vehicle?
no
Find segment
number Si'

Send CTB
Received CTB
within
SIFS+ i-1 Ts?
no
yes

Check mode

Continue session



66

The node that reaches zero initiates a CTB message including its MAC address and continues
the session with the transmitting node.

This completes the analytical analysis of the proposed protocol. The next chapter will
further investigate the protocol with some simulation programs.






67

Chapter 5
Simulation Results
The only possible method of evaluating new VANET protocols is by simulation (Sec 2.4).
In order to study the performance of the proposed protocol, it is required to implement two
new models;
- A mobility model that supports generating of vehicles with positions that change
according to random speeds, keeping the headway difference as a random with the
Semi-Poisson distribution.
- A new network model where decisions are taken based on the current location, speed
and direction of travel of vehicles i.e. any simulation programs that implements the
mobility model before stating the simulation of the network model will not satisfy our
needs.
According to the complexity of both tasks, we preferred using the well-known Matlab
commercial program for being popular, intuitive and easy to use. Matlab offers a full control
of all simulation parameters.
5.1 Performance Metrics
The performance metrics used to validate the proposed protocol are:
Latency: the total time required measured from the first attempt to broadcast to the
complete of the broadcasting phase.
Collision Probability: the average probability of collision in the ACK messages in each
segment of the transmission range.



68

5.2 Measurement Methodology
Latency is measured in the simulation program using four distinct delay sources (Fig. 5-1);
1- Contention starting time: that is equal to DIFS + RTB transmission time
2- Success broadcasting time: in case of a following node replied with a valid CTB
message, the time required for the rest of the broadcasting phase, equals to SIFS + CTB
transmission time + SIFS + message transmission time + SIFS + ACK transmission time.
3- Collision time: in case of a collision in the CTB packet, the time wasted will be equal to
DIFS + RTB transmission time + SIFS + CTB transmission time + SIFS. Note that a collision
destroys a complete phase.
4- Waiting time: the time taken by a node to decide either it will reply with a CTB message
or not, each instance of this time equals to a single time slot.



Fig. 5-1. RTB/CTB/data/ACK timeline

Note that: event (1) and (2) must happen once per any broadcasting phase, however, event
(3) and (4) may happen with a variable number according to the protocol design and the
network condition. There is a tradeoff between events (3) and (4) in all broadcasting protocols
that depends on segmentation of the transmission area (e.g. [12] and [30]); increasing the
number of segments leads to smaller number of vehicles in each segment, hence, decreases
collision probability while increases waiting times, and vice versa.
Assuming that both CW
min
and CW
max
equal to one (i.e. there is no contention or random
backoff), collision probability is measured in the simulation program by dividing the number
of broadcast phases that happen to have more than one node in the first non-empty segment
by the total number of broadcast phases.
SIFS
DIFS
SIFS
SIFS
RTB DATA Transmitter
Receiver
CTB ACK



69

5.3 Simulation Parameters
Table 5-1 summarizes parameters that are taken during simulation. These parameters are
quoted from the 802.11p [22] standard. It is assumed that the length of ACK message is the
same as the original broadcast. The ACK message is a mere repeat of the original broadcast
setting the ACK field, which is considered as a compensation to the expected collisions at far
range nodes (Sec. 4.3.1).

Table 5-1 Matlab parameters
Time-Slot 16 s CTB 14 bytes
SIFS 32 s Messages 512 bytes
DIFS 64 s ACK 512 bytes
RTB 20 bytes Data rate 3 Mbps
5.4 Random Number Generator
The problem of a new mobility model is treated with a random number generator to be
used as the headway between vehicles of the same lane. The histogram of one of these
variables is shown in Fig. 5-2 (at 2000 sample) indicating the flexibility of Matlab
environment in the simulation. This headway distribution is considered at a traffic volume of
approximately 330 veh/h to be identical to the one used for the analytical analysis (Fig. 4-10).
5.5 Simulation Results
Using these random variables, a simulation program was conducted for estimating the
probability of collisions and the average latency within each segment of the communication
range (10 sec). The width of each segment is taken according to Table 4-1. The probability of
collision is shown in Fig. 5-3 while the average latency is shown in Fig. 5-4.
For clearness, we have to repeat the following note: In these figures, values are measured
as an average on the segment, and are plotted on the last point of it.
Comparing simulation results with analytical results (Fig. 4-15) proves the correctness of
the analytical analysis.



70


Fig. 5-2. Histogram of one of the variables

Fig. 5-3. Simulated calculation of Pc for best segmentation
0 2 4 6 8 10 12
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
headway (sec)
P
r
o
b
a
b
i
l
i
t
y

o
f

C
o
l
l
i
s
i
o
n
Pc for non-uniform seg at 330v/h (simulation)


10 seg
9 seg
8 seg
7 seg
6 seg
5 seg
4 seg
0 2 4 6 8 10
0
10
20
30
40
50
60
headway (sec)
N
u
m
b
e
r

o
f

s
a
m
p
l
e
s
Histogram of X



71


Fig. 5-4. Simulated calculation of latency at best segmentation

In Fig. 5-4, the curves are close to each other, and thus, the curves were plotted only for
cases with even number of segments. The average latency associated with each segment
reveals that, the case of 6-seg gives the minimum latency (best performance) before over-
segmentation begins to take place with 8 and 10 segments.
5.6 Robustness at Different Traffic Volumes
The performance of almost all previously published protocols, changes drastically with
changing the traffic volume. However, the proposed protocol possesses unique robustness at
different traffic volumes. In this section, we will study the performance of the same protocol
under different traffic volumes. The headway distribution at traffic volume of 330veh/h and
1300 veh/h (very low vs. very high) are shown in Fig. 5-5 [32].
It could be seen that increasing the traffic volume results in increasing the ratio of short
headways and decreasing that of long headways.
0 2 4 6 8 10 12
2850
2900
2950
3000
3050
headway (sec)
L
a
t
e
n
c
y

(

s
e
c
)
Latency for non-uni seg at 330veh/h (simulation)


