A COMPARATIVE ANALYSIS OF ROUTING
PROTOCOLS IN VANET ENVIROMNENT USING
REALISTIC VEHICULAR TRACES
A thesis submitted in partial fulfillment
of the requirements for the degree of
Master of Science (Computer and Information Science)
by
Junaid Mehboob Shaikh
Supervisor
Dr. Ivan Lee
School of Computer and Information Science
University of South Australia
JUNE 2010
DECLARATION
I declare the following to be my own work, unless otherwise referenced, as
defined by the University’s policy on plagiarism.
…………………………………..
Junaid Mehboob Shaikh
June 2010
ii
ABSTRACT
This research project is exploring the special application of infrastructure-less
wireless system called ‘Vehicular ad-hoc networks (VANET)’ which is derived
from its parent network of ‘Mobile ad hoc networks (MANET).’ The key focus of
this work is considering the investigative and analysis study of particular routing
protocols in VANET environment. The accumulated issues of these routing
protocols at network layer like: network formation, traffic movements, and route
hopping are examined thoroughly by involving the realistic traces of VANET
mobility patterns using network simulator ns-2.
The overall purpose of this study is also associated with the understanding of
available routing protocols along their effectiveness and underlying limitations
within certain density levels of city and highway scenarios. The careful
consideration of these ad hoc routing protocols: AODV, AOMDV, DSR, and
DSDV are separately incorporated by simulation models of low, medium, and
high concentration phases along with the precise parametric values of defined
scenes. The evaluative metrics for the examination of these protocols are
measured by packet delivery ratio (PDR) and average end-to-end delay
respectively.
The generated and analyzed outputs as a result of the specified inputs by mean
of extensive and rigorous simulations are reasonably rational for some and
irrational for others. AOMDV and AODV are on the rational side and originated
as the proper selection of ad hoc routing protocols for the given cases of city
and highway models of VANET with varying traffic concentration.
iii
ACKNOWLEDGEMENTS
First of all, I praise Allah Almighty for His favor to me in completing this study
work. Secondly, I would like to express thanks to my supervisor Dr. Ivan Lee for
his munificent support and encouragement throughout the research exercise. In
addition, I’m also grateful to LST research group at ETH Zurich for providing the
generated vehicular traces for further incorporation by the investigative
community. Finally, my tremendous appreciations and gratitude are for my
parents and brothers whose blessings and moral support always remained with
me in every regard.
iv
ABBREVIATIONS & SYMBOLS
Symbol/Abbreviation
Term
AODV -------------------------------------------------Ad Hoc on Demand Distance Vector
AOMDV ---------------------------------Ad Hoc on Demand Multipath Distance Vector
CBR ---------------------------------------------------------------------------Constant Bit Rate
DSDV--------------------------------------------Destination-Sequenced Distance-Vector
DSR ------------------------------------------------------------------Dynamic Source Routing
E2E ------------------------------------------------------------------------------------End-to-End
MANET ---------------------------------------------------------------Mobile Ad Hoc Network
NS-2 -----------------------------------------------------------------------Network Simulator 2
PDR ----------------------------------------------------------------------Packet Delivery Ratio
PERL---------------------------------------Practical Extraction and Reporting Language
QOS ---------------------------------------------------------------------------Quality of Service
TCL -----------------------------------------------------------------Tool Command Language
TCP -----------------------------------------------------------Transmission Control Protocol
UDP -------------------------------------------------------------------User Datagram Protocol
VANET ------------------------------------------------------------Vehicular Ad Hoc Network
v
TABLE OF CONTENTS
----------------------------------------------------- ----------------------------------------------- Page
DECLARATION---- --------------------------------------------------------------------------------- ii
ABSTRACT---------------------- ------------------------------------------------------------------- iii
ACKNOWLEDGEMENTS ------------------------------------------------------------------ -----iv
ABBREVIATIONS & SYMBOLS ---------------------------------------------------------------- v
TABLE OF CONTENTS -------------------------------------------------------------------------- vi
LIST OF TABLES --------------------------- ----------------------------------------------------- viii
LIST OF FIGURES -------------------------------------------------------------------------------- ix
CHAPTER 1 – INTRODUCTION ------------------------------------------------------------- 01
1.1 General------------------ ----------------------------------------------------------------------- 01
1.2 Problem Definition --------------------------------------------------------------------------- 03
1.3 Study Objectives ----------------------------------------------------------------------------- 04
1.4 Scope of Work
--- -------------------------------------------------------------------------05
1.5 Organization of Thesis --------------------------------------------------------------------- 05
CHAPTER 2 – LITERATURE REVIEW ----------------------------------------------------- 07
2.1 Vehicular Ad hoc Networks --------------------------------------------------------------- 08
2.1.1 Mobile Models and Patterns in VANET ------------------------------------------ 08
2.1.2 Reliability and Congestion Issues in VANET ----------------------------------- 11
2.2 Network Layer and Routing Protocols-------------------------------------------------- 13
2.2.1 Routing Protocols in VANET -------------------------------------------------------- 13
2.2.2 Selected Routing Protocols --------------------------------------------------------- 16
2.3 The Simulator---- ----------------------------------------------------------------------------- 17
2.3.1 Network Simulator 2 (NS-2) --------------------------------------------------------- 18
2.3.2 Simulaiton Components -------------------------------------------------------------- 18
2.3.3 Simulaiton Operations ---------------------------------------------------------------- 19
vi
CHAPTER 3 – RESEARCH METHODOLOGY ------------------------------------------- 20
3.1 Simulations--------- --------------------------------------------------------------------------- 20
3.1.1 Tool Command Language ----------------------------------------------------------- 22
3.1.2 Network Animator and Trace Files ------------------------------------------------ 22
3.1.3 Text Analyzer --------------------------------------------------------------------------- 24
3.2 Observations---- ------------------------------------------------------------------------------ 24
3.2.1 Routing Metrics ----- ------------------------------------------------------------------- 24
CHAPTER 4 – SIMULATIONS AND RESULTS ------------------------------------------ 26
4.1 City Scene--------- ---------------------------------------------------------------------------- 26
4.1.1 Low Density Model -------------------------------------------------------------------- 29
4.1.2 Medium Density Model --------------------------------------------------------------- 32
4.1.3 High Density Model-------------------------------------------------------------------- 34
4.2 Highway Scene------------------------------------------------------------------------------- 36
4.2.1 Low Density Model -------------------------------------------------------------------- 39
4.2.2 Medium Density Model --------------------------------------------------------------- 41
4.2.3 High Density Model-------------------------------------------------------------------- 43
CHAPTER 5 – ANALYSIS AND DISCUSSION ------------------------------------------- 46
5.1 City Results------------------------------------------------------------------------------------ 47
5.2 Highway Results ----------------------------------------------------------------------------- 49
5.3 Overall Results ------------------------------------------------------------------------------- 50
CHAPTER 6 – CONCLUSIONS AND FUTURE WORK-------------------------------- 54
REFERENCES----------- ------------------------------------------------------------------------- 57
APPENDIX A – TCL & AWK SCRIPTS WITH TRAFFIC PATTERN FILE -------- 61
APPENDIX B – ANALYZED SIMULATION RESULTS --------------------------------- 70
vii
LIST OF TABLES
Figure-------------------------------------------------------------------------------------------- Page
4.1
Common variables in city model ----------------------------------------------------- 28
4.2
City (low density) variables ------------------------------------------------------------ 29
4.3
Analyzed data of city low density----------------------------------------------------- 30
4.4
City (medium density) variables ------------------------------------------------------ 32
4.5
Analyzed data of city medium density ---------------------------------------------- 32
4.6
City (high density) variables ----------------------------------------------------------- 34
4.7
Analyzed data of city high density --------------------------------------------------- 35
4.8
Common variables in highway model ----------------------------------------------- 38
4.9
Highway (low density) varialbles ----------------------------------------------------- 39
4.10
Analyzed data of highway low density --------------------------------------------- 39
4.11
Highway (medium density) variables ----------------------------------------------- 41
4.12
Analyzed data of highway medium density --------------------------------------- 42
4.13 Highway (high density) variables ----------------------------------------------------- 43
4.14 Analyzed data of highway high density --------------------------------------------- 44
5.1
Overall evaluation matrix --------------------------------------------------------------- 51
viii
LIST OF FIGURES
Figure-------------------------------------------------------------------------------------------- Page
1.1
Wireless infrastructure network ------------------------------------------------------- 01
1.2
Wireless ad hoc network --------------------------------------------------------------- 02
3.1
Methodology flow ------------------------------------------------------------------------- 21
3.2
NAM file output ---------------------------------------------------------------------------- 23
3.3
Trace file output --------------------------------------------------------------------------- 23
4.1
City movement traces on Google map---------------------------------------------- 27
4.2
City movement traces on network animator --------------------------------------- 27
4.3
PDR at city low density ----------------------------------------------------------------- 31
4,4
Average E2E at city low density ------------------------------------------------------ 31
4.5
PDR at city medium density ----------------------------------------------------------- 33
4.6
Average E2E at city medium density ------------------------------------------------ 34
4.7
PDR at city high density ----------------------------------------------------------------- 35
4.8
Average E2E at city high density ------------------------------------------------------ 36
4.9
Highway movement traces on Google map ---------------------------------------- 37
4.10 Highway movement traces on network animator --------------------------------- 37
4.11 PDR at highway low density ------------------------------------------------------------ 40
4.12 Average E2E at highway low density ------------------------------------------------ 41
4.13 PDR at highway medium density ----------------------------------------------------- 42
4.14 Average E2E at highway medium density ------------------------------------------ 43
4.15 PDR at highway high density ---------------------------------------------------------- 44
4.16 Average E2E at highway high density ---------------------------------------------- 45
ix
5.1
Generic review ----------------------------------------------------------------------------- 46
5.2
PDR of routing protocols in city ------------------------------------------------------- 47
5.3
Average end-to-end delay of routing protocols in city -------------------------- 48
5.4
PDR of routing protocols on highway ----------------------------------------------- 49
5.5
Average end-to-end delay of routing protocols on highway ------------------- 50
5.6
Graphical representation of overall evaluation matrix --------------------------- 53
x
CHAPTER 1
INTRODUCTION
1.1 General
The advent of wireless networking is responsible for the entire drift of the
communication paradigm we observe today. This is because of its easy
deployment and setup phases. Devices are simply required to be powered by
some source of energy and having their availability within the specified ranges
to form a network and start communicating with each others with no wires or
ducts. Furthermore, these wireless networks are classified depending on their
deployment modes of fixed and flexible mobile scenarios, and therefore termed
as wireless infrastructure and wireless ad-hoc networks respectively.
