ijaz master thesis 19march

8/13/2019 Ijaz Master Thesis 19march

1/61

1

MEE 09: 17

MASTER THESIS REPORT

MSc IN ELECTRICAL ENGINEERING

WITH EMPHASIS ON TELECOMMUNICATION

TRASMISSION CONTROL PROTOCOL (TCP)

PERFORMANCE EVALUATION IN MANET

BLEKINGE INSTITUTE OF TECHNOLOGYMARCH 2009

Author:Muhammad [email protected]

Supervisor & Examiner:Adrian [email protected]
mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]


2/61

2

ACKNOWLEDGEMENT

In the name of greatest All mighty ALLAH who has always bless me with potentialknowledge and success.

I am thankful to my supervisor Adrain Popescu, my friends who help me during my hardtimes when I need their assistance during simulation.

I am especially thankful to my Parents, Uncles and Grandmother, who had always providedme the courage, strength, best wishes, moral and financial support during my whole career.

I also have best regards for BTH faculty including Mikeal sman and Lena Magnusson whohad been helpful throughout my maters degree.


3/61

3

Abstract

Mobile Ad hoc network routing protocols have been divided in several different categoriessuch as Reactive and Proactive Routing Protocol. The performances of these categories areevaluated in different scenario with TCP variants. We present a comprehensive TCP

performance evaluation study to understand the nature of the TCP performance in differentscenarios with variable amount of payload and number of nodes. The traffic consists of threedifferent packet sizes i.e. 512, 1000, 1500 bytes each. Three different routing protocols(AODV, DSR and TORA) are to be evaluated with three different TCP variants (Tahoe, Renoand New Reno) in three different scenarios having 3, 5 and 8 nodes. The performances

parameters on the basis of which routing protocols are to be graded are mainly throughput,congestion window and delay. Conclusions are drawn based on the simulation results and thecomparisons between them have been elaborated.


4/61

4

LIST OF OBSERVATIONS

ABR Associativity -Based Routing

ACK Acknowledgement

ALP Adaptive Link-State Protocol

AODV Ad hoc On Demand Distance Vector

BER Bit Error Rate

CLR Clear

CWND Congestion Window

DAG Directed Acyclic Graph

DREAM Distance Routing Effect Algorithm for Mobility

DSDV Destination Sequence Distance VectorDSN Distance Sequence numbers

DSR Dynamic Source Routing

FORP Flow Oriented Routing Protocol

FSR Fisheye State Routing

GLS Grid Location Service

GPS Global Positioning System

GPSR Greedy Perimeter Stateless Routing

GSR Global State Routing

IEEE Institute of Electronics and Electrical Engineers

IP Internet Protocol

LAN Local Area Network

LANMAR Landmark Ad Hoc Routing

LAR Location-Aided Routing

LMR Lightweight Mobile Routing

LRR Least Resistance RoutingMANET Mobile Ad hoc Network

MAC Medium Access Control

MDSR Multipath Dynamic Source Routing

MSS Maximum Segment Size

OLSR Optimized Link State Routing

OPNET Optimized Network Evaluation Tool

PHY Physical Layer

RABR Route-Lifetime Assessment Based Routing


5/61

5

RERRs Route Errors

RIP Routing Internet Protocol

QRY Query

RREQs Route Requests

RREP Route Reply

RTO Retransmission Time Out

RDMAR Relative Distance Micro-discovery Ad Hoc Routing

ROAM Routing On demand Acyclic Multipath

RWND Advertize Window

SSA Signal Stability-Based Adaptive

STAR Source Tree Adaptive Routing

SWND Send WindowTCP Transmission Control Protocol

TORA Temporally Ordered Routing Algorithm

TTL Time-To-Live

UDP User Datagram Protocol

WRP Wireless Routing Protocol

ZHLS Zone-Based Hierarchical Link State


6/61

6

LIST OF FIGURES

FIGURE DESCRIPTION PAGE

Figure 2.1 Slow Start and Congestion Avoidance Mechanism 15Figure 2.2 Hidden and Exposed Node Terminal Problem 18Figure 3.1 Route discovery for target node 23Figure 3.2 Maintenance for Error Route 24Figure 3.3 Route Discoveries in TORAQRY Message 26

Figure 3.4 Route Discoveries in TORAUpdate Message 26

Figure 4.1 Research Methodology 30Figure 5.1 Simulation Model 32Figure 5.2 MANET eight node scenarios 33Figure 5.3 Simulation Flow 34Figure 5.4 Throughput comparison in three nodes scenario 36

Figure 5.5 Throughput comparison in five nodes scenario 37Figure 5.6 Throughput comparison in eight nodes scenario 38Figure 5.7 Delay Comparisons in Different Scenario 39Figure 5.8 Congestion Window dynamics 40

LIST OF TABLES

TABLE DESCRIPTION PAGE

Table 5.1 Detail of TCP Variants and Routing Protocol for Three Nodes 35

Table 5.2 Detail of TCP Variants and Routing Protocol for Five 36Table 5.3 Detail of TCP Variants and Routing Protocol for Eight Nodes 37


7/61

7

LIST OF CONTENTS

CHAPTER 1 Introduction 9

1.1 Background 11

1.2 Aims/ Objectives 111.3 Research Questions 111.4 Thesis Structure 111.5 Expected Contribution 11

CHAPTER 2 TCP and Ad hoc Networks 122.1 Types of Wireless Networks 132.1.1 Infrastructure Networks 132.1.2 Ad hoc Networks 132.2 TCP working mechanism and Variants of TCP 14

2.2.1 Working Mechanism of TCP 142.2.2 Various Flavor of TCP 152.2.2.1 Tahoe 152.2.2.2 Reno 162.2.2.3 New Reno 162.3 Challenges and Limitation of TCP in Wireless Network 162.3.1 Dynamic Topology/Mobility 162.3.2 Multi Hopping 172.3.3 Differentiating Transmission and Congestion Losses 172.3.4 High Bit Errors 17