10 seg
8 seg
6 seg
4 seg



72


Fig. 5-5. Headway distribution at 330v/h and 1300v/h
Assuming a traffic volume of 1300 veh/h, we implemented the same simulation program
with the same segmentation table (Table 4-1) which was extracted for the best operation
under 330 veh/h. We compare the performance of this table with a new table calculated
especially for the new traffic volume (Table 5-2). By this comparison, we answer the question
should we change the segmentation table according to the current traffic volume or not?
Table 5-2 Best segmentation points for 1300 vehicle /h (in headway sec)
10-seg = [ 1.16 1.93 2.72 3.55 4.45 5.43 6.5 7.63 8.81 10];
9-seg = [ 1.28 2.13 3.01 3.95 4.98 6.13 7.36 8.67 10];
8-seg = [ 1.42 2.36 3.35 4.42 5.64 7 8.46 10];
7-seg = [ 1.59 2.66 3.79 5.07 6.57 8.22 10];
6-seg = [ 1.79 3.01 4.34 5.95 7.84 10];
5-seg = [ 2.04 3.47 5.14 7.33 10];
4-seg = [ 2.34 4.06 6.42 10];

Fig. 5-6 and Fig. 5-7 show the PC and latency at 1300veh/h for 6-seg computed with,
- Segmentation points that result in best operation under a traffic volume of 330v/h
- Newly calculated segmentation points for best operation under a traffic volume of 1300v/h
0 2 4 6 8 10
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
Headway (sec)
D
e
n
s
i
t
y
Headway model for 330 V/h and 1300 V/h


1300 V/h
330 V/h



73


Fig. 5-6. P
C
for 6-seg at 1300v/h

Fig. 5-7. Latency for 6-seg at 1300v/h

0 2 4 6 8 10 12
0
0.05
0.1
0.15
0.2
0.25
Headway (sec)
P
r
o
b
a
b
i
l
i
t
y

o
f

c
o
l
l
i
s
i
o
n
Pc for 6-seg at 1300 V/h


Best for 1300 V/h
Best for 330 V/h
0 2 4 6 8 10 12
2850
2900
2950
3000
3050
3100
Headway (sec)
L
a
t
e
n
c
y

(

s
e
c
)
Average latency for 6-seg at 1300 V/h


Best for 1300 V/h
Best for 330 V/h



74

The difference in P
C
is in a range of only 0.025 and the difference in latency is in a range
of 10sec. Hence, the answer to the above question is simply no. This protocol possesses
high robustness at different traffic volumes that we do not need to recalculate the table of best
segmentation points for each traffic volume.
5.7 Protocol Comparison
The last section of this chapter is about comparing the performance of the proposed
protocol with previously published ones. We performed the same simulation analysis to
compare the performance of the proposed protocol with The Urban Multihop Broadcast
Protocol (UMB) [30] and The Smart Broadcasting Protocol (SB) [12] protocols. The
difficulty with this comparison is that the proposed protocol is the first one to use the concept
of headway-based segmentation. To accomplish this comparison, we assumed that the speed
is uniform (i.e., neglecting the effect of the proposed headway-based segmentation). The
objective of this comparison becomes only to study the effect of non-uniform segmentation
on the performance. This is to illustrate that the proposed protocol has uniquely succeeded in
achieving a linearly increasing latency with a minimum slope. Despite the neglecting of
headway, the proposed protocol has two new features: it uses slightly different positions for
segmentation and a reversed order of priority.
Fig. 5-8 shows the collision probability per each segment with respect to each protocol,
while Fig. 5-9 shows the average latency per segment.
Fig. 5-8 shows the superior performance of our protocol with respect to P
C
while lines of
both SB and UMB coincide on each other. Fig. 5-9 shows that both versions (uniform and
non-uniform segmentation) of our protocol perform better than SB and UMB. Note that
the non-uniform segmentation mimics danger situation better than uniform segmentation,
where the latency is required to be directly proportional to the headway.
This section showed that, even if we neglected the effect headway-based segmentation (the
core contribution), the proposed protocol still outperforms previously presented ones.




75


Fig. 5-8. Probability of Collision (protocol comparison)

Fig. 5-9. Latency (protocol comparison)

0 2 4 6 8 10 12
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
headway (sec)
P
r
o
b
a
b
i
l
i
t
y

o
f

C
o
l
l
i
s
i
o
n
Comparison of Pc at 330 V/h


6-seg non-uni
6-seg uni
6-seg SB
6-seg UMB
0 2 4 6 8 10 12
2850
2900
2950
3000
headway (sec)
L
a
t
e
n
c
y

(

s
e
c
)
Comparison of average latency at 330 V/h


6-seg non-uni
6-seg uni
6-seg SB
6-seg UMB



76

Chapter 6
Conclusion
In this research, we introduced a novel broadcasting protocol in VANET environments
with these new features:
The first protocol to use the concept of headway-based segmentation and to include
effects of human behaviors in its design with the headway model.
Non-uniform segmentation achieving a unique a minimum slope linearly increasing
latency distribution.
Unique robustness at different speeds and traffic volumes rooted to the headway
robustness at different traffic volume variations. For example the latency difference between
the traffic volume of 330veh/h and 1300veh/h is in a range of 10sec.
Superior minimum latency for public safety applications.
Application adaptability with special multi-mode operations.
Considered offering a solution to applications never discussed in literature, like
Approaching Emergency Vehicle.



77

Appendix A - List of Co-authored Publications

[1] VANET-DSRC Protocol for Reliable Broadcasting of Life Safety Messages, In Proc.
of the 7
th
IEEE Int. Symposium on Signal Processing and Information Technology ISSPIT07,
Cairo, EGYPT, pp. 104-109, Dec.2007.
Available: http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=4458046
[2] Integrated intra-vehicle VANET system for increasing the road safety, in Proc. of
the Global Knowledge Forum NOOR-2008, Almadina, KSA, June 2008.
[3] A Novel Headway-Based Vehicle-to-Vehicle Multi-Mode Broadcasting Protocol
accepted for publication in Proc. of the 68
th
IEEE Vehicular Technology Conf. VTC2008-
Fall, Calgary, Alberta, CANADA, 2124 September 2008.