The fixed mode in wireless requires a firm infrastructure to be arranged before
devices can start communication with each other. These devices (commonly
known as nodes) are then come in contact with those centrally installed bridges
or routers to forward and receive the data. This depending feature of nodes on
fixed hardware devices is an example of an infrastructure mode of wireless
network (Figure 1.1).
Figure 1.1: Wireless infrastructure network
1
On the other hand, there are circumstances where nodes do not require any
preinstalled setup due to various reasons and can directly establish their
communication by using the services of other co-joined nodes as a router for
forwarding and receiving of data in between nodes. This is an example of
wireless ad-hoc network (Figure 1.2). Moreover, ad-hoc networks with the
manipulation of WLAN (802.11) standard and its built-in support in various
devices of daily usage has introduced the notion of inexpensive and cheap
communication models. It is the most obvious reason for wireless popularity
nowadays.
Figure 1.2: Wireless ad hoc network
The communication area which is related with the scope of this thesis is an
emerging and exciting application of an ad-hoc network where vehicles are
serving as nodes. This area has certain promised aspects and activities to be
offered, which are broadly related with the safety, convenience, entertainment,
and various other topics of interest. It is an ad hoc network of vehicles, known
as ‘Vehicular Ad Hoc Network (VANET).’
2
1.2 Problem Definition
Movement of vehicles on roads is constrained by different conditions, suggested
in James (2009). These conditions are related with speed zones, traffic
congestions, weather conditions, road works, etc. These kinds of limitations
allow vehicles to form a group of clusters among them to manage traffic flow in
all directions fairly and smoothly. Another condition discussed in Victor (2009)
regarding the varying velocities of vehicles and abrupt move of paths without
any notification. With these conditions and limitations it is sometimes not
possible for vehicles to establish direct link between one another with the help
of single hop, which is related with the specified area of coverage. Hence,
internetworking among different clusters needs to be considered. To manage
the communication link between out of ranged vehicles (nodes), different routing
protocols are involved. Through relevant studies in Victor (2009), Tarik (2006),
Chung (2006), and many other associated works, it is found that the available
routing protocols for ad-hoc networks already proposed and implemented are
not majorly compatible within VANET scenario due to above conditions. Hence,
certain adaption and improvements being made with respect to available
conditions of the said network and are still in focus of many researchers for
revision.
This research work is therefore (an effort to) highlighting the importance of
routing protocols in VANET environments under different conditions (especially
through pragmatic scenarios) and to observe and analyze their effects
3
accordingly by mean of rigorous simulation test cases and comparative
analyses.
1.3 Study Objectives
The objectives of this research are devoted with the analyses of routing
protocols in vehicular networking environment. This is done by considering the
performance metrics of routes within various mobility models and densities of
vehicles; also involving their communication paradigm and hoping techniques.
The mobility models are actually the movement patterns for vehicular network
which are replicating the physical roads for simulation prototypes. Many of them
are already available and could be converged according to their particular
VANET scenarios. So this work will be emphasizing the study of most viable
routing protocols which are self converged and flexible enough within such
network situations. It also analyzes and examines the selected proposed
protocols with their mentioned future deployments on top of existing routing
protocols to observe their results. The major prominence for these analyses
would be varying according to the conditions of routing metrics like: packet
delivery ratios, average end-to-end delay, number of hop counts, and likewise.
Since many of the VANET research works are still simulation based but there
are some quite exciting and upcoming projects expected to be available before
long in the real world situations. Therefore, the study objectives highlighted here
are also depending upon various rigorous simulation scenarios. In outcome, the
analyzed results will show and differentiate the appropriate selection(s) of
4
network layer protocols i.e. routing protocols within the given circumstances of
feasible path selection in the vehicular traffic scenarios.
1.4 Scope of Work
The scope of this work is associated with the coverage of complete
communication and routing paradigm in vehicular ad hoc networks.
Initially, the simulation schemes would be considering the generic mobility
patterns of the road networks. These patterns will then be extended for the
specified cases of traffic scenarios, like city movements and highway flows
depending on the saturation capacity of the provided scenes. These aspects will
then be broadly dealing with message broadcasting, multicasting, and
unicasting
requirements,
depending
on
the
nature
of
communication.
Additionally, the routing investigations at variable densities of these mobility
models would be simulated and formulated at their certain generated test-beds
setup in the tool called Network Simulator (NS2).
1.5 Organization of Thesis
This report is written in chapter wise format with the total of six chapters. The
first chapter gives the general idea regarding the motif and theme of the work. It
also discusses the brief and concise overview of problems involved, study
objectives and scope of work. The second chapter is of literature review which
will be covering all the related topics in more detail for further understanding.
The reason of this chapter is to make the grounding of work in depth through all
5
the searched and reviewed stuff so that everything should be clear beforehand.
The concerned topics for this chapter will be mainly related with finding of issues
in VANETs and their proposed, deployed, and corresponding routing schemes
along with the simulation tool. The third chapter is of research methodology
where the useful working way will be conferred in the step wise layout. This
chapter discusses the procedures and approaches to be adopted for achieving
the objectives. The fourth chapter is the practical representation of its previous
chapter in which the Simulations’ outcome and their Results are highlighted in
both tabulated and graphical formats. This will then be followed by the
comprehensive analysis and discussion section of chapter five. By academic
tradition, the final chapter will be of conclusions and future work.
6
CHAPTER 2
LITERATURE REVIEW
This chapter covers the initial literature survey done for VANETs and their
associated schemes including network layer issues in particular. It further
followed by narrowing down to the specific literature reviews toward the selected
topic and its achieving objectives.
As mentioned earlier, due to the inherited form of MANET, most of the
operational phases of VANET are derived as well as adapted from the previous
type of network in one way or the other. Not completely relying, there are some
characteristics and distinctiveness differences from its classical beginnings.
According to the explicit discussion in Abedi (2008), the infeasible routing
criteria of well-known MANET protocols not fully compatible within VANET’s
scenario, could be due to its mobility differences. In fact, adaptability approach
remains there for some achievable results. Before discussing the routing
contemplation (the core scheme of this proposition), there are many other
interrelated areas (actually subareas) with their issues and proposed solutions
are explored during the phase of literature surveying. The major differentiation
of these sub areas are identified according to the study of mobility patterns and
their associated models, and reliability concerns with traffic flow and congestion
controls. This will then leads toward the actual progression of routing scope
from network layer.
7
2.1 Vehicular Ad hoc Networks
This section covers the related reviews on fundamental concepts and issues of
VANET mobility and reliability. It will be covering the topological formations
including various mobility models, traces, and patterns along with their available
and proposed tools. Further, some reliability schemes and congestion concerns
will be discussed.
2.1.1 Mobility Models and Patterns in VANET
The study of mobility patterns in different scenarios are depending on the
movement of node with respect to their speed and velocity. The linked
parameters remained part of study in separation for any working ambiguity.
Most of these working patterns and models are highly regarded in the following
literatures.
According to Chia-Chen (2008), the architectural model for carrying reliable
vehicle-to-vehicle services in an unreliable VANET environment has a range of
factors. These variable factors of VANET are due to its multi-hop delivery
mechanism with different network involvements. Therefore, emphasizing on
single network, instead, concept of heterogeneous vehicular network (HVN) is
proposed. The clusters of VANET with WMAN (wireless metropolitan area
network) are incorporated with each other for observing different mobility
patterns. The induction of Mobility Pattern Aware Routing Protocol (MPARP)
and HVN gave progressive outcomes. The simulation results also show the
8
credibility of enhancement in terms of packet delivery ratio (PDR), number of
links break, and instant throughput and
delay performances of
the
communication mediums. Similarly in Choffnes (2005), another mobility model
is proposed for dynamic mobile nodes movement with concentration on
metropolitan areas. It provided some achievable trials for C3 (car-to-car
cooperation) project. To include the level of realistic features, actual city maps
were used for real-time consequences in the metropolitan areas. The proposed
model named STRAW (STreet RAndom Waypoint) also evaluated the routing
performance in the ad-hoc networks. In comparison of two main routing
protocols DSR and AODV with respect to packet delivery ratio gives the clear
picture of vehicular diversification of on the road networks. There is also a
comparative study available in Djenouri (2008) discussing different VANET
mobility models like: Freeway, Manhattan, City Section Model (CSM), Stop Sign
Model (SSM), and STRAW for some positive mobility considerations with
different tools.
A simulation tool is discussed in Karnadi (2007) for generating rapid mobility
pattern and experimenting real world mobility scenarios of VANET. The tool is
named as MOVE (MObility model generator for VEhicular networks) which
provides rapid realistic simulation environment for VANET cases. The
development of this tool (in Java) was made on top of open source micro traffic
simulator SUMO which includes the functionality of map and vehicle movements
editing. The GUI based approach fascinates the simulation environment with
encouraging thing for researcher is the public availability of MOVE with certain
9
open issues need to be considered like node’s random configuration and tight
coupling via proper interfaces with other advanced tools like ns-2 and Qualnet. It
provides the options to new researchers to study serious issues of location,
speed, and directions with dynamically adjustment of real time traffic.