2.3.5 Channel Contention 172.3.6 Hidden and exposed terminal problem 182.3.7 Out of Order Packet 19

CHAPTER 3 Ad hoc Network Routing Protocols 203.1 Proactive routing protocols 223.2 Reactive routing protocols 223.2.1 Dynamic Source Routing (DSR) 233.2.2 Ad hoc on Demand Distance Vector (AODV) 243.2.3 Temporarily Order Routing Algorithm (TORA) 25

CHAPTER 4 Research Methodology 284.1 Research Methodology 294.2 Author Approach 294.3 Research Design 294.3.1 Problem Selection / Identification 294.3.2 Literature Review for TCP and MANET 294.3.3 Building Simulation Environment 304.3.4 Simulation Results 304.4 Simulation Tools 30


8/61

8

CHAPTER 5 Simulation and Empirical Study 315.1 Simulation Environment 325.1.1 Model 325.1.2 OPNET 325.1.3 Network Entities and Functions 32

5.1.3.1 Application Definition 335.1.3.2 Profile Configuration 335.1.3.3 Mobility Configuration 335.1.3.4 Server 335.1.3.5 Mobile Nodes 335.2 Simulation Configuration 335.3 Simulation Flow 335.4 Scenarios 34

CHAPTER 6 Discussion/ Results 426.1 Conclusions 436.2 Future Work 43

Appendix 44

Reference 60


9/61

9

Chapter 1

INTRODUCTION


10/61

10

1.1 Background

Internet evolution thrives with time and development in the diverse technologyimprovements. Communication devices and medium have significant impact on overallinternet. Numerous networks mixtures are present at the internet and even within these

networks different preposition and parametric approaches for implementations are deployedsubject organization needs. Overall internet traffic is somehow dependent upon underlyingnetworks and vice versa. To cope with multiple simultaneous data networks internet requiresand efficient algorithmic techniques to support data transmission successfully. TransmissionControl Protocol/Internet Protocol (TCP/IP) has been very effective in data transmissiondelivery and have also developed variants to possess the possibility to increase performanceand multiple packet loss recovery. An important aspect form [1] defines the data flow methodthat applications are responsible for the selection of the protocols and IP plays a central roleall TCP and user datagram protocol (UDP) for data transmission.

TCP is reliable and connection oriented protocol developed in 1981. Its based on simple

sliding window flow control, during the early stages congestion collapses occurred because oflack of congestion control mechanism. With the advent of Jacobson congestion algorithms forTCP as a remedy, TCP is updated to its new version known as TCP Tahoe. Now currentlyTCP Reno is widely used in Internet. TCP Tahoe congestion Control includes slow start,congestion avoidance and fast retransmission. While in addition to these three algorithms,TCP Reno also adds the fast recovery algorithm. TCP is design for wired network, but withthe technology emerging towards wireless medium, the need to implement TCP is of greatimportant but it faces many problems especially in an ad hoc networks. Mobile ad hocnetwork (MANET) is a famous ad hoc network can be utilized well for emergency situationand military applications. TCP has poor performance in MANTE due to dynamic topology,shared medium, high error ratio; channel connotation and multi hop architecture.

As an evident, ad hoc networks are growing dynamically and its true development fact lies inthe problems associated with the seamless internet access and connectivity. With evolution inthe hardware devices and software that offers wireless access to the internet by means of anintermediate connection through access point ad hoc tends to improve and increase gradually.Mobile ad hoc network is sub field with enormous commercial, industrial and military growthand realistic practical approach for internet access. Most important things to know about thetechnology is the ability of autonomously establishing and managing the network [2] whilethe an effective military consensus applications, data exchange based on mobility networksfor commercial purpose i.e. business organization or it could be better utilized during

emergency within and out of remote areas. This IP based dynamic wireless networks workseffectively, reliable and robust data transmissions.

Problems are associated with in the MANET performance, an evaluation and optimizationtechniques are necessary to opt and adhere for the better execution of the transmissionmedium. Pledged data delivery is TCP utmost drawback in wireless networks but possiblesolutions are available to recognize the data transmission effects. To measure the performanceof different TCP variants, simulation study has been conducted in practice. MANET utilizesTCP and UDP for data transmission and our study focus on different variants of the TCP i.e.

particularly Tahoe, Reno and New Reno explicitly using AODV, DSR and TORA protocolsin focus.


11/61

11

1.2 Aims/ Objectives

Thesis aims to understand the performance of the TCP variants in MANET and its behaviorwith respect to different protocols. It focuses following issues in the thesis:

Detail literature study for state of the arttechnology in TCP and MANET.Performance comparison in delay and throughput by increasing number of nodes.Performance comparison between the different protocols used in MANET throughcongestion window, delay and throughput.Suggest better protocol and TCP variant for MANET.

1.3 Research Questions

Q1: What are different TCP variants and its need?Q2: What are important routing protocols used in MANET?

Q3: What is the impact on TCP performance during different protocol selection/implementation in MANET?

Q4: Which factors influence the TCP performance and why?

1.4 Thesis Structure

Chapter 2defines wireless networks and an extension to ad hoc networks and its working, italso include the detail study about the TCP flavors and its limitations.

Chapter 3 is associated with the study of MANET routing protocols specific to ad hoc

networks, their types, working and algorithms.Chapter 4 depicts the overall project flow explicitly starting form problem definition andfollowing it towards solution, implementation, simulation and results hierarchy.

Chapter 5contains simulation environment description, list of figures and assumptions.

Chapter 6holds the brief discussion about the simulation results and future work.

1.5 Expected Contribution

The thesis will address the scientific society about the TCP and MANET implementations inreality. It will contribute through using multiple different numbers of nodes to describe the

behavior of the TCP performance over different variants; it should reveal the idea todeveloping commercial and non commercial business project for MANET deployments andthe results in reality should have utmost significance.