78

Appendix B - Word-Wide VANET Projects

USA:
DSRC http://www.leearmstrong.com/DSRC/DSRCHomeset.htm
http://grouper.ieee.org/groups/scc32/index.html
California PATH http://www-path.eecs.berkeley.edu/
DynaMIT http://mit.edu/its/index.html
ITS Research Program http://www.ivhs.washington.edu/
CITranS http://citrans.pti.psu.edu/
CISR http://www.cisr.gwu.edu/
TIDE http://www.njtide.org/
ODU VANET http://www.cs.odu.edu/~vanet/index.html

Europe:
AIDE http://www.aide-eu.org/
Car to Car http://www.car-2-car.org/
CarTALK 2000 http://www.cartalk2000.net/
CVIS http://www.cvisproject.org/
FleetNet http://www.et2.tu-harburg.de/fleetnet/index.html
Invent http://www.invent-online.de/
Network on Wheels http://www.network-on-wheels.de/
SEVECOM http://www.sevecom.org/
SAFESPOT http://www.safespot-eu.org/pages/page.php
SmartPark http://smartpark.epfl.ch/
WATCH-OVER http://www.watchover-eu.org/
WILLWARN http://www.prevent-ip.org/en/prevent_subprojects/
safe_speed_and_safe_following/willwarn/

Japan:
ITS Consortium http://www.its-jp.org/english/



79

Appendix C - VANET Simulation Programs

Network Simulators
NS-2 http://www.isi.edu/nsnam/ns/
GloMoSim http://pcl.cs.ucla.edu/projects/glomosim/
QualNet http://www.scalable-networks.com/
OPNet http://www.opnet.com/
NCTUns http://nsl.csie.nctu.edu.tw/nctuns.html
MATLAB http://www.mathworks.com/

VANET Mobility Generators
VanetMobiSim http://vanet.eurecom.fr/
CanuMobiSim http://canu.informatik.uni-stuttgart.de/mobisim/


Joint Mobility and network simulators for VANET
TraNS http://trans.epfl.ch/
MOVE http://www.csie.ncku.edu.tw/~klan/move/index.htm



80

Appendix D - MATLAB Scripts

Program set #1:
%Thi s f unct i on gener at es and pl ot s t he Semi - Poi sson Di st r i but i on
%x must be equal t o ei t her 330 or 1300
%- - - - - - - - - - - - - - - - - - - - - - - - - - - - -
x=i nput ( ' Ent er t he r equi r ed vehi cl e di st r i but i on 330 or 1300: ' ) ;

i f ( x==330)
P=0. 14; %par amet er s f or appr oxi mat e 330 veh/ h
B=2. 929;
A=5. 488;
T=0. 06;
el sei f ( x==1300)
P=. 64; %par amet er s f or appr oxi mat e 1300 veh/ h
B=2. 929;
A=5. 488;
T=. 18;
el se
er r or ( ' t he val i d i nput s ar e onl y 330 or 1300' ) ;
end

f =[ ] ;
f or t =0: 0. 01: 9. 99 %1000 sampl e i n 10 sec
r =P*( B*t ) ^( A- 1) / gamma( A) *B*exp( - B*t ) +( 1-
P) *gammai nc( A*t , B) / gamma( B) *( 1+T/ A) ^B*T*exp( - T*t ) ;
f =[ f r ] ;
end
f =f / ( . 01*t r apz( f , 2) ) ; %nor mal i zi ng t he sumt o 1
pl ot ( [ 0: . 01: 9. 99] , f , ' - b' )

xl abel ( ' headway ( sec) ' )
yl abel ( ' Densi t y %' )



81

Program set #2:
%Thi s f unct i on cal cul at es t he best poi nt s of segment at i on
%- - - - - - - - - - - - - - - - - - headway pdf - - - - - - - - - - - - - - - -
x=i nput ( ' Ent er t he r equi r ed vehi cl e di st r i but i on 330 or 1300: ' ) ;

i f ( x==330)
P=0. 14; %par amet er s f or appr oxi mat e 330 veh/ h
B=2. 929;
A=5. 488;
T=0. 06;
f 0=0. 33864;
el sei f ( x==1300)
P=. 64; %par amet er s f or appr oxi mat e 1300 veh/ h
B=2. 929;
A=5. 488;
T=. 18;
f 0=0. 79708;
el se
er r or ( ' t he val i d i nput s ar e onl y 330 or 1300' ) ;
end

f =[ ] ;
f or t =0: 0. 01: 29. 99 %3000 sampl e i n 10 sec
r =P*( B*t ) ^( A- 1) / gamma( A) *B*exp( - B*t ) +( 1-
P) *gammai nc( A*t , B) / gamma( B) *( 1+T/ A) ^B*T*exp( - T*t ) ;
f =[ f r ] ;
end
f =f / f 0; %nor mal i zi ng t he summat i on t o 1

%%- - - - - - - - - - - - - - - - Run f or al l
opt i ons = opt i mset ( ' Di spl ay' , ' of f ' ) ; %opt i mi zat i on par amet er s
H_new10=0;
Ns=10; %number of segment s
Best _H1=f mi nsear ch( @opt mm, 1, opt i ons, f , Ns) ; %mi n. f unc. on t he f i r st poi nt
H_new10=ext r act _H( Best _H1, f , Ns) ; %t he compl et e set of poi nt s
di sp ( ' done f or 10- seg' ) ;
Pc10=ext r act _Pc( H_new10, f ) ; %Pc f or t he set of poi nt s

H_new9=0;
Ns=9; %same f unct i on used f or 9- seg
Best _H1=f mi nsear ch( @opt mm, 1, opt i ons, f , Ns) ;
H_new9=ext r act _H( Best _H1, f , Ns) ;
di sp ( ' done f or 9- seg' ) ;
Pc9=ext r act _Pc( H_new9, f ) ;

H_new8=0;
Ns=8; %same f unct i on used f or 8- seg
Best _H1=f mi nsear ch( @opt mm, 1, opt i ons, f , Ns) ;
H_new8=ext r act _H( Best _H1, f , Ns) ;
di sp ( ' done f or 8- seg' ) ;
Pc8=ext r act _Pc( H_new8, f ) ;

H_new7=0;



82

Ns=7; %same f unct i on used f or 7- seg
Best _H1=f mi nsear ch( @opt mm, 1, opt i ons, f , Ns) ;
H_new7=ext r act _H( Best _H1, f , Ns) ;
di sp ( ' done f or 7- seg' ) ;
Pc7=ext r act _Pc( H_new7, f ) ;

H_new6=0;
Ns=6; %same f unct i on used f or 6- seg
Best _H1=f mi nsear ch( @opt mm, 1, opt i ons, f , Ns) ;
H_new6=ext r act _H( Best _H1, f , Ns) ;
di sp ( ' done f or 6- seg' ) ;
Pc6=ext r act _Pc( H_new6, f ) ;

H_new5=0;
Ns=5; %same f unct i on used f or 5- seg
Best _H1=f mi nsear ch( @opt mm, 2, opt i ons, f , Ns) ;
H_new5=ext r act _H( Best _H1, f , Ns) ;
di sp ( ' done f or 5- seg' ) ;
Pc5=ext r act _Pc( H_new5, f ) ;