Another scope of realistic mobility model for VANET is described in Saha (2004)
with multiple similarities of previous works on mobility issues. It is also focusing
on real map usage for realist results commented. The proposed model is
applicable in ns-2 and therefore contributes the open research communities in
specified domain. The presented model is identifying and evaluating the work
done by Karnadi (2007). This similar nature of work is already related with some
further clarification and having suggestions to consider the mobility scenarios of
residential and business area. It is also partially done by Rybicki (2007) in their
urban pedestrian flows models.
In Wenjing (2007), much of the focus was made on safety related problems
faced by vehicular ad hoc network (VANETs) with certain limitations observed
within and general traffic monitoring. In addition, another vehicular mobility
model is proposed that reflects real world vehicle movement on road and
performance of present network. The networking performance of VANET is
directly affected by traffic rules (physical) road layouts and traffic regulations.
Keeping this fact in mind, process of VANETs requires careful investigations.
The observation leads to the drawbacks of the MANET protocols which modified
with certain changes and an investigation made to large scale VANETs phases
10
of routing protocols with the incorporations of map information and overlay road
graphs.
2.1.2 Reliability and Congestion Issues in VANET
The mode of communication paradigm related with reliability, like congestion
formation and their controlling mechanisms, are important considerations for
consistent messaging. They are greatly discussed in the following literatures.
In Chen (2007), the performance of Transmission Control Protocol (TCP) in
VANET environment with respect to transmission power of the network is
elaborated. It is highlighting, how increase in transmission power can boost the
poor performance of TCP in ad hoc networks to the level of improvement. In
VANET there are cases mostly within non-safety applications, where on-the-fly
internet connective, file sharing, message forwarding, and other likewise
communication facilities are desirable. In all of these enforcements, a reliable
connectivity via TCP is important. For this purpose, the authors have used
SWAN simulator to gather network statistics and then analyze the susceptible of
power loss. Their scope of maximum power transmission is limited to the vehicle
and the road side access point. According to similar studies, in VANETs, power
issues are more manageable via internal vehicle sources (Maeda, 2005).
According to Fahmy (2008), various interesting terms relating with specific
application oriented approach of vehicular networks are discussed. They have
proposed a mechanism for finding the relative traffic congestion and its specified
volume within the given scenario. The mode of beaconing is used to identify
11
whether the congestion is occurring or not, depending on the reply of receiving
node within the certain interval of time. The proposed idea is fairly distributed in
three algorithm calls:
1.
Discovering congestion, constructing tree, and counting nodes
2.
Construct tree and counting when node receives beacon
3.
Node loses its neighbor.
Depending on various simulation parameters, the justification of the scope with
respect to various involved complexities like message and time is readily shown
and described accordingly.
In Shie-Yuan (2008) and Wischhof (2005), emerging standard of IEEE 802.11
family for vehcles i.e. 802.11(p)/1609 along with the throttling effects are
discussed. It is fairly outlining the amendments of IEEE 802.11-2007 standard
and defines a new WAVE operational mode for vehicular environments.
The throttling issue in networks like congestion and its control always remains a
researchable topic under different circumstances. Here it is specifically focusing
vehicular networks. According to the authors in Wischhof (2005), congestion
control study within VANET is not fully identified. This is due to the high mobility
issues of node where connection loss is on high than proper connection
establishment. The proposed concept to deal with such scenario is termed as
utility-based congestion control and packet forwarding in VANETs. It is a utility
based control algorithm that encodes the quantitative information of the data
packets in a local environment. The after approach is of calling a decentralized
12
algorithm which is responsible for calculating the average utility value of the
transmitted data. Its other propositions, decentralized Utility-Based Packet
Forwarding and Congestion Control (UBPFCC) are fairly research oriented.
2.2 Network Layer and Routing Protocols
The network layer plays a vital role in any communication scheme for organizing
and delivering certain packets from the source to the destination and vice versa,
if required. This is happened by mean of specified routing protocols which help
in calculating and identifying reasonable route(s) within the required set-ups of
networking and internetworking approaches.
This section will be discussing the reviews of some network layer protocols
suitable for routing in the environment of ad hoc networks specifically of
vehicles.
2.2.1 Routing Protocols in VANET
Routing is the crux of this formal thesis project. It manifests the working grounds
of complete forwarding and sharing approaches and is reviewed along with their
issues in formation, deployment, and implementations from major studies of the
following literatures.
According to Victor (2009), movements of vehicles are described in the form of
distinct clusters, which is also an observable scene. Therefore, communication
within a single group of vehicles depends on the range of wireless coverage
being used; commonly deal with physical aspects of the networks. When it
comes to the internetworking, more generic - inter-cluster communication,
13
where source vehicle from a group is unreachable in contact with destination
vehicle from other (could be same) group creates the communication
discontinuity. To overcome the problem like these terminations, various routing
protocols are employed for continuous communication manageability purposes.
Hence, the concept of hopping where destinations achieved by mean of relaying
on other intermediate node is used specifically.
The various experimentations done in Abedi (2008) have identified that the
popular Mobile ad-hoc network (MANET) routing protocol AODV is not much
suitable for Vehicular ad-hoc network (VANET) environment. Therefore they
have proposed its modified version considering its core parameter- direction - as
a focal point for route discovery and named it as DAODV. This proposed
algorithm via theoretical simulation results has shown the reduction of overall
overhead of route discovery than AODV in comparison of highly performance
measures. The results like recovery from broken links and route expiration time
are improved with respect to number of hops, nodes, and speed with the
projected version. The logical aspect of the intended protocol is only positive
while focusing on a particular mobility model for example, Manhattan, in this
case. Whereas, there are diversified models available depending on their
mobility constrains which need to be plotted for generic acceptance of DAODV.
An application oriented study in Baldessari (2007) done by NEC Europe
researchers for analyzing the networks mobility and their deployable ability
approaches in VANET environment. The convergence of IP mobility and
Network mobility within VANET is clearly highlighted with the term ‘VANEMO.’ It
14
differentiates the related applications of VANETs by mean of ‘safety’ and ‘nonsafety’ grouping with latter’s classifications into economical and functional
performances, and deploy ability requirements. The two main approaches of
connectivity i.e. MANET-centric and NEMO-centric are discussed. With previous
approach (which agrees the scope), mobile nodes communicate with each other
by utilizing MANET layer for ad hoc routing protocols towards NEMO for high
end (infrastructure) connectivity. The latter approach is directly connecting the
physical layer of nodes with the NEMO layer. This difference is required to
maintain the connectivity tasks of safety (reliable) and non-safety (general ad
hoc) applications within different mobility scenarios. Similarly in Biswa (2006),
the application aspect of public safety issues in transportation system and how
to mitigate the cases of accident (through chain collision) cases by using vehicle
to vehicle communication paradigm are discussed. The major concepts
discussed are of Cooperative Collision Avoidance (CCA) with the help of
Wireless Collision Warning Messages (W-CWM) concept. After analyzing
various levels of connections establishment through MAC level and respective
scheduling protocols, a conclusive remark for broadcast oriented approach with
packet forwarding on geographical and temporal context being preferred. In
regard of safety application known doubts, such preferable measures are
required for possible connectivity but these may be exposed to security
concerns. The given pseudo code with plotted graphs show impressive outputs,
whether applicable for an unpredicted case is not identified. The performance
analysis of routing protocols in Garcia (2007) analyzes two location-based
routing protocols. The scope of this work is limited to SIFT (SImple Forwarding
15
over Trajectory), and
DREAM
(Distance
Routing Effect Algorithm for
Mobility) routing mechanisms. To make a realistic approach like most of the
VANET researchers prefer, a real mobility model already discussed in Djenouri
(2008) is used for carrying a comparative simulation study. According to similar
author’s view point, routing schemes for actual ad hoc networks (MANET) are
not suitable for vehicular ad hoc networks (VANET). Hence the location-based
schemes like SIFT and DREAM are more appropriate for such scenarios for
enhanced efficiency. Further within these schemes certain issues like lowconnectivity zone, spatial-awareness, and other are discussed and studied
through simulations. At the end of conclusive remark, SIFT wins the game for
efficiently resolving these issues. In addition, the considerations of routing in
large scale are further studied in (Wenjing, 2007).
With the help of all these major interrelated sections of mobility and routing
explored in this literature review, a formal methodology be derived to consider
the overall aspects of mobility and reliability with different VANET routing
schemes in a pragmatic simulation environment with the help of NS2.
2.2.2 Selected Routing Protocols
Depending on these study reviews (above) and associated literatures, there are
various routing protocols which are proposed with their suitability in MANET and
VANET perspectives. After careful consideration, following are some chosen
ones for the exploration of city and highway’s density levels of this study:
16

DSDV: Destination-Sequence Distance-Vector (Perkins, 2007), is a table
driven routing protocol where every node maintains a table of information
(which updates periodically or when change occurred in the network) of
presence of every other node within the network. Any change in network
is broadcasted to every node of the network.

AODV: Ad hoc On-demand Distance Vector (Perkins, 2007), an
improved version of DSDV, as its name suggest, establishes the route
only when demanded or required for the transmission of data. By this
mean, it only updates the relevant neighboring node(s) instead of
broadcasting every node of the network i.e. it does not make source
routing to the entire node for the entire network.

AOMDV: Ad hoc On-demand Multipath Distance Vector, an extension of
AODV with an additional feature of multipath route discovery which
prevents this on-demand routing protocol to form any loop or alternative
paths (Padmani, 2008).

DSR: Dynamic Source Routing (Perkins, 2007), an on demand routing
protocol like AOD(M)V. It maintains the source routing, in which, every
neighbor maintains the entire network route from source to the
destination.
2.3 The Simulator
This section will be reviewing the important aspect of this research including the
parts concerned with the simulation tool.
17
2.3.1 Network Simulator – 2 (NS-2)
The network simulator (NS2), as its name suggests, is a simulation tool for
replicating real life networking environment and their working and adjoining
standards respectively. It works with the combinations of different development
tools
and
languages
because
of
its
environment
of
open
source
possessiveness. Mainly by default, the backend object oriented and scripting
languages used by this simulator are the ‘C++’ and ‘TCL.’ The previous is used
for the development and implementation of low level operations and algorithms,
whereas, the latter is used for the actual scripting codes for the simulations
output scenarios. There are some associated tools with NS, like Network
Animator (NAM) and Ad-Hockey; these associates are majorly used for
visualization purposes.