12/61

12

Chapter 2

TCP AND AD HOC

NETWORKS


13/61

13

In this chapter first to discuss the taxonomy of wireless networks such is infrastructure

networks and ad hoc networks or infrastructure-less networks. Then before to discuss the

different challenges transmission control protocol (TCP) is facing in wireless networks, it is

better to give a look to the TCP working mechanism and its different variants.

2.1. Types of Wireless Networks

One of the unique features of wireless networks is compare to wire network is that data is

transmitted from one point to another through wireless links i.e. there is no need of wired link

between the two nodes for transmission. They just need to be in the transmission range of

each other. Wireless networks or divided into two categories. Infrastructure wireless network

and infrastructure less or ad hoc wireless network.

2.1.1. Infrastructure Networks

Infrastructure network have fixed network topology. Wireless nodes connect through the

fixed point known as base station or access point. In most cases the access point or base

station or connected to the main network through wired link. The base station, or access point,

is one of the important elements in such types of networks. All of the wireless connections

must pass from the base station. Whenever a node is in the range of several base stations thenit connect to any one of them on the bases of some criteria [3].

2.1.2. Ad hoc Networks

Ad hoc networks also called infrastructure less networks are complex distributed systems

consist of wireless links between the nodes and each node also works as a router to forwards

the data on behalf of other nodes. The nodes are free to join or left the network without any

restriction. Thus the networks have no permanent infrastructure. In ad hoc networks the nodescan be stationary or mobile. Therefore one can say that ad hoc networks basically have two

forms, one is static ad hoc networks (SANET) and the other one is called mobile ad hoc

networks (MANET). From the introduction of new technologies such as IEEE 802.11[4] the

commercial implementation of ad hoc network becomes possible [5].

One of the good features of such networks is the flexibility and can be deployed very easily.

Thus it is suitable for the emergency situation. But on the other side it is also very difficult to

handle the operation of ad hoc networks. Each node is responsible to handle its operation


14/61

14

independently. Topology changes are very frequent and thus there will be need of an efficient

routing protocol, whose construction is a complex task. TCP performances are also very poor

in mobile ad hoc network. In coming sections we are discussing the TCP working mechanism

and challenges for TCP in ad hoc networks in more detail.

2.2 TCP working mechanism and Variants of TCP

Transmission Control Protocol (TCP) [6] is a Transport Layer Protocol and originally

designed for wired network in 1981. The basic responsibility of TCP is to provide reliable

transfer of data between the nodes i.e. to ensure that the data is reached the destination

correctly without any loss or damage. The data is transmitted in the form of continuous stream

of octets. The mechanism is adopted to assign a sequence number to each octet of data andreceiver respond with positive acknowledgement to ensure that the data is received correctly.

Another aspect of TCP is the tree way handshakes mechanism to establish a connection

between the end points (nodes). In case if there are two or more TCP connection between the

end points for this TCP uses the port assignment mechanism to differentiate each connection

from others.

2.2.1 Working Mechanism of TCP

The congestion control algorithm employed by TCP is window based. It uses three types of

windows called congestion window (cwnd), advertize window (rwnd) and send window

(swnd). Congestion windows shows the total amount of data the sender is allowed to output to

the network without any Acknowledgement (ACK). While on the other hand the advertise

window indicates the amount of data the receiver is ready to accept. It is equal to the available

buffer size on the receiver. Now the size of the flight window or send window is set to the

minimum of the congestion window and advertize window.

Basically the congestion control mechanism has two phases the slow start and congestion

avoidance phase. When a connection is established the initial size of congestion window is set

to One MSS (Maximum Segment Size). Receiving data from the sender the receiver ACK the

reception of data to the sender, the receiver is actually indicating the sequence number of next

expected data segment in ACK, from this the sender conclude that all the data segments

which have sequence number less than the indicated one are delivered correctly. While in case

of packet loss, out-of-order packets arrive at the receiver. Then the receiver sends a duplicate


15/61

15

ACK to the sender in response of each out-of-order packet arrival. When sender receives three

duplicates ACK, it is concluded that the packet is lost and retransmission of data packet take

place.

0

2

4

6

8

10

12

14

16

18

20

22

24

0 1 2 3 4 5 6 7 8 9 1 0 11 12 13 14 15 16

Slow Start threshold

Time out Occured

Slow Start threshold

Congestion Avoidance

Phase

C

ongestionWindow

Number of Acknowledgement

Slow Start

Figure 2.1 Slow Start and Congestion Avoidance Mechanism [10]

If sender receives non-duplicate ACK Then the congestion window is incremented. In slow

start phase exponential incrimination occurs until it reaches to the slow start threshold. While

in congestion avoidance phase the cwnd is incremented by one packet. This window growth is

limited to a maximum window size. This whole process is depicted in Fig. 2.1.

2.2.2. Various Flavor of TCP

After the introduction of first version of TCP several different flavors exist, here we are

discussing the most famous implementation of TCP called Tahoe, Reno, New Reno and

Vegas.

2.2.2.1. Tahoe

In the first version of TCP there was no congestion control mechanism. So after observing the

congestion collapses 1988 Jacobson introduced several Congestion Control algorithms [7] and

this version is called TCP-Tahoe. The congestion control algorithms introduced in this version

are:


16/61

16

a) Slow startb) Congestion Avoidancec) Fast Retransmit

2.2.2.2. Reno

The fast retransmit phase was first introduced in TCP-Tahoe followed by Slow Start. But

TCP-Reno [8] also added the algorithm of Fast Recovery, so that Fast Recovery dictates the

sender to perform congestion avoidance directly after fast retransmission rather than

immediately reducing the data flow using slow start mechanism.

2.2.2.3. New Reno

TCP Reno recovers only one lost packet during the recovery process. So TCP-New Reno [9]

is just adding the capability to TCP Reno to deals with multiple packets losses to recovery in a

single transmission window.