H_new4=0;
Ns=4; %same f unct i on used f or 4- seg
Best _H1=f mi nsear ch( @opt mm, 2, opt i ons, f , Ns) ;
H_new4=ext r act _H( Best _H1, f , Ns) ;
di sp ( ' done f or 4- seg' ) ;
Pc4=ext r act _Pc( H_new4, f ) ;

di sp ( ' The best poi nt s of segment at i on ar e' )
H_new10
H_new9
H_new8
H_new7
H_new6
H_new5
H_new4

f i gur e
hol d
pl ot ( H_new10, Pc10, ' - bo' )
pl ot ( H_new9, Pc9, ' - gx' )
pl ot ( H_new8, Pc8, ' - r +' )
pl ot ( H_new7, Pc7, ' - c*' )
pl ot ( H_new6, Pc6, ' - ms' )
pl ot ( H_new5, Pc5, ' - yd' )
pl ot ( H_new4, Pc4, ' - kv' )

l egend( ' 10- seg' , ' 9- seg' , ' 8- seg' , ' 7- seg' , ' 6- seg' , ' 5- seg' , ' 4-
seg' , ' Locat i on' , ' best ' )
xl abel ( ' Headway ( sec) ' )
yl abel ( ' Pr obabi l i t y of Col l i si on' )





83


f unct i on D=opt mm( H1, f , Ns)
%Thi s f unct i on i s used by t he Mat l ab opt i mi zat i on pr ogr am
%I t accept s t he f i r st poi nt of segment at i on ( H1) and t he used
%di st r i but i on ( f ) and t he number of segment s( Ns)
%I t t r i es t o f i nd a l i near l y i ncr easi ng sl op st ar t i ng at H1 and ends at
%t he exat l y 10
H=zer os( 1, Ns) ;
H( 1) =r ound( H1*100) ;

Px=sum( f ( 1: H( 1) ) ) ;
Pc=0; %f i nd sl op of Pc f or f i r st segment
f or x=1: H( 1) - 1
Pc=Pc+( f ( x) / Px) *sum( f ( 1: H( 1) - x) ) ;
end
S=Pc/ H( 1) ;

f or i =1: Ns- 1;
f or Next _H=H( i ) +1: 2000 %l oop on f i ndi ng next H
Px=sum( f ( H( i ) : Next _H) ) ;
Pc=0;
f or x=H( i ) : Next _H- 1
Pc=Pc+( f ( x) / Px) *sum( f ( 1: Next _H- x) ) ;
end
i f ( Pc/ Next _H) >= S %f i nd next H havi ng same sl ope
H( i +1) =Next _H;
br eak
end
end
end

D=abs( H( Ns) - 1000) ;

f unct i on Pc=ext r act _Pc ( H, f )
%ext r ac Pc f or Best _H
L=l engt h( H) ;
H=r ound( H. *100) ;
H=[ 1 H] ;
Pc=zer os( 1, L) ;

f or i =1: L
Px=sum( f ( H( i ) : H( i +1) ) ) ;
f or x=H( i ) : H( i +1)
Pc( i ) =Pc( i ) +( ( f ( x) / Px) *sum( f ( 1: ( H( i +1) - x) ) ) ) ;
end
end

Pc=Pc/ 100;

f unct i on H=ext r act _H( H1, f , Ns)
%ext r act al l poi nt s of segment at i on f r omt he f i r st one
H=zer os( 1, Ns) ;



84

H( 1) =r ound( H1*100) ;

Px=sum( f ( 1: H( 1) ) ) ;
Pc=0; %f i nd sl op of Pc f or f i r st segment
f or x=1: H( 1) - 1
Pc=Pc+( f ( x) / Px) *sum( f ( 1: H( 1) - x) ) ;
end
S=Pc/ H( 1) ;


f or i =1: Ns- 1;
f or Next _H=H( i ) +1: 2000 %l oop on f i ndi ng next H
Px=sum( f ( H( i ) : Next _H) ) ;
Pc=0;
f or x=H( i ) : Next _H- 1
Pc=Pc+( f ( x) / Px) *sum( f ( 1: Next _H- x) ) ;
end
i f ( Pc/ Next _H) >= S %f i nd next H havi ng same sl ope
H( i +1) =Next _H;
br eak
end
end
end

H=H/ 100;



85

Program set #3:
%Thi s f unct i on comput e t he aver age PC and l at ency f or t he best poi nt s of
%segment at i on i n 330 and 1300 v/ h
x=i nput ( ' Ent er t he r equi r ed vehi cl e di st r i but i on 330 or 1300: ' ) ;
i f ( x==330)
seg10=[ 1. 01 1. 74 2. 51 3. 34 4. 26 5. 27 6. 36 7. 52
8. 75 10. 05] ;
seg9 =[ 1. 12 1. 93 2. 79 3. 73 4. 79 5. 96 7. 23 8. 58
10. 02] ;
seg8 =[ 1. 25 2. 15 3. 12 4. 22 5. 47 6. 85 8. 35 9. 96] ;
seg7 =[ 1. 42 2. 45 3. 59 4. 93 6. 47 8. 18 10. 06] ;
seg6 =[ 1. 62 2. 81 4. 18 5. 86 7. 81 10. 04] ;
seg5 =[ 1. 86 3. 27 5. 03 7. 28 10. 02] ;
seg4 =[ 2. 15 3. 88 6. 38 9. 95] ;
el sei f ( x==1300)
seg10=[ 1. 16 1. 93 2. 72 3. 55 4. 45 5. 43 6. 5 7. 63 8. 81
10] ;
seg9 =[ 1. 28 2. 13 3. 01 3. 95 4. 98 6. 13 7. 36 8. 67 10] ;
seg8 =[ 1. 42 2. 36 3. 35 4. 42 5. 64 7 8. 46 10] ;
seg7 =[ 1. 59 2. 66 3. 79 5. 07 6. 57 8. 22 10] ;
seg6 =[ 1. 79 3. 01 4. 34 5. 95 7. 84 10] ;
seg5 =[ 2. 04 3. 47 5. 14 7. 33 10] ;
seg4 =[ 2. 34 4. 06 6. 42 10] ;
el se
er r or ( ' t he val i d i nput s ar e onl y 330 or 1300' ) ;
end