2.3.2 Simulation Components
There are some very basic and generic components used by NS to establish
various special and diverse simulation scenarios. The most common (but not
limited) are the Nodes, Agents, and Links.
The nodes are the participating objects within the simulation environment.
Vehicles are the appropriate example in case of simulation scene for VANET.
These nodes can further be classified with the attributes of source and sink
depending on their traffic generator and/or receptor functions respectively.
Agents on the other hand are the dependent elements. They rely on nodes for
specifying the traffic type between their communication processes. And finally,
18
links are used to specify the medium of connection i.e. wired or wireless
between the participating nodes.
2.3.3 Simulation Operations
The simulation operations performed by the NS-2 after employing the
components (mentioned above) can be broadly categorized as follow:

Creating the event scheduler: in this operation different event related
activities being done. For example: create scheduler, schedule event(s)
and start scheduler.

Creating network: in this operation the required nodes with their linkage
and queuing operations are created.

Creating connection: in this operation the actual connection scheme e.g.
TCP or UDP is given (this work deals with TCP connection).

Creating traffic: in this operation traffic flow is being mentioned i.e. how
much traffic is needed for the simulated network. The common traffic
creation criterion is Constant Bit Rate (CBR) where constantly bits of
traffic are supplied to the network.

Tracing: this is the crucial operation which reads the NS-2 simulation
generated output file and shows different output results in the form of text
or graph.
19
CHAPTER 3
RESEARCH METHODOLOGY
The scope of Vehicular Ad Hoc Network (VANET) and its related research
studies are still in progression phases to a major extend. The limited practical
deployable options under different projects are purely simulation based before
their actual implementations in the real scenarios. The list of all major projects
along with some related developments could be found in (VANET Projects). The
collaboration of imminent research objectives and its related scope in this study
are also collapsed into same influence of simulation environment for generating
some authenticated outcomes. For this purpose, the adopted methodology for
the results of this research work (specifically comparative routing analyses) is
based on simulations near to the real time packages before any actual
implementation.
3.1 Simulations
The most reliable and authenticated tools used and preferred by most of the
researchers for these kinds of simulations are: NS-2 (Kevin, 2009) and/or
OPNET for real looking simulations according to their parameter precisions. For
vehicular movements on roads, as discussed in previous chapter, another
particular tool and its extendable variant ‘SUMO,’ the helping tool for traffic
mobility patterns generation for network simulator is used.
The generic and experimental simulation runs of this methodology adoption are
illustrated in Figure 3.1 below:
20
Mobility and Traffic Generator
City
Scenario
Highway
Scenario
TCL File with support of Mobility Patterns,
Comm. Paradigms, Reliability constraints,
and Related Parameters
NS-2
Simulator
Compile
AODV
AOMDV
DSR
DSDV
Multiple Trace & NAM Files
(According to various parameters)
Trace File Analysis
(Preferably AWK Script)
Figure 3.1: Methodology flow
21
According to the objectives of this research, the major emphasis of this study
depends on the analysis of VANET routing protocols.
These protocols are
needed to be defined individually within their specified TCL file along with their
supporting components. The movement and traffic files are generated and
compiled separately before associating with NS-2 simulation, which would then
be in the receiving format for NS2 to amalgamate with the body of actual TCL.
3.1.1 Tool Command Language
The Tool Command Language (TCL) file is the scripting representation for
coding and developing the desired networking scenarios (wired/wireless) – in
this particular case, ad hoc vehicular network flow on the road is scripted for
generating and associating relevant file. These scenarios are based on various
parameters and their settings of generated traffics along with their mobility,
reliability, and likewise constraints as discussed in previous chapter.
Initializing the routing protocol within a TCL file as inputs in association of
particular traffic and movement files, the NS-2 simulates accordingly. Ultimately,
as a result, it generates two files i.e. Network Animator File (*.nam) and a Trace
files (*.tr) as the outputs.
3.1.2 Network Animator and Trace Files
The NAM file consists of all the operations to be performed at the time of
simulation with all the positioning and graphical information and their defined
parameters. This NAM file then can be called or executed by its built-in “nam”
22
command from the operation component of NS itself. The output example of this
file is shown in Figure 3.2.
Figure 3.2: NAM file output
On the other hand, the trace file contains all of the data e.g. how many packets
are sent, received, dropped and with what sequence number, type, size, etc.
The trace file is simply available in a text format and could be called as a log file
of the simulation with all the information logged in columns format (Figure 3.3).
M 0.01000 7 (3076.65, 4672.97, 0.00), (3198.59, 4629.61), 13.65
s 2.556838879 _1_ AGT --- 0 cbr 512 [0 0 0 0] ------- [1:0 2:0 32 0] [0] 0 0
r 2.556838879 _1_ RTR --- 0 cbr 512 [0 0 0 0] ------- [1:0 2:0 32 0] [0] 0 0
s 2.560742394 _1_ RTR --- 1 DSR 32 [0 0 0 0] ------- [1:255 2:255 32 0] 1 [1
1] [0 1 0 0->0] [0 0 0 0->0]
r 2.561962728 _4_ RTR --- 1 DSR 32 [0 ffffffff 1 800] ------- [1:255 2:255
32 0] 1 [1 1] [0 1 0 0->0] [0 0 0 0->0]
r 2.561963021 _6_ RTR --- 1 DSR 32 [0 ffffffff 1 800] ------- [1:255 2:255
32 0] 1 [1 1] [0 1 0 0->0] [0 0 0 0->0]
s 2.604736825 _1_ RTR --- 2 DSR 32 [0 0 0 0] ------- [1:255 2:255 32 0] 1 [1
2] [0 2 0 0->16] [0 0 0 0->0]
Figure 3.3: Trace file output
23
3.1.3 Text Analyzer
The next task is to analyze the trace file(s). This could be done by mean of
various analyzing methods and scripting codes, for example: PERL (Practical
Extraction and Reporting Language), AWK (named after their writers, Alfred
Aho, Peter Weinberger, and Brian Kernighan) and some other third parties text
search software. For this study, AWA is used to extract meaningful values from
the generated trace files.
The sample of coding scripts for TCL, Traffic, and AWK are provided in
Appendix A.
3.2 Observations
The main observation factors are related with the calculations of particular
routing metrics. They identify the accumulated results from the output trace files
which are generated by the simulator upon their specified inputs from mobility
and traffic files.
3.2.1 Routing Metrics
There are various routing metrics devised in different literatures to signify the
importance and measuring purposes of numerous routing protocols. In (Realistic
Traces), complete surveys along with the taxonomy of these metrics with their
particular classifications are discussed in detail. The two highly discussed
metrics which are very useful in differentiating the performing trends of routing;
and specially picked by similar assessments of such protocols while analyzing
24
ns-2 traces are used for results generation in this research project. They are
‘packet delivery ratio’ and ‘average end-to-end delay.’
Packet Delivery Ratio (PDR): It is the fraction of generated packets by
received packets. That is, the ratios of packets received at the destination to
those of the packets generated by the source. As of relative amount, the usual
calculation of this system of measurement is in percentage (%) form. Higher the
percentage, more privileged is the routing protocol.
Average End-to-End Delay (E2E Delay): It is the calculation of typical time
taken by packet (in average packets) to cover its journey from the source end to
the destination end. In other words, it covers all of the potential delays such as
route discovery, buffering processes, various in-between queuing stays, etc,
during the entire trip of transmission of the packet. The classical unit of this
metric is millisecond (ms). For this metric, lower the time taken, more privileged
the routing protocol is considered.
25
CHAPTER 4
SIMULAITONS AND RESULTS
This chapter comprises of complete simulation criteria for considering the
resolution of specified objectives and their problem reports simultaneously, that
is, the behavior of routing protocols in VANETs by considering the realistic
vehicular traces. The outlining traffic mobility traces associated with simulations
and results in this chapter are the manageable efforts done by the respective
researchers of VANET domain. They are available as an attempt of contribution
and are also made public for open research communities to couple them within
their scope of study. The downloadable contents could be found at Rainer et al.
(2007) along with their relevant descriptive formation approaches and major
focus on real maps in Saha (2004). On the basis of these traces, the simulation
models are developed and used here for VANET scenarios which are classified
into two pragmatic scenes:
1. City Scene
2. Highway Scene
4.1 City Scene
The city model is considering the road patterns of the bona fide city settings of
Switzerland expanse. The specified regions within the movement files and their
masking outlines from the Google Maps to Network Animator are shown in
Figure 4.1 and Figure 4.2 respectively.
26
Figure 4.1: City movement traces on Google map (source: maps.google.com)
Figure 4.2: City movement traces on network animator
27
To study and analyze the comparative study of selected routing behavior in their
respective model, an approach of density formulation among traffic flow is used.
For this reason, city scene is further sub-classified on the basis of their
participating vehicles in a low, medium, and high density phases.
As discussed earlier, to generate simulation instance, there are certain number
of variables required to be defined within the simulation script to take action. For
different densities of city model, the common variables defined are shown in
Table 4.1 below:
Table 4.1: Common variables in city model
Variable
Value
Simulation time
300 s
Topology size
4000 m x 7000 m
Routing Protocols
AODV, AOMDV, DSR, DSDV
Traffic Type
TCP
In this table, the variables along with their appropriate values are highlighting
the simulation’s parameters for all the density phases of city model. According
to this, the approximated area of simulation (topology size) is defined by 4000m
x 7000m based on the selected city region with a typical simulation running time
of 300 seconds. Further, the routing protocols deployed individually for the
28
analysis of each simulation are: AODV, AOMDV, DSDV, and DSR. Lastly, the
traffic type for nodes’ communication is commonly defined by TCP agent source
for considering reliable communication instead of UDP.