2.3. Challenges and Limitation of TCP in Wireless Network

As compare to wired network wireless network have some special characteristic such as

Mobility, share medium and signal fading. All these factors are responsible to create different

types of challenges that restrict the range, data rate and reliability of the wireless transmission.

These challenges are as follow.

2.3.1. Dynamic Topology/Mobility

In contrast to wired network, in wireless ad hoc networks the devices are free to move

which leads to frequent topology changes. Then two types of problem occurrence are

possible [11].

(a)Path loss(b)Network PartitionPath loss leads to the path re-computation at sender side and during this phase there will

be no transmission which ultimately causes throughput degradation. Another possibility

during this phase is that if path re-computation take more time then may be retransmission

time out (RTO) occurs. Thus there will be data retransmission, RTO is increasedexponentially and TCP enter to slow start phase. This affect will be more serious in high


17/61

17

mobility environment. In case of network partition the sender and receiver remains at

different network and all the packets will be dropped. In such situation multiple

consecutive transmission of the same segment are possible, while the receiver will be

disconnected and it is called serial time out.

2.3.2.Multi-hopingIn Ad hoc Network Environment every node is also supporting the responsibility of router to

forwards packets on behalf of other nodes. Thus longer flow, in since of hops, having longer

round trip time and higher packet dropping probability and high fluctuating end-to-end

throughput as compare to flow having small number of hops.

2.3.3.Differentiating Transmission and Congestion Losses

Since wireless network have open medium, signal passing through the medium they have to

face different blockage such as building, due to these blockages the transmission signals are

reflected, diffracted and scattered and thus causes packets loss. TCP assume that this loss is

occurred due to congestion and it activate its congestion control and result in congestion

window reduction, which leads to low throughput and low utilization of available bandwidth.

So TCP has no such mechanism to differentiate between congestion losses and transmissionlosses.

2.3.4 High Bit Errors

In wired network the Bit Error Rate (BER) is ranging from 10 -6 to 10-8, while in wireless

network this range is from 10-3 to 10-1. Since due to high BER the packet losses are very

frequent and TCP reaction to these losses reduces the congestion window. Thus leads to non-

optimal performance [10].

2.3.5 Channel ContentionAnother reason of TCP performance degradation is the channel contention due to increasing

number of nodes. It can occur between different flows passing through the same vicinity or

between different packets in the same flow. In IEEE802.11, when the number of try for

channel access exceeds the predefined limit, then cause to drop the packets and the Medium

Access Control (MAC) Protocol notifies (wrongly) the upper layer that the path is


18/61

18

unavailable. In response the upper layer starts the route recovery procedure [12] and TCP stop

its transmission and the throughput drop to zero during route recovery process.

This channel contention also leads to unfairness problem. The unfairness also occurs between

the nodes so that each node has or not the equal access to the medium as compare to other

node. If there is no equal access to the medium for each node then the unfairness occurs

between the flows passing from different nodes. This unfairness also happens among the

flows passing from the same path.

2.3.6 Hidden and exposed terminal problem

Due to the share medium and multi hoping capability the nodes facing the hidden and exposed

node problem. Fig: 2.2 representing the hidden and exposed terminal problem of IEEE

802.11 standard. The circles show the transmission range of A and B, where C is in the

transmission range of both A and B. Let A and B both want to transmit data to C, so there will

be collision at C, because A and B do not know about the transmission of each other due to

hidden node problem.

Figure 2.2: Hidden and Exposed Node Terminal Problem

Now let that there is a transmission between C and B, while at the same time A wants to

transmit data to D, but when A senses the medium, it find that the medium is busy due to C

transmission and thus A stops its transmission. Actually in this situation A transmission for D

will not going to collide with C transmission, This problem exist in IEEE802.11 standard and

known as exposed terminal problem.


19/61

19

2.3.7 Out of Order Packet

When a receiver receives out of order packets, the receiver transmits duplicate

acknowledgement, after receiving three duplicate ACK the sender retransmit the packets and

congestion control is activated. But the problem is that congestion control is activated

wrongly most of the time, because out-of-order packet occurrence take place due to different

reasons such as multipath routing protocol and rout failure and not only due to congestion.


20/61

20

Chapter 3

Ad hoc Network Routing

Protocols


21/61

21

In this chapter routing protocols are described. In general route means the way and protocol is

the set of rules through which two or more devices (computers, mobile nodes) are

communicating with each others. Routes are multi hop in ad hoc networks because the

propagation range (250 meters in an open field) of wireless radio is limited. Nodes travel

freely and randomly in the network and routes are often find connection or disconnection.

Establishing strong routes, maintaining and reconstruction in time are the main task for

routing protocols. All the above responsibilities are performed by the routing protocol, except

generating excessive control message overhead. Data packets send efficiently must be utilized

by control packets and be generated only when needed. Routing protocol efficiency in

bandwidth and energy consumption could be made by reducing the control overhead. [13]

In generally mobile ad hoc wireless networks routing protocols can be classified by their

routing strategy.

Pure distance vector algorithms (e.g., Distributed Bellman Ford, Routing Internet

Protocol (RIP), etc.) do not give a good result in mobile networks because of some

limitation. Then some new protocols were proposed to modify and enhance the

distance vector algorithm. Protocols such as Wireless Routing Protocol (WRP), Least

Resistance Routing (LRR), Destination Sequence Distance Vector (DSDV) routing

protocol, and the protocol by Lin and Liu.

The protocols which are based on link state algorithms. Theses protocols include

Global State Routing (GSR), Landmark Ad Hoc Routing (LANMAR) protocol,

Optimized Link State Routing (OLSR) protocol, Adaptive Link-State Protocol (ALP),

Fisheye State Routing (FSR) protocol, and Source Tree Adaptive Routing (STAR)

protocol.