d10=zer os( 50, l engt h( seg10) ) ; %pr e al l ocat i on f or bet t er speed
c10=zer os( 50, l engt h( seg10) ) ;
d9=zer os( 50, l engt h( seg9) ) ; %d: del ay
c9=zer os( 50, l engt h( seg9) ) ; %c: col l i si ons
d8=zer os( 50, l engt h( seg8) ) ;
c8=zer os( 50, l engt h( seg8) ) ;
d7=zer os( 50, l engt h( seg7) ) ;
c7=zer os( 50, l engt h( seg7) ) ;
d6=zer os( 50, l engt h( seg6) ) ;
c6=zer os( 50, l engt h( seg6) ) ;
d5=zer os( 50, l engt h( seg5) ) ;
c5=zer os( 50, l engt h( seg5) ) ;
d4=zer os( 50, l engt h( seg4) ) ;
c4=zer os( 50, l engt h( seg4) ) ;

f or i =1: 50
[ x1] =headway( x) ;
[ d10( i , : ) c10( i , : ) ] =MMB( x1, seg10) ;
[ d9( i , : ) c9( i , : ) ] =MMB( x1, seg9) ;
[ d8( i , : ) c8( i , : ) ] =MMB( x1, seg8) ;
[ d7( i , : ) c7( i , : ) ] =MMB( x1, seg7) ;
[ d6( i , : ) c6( i , : ) ] =MMB( x1, seg6) ;
[ d5( i , : ) c5( i , : ) ] =MMB( x1, seg5) ;
[ d4( i , : ) c4( i , : ) ] =MMB( x1, seg4) ;
end



86


Del ay10=( mean( d10) ) ' ;
Pc10=( mean( c10) ) ' ;
Del ay9=( mean( d9) ) ' ;
Pc9=( mean( c9) ) ' ;
Del ay8=( mean( d8) ) ' ;
Pc8=( mean( c8) ) ' ;
Del ay7=( mean( d7) ) ' ;
Pc7=( mean( c7) ) ' ;
Del ay6=( mean( d6) ) ' ;
Pc6=( mean( c6) ) ' ;
Del ay5=( mean( d5) ) ' ;
Pc5=( mean( c5) ) ' ;
Del ay4=( mean( d4) ) ' ;
Pc4=( mean( c4) ) ' ;

f i gur e
hol d
pl ot ( seg10, Del ay10, ' - bo' )
pl ot ( seg9, Del ay9, ' - gx' )
pl ot ( seg8, Del ay8, ' - r +' )
pl ot ( seg7, Del ay7, ' - c*' )
pl ot ( seg6, Del ay6, ' - ms' )
pl ot ( seg5, Del ay5, ' - yd' )
pl ot ( seg4, Del ay4, ' - kv' )

l egend( ' 10 seg' , ' 9 seg' , ' 8 seg' , ' 7 seg' , ' 6 seg' , ' 5 seg' , ' 4
seg' , ' Locat i on' , ' best ' )
xl abel ( ' headway ( sec) ' )
yl abel ( ' Lat ency ( usec) ' )

f i gur e
hol d
pl ot ( seg10, Pc10, ' - bo' )
pl ot ( seg9, Pc9, ' - gx' )
pl ot ( seg8, Pc8, ' - r +' )
pl ot ( seg7, Pc7, ' - c*' )
pl ot ( seg6, Pc6, ' - ms' )
pl ot ( seg5, Pc5, ' - yd' )
pl ot ( seg4, Pc4, ' - kv' )

l egend( ' 10 seg' , ' 9 seg' , ' 8 seg' , ' 7 seg' , ' 6 seg' , ' 5 seg' , ' 4
seg' , ' Locat i on' , ' best ' )
xl abel ( ' headway ( sec) ' )
yl abel ( ' Pr obabi l i t y of Col l i si on' )

f unct i on [ x1] =headway( x)
%Headway Model and r andomnumber gener at or
%- - - - - - - - - - - - - - - - - - - - - - - - - - - - -
i f ( x==330)
P=0. 14; %par amet er s f or f or appr oxi mat e 330 veh/ h f or t est i ng of MMB
B=2. 929;
A=5. 488;



87

T=0. 06;
el sei f ( x==1300)
P=. 64; %par amet er s f or f or appr oxi mat e 1300 veh/ h f or t est i ng of MMB
B=2. 929;
A=5. 488;
T=. 18;
end

f =zer os( 1, 101) ;
i =1;
f or t =0: 0. 1: 10
f ( i ) =P*( B*t ) ^( A- 1) / gamma( A) *B*exp( - B*t ) +( 1-
P) *gammai nc( A*t , B) / gamma( B) *( 1+T/ A) ^B*T*exp( - T*t ) ;
i =i +1;
end
f =f / t r apz( f , 2) ;
%- - - - - - - - - - - - - - - - - - - - - - - - - - - - -
%Randomnumber gener at or
%- - - - - - - - - - - - - - - - - - - - - - - - - - - - -
f =f *10;
x1=zer os( 2000, 2) ;
y=zer os( 101, 5) ;
f or j =1: 2
i =1;
whi l e i <2000
r =10*r and;
c=cei l ( 10*r ) ;
i f y( c, j ) < 200*f ( c)
y( c, j ) =y( c, j ) +1;
x1( i , j ) =r ;
i =i +1;
end
end
x1( : , j ) =shuf f l e( x1( : , j ) ) ;
end

x1( : , 2) =x1( : , 1) +x1( : , 2) ;

f unct i on Y = shuf f l e( X)
%t hi s f unct i on i s used t o shuf f l e t he r andomnumber
Y=zer os( 1, l engt h( X) ) ;
f or I =1: l engt h( X)
A=cei l ( l engt h( X) *r and) ;
Y( I ) =X( A) ;
X( A) =[ ] ;
end

f unct i on [ d, c] =MMB( x1, seg)
%t he pr oposed pr ot ocol
%I t accept s t he r andomnumber and t he segment at i on
%- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -




88

T0=64+50. 86; %DI FS+RTB RTB=20byt e/ 3Mbps
TB=32+35. 604+32+1302. 1+32+1302. 1; %SI FS+CTB+SI FS+Dat a+SI FS+ACK
CTB=14byt e/ 3Mbps message=512byt e / 3Mbps
TI =16; %si ngl e t i me- sl ot
TC=64+50. 86+32+35. 604; %DI FS+RTB+SI FS+CTB