4.1.1 Low Density Model
Simulation Parameters: The low density model along with previous common
parameters comprises of 12 vehicles as nodes. These nodes are deployed and
arranged according to the provided patterns of mobility traces with a maximum
of 8 intercommunication connections, as shown in Table 4.2.
Table 4.2: City (low density) variables
Variable
Value
No. of nodes
12
Max. Connections
8
Simulation Results: The analyzed results from the particular trace file of the
city’s low density VANET scenario are assessed on the basis of two diligent
metrics of routing protocol (Section 3.2.1), i.e. ‘PDR’ and ‘Average E2E Delay.’
The outputs of their numerical calculations are placed in Table 4.3. Whereas the
logical representation for filtering and extracting the required information from
raw trace file data by mean of AWK script is given in Appendix A at the end of
the report. It is followed by detailed analyzed results in Appendix B.
29
Table 4.3: Analyzed data of city low density
CITY LOW DENSITY
AODV
Packet Delivery Ratio
Average End-to-End Delay
AOMDV
DSR
DSDV
99.7861%
99.7767%
52.4948%
98.9038%
66.2026ms
79.5885ms
23.7363ms
68.3394ms
The observable percentage of PDR in city’s low density model are quiet
acceptable by AODV, AOMDV, and DSDV. The results generated by DSR
routing protocol are declined sharply as compared to others. Oppositely, the
Average E2E Delay is observed very low by DSR compared to other protocol
results which are still is acceptable range of less than 150ms to 250ms.
The graphical representation of the analyzed metrics of ‘PDR’ and ‘Average
E2E Delay’ along with their deliverable values on top of the specified routing
protocols of this comparative study is shown in Figure 4.3 and Figure 4.4
respectively.
30
Packet Delivery Ratio
Pc. (%) of Packet Received
(Rec. Pkt./Sent Pkt. *100)
120.00%
100.00%
80.00%
60.00%
40.00%
20.00%
0.00%
AODV
AOMDV
DSR
DSDV
Routing Protocols
Figure 4.3: PDR at city low density
Time (msec.)
Average End-to-End Delay
90
80
70
60
50
40
30
20
10
0
AODV
AOMDV
DSR
Routiong Protocols
Figure 4.4: Average E2E at city low density
31
DSDV
4.1.2 Medium Density Model
Simulation Parameters: The medium density model comprises of 260 vehicles
as nodes and intercommunication source type of TCP with a maximum of 150
connections. Rests of the parametric values are similar to that of previous low
density model declarations. The tabulated view of these variables and their
corresponding values are arranged in Table 4.4.
Table 4.4: City (medium density) variables
Variable
Value
No. of nodes
260
Max. Connections
150
Simulation Results: The analyzed numerical results for city’s medium density
VANET scenario from the generated trace files for individual routing protocols
are given in Table 4.5.
Table 4.5: Analyzed data of city medium density
CITY MEDIUM DENSITY
AODV
Packet Delivery Ratio
Average End-to-End Delay
AOMDV
98.4116%
98.3373% 13.5701%
130.047ms 218.537ms
32
DSR
DSDV
68.3788%
79.367ms 195.119ms
The analyzed results for this density model with 260 nodes and 150 TCP
connections, the PDR of routing protocols in AODV and AOMDV remains stable
comparing to DSDV with approximately quarter percentage of slash and DSR
with really poor performance with a very brief minor chunk of ratio. For average
E2E delay, AODV and DSR results are reasonable with an acceptable eye on
DSDV than AOMDV.
Similarly, the graphical representations of these analyzed results are plotted in
Figure 4.5 and Figure 4.6 respectively.
Packet Delivery Ratio
Pc. (%) of Packet Received
(Rec. Pkt./Sent Pkt. *100)
120.00%
100.00%
80.00%
60.00%
40.00%
20.00%
0.00%
AODV
AOMDV
DSR
Routing Protocols
Figure 4.5: PDR at city medium density
33
DSDV
Average End-to-End Delay
250
Time (msec.)
200
150
100
50
0
AODV
AOMDV
DSR
DSDV
Routing Protocols
Figure 4.6: Average E2E at city medium density
4.1.3 High Density Model
Simulation Parameters: The high density model comprises of 812 vehicles as
nodes with all other variable values same as that of previous medium density
model. The table for variable ~ value representation is given by Table 4.6.
Table 4.6: City (high density) variables
Variable
Value
No. of nodes
812
Max. Connections
150
34
Simulation Results: The analyzed results for city’s high density VANET
scenario with similar to previous routing metrics of ‘packet delivery ratio’ and
‘average end-to-end delay’ given in Table 4.7.
Table 4.7: Analyzed data of city high density
CITY MEDIUM DENSITY
AODV
Packet Delivery Ratio
Average End-to-End Delay
AOMDV
DSR
DSDV
98.468%
98.9253%
2.29505%
33.6591%
106.343ms
97.5508ms
17.799ms
40.795ms
The corresponding graphical representation of ‘PDR’ and ‘Average E2E Delay’
are illustrated in Figure 4.7 and Figure 4.8 respectively.
Packet Delivery Ratio
Pc. (%) of Packet Received
(Rec. Pkt./Sent Pkt. *100)
120.00%
100.00%
80.00%
60.00%
40.00%
20.00%
0.00%
AODV
AOMDV
DSR
Routing Protocols
Figure 4.7: PDR at city high density
35
DSDV
Average End-to-End Delay
120
Time (msec.)
100
80
60
40
20
0
AODV
AOMDV
DSR
DSDV
Routing Protocols
Figure 4.8: Average E2E at city high density
4.2 Highway Scene
Similar to the city outlook, highway scene is also considering the road pattern of
another bona fide trace from the Switzerland expanse. Again, the specified
regions within the movement files and their masking outlined from the Google
Maps to Network Animator are shown in Figure 4.9 and Figure 4.10
respectively.
36
Figure 4.9: Highway movement traces on Google map (source: maps.google)
Figure 4.10: Highway movement traces on network animator
37
Like city model, to make comparative study of selected routing behavior in their
respective scenes, an approach of density formulation among traffic flow is
reused. The highway model is further sub-classified on the basis of their
participating vehicles in a low, medium, and high density phases. For these
different densities, the common variables defined are shown in Table 4.8.
Table 4.8: Common variable in highway model
Variable
Value
Simulation time
300 s
Topology size
14000 m x 10000 m
Routing Protocols
AODV, AOMDV, DSR, DSDV
Traffic Type
TCP
Most of the common density variables in Table 4.8 are similar to those of city’s
densities with only exception of topology size. The size of simulation area is
expanded according to the highway’s spread out. Therefore, the approximated
topology size is defined by 14000m x10000m based on the selected highway
region. The value of simulation running time is of 300 seconds, routing protocols
deployed as AODV, AOMDV, DSDV, DSR, and the traffic connection type of
TCP are kept unchanged.
38
4.2.1 Low Density Model
Simulation Parameters: The highway low density model comprises of 370
vehicles as nodes. These nodes are deployed and arranged according to the
provided patterns of mobility traces with maximum connections of 150, they are
given in the following (Table 4.9).
Table 4.9: Highway (low density) variables
Variable
Value
No. of nodes
370
Max. Connections
150
Simulation Results: The analyzed results of routing protocols for highway’s low
density VANET scenario with their assessment metrics of ‘PDR’ and ‘Average
E2E Delay’ are placed in Table 4.10. The analyzing AWK script could be found
in Appendix A at the end of the report.
Table 4.10: Analyzed data of highway low density
HIGHWAY LOW DENSITY
Packet Delivery Ratio
Average End-to-End Delay
AODV
AOMDV
DSR
DSDV
99.43%
99.52%
29.05%
70.38%
74.4928ms 68.8368ms 20.3766ms 27.7637ms
39
The PDR percentage of low density of highway is perfectly achievable by
AODV, and AOMDV. DSDV with quarter less and DSR with only single quarter
results are suspicious in their performance. In contrast, results of latter protocols
pair for average E2E delay are remarkable along with the previous one. The
graphical representations are shown in Figure 4.11 and Figure 4.12
respectively.
Packet Delivery Ratio
Pc. (%) of Packet Received
(Rec. Pkt./Sent Pkt. *100)
120.00%
100.00%
80.00%
60.00%
40.00%
20.00%
0.00%
AODV
AOMDV
DSR
Routing Protocols
Figure 4.11: PDR at highway low density
40
DSDV
Average End-to-End Delay
80
70
Time (msec.)
60
50
40
30
20
10
0
AODV
AOMDV
DSR
DSDV
Routing Protocols
Figure 4.12: Average E2E at highway low density
4.2.2 Medium Density Model
Simulation Parameters: The medium density model for highway comprises of
837 vehicles as nodes and the intercommunication source type of TCP with a
maximum of 150 connections (Table 4.11). Remaining variable values remained
unchanged.
Table 4.11: Highway (medium density) variables
Variable
Value
No. of nodes
837
Max. Connections
150
41
Simulation Results: The analyzed numerical results for highway’s medium
density VANET scenario from the generated trace files for individual routing
protocols are given in Table 4.12 below:
Table 4.12: Analyzed data of highway medium density
HIGHWAY MEDIUM DENSITY
AODV
Packet Delivery Ratio
Average End-to-End Delay
AOMDV
97.30%
DSR
98.11%
DSDV
6.17%
12.22%
90.3927ms 70.5107ms 15.0604ms 23.2978ms
The PDR of routing protocols in AODV and AOMDV remains usual with
maximum rate. DSDV and DSR are observed to be extremely short sighted in
their performance. The average E2E delay remains under control for certain
threshold value of 150ms. The graphical representations of these analyzed
results are plotted in Figure 4.13 and Figure 4.14 respectively.
Packet Delivery Ratio
Pc. (%) of Packet Received
(Rec. Pkt./Sent Pkt. *100)
120.00%
100.00%
80.00%
60.00%
40.00%
20.00%
0.00%
AODV
AOMDV
DSR
Routing Protocols
Figure 4.3: PDR at highway medium density
42
DSDV
Average End-to-End Delay
140
Time (msec.)