The third one is on-demand routing protocols which are planned only for ad hocnetwork. Route to every destination of the networks on a regular basis is not

maintained by on-demand routing protocols. The source establishes routes on demand.

The source floods a route request packet to construct a route when it needed. The

destination use route selection algorithm and select the best route for which

destination receives request. Then route reply packet is sent to the source through new

best route. There is no requirements of periodic exchange of route tables and control

traffic overhead is greatly reduce by on-demand routing protocols. Several protocolsof this type have been propose d. Ad-Hoc On Demand Distance Vector (AODV)


22/61

22

routing, Dynamic Source Routing (DSR), Lightweight Mobile Routing (LMR),

Temporarily Ordered Routing Algorithm (TORA), Route-Lifetime Assessment Based

Routing (RABR), Associativity-Based Routing (ABR), Relative Distance Micro-

discovery Ad Hoc Routing (RDMAR) protocol, Signal Stability-Based Adaptive

(SSA) routing, Multipath Dynamic Source Routing (MDSR), and Routing On

demand Acyclic Multipath (ROAM) algorithm are on demand routing protocol.

The fourth category is GPS (Global Positioning System), in the early stages protocols

was using node location information while building routes have been proposed

recently. Through information node position, routing can require more cost to

exchange location information. GPS routing protocols are Greedy Perimeter Stateless

Routing (GPSR), Distance Routing Effect Algorithm for Mobility (DREAM), GridLocation Service (GLS), Location-Aided Routing (LAR), Flow Oriented Routing

Protocol (FORP), and Zone-Based Hierarchical Link State (ZHLS).

In addition we can say routing protocols are categorized as reactive, proactive and hybrid

(combination of both is called hybrid). [13]

3.1 Proactive Routing Protocols

This type of protocols has to construct and maintain fresh routing information to all the nodes.This is free of whether or not the route is needed. For this achievement control message are

transmitted periodically. Proactive routing protocols are not bandwidth well-organize. Even if

there is no data flow, the control message is broadcasted. This type of protocols has some

advantages and disadvantages. One of the main advantages is that nodes can get easily routing

information and easy to set up a session. The disadvantage is:

There is too much data kept by the nodes for route protection and restructure is slow when

there is a failure in exacting link. DSDV is an example of proactive routing protocol.

3.2 Reactive Routing Protocols

Reactive routing protocols are specially planned for ad hoc networks. Permanent routing

information is not kept by these protocols. Routes are built when the source needed. Route

request is sending across the network to achieve this. In reactive routing protocols we describe

in detail DSR (Dynamic Source Routing) protocol, AODV (Ad Hoc On-demand Distance

Vector) protocol, and TORA (Temporally Ordered Routing Algorithm) protocol. [13]


23/61

23

3.2.1 Dynamic Source Routing (DSR)

DSR was developed at Carnegie Mellon University. It is simple and efficient reactive routing

protocol which is specially designed for multi-hop ad hoc network of mobile nodes. The

nodes in the network easily join or leave the network without any information. The networkusing DSR is not requiring existing network infrastructure or administration. The node

desiring to transmit a packet define route for the packet because it is based on source routing.

DSR works for ad-hoc network of approximately 200 nodes.Each node participating in ad-

hoc network should forward packets and discard the erroneous packets (corrupted). DSR has

two mechanisms: route discovery and route maintenance. [13]

Route discovery

The source starts a route discovery when sending data packet to the destinationbut have no routing information. To set up a route, the source floods RREQs

message with a distinctive request ID. When the destination receives this

request message or a node which has destination route information then it

transmits RREP message back to the source with route information. Figure 3.1

shows route discovery of DSR. Node 2 is the initiator and node 9 is the target.

Figure 3.1: Route discovery for target node, [19]

Route Maintenance

In LAN routing the main improvement of DSR is in route maintenance and

monitoring in the attendance of mobility. DSR based on the acknowledgments of

data packets sent to adjacent nodes to monitors the validity of existing routes. This

monitoring is achieved by inactively listening for communication of the adjacent to

the next hop or sitting a bit in a packet to ask for open acknowledgment. The

RERRs packet is sent to the creative sender to raise a new route discovery stage

when a node fails to accept an acknowledgment. Nodes receive a REERs message

remove any route entry (from their route cache) which uses the out of order link.


24/61

24

When a node has problem transferring packet during that link then REER message

is propagated. So this selective transmission reduces control overhead (if no packets

pass through a link), it yields a long delay when a packet wants to go through a new

link. [13]. Node 9 cannot be reached by node 6 anymore and a REER is returned to

node 2.

Figure 3.2: Maintenance for Error Route, [19]

DSR main advantages are that it reduce routing overhead and does not need to discover routes

to all the nodes in the network. The disadvantage of DSR is low mobility and static networks.

Its performance is reduced by high mobility.

3.2.2 Ad Hoc On-demand Distance Vector (AODV)The Ad hoc On-Demand Distance Vector (AODV) is a routing protocol. AODV is designed

for ad hoc mobile networks and of both routing, that is unicast and multicast routing. AODV

establish routes between different nodes as needed by source nodes. AODV maintain these

routes as well as form trees which connect different multicast group members. The group

members compose the trees and the members are connected by the nodes. In an ad hoc

network when two nodes want to make a connection with each other, AODV enable multi-

hop routes within the nodes. Ad hoc On-demand distance vector is free loop. DSN (DistanceSequence numbers) is used by the AODV to avoid counting to infinity, and this is one of the

most important quality and feature of this algorithm. In a network the requested nodes send

the DSN with other routing information from the source to the destination. It has also the

feature to select optional route which is based on the sequence number. [14]

There are three messages which are defined by AODV. These messages are Route Errors

(RERRs), Route Request (RREQs) and Route Replies (RREPs). [15].