L_seg=l engt h( seg) ;

seg=[ 0 seg] ;
Count er =zer os( 1, L_seg) ; %same l engt h as f
Del ay=zer os( 1, L_seg) ;
C_col l =zer os( 1, L_seg) ;

f or xc1=1: 2000 %l engt h of x1
t =T0;
vehi cl e=x1( xc1, : ) ;

f or i =1: L_seg
n=sum( ( vehi cl e>seg( i ) ) & ( vehi cl e<seg( i +1) ) ) ;
i f ( n==0)
t =t +TI ;
el sei f ( n==1)
t =t +TB;
Del ay( i ) =Del ay( i ) +t ;
Count er ( i ) =Count er ( i ) +1;
br eak
el sei f ( n>1)
t =t +TC;
Del ay( i ) =Del ay( i ) +t ;
C_col l ( i ) =C_col l ( i ) +1;
br eak
end
end

end
d=Del ay. / Count er ;
c=C_col l . / ( Count er +C_col l ) ;



89

Pr ogr amset #4:
%Thi s f unct i on i s t p compar e t he per f or mance of t he pr oposed pr ot ocol wi t h
%ot her publ i shed ones

seg6=[ 1. 62 2. 81 4. 18 5. 86 7. 81 10. 04] ;
segu=[ 10/ 6 20/ 6 30/ 6 40/ 6 50/ 6 10] ;

d6=zer os( 200, l engt h( seg6) ) ;
c6=zer os( 200, l engt h( seg6) ) ;
du=zer os( 200, l engt h( segu) ) ;
cu=zer os( 200, l engt h( segu) ) ;
ds=zer os( 200, l engt h( segu) ) ;
cs=zer os( 200, l engt h( segu) ) ;
dU=zer os( 200, l engt h( segu) ) ;
cU=zer os( 200, l engt h( segu) ) ;

f or i =1: 200 %r edo t he pr ogr amf or 200 t i mes
[ x1] =headwaySB;
[ d6( i , : ) c6( i , : ) ] =MMB( x1, seg6) ;
[ du( i , : ) cu( i , : ) ] =MMB( x1, segu) ;
[ ds( i , : ) cs( i , : ) ] =SB( x1, segu) ;
[ dU( i , : ) cU( i , : ) ] =UMB( x1, segu) ;
end

Del ay6=( mean( d6) ) ' ;
Pc6=( mean( c6) ) ' ;
Del ayu=( mean( du) ) ' ;
Pcu=( mean( cu) ) ' ;
Del ays=( mean( ds) ) ' ;
Pcs=( mean( cs) ) ' ;
Del ayU=( mean( dU) ) ' ;
PcU=( mean( cU) ) ' ;

f i gur e
hol d
pl ot ( seg6, Del ay6, ' - ms' )
pl ot ( segu, Del ayu, ' - kd' )
pl ot ( segu, Del ays, ' - c*' )
pl ot ( segu, Del ayU, ' - gx' )
l egend( ' 6- seg non- uni ' , ' 6- seg uni ' , ' 6- seg SB' , ' 6- seg
UMB' , ' Locat i on' , ' best ' )
xl abel ( ' headway ( sec) ' )
yl abel ( ' Lat ency ( usec) ' )

f i gur e
hol d
pl ot ( seg6, Pc6, ' - ms' )
pl ot ( segu, Pcu, ' - kd' )
pl ot ( segu, Pcs, ' - c*' )
pl ot ( segu, PcU, ' - gx' )
l egend( ' 6- seg non- uni ' , ' 6- seg uni ' , ' 6- seg SB' , ' 6- seg
UMB' , ' Locat i on' , ' best ' )
xl abel ( ' headwaySB ( sec) ' )
yl abel ( ' Pr obabi l i t y of Col l i si on' )



90


f unct i on [ x1] =headwaySB( )
%Headway Model
%- - - - - - - - - - - - - - - - - - - - - - - - - - - - -

P=0. 14; %par amet er s f or f or appr oxi mat e 330 veh/ h f or t est i ng of MMB
B=2. 929;
A=5. 488;
T=0. 06;

f =zer os( 1, 101) ;
i =1;
f or t =0: 0. 1: 10
f ( i ) =P*( B*t ) ^( A- 1) / gamma( A) *B*exp( - B*t ) +( 1-
P) *gammai nc( A*t , B) / gamma( B) *( 1+T/ A) ^B*T*exp( - T*t ) ;
i =i +1;
end
f =f / t r apz( f , 2) ;
%t =0: 0. 1: 38;
%pl ot ( t , f )
%- - - - - - - - - - - - - - - - - - - - - - - - - - - - -
%Randomnumber gener at or
%- - - - - - - - - - - - - - - - - - - - - - - - - - - - -
f =f *10;
x1=zer os( 2000, 10) ;
y=zer os( 101, 10) ;
f or j =1: 10
i =1;
whi l e i <2000
r =10*r and;
c=cei l ( 10*r ) ;
i f y( c, j ) < 200*f ( c)
y( c, j ) =y( c, j ) +1;
x1( i , j ) =r ;
i =i +1;
end
end
x1( : , j ) =shuf f l e( x1( : , j ) ) ;
end

x1( : , 2) =x1( : , 1) +x1( : , 2) ; %f or meach 10 val ues ser i al l y
x1( : , 3) =x1( : , 2) +x1( : , 3) ;
x1( : , 4) =x1( : , 3) +x1( : , 4) ;
x1( : , 5) =x1( : , 4) +x1( : , 5) ;
x1( : , 6) =x1( : , 5) +x1( : , 6) ;
x1( : , 7) =x1( : , 6) +x1( : , 7) ;
x1( : , 8) =x1( : , 7) +x1( : , 8) ;
x1( : , 9) =x1( : , 8) +x1( : , 9) ;
x1( : , 10) =x1( : , 9) +x1( : , 10) ; %f our l ane

f unct i on Y = shuf f l e( X)
Y=zer os( 1, l engt h( X) ) ;
f or I =1: l engt h( X)
A=cei l ( l engt h( X) *r and) ;



91

Y( I ) =X( A) ;
X( A) =[ ] ;
end

f unct i on [ d, c] =MMB( x1, seg)
%The pr oposed pr ot ocol
%- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

T0=64+50. 86; %DI FS+RTB RTB=20byt e/ 3Mbps
TB=32+35. 604+32+1302. 1+32+1302. 1; %SI FS+CTB+SI FS+message+SI FS+ACK
CTB=14byt e/ 3Mbps message=512byt e / 3Mbps
TI =16; %si ngl e t i me- sl ot
TC=64+50. 86+32+35. 604; %DI FS+RTB+SI FS+CTB