120
100
80
60
40
20
0
AODV
AOMDV
DSR
DSDV
Routing Protocols
Figure 4.4: Average E2E at highway medium density
4.2.3 High Density Model
Simulation Parameters: The high density model comprises of 1112 vehicles as
nodes with 150 traffic connections of TCP type (Table 4.13).
Table 4.13: Highway (high density) variables
Variable
Value
No. of nodes
1112
Max. Connections
150
Simulation Results: The analyzed results for highway’s high density VANET
scenario with routing metrics of ‘packet delivery ratio’ and ‘average end-to-end
delay’ are given in Table 4.14.
43
Table 4.14: Analyzed data of highway high density
HIGHWAY HIGH DENSITY
AODV
Packet Delivery Ratio
AOMDV
97.61%
Average End-to-End Delay
DSR
98.15%
DSDV
6.48%
8.16%
116.572ms 94.2247ms 17.3722ms 11.0122ms
The analyzed results for PDR are nearly similar to medium density case with
minor variations observed in average E2E delay values. They are graphically
represented in Figure 4.15 and Figure 4.16 respectively.
Packet Delivery Ratio
Pc. (%) of Packet Received
(Rec. Pkt./Sent Pkt. *100)
120.00%
100.00%
80.00%
60.00%
40.00%
20.00%
0.00%
AODV
AOMDV
DSR
Routing Protocols
Figure 4.15: PDR at highway high density
44
DSDV
Average End-to-End Delay
140
Time (msec.)
120
100
80
60
40
20
0
AODV
AOMDV
DSR
DSDV
Routing Protocols
Figure 4.16: Average E2E at highway high density
45
CHAPTER 5
ANALYSIS AND DISCUSSION
In this chapter a comparative analysis of the selected routing protocols based
on their investigative metrics are thoroughly discussed. In a brief review, the
complete extent of the overall study comprises of two decisive factors: ‘Packet
Delivery Ratio’ and ‘Average End-to-End Delay.’ These features are directly
dependant on the formulation of their respective density models which are
classified from their parent models of City and Highway scenes. The hierarchical
designs of these momentous objectives are shown here:
City Scene



Highway Scene

Low Density Model
o Packet Delivery Ratio
o Average End-to-End Delay
Medium Density Model
o Packet Delivery Ratio
o Average End-to-End Delay
High Density Model
o Packet Delivery Ratio
o Average End-to-End Delay


Low Density Model
o Packet Delivery Ratio
o Average End-to-End Delay
Medium Density Model
o Packet Delivery Ratio
o Average End-to-End Delay
High Density Model
o Packet Delivery Ratio
o Average End-to-End Delay
Figure 5.1: Generic review
Considering the basis of the given streams along with their simulation outputs,
the enhanced plotting taken as a whole from the individually analyzed factors
(pervious chapter) with respect of routing protocols are discussed graphically
within their panoramas of city and highway results.
46
5.1 City Results
The city observations from different approaches of varying densities with
respect to packet delivery ratio and average end-to-end delay are looking
rational for some instances while irrational for the others. These are due to the
depending consequences of routines and performances of the routing protocol.
The big pictures of these remarks are graphically plotted in Figure 5.2 and
Figure 5.3 respectively.
Packet Delivery Ratio
Delivery Percentage (%)
120.00%
100.00%
80.00%
60.00%
City Low Density
40.00%
City Medium Density
City High Density
20.00%
0.00%
AODV
AOMDV
DSR
DSDV
Routing Protocols
Figure 5.2: PDR of routing protocols in city
47
Time (msec)
Average End-to-End Delay
450
400
350
300
250
200
150
100
50
0
City High Density
City Medium Density
City Low Density
AODV
AOMDV
DSR
DSDV
Routing Protocols
Figure 5.2: Average end-to-end delay of routing protocols in city
In packet delivery ratio (PDR) plotting (Figure 5.1), the more hype in the value of
percentage represents an added achievable performance of their respective
protocol. In the consideration of careful simulation results, AODV and AOMDV
made their obvious marks in all of the VANET density schemes of the given city
scenario. It is also apparent that DSDV has its acceptability for city’s low density
variation only. To end with, DSR and DSDV (in medium and high density)
remained unsuitable for the given case of city models from showing their
approachable PDR results.
Conversely, for average end-to-end (Figure 5.2), where minimal timing values
are required, DSR majorly wins the game for all of the density levels. In addition,
for low density level, all of the routing protocols had equally acceptable values.
These rates still remained intact (in general acceptability) by AODV and DSDV
48
than AOMDV for medium density. Finally, the admirable limits are crossed by
every routing protocol (except DSR) for high density phase.
5.2 Highway Results
Similar to city observations, highway results have also formed mixed variations
of balanced and imbalanced approaches inside varying densities with respect to
packet delivery ratio and average end-to-end delay. The facts of their
accumulated results by mean of graphical plotting are shown in Figure 5.3 and
Figure 5.4 respectively.
Packet Delivery Ratio
Delivery Percentage (%)
120.00%
100.00%
80.00%
60.00%
Highway Low Density
40.00%
Highway Medium Density
Highway High Density
20.00%
0.00%
AODV
AOMDV
DSR
DSDV
Routing Protocols
Figure 5.4: PDR of routing protocols on highway
49
Average End-to-End Delay
300
Time (msec.)
250
200
150
Highway High Density
100
Highway Medium Density
Highway Low Density
50
0
AODV
AOMDV
DSR
DSDV
Routing Protocols
Figure 5.5: Average end-to-end delay of routing protocols on highway
For packet delivery ratio (PDR) (Figure 5.3), elevate percentage means higher
performance. In view of this fact, AODV and AOMDV remained the highly
acceptable approaches for all of the density levels of highway model. In
contrast, DSR and DSDV with no chance of assumption are bearing the
comportment for this measuring metric in the specified scene of VANET.
On the other hand, the average end-to-end (Figure 5.2) remained in approach
by all of the routing protocols with a minor doubt of variation by AODV in high
density model.
5.3 Overall Evaluation
The overall evaluation is based on the findings of simulation results discussed in
previous sections of city and highway results. It is basically the formation of
50
evaluation matrix (Vehar, 1996), where rows and columns are scored according
to their corresponding weighting factors and rating values to make final score.
Carrying this procedure, the matrix is transformed accordingly for the evaluation
of individual performance of routing protocols with respect to their measuring
metrics of PDR and Average end-to-end delay. Following is table of such
generation:
Table 5.1: Overall Evaluation Matrix
Evaluative Routing Metrics
Packet Delivery Ratio
Routing
Average End-to-End Delay
Weighting
Total
Rating
Protocols
Score
Rating
Score
Factor
Score
AODV
4
4
AOMDV
4
4
DSDV
3
2
DSR
2
1
16
2
8
24
2
8
24
6
3
9
15
2
4
8
10
16
The major scope of this matrix is associated with the cross relation of routing
protocols and their evaluative metrics. They are distributed in the table by first
column and part of first row respectively. The weighting factor column is
assigned to each criterion based on its effect on the success of its consideration
depending on their generated results from previous chapter. They are classified
as: AODV - 4, AOMDV – 4, DSDV – 3, and DSR – 2. There is a quantitative
51
rating to each qualitative value such as: poor, fair, good, and very-good. These
values are further associated with their respective numerical values of 1, 2, 3,
and 4 correspondingly. The assignments of these values are done by
characterizing the overall performance of all the simulation results according to
their output levels of the generated results. After the weights are assigned to
each category, the calculation phase is established. It is where the assigned
weights are multiplied by their relative rating to determine the specified
operation score (as in case of PDR and Average E2E delay). At the end of each
score the summation is being performed on to get the total score of every
individual participant from their operative scores. The member with high score in
total will be assigned the ranking accordingly and fairly on the basis of its overall
performance results mentioned earlier.
Total score column of the given table clearly shows up to standard
performances of AODV and AOMDV on equality basis, despite of extra
enhancement by AOMDV discussed in literature review. Furthermore, DSDV
and DSR lagged behind their performance (especially in PDR than E2E) and
remained unsuitable for the given case of VANET environment of city and
highway with varying densities.
The graphical demonstration of this overall
evaluation matrix approach is shown in Figure 5.6.
52
Total Score
30
25
20
15
Total Score
10
5
0
AODV
AOMDV
DSDV
DSR
Figure 5.6: Graphical representation of overall evaluation matrix
53
CHAPTER 6
CONCLUSIONS AND FUTURE WORK
In this proportional study of specified routing protocols in VANET environment,
certain considerable and realistic approaches of ad hoc networks are thoroughly
examined. The involvement of vehicular traces of city and highway models
really played their vital role. Moreover, the representations of these models
along with their variable density levels have generated faction wise numerous
practical results. These results now could be used for the assortment and
selection of routing protocols for specified domains.
The generic choice of routing protocols: AODV, AOMDV, DSDV, and DSR are
made on the basis of vigilant thoughts of related work in the same domain of ad
hoc networks. Similarly, for analyses purpose among the variety of available
routing metrics, the options are prepared for packet delivery ratio (PDR) and
average end-to-end delay. It is due to their classical and comprehensible
differentiation in the context of routing measures.
The simulation test beds for two scenarios of city and highway are deployed by
ns-2 along with their injecting level of densities. The density scale was defined
by the number of participating vehicles within the projected scene of city and
highway. They are classified as: low, medium, and high, depending on the
vehicular strength and devises results.
The generated and analyzed results for city and highway models apparently
remained competitive in between AODV and AOMDV routing protocols. The
54
PDR values of both of them were tussling in between 97% to 99%
approximately, whereas, the Average E2E Delay remained within the threshold
limit of 250ms by all of the participating protocols.
In a nutshell, the end results by mean of extensive and rigorous simulations
within the particular deployable test beds are quite rational and pragmatic in
comparison to the real roads and traffic scenarios. As an outcome, AOMDV and
AODV - two favorite ad hoc routing protocols - are found the most appropriate
selection for reasonably adjustable protocols at the network layer of given
cases, i.e. city and highway models in VANET with varying traffic concentration.