25/61

25

For discovering and maintaining routes in the network these three messages are used, by using

UDP packets from source to destination. A node uses its IP address as the source address in

the IP header of a message when it request for a route, and for broadcast 255.255.255.255. In

the ad hoc network the number of hops a particular routing message is determined by the TTL

(Time-To-Live). The RREQ is broadcasted by the requested node when a route is needed to

be created to the destination. When the next hop node received the message a new route is

determined, or itself by the destination [16]. Routes of the PREQ from the originator to all the

stations that receive message are cached in these stations. A RERR message is delivered or

generated when there is a failure in the link. The message has all the important information

about the node which is not reachable because of the failure. The IP address of the nodes is

also mentioned in the message as the next hop for the destination.

AODV is table based. All the information about the routes in the network is stored in this

table. The routing table has the following entries i.e. DSN, flag, next hop, IP address, State,

hop count, the list of precursors, Life time and network interface.

3.2.3 Temporally Ordered Routing Algorithm (TORA)

TORA is a reactive routing algorithm based on the conception of link reversal and used in

MANETs to improve the scalability. Highly dynamic mobile ad hoc networks can be used by

TORA. It is an adaptive routing protocol and used in multi-hop networks. It makes scaled

routes between source and destinationand Directed Acyclic Graph (DAG) is used to build in

the destination node. A DAG is accomplished for the network by handing over each node i

witha height metric i.e. hi. Directional from the link ito jmeans hi > hi. The height of the

node in TORA is defined as quintuple, which contains the unique ID of the node, a

propagation ordering parameter, the logical time of a link failure, a reflection indicator bit and

a unique ID of the node which defines the new reference level. The first two values identify

an offset with respect to the reference level and the last three values jointly represent the

reference level. A data packet goes from up flow to down flow according the height

difference between nodes. TORA has the capacity that many nodes can send packets to a

given destination is provided by DAG. It also guarantees that all routes are loop free. There

are three basic functions in TORA, [17 and 20].

Route Creation

Route maintenance


26/61

26

Route erasure

A route creation operation start all the nodes height is set to NULL i.e. (-,-,-,-,i) and

destination is set to ZERO i.e. (0,0,0,0,dest). The nodes height is flexible according to the

topology. A QRY message containing the destinations ID is broadcasted by the source. A

non-NULL height node responds with UPD packet containing its own height. With the

reception of UPD packet, a node increment its height with one than that of the UPD generator.

The higher height of a node is measured as up flow and lower height of a node is measured as

down flow. Then the DAG from source to destination is constructed. Route maintenance

operation is a main part of TORA. It has an exclusive quality that control message are

contained into a small set of nods near the happening of topology changes. A node generates a

new reference level and broadcasts the reference to its neighbours when it loses its last down

flow link. Then the links reproduce the topology change and adjust to the new reference level.

The route erasure operation in TORA floods CLR packets through the network and remove

unacceptable routes, [18].

Figure 3.3 Route Discoveries in TORAQRY Message

In this figure node A shows and node B shows destination. Node D and node G is one hop

neighbours away from the destination. So they will propagate updates as shown in next figure.

Figure 3.4 Route Discoveries in TORAUpdate Message


27/61

27

The algorithm [21], contains many flows with most severe in which algorithm extremely

depend upon the numbers of nodes during initial stage, which causes dependence on the

traffic requirement. It is measured as change in rate with regard to total number of nodes

participating in network traffic. In case of gradually positive increase in the network volume,

TORA is not considered as better solution.


28/61

28

Chapter 4

Research Methodology


29/61

29

4.1 Research Methodology

It illustrates the overall thesis work to be carried out in as in the form of research activity. Itgeneralizes the understanding about author concept in developing multiple steps or stages forwhole work.

Creswell [22] enlighten two major form of research i.e. quantitative and qualitative.Quantitative research is carried out by means of inquiry strategies including certainexperiments and simulations which produce some form of statistical data for analysis ordecisions. Qualitative research is based on the researcher knowledge through participatoryand /or constructive perspectives.

4.2 Author approach

Thesis is carried out through multiple stages starting form literature review about the state ofart technology specific to TCP and MANET and then performing the simulation in OPNETv14.5 tool. So both molds of approaches are necessary to complete the thesis work and have

been administered.

4.3 Research Design

Author believes that the most of the time is spent in identifying and selecting the problem andconduction literature review. It is also vital to determine that simulation tool also requiressome time during configuration and simulation, as results are often indispensable. Author

divided the overall research into four stages.

4.3.1 Problem Selection / Identification

In problem selection stage author adhere to only MANET and TCP as focus area of researchinterest. Problem has been selected in consideration to data performance issues in MANETdue to wireless nature of the network.

4.3.2 Literature Review for TCP and MANETIt is necessary to conduct literature study for understand basic and expertise concept regardingTCP variants, Ad hoc networks, MANET protocols and related work. OPNET simulator studyis also important to develop the scenario, run simulation and collect result.


30/61

30

Figure 4.1 Research Methodology

4.3.3 Building Simulation EnvironmentThis stage is critical to understand as it exhibit deep understanding of how and why

these specific nodes and protocols are chosen. Also it conceptually accelerates the

whole simulation work efficiently and smoothly.

4.3.4 Simulation Results

The last stage is to grasp the relevant results of interest for research study. Simulationgraph results consist of TCP performance for Tahoe, Reno and New Reno relative toeach selectable MANET protocol i.e. AODV, DSR and TORA.

4.4 Simulation Tool

OPNET is selected as it offer easy graphical interface, possibility to develop and run

this simulation environment, validity of the simulation results and the tools is highly

reliable, robust and efficient. Another factor of choosing OPNET is the authorsfamiliarity and trust over the tool.