L_seg=l engt h( seg) ;

seg=[ 0 seg] ;
Count er =zer os( 1, L_seg) ; %same l engt h as f
Del ay=zer os( 1, L_seg) ;
C_col l =zer os( 1, L_seg) ;

f or xc1=1: 2000 %l engt h of x1
t =T0;
vehi cl e=x1( xc1, : ) ;

f or i =1: L_seg
n=sum( ( vehi cl e>seg( i ) ) & ( vehi cl e<seg( i +1) ) ) ;
i f ( n==0)
t =t +TI ;
el sei f ( n==1)
t =t +TB;
Del ay( i ) =Del ay( i ) +t ;
Count er ( i ) =Count er ( i ) +1;
br eak
el sei f ( n>1)
t =t +TC;
Del ay( i ) =Del ay( i ) +t ;
C_col l ( i ) =C_col l ( i ) +1;
br eak
end
end

end
d=Del ay. / Count er ;
c=C_col l . / ( Count er +C_col l ) ;

f unct i on [ d, c] =SB( x1, seg)
%The Smar t pr ot ocol
%- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

T0=64+50. 86; %DI FS+RTB RTB=20byt e/ 3Mbps
TB=32+35. 604+32+1302. 1+32+1302. 1; %SI FS+CTB+SI FS+message+SI FS+ACK
CTB=14byt e/ 3Mbps message=512byt e / 3Mbps



92

TI =16; %si ngl e t i me- sl ot
TC=64+50. 86+32+35. 604; %DI FS+RTB+SI FS+CTB

L_seg=l engt h( seg) ;

seg=[ 0 seg] ;
Count er =zer os( 1, L_seg) ; %same l engt h as f
Del ay=zer os( 1, L_seg) ;
C_col l =zer os( 1, L_seg) ;

f or xc1=1: 2000 %l engt h of x1
t =T0;
vehi cl e=x1( xc1, : ) ;

f or i =L_seg+1: - 1: 2
n=sum( ( vehi cl e<seg( i ) ) & ( vehi cl e>seg( i - 1) ) ) ;
i f ( n==0)
t =t +TI ;
el sei f ( n==1)
t =t +TB;
Del ay( i - 1) =Del ay( i - 1) +t ;
Count er ( i - 1) =Count er ( i - 1) +1;
br eak
el sei f ( n>1)
t =t +TC;
Del ay( i - 1) =Del ay( i - 1) +t ;
C_col l ( i - 1) =C_col l ( i - 1) +1;
br eak
end
end

end
d=Del ay. / Count er ;
c=C_col l . / ( Count er +C_col l ) ;

f unct i on [ d, c] =UMB( x1, seg)
%The UMB pr ot ocol
%- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

T0=64+50. 86; %DI FS+RTB RTB=20byt e/ 3Mbps
TB=32+35. 604+32+1302. 1+32+1302. 1; %SI FS+CTB+SI FS+message+SI FS+ACK
CTB=14byt e/ 3Mbps message=512byt e / 3Mbps
TI =16; %si ngl e t i me- sl ot
TC=64+50. 86+32+35. 604; %DI FS+RTB+SI FS+CTB

L_seg=l engt h( seg) ;

seg=[ 0 seg] ;
Count er =zer os( 1, L_seg) ; %same l engt h as f
Del ay=zer os( 1, L_seg) ;
C_col l =zer os( 1, L_seg) ;

f or xc1=1: 2000 %l engt h of x1



93

t =T0;
vehi cl e=x1( xc1, : ) ;

f or i =L_seg+1: - 1: 2
n=sum( ( vehi cl e<seg( i ) ) & ( vehi cl e>seg( i - 1) ) ) ;
i f ( n==1)
t =t +( i - 1) *TI ;
t =t +TB;
Del ay( i - 1) =Del ay( i - 1) +t ;
Count er ( i - 1) =Count er ( i - 1) +1;
br eak
el sei f ( n>1)
t =t +( i - 1) *TI ;
t =t +TC;
Del ay( i - 1) =Del ay( i - 1) +t ;
C_col l ( i - 1) =C_col l ( i - 1) +1;
br eak
end
end

end
d=Del ay. / Count er ;
c=C_col l . / ( Count er +C_col l ) ;




94

Appendix E - References

[1] ARIB official homepage http://www.arib.or.jp/english/
[2] N. Balon, and J. Guo, Increasing Broadcast Reliability in Vehicular Ad Hoc Networks,
In Proc. of the 3
rd
ACM Int. Workshop on Vehicular Ad Hoc Networks VANET'06, NY,
USA, pp. 104-105, 2006.
[3] Bibliography on Secure Vehicular Communications
http://bbcr.uwaterloo.ca/~rxlu/sevecombib.htm
[4] B. Parno, and A. Perrig Challenges in securing vehicular networks, In Proc. of the Int.
Workshop on Hot Topics in Networks (HotNets-IV), 2005.
[5] Assessment of the applicability of cooperative vehicle-highway automation systems to
bus transit and intermodal freight: case study, California Partners for Advanced Transit
and Highways (PATH), 2004.
http://repositories.cdlib.org/cgi/viewcontent.cgi?article=1629&context=its/path
[6] Car-to-Car communication consortium http://www.car-2-car.org/
[7] CEN official homepage http://www.cen.eu/cenorm/homepage.htm
[8] Dedicated Short Range Communications (DSRC) Home,
http://www.leearmstrong.com/DSRC/DSRCHomeset.htm
http://grouper.ieee.org/groups/scc32/index.html
[9] "The official highway code," Department of Transport (DOT), Driving Standards Agency
(DSA), The Stationery Office, 2007.
[10] Statewide Plan, Department of Transportation (DOT), Intelligent Transportation
Systems (ITS), September 2005.
[11] Projects Book, Department of Transportation (DOT), Intelligent Transportation
Systems (ITS), January 2000.
[12] E. Fasolo, A. Zanella, and M. Zorzi, An Effective Broadcast Scheme for Alert Message
Propagation in Vehicular Ad hoc Networks, In Proc. of the IEEE Int. Conf. on
Communications ICC'06, vol. 9, pp. 3960-3965, 2006.
[13] GM advanced technology, Vehicle-to-Vehicle technology,
http://media.gm.com:8221/us/gm/en/technology/advanced_technology/safety_telematics/
V2V_communications.htm
[14] M. Green, How long does it take to stop? Methodological analysis of driver perception-
brake times, In Transportation Human Factors, vol.2, pp.195-216, 2000.
[15] H. Alshaer and E. Horlait, An optimized adaptive broadcast scheme for inter-vehicle
communication, In Proc. of the 61
st
Int. IEEE Vehicular Technology Conf. VTC05,
vol.5, pp. 28402844, 2005.
[16] L. Huang, A. Arora, and TH. Lai, Reliable MAC layer multicast in IEEE 802.11
wireless networks, In Proc. of the Int. Conf. on Parallel Processing ICPP'02, USA, 2002.