The test bed(s) created as an endurance of this work can be utilize in a variety
of manners for the future work, both in an immediate and steady ways.
The two main instantaneous diversions can be drive through from:
1. Investigative study of numerous protocols of OSI layers by mean of their
corresponding parametric value, similar or revolutionize as mentioned in
this work.
2. Formation of practical traces with the help of coordination modeling. The
city, highway, and relevant coordinates can be obtained from subsequent
research schools of regional roads and transport studies.
The other move of gradual and continuing approach of future work can be made
with the help exponential mapping; utilizing Google Maps can be a good start
from first to last. This could be comprises of plans from regional to national, and
55
from national to international road maps and traffic study. These mapping
schemes will further lead towards the geographical constraints in-between
digital and physical scopes, for example, Quality of Service (QoS) issues,
reliability check, security concerns, and likewise reflections to ponder upon in
VANETs.
56
REFERENCES
Abedi, O., M. Fathy, et al. 2008, Enhancing AODV routing protocol using
mobility parameters in VANET, IEEE/ACS International Conference on
Computer Systems and Applications, AICCSA 2008.
Baldessari, R., A. Festag, et al. 2007, NEMO meets VANET: A Deployability
Analysis of Network Mobility in Vehicular Communication, 7th International
Conference on ITS Telecommunications, ITST '07.
Biswas, S., R. Tatchikou, et al. 2006, "Vehicle-to-vehicle wireless
communication protocols for enhancing highway traffic safety," Communications
Magazine, IEEE 44(1): 74-82.
Chen, A., B. Khorashadi, et al. 2007, Impact of transmission power on TCP
performance in vehicular ad hoc networks, Fourth Annual Conference on
Wireless on Demand Network Systems and Services, WONS '07.
CHEN, T.S., CHIH, Y.C., and YUH, S.C. 2005, Wireless Ad Hoc and Sensor
Networks, Journal of Internet Technology, Vol. 6, No. 1.
Chia-Chen, H., H. Chan, et al. 2008, Mobility Pattern Aware Routing for
Heterogeneous Vehicular Networks. IEEE Wireless Communications and
Networking Conference, WCNC 2008.
Choffnes, D. R., Fabin, et al. 2005, An integrated mobility and traffic model for
vehicular wireless networks, Proceedings of the 2nd ACM international
workshop on Vehicular ad hoc networks, Cologne, Germany, ACM.
Chung, W., Kuo, S., Chen, S. 2002, Direction-Aware Routing Protocol for Mobile
Ad Hoc Networks, Department of Electrical Engineering, National Taiwan
University. IEEE Publication, 2002.
COMPUTER MAGAZINE, Institute of Electrical and Electronic Engineering
(IEEE) Press, Issue February 2004.
57
Djenouri, D., W. Soualhi, et al. 2008, VANET's Mobility Models and Overtaking:
An Overview, 3rd International Conference on Information and Communication
Technologies: From Theory to Applications, ICTTA 2008.
Fahmy, M. F. and D. N. Ranasinghe 2008, Discovering automobile congestion
and volume using vanet's, 8th International Conference on ITS
Telecommunications, ITST 2008.
HALDAR,
P.
2002
Wireless
World
in
<http://www.isi.edu/nsnam/ns/ns-tutorial/tutorial-02>
NS.
USC/ISI,
Garcia de la Fuente, M. and H. Labiod 2007, Performance analysis of positionbased routing approaches in VANETS, 9th IFIP International Conference on
Mobile Wireless Communications Networks.
James, B., Manivanna, D., 2009, Unicast routing protocols for vehicular ad hoc
networks: A criticla comparison and classification. Science Direct, Passiver and
Mobile Computing.
Karnadi, F. K., M. Zhi Hai, et al. 2007, Rapid Generation of Realistic Mobility
Models for VANET, IEEE Wireless Communications and Networking
Conference, WCNC 2007.
Kevin, F., Kannan, V., 2009, The ns Manual. The VINT Project, January 06,
2009.
KUROSE, J.F., and ROSS, K.W. 2004, Computer Networking: A Top Down
Approach Featuring The Internet, 3rd ed., Addison Wesley.
LANG, D. 2003, A Comprehensive Overview About Selected Ad Hoc
Networking Routing Protocols, DCS, Technische Uni. Munich, Germany, pp. 47.
Maeda, K., K. Sato, et al. 2005, Getting urban pedestrian flow from simple
observation: realistic mobility generation in wireless network simulation,
Proceedings of the 8th ACM international symposium on Modeling, analysis and
simulation of wireless and mobile systems, Montral, Quebec, Canada, ACM.
58
Mahesh, K. and R. Samir, Ad hoc on-demand multipath distance vector routing,
Computer Science Department, University of California, Los Angeles, CA
90095-1596, U.S.A.
Padmini, M, Routing protocols for Ad Hoc Mobile Wireless Networks, available
at: < http://www.cse.wustl.edu/~jain/cis788-99/ftp/adhoc_routing/>
Perkins, C., Belding, E., and Das, S. 2003, Ad hoc On-Demand Distance
Vector (AODV) Routing. Request for Comments: 3561.
Rainer, B., et al. 2007, A Survey on Routing Metrics, TIK Report 262, Computer
Engineering and Networks Laboratory, ETH-Zentrum, Switzerland, February 10,
2007.
Realistic Vehicular Traces, available at: <www.lst.inf.ethz.ch/research/adhoc/car-traces>
Rybicki, J., et al. 2007, Challenge: peers on wheels - a road to new traffic
information systems, in Proceedings of the 13th annual ACM international
conference on Mobile computing and networking.
Saha, A. K. and D. B. Johnson 2004, Modeling mobility for vehicular ad-hoc
networks, Proceedings of the 1st ACM international workshop on Vehicular ad
hoc networks, Philadelphia, PA, USA, ACM
Shie-Yuan, W., C. Hsi-Lu, et al. 2008, Evaluating and improving the TCP/UDP
performances of IEEE 802.11(p)/1609 networks, Computers and
Communications, ISCC 2008.
Tarik, T., Mitsuru, O., et al. 2006, An Efficient Vehicle-Heading Based Routing
Protocol for VANET Networks, IEEE WCNC Proceedings.
VANET Projects List available at <www.vanet.info/projects>
Vehar, J and Firestein, R 1996, Creative Thinking and Creative Problem
Solving, Buffalo: Creative Education Foundation.
59
Victor, C., Francisco, J., Pedro, M. 2009, Simulation-based Study of Common
Issues in VANET Routing Protocols. IEEE 69th Vehicular Technology
Conference, VTC2000.
Wenjing, W., X. Fei, et al. 2007, TOPO: Routing in Large Scale Vehicular
Networks, IEEE 66th Vehicular Technology Conference, VTC-2007.
Wenjing, W., X. Fei, et al. 2007, An Integrated Study on Mobility Models and
Scalable Routing Protocols in VANETs. 2007 Mobile Networking for Vehicular
Environments.
Wischhof, L. and H. Rohling 2005, Congestion control in vehicular ad hoc
networks, IEEE International Conference on Vehicular Electronics and Safety,
2005.
60
APPENDIX A
TCL & AWK SCRIPTS WITH TRAFFIC PATTERN FILE
Page
Sample A – 1
MAIN TCL script file
62
Sample A – 2
Traffic pattern file
66
Sample A – 4
AWK script
68
61
# ===========================================================
# MAIN TCL script file
# ===========================================================
#===========================================================
#
#Simple VANET Scenario
#
#Junaid M. Shaikh
#
#Courtesy: Marc Gries Tutorial
#
#===========================================================
#===========================================================
# Define Options
#===========================================================
set opt(chan)
set opt(prop)
model
set opt(netif)
set opt(mac)
set opt(ifq)
Channel/WirelessChannel
Propagation/TwoRayGround
;# channel type
;# radio-propagation
Phy/WirelessPhy
Mac/802_11
Queue/DropTail/PriQueue
;# network interface type
;# MAC type
;# Interface queue type
##Following option should be enabled for DRS as (rp)
#set opt(ifq)
set opt(ll)
set opt(ant)
set opt(ifqlen)
set opt(nn)
mobilenodes
#set opt(rp)
#set opt(rp)
##set opt(rp)
set opt(rp)
set opt(sc)
set opt(cp)
CMUPriQueue
LL
Antenna/OmniAntenna
100
812
;# Interface queue type
;# Link layer type
;# Antenna type
;# max packet in ifq
;# number of
AODV
AOMDV
DSR
DSDV
;# ad-hoc routing protocol
;# ad-hoc routing protocol
;# ad-hoc routing protocol
;# ad-hoc routing protocol
chd.tcl"
"tcp-812-150-ch"
;# node movement file
;# traffic file
62
# ============================
# Define simulation area range
# ============================
#set opt(x)
#set opt(y)
set opt(x)
set opt(y)
3000
3000
4011
7011
#===========================================================
# Main Program
#==========================================================
# ===========================
# Creating simulator instance
# ===========================
set ns_ [new Simulator]
$ns_ color 0 Brown
#$ns_ color 0 Blue
# ====================================================
# Creating network animator (NAM) and Trace (tr) files
# ====================================================
set nf [open vanet.nam w]
set tf [open vanet.tr w]
$ns_ namtrace-all-wireless $nf $opt(x) $opt(y)
$ns_ trace-all $tf
# =============================
# Creating topological boundary
# =============================
set topo [new Topography]
$topo load_flatgrid $opt(x) $opt(y)
# ========================================
# Create general operations director (god)
# ========================================
set god_ [create-god $opt(nn)]
63
# ==================
# Node configuration
# ==================
$ns_ node-config -adhocRouting $opt(rp) \
-llType $opt(ll) \
-macType $opt(mac) \
-ifqType $opt(ifq) \
-ifqLen $opt(ifqlen) \
-antType $opt(ant) \
-propType $opt(prop) \
-phyType $opt(netif) \
-topoInstance $topo \
-agentTrace ON \
-routerTrace ON \
-macTrace OFF \
-movementTrace ON \
-channel [new $opt(chan)]
# =================
# Creation of nodes
# =================
#set node_(0) [$ns_ node]
#set node_(1) [$ns_ node]
for {set j 0} {$j < 812} {incr j} {
set node_($j) [$ns_ node]
}
#$ns_ color 1 white
#$ns_ color 2 red
#$ns_ color 3 green
# =================
# Creating mobility
# =================
source $opt(sc)
for {set i 0} {$i < 812} {incr i} {
$ns_ initial_node_pos $node_($i) 812
}
#$ns_ at 0.1 "$node_(1) setdest 1.0 70.0 50.0"
#$ns_ at 0.1 "$node_(0) setdest 299.0 50.0 50.0"
64
# ================
# Creating traffic
# ================
source $opt(cp)
#set tcp [new Agent/TCP]
#$tcp set class_ 2
#set sink [new Agent/TCPSink]
#$ns_ attach-agent $node_(0) $tcp
#$ns_ attach-agent $node_(1) $sink
#$ns_ connect $tcp $sink
#set ftp [new Application/FTP]
#$ftp attach-agent $tcp
#$ns_ at 0.1 "$ftp start"
# =====================
# Simulation end timing
# =====================
$ns_ at 300 "finish"
#===========================================================
# End of Simulation
#===========================================================
proc finish {} {
global ns_ nf tf
$ns_ flush-trace
close $nf
close $tf
exec nam vanet.nam &
#exec xgraph simple-vanet.tr -geometry 800x400 &
exit 0
}
#===========================================================
# Running Simulation
#===========================================================
puts "Starting Simulation..."