ProblemIdentificationand Selection

LiteratureReview ofTCP andMANET

ProblemSimulation

Implementation

SimulationResults

State of art

Real

Problem

Technology

Evolution

Tool

Configuration

Internet

State of

art study

Author

InterestTool

Selection

Building

Scenario

Protocol

Selection

Graph

Results

Performance

Analysis

Assumption and

Conclusions

TCP Variant

Selection

Books


31/61

31

Chapter 5

Simulation and Empirical

Study


32/61

32

5.1 Simulation Environment

This section is the major portion of the thesis, it is important to setup simulation environmentto observer TCP behavior over MANET. Quantitative analysis is conducted to with the helpof OPNET tool.

5.1.1 Model

Actual problems are simulated through computers by modeling design for actual systems andtheoretical described systems, simulations are executed through computers for the collectionof relative informational results. Learning by doing [23], is a primary concept about systemsin study which requires modeling and operating them.

Figure 5.1: Simulation Model, [23]

5.1.2 OPNETOptimized Network Evaluation Tool (OPNET version 14.5) modeler is used infor simulation network design and execution. OPNET technologies Inc [24],

offers drag and drop communication devices, maximum technologies, multipleprotocols and virtual environment for simulation.

5.1.3 Network Entities and FunctionsSimulation network model contain important entities as depicted in the figure

below:

Model

Design

Model

Execution

Execution

Analysis


33/61

33

Figure 5.2 MANET eight node scenarios

5.1.3.1 Application DefinitionIn our thesis, this module has been configured to achieve the desired results.Basically we administer application name as FTP_Application and madenecessary configurations, [Appendix].

5.1.3.2 Profile ConfigurationIn this configuration I set the profile name with start time offset and starts time,[Appendix].

5.1.3.3 Mobility ConfigurationThis module is required to produce the mobility options for the mobile nodes; Ihave set default random waypoint and adjusted some parameters, [Appendix].

5.1.3.4 ServerThis module controls the traffic for MANET, protocol selection and TCP flavorhas been selected through this module, [Appendix].

5.1.3.5 Mobile NodesI have set the mobility options; dynamically allocate the IP addresses to allmobile nodes including server as well as different routing protocols,[Appendix].

5.2 Simulation ConfigurationA detail configure is described in [Appendix].

5.3 Simulation FlowStep by step approach is used to keep the simulation flow clear and easy.Following model for simulation flow is drawn.


34/61

34

Figure 5.3: Simulation Flow

MANET network is designed for the simulating the model and configuration statisticsare applied on the whole model separately and the with regard to each individualscenario.

Simulation is executed multiple times to obtain accurate results describe in eachscenario.

Results from each scenario are gathered as graphs and stored in .jpg files to further use

in the thesis report. Some graphs were plotted one over another and with differentprotocol selections and TCP flavors to observe the require statistics.

5.4 Scenarios

The results obtained during the simulation are depicted in this area. In simulation there are

three type of different scenarios based on the number of nodes. Mainly we are concern with

throughput of each variant of TCP and to find out which variants out perform with what type

of routing protocol. To keep clear analysis let us consider each scenario separately, starting

from the three node scenario.

Three nodes Scenario

The table 5.1 shows the detail of routing protocol and TCP variants in three nodes scenario.

Without these three nodes there is a fixed node which working as source. A connection has

been established from each node to transfer a file of the same size over each connection. Then

to find out that which variants of TCP perform well and what was the underlying protocol.

MANET

Network

design

Running/

Executing

Simulation

Collecting

Results

Apply

Configuration

and Statistics


35/61

35

Table5.1 Detail of TCP Variants and Routing Protocol for Three Nodes

THROUGHPUT/ DELAY/ CONGESTION WINDOW MEASUREMENT

Stages TCP Variants Protocol Number of nodes Speed of Nodes m/sA New Reno AODV 3 10A New Reno DSR 3 10A New Reno TORA 3 10B Reno AODV 3 10B Reno DSR 3 10B Reno TORA 3 10C Tahoe AODV 3 10

C Tahoe DSR 3 10C Tahoe TORA 3 10

Throughput

From figure 5.4, by looking at the throughput performance between AODV and DSR at

approximately 25 seconds, AODV has better throughput performance over DSR. Comparing

AODV with TORA at the approximation of 95 sec, AODV provides better throughput

performance, in all the three graphs.

Total time consumed by each protocol for send the data through New Reno, Reno and Tahoe

is relatively smaller for AODV followed by DSR and than TORA. It means that TORA has

worst throughput performance in each of the three cases for MANET 3 nodes.The cause

behind the TORA less performance is considered as the reality of TORA working for route

recreation, maintenance and erasure, if dropping of the route occurs, which requires more

time and have bad impact in the data performance.


36/61

36

A B C

Figure 5.4: Throughput comparison in three nodes scenario

Five Nodes Scenario.

For five nodes scenario the detail of different protocols as shown in Table 5.2. There are five

nodes working as clients to establish connection with a fixed node working as source, and to

transfer a file of the same size over each connection.

Table5.2 Detail of TCP Variants and Routing Protocol for Five Nodes


Stages TCP Variants Protocol Number of nodes Speed of Nodes m/sA New Reno AODV 5 10A New Reno DSR 5 10A New Reno TORA 5 10B Reno AODV 5 10B Reno DSR 5 10B Reno TORA 5 10C Tahoe AODV 5 10C Tahoe DSR 5 10C Tahoe TORA 5 10

Throughput

Form figure 5.5, comparing all the three graphs with each other I have more or less same

assumption as for node three scenario, but when comparing figure 5.5 with figure 5.4, we

observe that as the number of the nodes are increased approximately 75% (3 to 5), throughput

has been decreased for every TCP variant and each desired protocols.


37/61

37

A B C

Figure 5.5:Throughput comparison in five nodes scenarioEight Nodes Scenario

Different protocols evaluated in this scenario are listed in Table 5.3. Like other scenario

depending on the number of clients eight simultaneous connections has been established with

a fixed source to download a file of the same size over each connection.