95

[17] IEEE 802.15.4a, Wireless medium access control (MAC) and physical layer (PHY)
specifications for low-rate wireless personal area networks (WPANs), amendment 1: add
alternate PHYs, August 2007.
[18] IEEE P1609.1, Trial-use standard for wireless access in vehicular environments
(WAVE) - resource manager, Draft D17, July 2006.
[19] IEEE P1609.2, Trial-use standard for wireless access in vehicular environments
(WAVE) - security services for applications and management messages, Draft D7, April
2006.
[20] IEEE P1609.3, Trial-use standard for wireless access in vehicular environments
(WAVE) - networking services, Draft D22, January 2007.
[21] IEEE P1609.4, Trial-use standard for wireless access in vehicular environments
(WAVE) - multi-channel operation, Draft D08, July 2006.
[22] IEEE P802.11p, Amendment 3: wireless access in vehicular environments (WAVE),
Draft D0.26, January 2006.
[23] IEEE P802.11-REVma
TM
/D7.0, Wireless LAN Medium Access Control (MAC) and
Physical Layer (PHY) Specifications, Revision of 802.11-1999, June 2006.
[24] IEEE P802.11s, Amendment: ESS Mesh Networking, Draft D1.03, April 2007.
[25] Internet Protocol Version 6 (IPv6) Specification, IETF Request for Comments, RFC
2460.
[26] Intelligent transportation systems (ITS) Home, http://www.its.dot.gov
[27] ITS Consortium http://www.its-jp.org/english
[28] J. Guo and N. Balon, "Vehicular ad hoc networks and dedicated short-range
communication," Book Chapter, 2006.
[29] J. Harri, F. Filali, and C. Bonnet, Mobility models for vehicular ad hoc networks: a
survey and taxonomy, The Institut Eurecom Department of Mobile Communications,
FRANCE, 2007.
[30] G. Korkmaz, E. Ekici, F. zgner, and U. zgner, "Urban multi-hop broadcast protocol
for inter-vehicle communication systems," In Proc. of the 1
st
ACM Int. Workshop on
Vehicular Ad Hoc Networks VANET'04, NY, USA, pp. 76-85, 2004.
[31] Largest Tiny Network Yet, http://webs.cs.berkeley.edu/800demo/
[32] T. Luttinen, Statistical analysis of vehicle time headways, Teknillinen korkeakoulu,
pp.155-172, 1996
[33] M. Heissenbttel, T. Braun, M. Wlchli, and T. Bernoulli, Optimized stateless
broadcasting in wireless multi-hop networks, in IEEE Infocom 2006, Barcelona, April
2006.
[34] M. Torrent-Moreno, D. Jiang, and H. Hartenstein, Broadcast reception rates and effects
of priority access in 802.11-based vehicular ad-hoc networks, in Proc. of the of the 1
st

ACM Int. Workshop on Vehicular Ad Hoc Networks VANET04, pp.10-18, 2004.



96

[35] M. Heron, D. Hoyert, J. Xu, C. Scott, and B. Tejada-Vera Deaths: Preliminary Data for
2006, Division of vital statistics, National Vital Statistics Reports, Vol.56, N.16, 2008.
[36] Fcc allocates spectrum in 5.9 ghz range for intelligent transportation systems uses,
News, Federal Communications Commission, 1999.
http://www.fcc.gov/Bureaus/Engineering_Technology/News_Releases/1999/nret9006.ht
ml
[37] S.Y. Ni, Y.C. Tseng, Y.S. Chen, and J.P. Sheu, "The broadcast storm problem in a
mobile ad hoc network," In Proc. of the 5th ACM/IEEE Int. Con. on Mobile Computing
and Networking MobiCom'99, NY, USA, pp. 151-162, 1999.
[38] Official IEEE 802.11 working group project timelines.
http://grouper.ieee.org/groups/802/11/Reports/802.11_Timelines.htm
[39] K. Rmer, and F. Mattern, The design space of wireless sensor networks, In Proc. of
the IEEE Wireless Communications and Networking Conf., vol.11, No.6, pp. 54-61,
2004.
[40] Status of Project IEEE 802.11 Task Group p.
http://grouper.ieee.org/groups/802/11/Reports/tgp_update.htm
[41] M. Taieb-Maimon, and D. Shinar, Minimum and Comfortable Driving Headways:
Reality versus Perception, In Human Factors, The Journal of the Human Factors and
Ergonomics Society, vol.43. pp.159-172, 2001.
[42] K. Tang, M. Gerla, MAC reliable broadcast in ad hoc networks, In Proc. of
Communications for Network-Centric Operations, Creating the Information Force, IEEE
Military Communications Conf. vol.2, pp.1008-1013, 2001.
[43] London Congestion Charging Technology Trials, Transport for London, Feb 2005.
http://www.tfl.gov.uk/assets/downloads/technology-trials.pdf
[44] USA National Transportation Statistics 2007, Bureau of Transportation Statistics,
USA, 2007.
[45] J. Xie, A. Das, S. Nandi, and A.K. Gupta, Improving the reliability of IEEE 802.11
broadcast scheme for multicasting in mobile ad hoc networks, In Proc. of the IEEE
Wireless Communications and Networking Conf., vol.1, pp.126-131. 2005.
[46] Q. Xu, T. Mak, J. Ko, and R. Sengupta, "Vehicle-to-vehicle safety messaging in DSRC,"
In Proc. of the 1
st
ACM Int. Workshop on Vehicular Ad Hoc Networks VANET'04, NY,
USA, pp.19-28, 2004.
[47] X. Yang, L. Liu, N.H. Vaidya, and F. Zhao, "A vehicle-to-vehicle communication
protocol for cooperative collision warning," In Proc. of the 1
st
Int. Conf. on Networking
and Services MOBIQUITOUS04, pp.114-123, 2004.
[48] A. Zanella, G. Pierobon, and S. Merlin, On the limiting performance of broadcast
algorithms over unidimensional ad-hoc radio networks, in Proc. of 7
th
Int. Symposium
on Wireless Personal Multimedia Communications, WPMC04, Abano Terme, Italy,
2004.






. ) - (
) VANETs (
.


) IEEE802.11 (
(IEEE802.11p)


.

.
: ) ACK (

.
) ACK ( .
.











) - (






/




2008


: -
. /
) (
. /
) (
. /
) (
/
) (
: -
. /
) (
/
) (
/
) (













) - (









/









2008