$ns_ run
#################
#------EOF------#
#################
65
#===========================================================
# Traffic Pattern File
#===========================================================
#
# nodes: 812, max conn: 150, send rate: 0.0, seed: 0.0
#
#
# 0 connecting to 1 at time 39.923569513449245
#
set tcp_(0) [$ns_ create-connection TCP $node_(0) TCPSink $node_(1) 0]
$tcp_(0) set window_ 32
$tcp_(0) set packetSize_ 512
set ftp_(0) [$tcp_(0) attach-source FTP]
$ns_ at 39.923569513449245 "$ftp_(0) start"
#
# 1 connecting to 2 at time 88.06065001900339
#
set tcp_(1) [$ns_ create-connection TCP $node_(1) TCPSink $node_(2) 0]
$tcp_(1) set window_ 32
$tcp_(1) set packetSize_ 512
set ftp_(1) [$tcp_(1) attach-source FTP]
$ns_ at 88.06065001900339 "$ftp_(1) start"
#
# 1 connecting to 3 at time 114.97263667870902
#
set tcp_(2) [$ns_ create-connection TCP $node_(1) TCPSink $node_(3) 0]
$tcp_(2) set window_ 32
$tcp_(2) set packetSize_ 512
set ftp_(2) [$tcp_(2) attach-source FTP]
$ns_ at 114.97263667870902 "$ftp_(2) start"
#
# 4 connecting to 5 at time 100.22347272384141
#
set tcp_(3) [$ns_ create-connection TCP $node_(4) TCPSink $node_(5) 0]
$tcp_(3) set window_ 32
$tcp_(3) set packetSize_ 512
set ftp_(3) [$tcp_(3) attach-source FTP]
$ns_ at 100.22347272384141 "$ftp_(3) start"
#
# 5 connecting to 6 at time 165.62555372045634
#
set tcp_(4) [$ns_ create-connection TCP $node_(5) TCPSink $node_(6) 0]
$tcp_(4) set window_ 32
66
$tcp_(4) set packetSize_ 512
set ftp_(4) [$tcp_(4) attach-source FTP]
$ns_ at 165.62555372045634 "$ftp_(4) start"
#
# 8 connecting to 9 at time 89.648865521721945
#
set tcp_(5) [$ns_ create-connection TCP $node_(8) TCPSink $node_(9) 0]
$tcp_(5) set window_ 32
$tcp_(5) set packetSize_ 512
set ftp_(5) [$tcp_(5) attach-source FTP]
$ns_ at 89.648865521721945 "$ftp_(5) start"
#
# 8 connecting to 10 at time 1.337946458411378
#
set tcp_(6) [$ns_ create-connection TCP $node_(8) TCPSink $node_(10) 0]
$tcp_(6) set window_ 32
$tcp_(6) set packetSize_ 512
set ftp_(6) [$tcp_(6) attach-source FTP]
$ns_ at 1.337946458411378 "$ftp_(6) start"
#
# 9 connecting to 10 at time 124.93961080207471
#
set tcp_(7) [$ns_ create-connection TCP $node_(9) TCPSink $node_(10) 0]
$tcp_(7) set window_ 32
$tcp_(7) set packetSize_ 512
set ftp_(7) [$tcp_(7) attach-source FTP]
$ns_ at 124.93961080207471 "$ftp_(7) start"
#
# 9 connecting to 11 at time 157.06980878350782
#
set tcp_(8) [$ns_ create-connection TCP $node_(9) TCPSink $node_(11) 0]
$tcp_(8) set window_ 32
$tcp_(8) set packetSize_ 512
set ftp_(8) [$tcp_(8) attach-source FTP]
$ns_ at 157.06980878350782 "$ftp_(8) start"
#
# 10 connecting to 11 at time 26.31274839225819
#
set tcp_(9) [$ns_ create-connection TCP $node_(10) TCPSink $node_(11) 0]
$tcp_(9) set window_ 32
$tcp_(9) set packetSize_ 512
set ftp_(9) [$tcp_(9) attach-source FTP]
$ns_ at 26.31274839225819 "$ftp_(9) start"
.
.
.
.
67
#===========================================================
# AWK Script for calculating: Packet Delivery Ratio and Average End-to-End Delay
#===========================================================
#! /bin/awk -f
#*****************************************************************************
#To run this AWK script type:
#./wireless_qos.tcl <trace file>
#*****************************************************************************
BEGIN {
seqno = -1;
count = 0;
spkt = 0;
rpkt = 0;
dpkt = 0;
fpkt = 0;
ack = 0;
}
{
#packet delivery ratio
# Sent tcp packets
if($4 == "AGT" && $1 == "s" && seqno < $6) {
seqno = $6;
}
# Received tcp packets
else if (($4 == "AGT") && ($1 == "r")){
rpkt++;
}
# Dropped tcp packets
else if ($1 == "D" && $7 == "tcp" && $8 > 512){
dpkt++;
}
# Forwarded tcp packets
if($1 == "f" && $7 == "tcp"){
fpkt++;
}
68
# end-to-end delay
if($4 == "AGT" && $1 == "s") {
start_time[$6] = $2;
} else if(($7 == "tcp") && ($1 == "r")) {
end_time[$6] = $2;
} else if($1 == "D" && $7 == "tcp") {
end_time[$6] = -1;
}
}
END {
for(i=0; i<=seqno; i++) {
if(end_time[i] > 0) {
delay[i] = end_time[i] - start_time[i];
count++;
}
else
{
delay[i] = -1;
}
}
for(i=0; i<count; i++) {
if(delay[i] > 0) {
n_to_n_delay = n_to_n_delay + delay[i];
}
}
n_to_n_delay = n_to_n_delay/count;
# Output
print " Generated/Sent TCP Packets
= " seqno+1;
print " Received TCP Packets
= " rpkt;
print " Dropped TCP Packets
= " dpkt;
print " Forwarded TCP Packets
= " fpkt;
print " Packet delivery ratio
= " rpkt/(seqno+1)*100 "%";
print " Average end-to-end delay
= " n_to_n_delay * 1000 " ms";
}
69
APPENDIX B
ANALYZED SIMULATION RESULTS
Page
Results B – 1
City Results
71
Results B – 2
Highway Results
72
70
CITY RESULTS
CITY LOW DENSITY
Generated TCP Packets
Received TCP Packets
Packet Delivery Ratio
Average End-to-End Delay
AODV
AOMDV
DSR
DSDV
39272
37619
71830
40593
39188
37535
37707
40148
99.79%
99.78%
52.49%
98.90%
66.2026
79.5885
23.7363
68.3394
CITY MEDIUM DENSITY
Generated TCP Packets
Received TCP Packets
Packet Delivery Ratio
Average End-to-End Delay
AODV
AOMDV
DSR
DSDV
194280
167200
1270724
29806
191194
164420
172438
20381
98.41%
98.34%
13.57%
68.38%
130.047
218.537
79.3674
195.119
CITY HIGH DENSITY
Generated TCP Packets
Received TCP Packets
Packet Delivery Ratio
Average End-to-End Delay
AODV
AOMDV
DSR
DSDV
231273
214099
4667263
772260
227730
211798
107116
26005
98.47%
98.93%
2.30%
33.66%
106.343
97.5508
17.7993
40.795
71
HIGHWAY RESULTS
HIGHWAY LOW DENSITY
Generated TCP Packets
Received TCP Packets
Packet Delivery Ratio
Average End-to-End Delay
AODV
AOMDV
DSR
DSDV
302008
294289
982200
26941
300293
292862
285308
18960
99.43%
99.52%
29.05%
70.38%
74.4928
68.8368
20.3766
27.7637
HIGHWAY MEDIUM DENSITY
Generated TCP Packets
Received TCP Packets
Packet Delivery Ratio
Average End-to-End Delay
AODV
AOMDV
DSR
DSDV
158363
135938
1911396
92388
154084
133375
117858
11291
97.30%
98.11%
6.17%
12.22%
90.3927
70.5107
15.0604
23.2978
HIGHWAY HIGH DENSITY
Generated TCP Packets
Received TCP Packets
Packet Delivery Ratio
Average End-to-End Delay
AODV
AOMDV
DSR
DSDV
162922
147759
1557112
143823
159031
145019
100977
11737
97.61%
98.15%
6.48%
8.16%
116.572
94.2247
17.3722
11.0122
72