Table5.3 Detail of TCP Variants and Routing Protocol for Eight Nodes


Stages TCP Variants Protocol Number of nodes Speed of Nodes m/sA New Reno AODV 8 10A New Reno DSR 8 10A New Reno TORA 8 10B Reno AODV 8 10B Reno DSR 8 10B Reno TORA 8 10C Tahoe AODV 8 10C Tahoe DSR 8 10

C Tahoe TORA 8 10

Throughput

From figure 5.6, we conclude that the graph behavior remains same and the assumption made

for figure 5.4 are also acknowledgeable. Further we observe that the AODV have the higher

throughput in the start in all the three figures (5.4, 5.5, and 5.6) as compared to the DSR and

TORA. A small change has been observed in the number of data packets when nodes are

increased to 8 in figure 5.6. The highest number of data packets are reduces from

approximately 750 to 575, which means that if more and more nodes are added in MANET,


38/61

38

throughput will reduce. Similarly TORA takes more time when the numbers of nodes are

increased to 8, as compared to 3 nodes and 5 nodes.

A B C

Table 5.6 Throughput comparison in eight nodes scenario

Assumptions

It has been concluded that throughput performance of DSR and TORA are minutely affected

with increase in the number of nodes and due to mobility. In general, MANET could have

dynamic number of nodes connectivity in mobility, so its important to realize that when the

number of nodes is higher, DSR and TORA would be avoided. AODV has better throughputperformance shown in all three figures as compared to DSR and TORA and is the best

solution for MANET. However simulation results for AODV with respect to New Reno, Reno

and Tahoe depict that throughput is the same in all the cases, so our proposed solution in this

case will be New Reno as it offers multiple packet loss recovery.

Delay

Figure 5.7 hold the simulation results of each and every TCP variant with respect different

routing protocols all together.

With three node scenario, TORA has the highest delay as compared to DSR and AODV

which validate our simulation results. With 5 node and 8 node scenario, TORA and DSR have

approximately same delay, which is not considered as a good perception in wireless networks,

as we encounter numerous losses and delays due to SNR, reflection, diffraction and inter

symbol interference, so the delay measure is considered highly sensitive. AODV on the other

hand have similar delay for 9 results, it maybe because the high resolution graph will show us

small difference in each graph subject to 3, 5 and 8 nodes respectively.


39/61

39

1 2 3

4 5 6

7 8 9

Figure 5.7 Delay Comparisons in Different Scenario


40/61

40

Congestion Window

Figure 5.8 hold 6 graphs define earlier for all the simulated scenarios; there are much

dissimilarity within each set of variants and routing protocols.

Considering 3 node scenario, congestion window of TORA reaches to the maximum of160,000 bytes and will remain there which depicts that the exponential increase in the

congestion window leads to the delay and lower performance, as there are not enough space

in the buffer to manage incoming data and the window remains constant.

1 2 3

4 5 6

Figure 2.8 Congestion Window dynamics

Secondly, DSR have different results in all the six cases, variations for congestion window are

found approximately between 65,000 bytes to 80,000 except for case 3, where it had reachedaround 11,000 bytes. So its likely to know that DSR congestion window is quite uncertain


41/61

41

and its use in the MANET network will have severe results. Congestion window will

ultimately influence throughput and delay parameters important for the performance

evaluation.

Third protocol, AODV have similar results during individual 3 nodes and 5 nodes (New

Reno, Reno and Tahoe). It explicitly defines that AODV has the formal behavior of TCP aswithin the normal networks. The congestion window is dynamically increasing and

decreasing after receiving reply for the receiver. It is however not much clear whether it

doubles the size of the congestion window after each successful data packet transmission or

not, but the behavior of the graph exactly resembles with the actual TCP congestion window

graph with increase and decrease. We suggest AODV as the best possible solution and

recommendation for MANET.


42/61

42

Chapter 6

Discussions/ Results


43/61

43

6.1 Conclusion

The overall research has been conducted in the well organized way as defined in chapter 4.Research questions 1 and 2 defined in chapter 1 are answered through state of art study andliterature review. Research question 3 and four are based on the simulation results and

explanations.

We adhere to the simulation results as an evident that TCP variants have minor affect on theoverall results except in few cases defined earlier, but the major dependence lies on MANETrouting protocols. Simulation observation based on AODV, DSR and TORA clearly describeus about the performance evaluation through measuring throughput, delay and congestionwindow that the best routing protocol for MANET is AODV.

6.2 Future Work

As I have selected these numerous MANET routing protocols of interest by simulation in anOPNET tool, another possibility of doing the same work can be done through another toollike NS-2. Also, selection of other routing protocols can be use for the performanceevaluation or other parameters of performance could be considered for simulation.


44/61

44

Appendix: Simulation Steps

Open the OPNET simulator.

Click on File and select New then click ok

Enter name to project and scenario then click ok.

Select Create empty scenario and click Next


45/61

45

Select Campus Network then Click Next

X span: 1000Y span: 1000

Units : Meters then Click Next

Select the Technology (MANET) then Click Next


46/61

46

Now Click Finish

From object Palette Tree select one by one

Application Config, Profile Config, Mobility Config, Wlan_Server (Fixd Node) and

Wlan_wkstn (Mobile node) with the requirements of the user.


47/61

47

Drag all these object one by one in the working area.

Right Click on Application Config, profile Config, mobility Config and Server to set

their name.


48/61

48

Click Ok

1 Application Configuration

Right Click on Application Config then Edit Attributes and configure the Application.


49/61

49


50/61

50

At last Click Ok

2 Profile Configurations

Right Click on Profile Config then Edit Attributes and configure the Profile.


51/61

51

Select all Nodes and Server Click on Protocol menuIPAddressingAuto-

Assign IPv4 Address.

Select all Nodes and ServerRight Click on ServerEdit AttributesSelect AD-

HOC Routing Protocol .e.g. AODV.

In TCP Select the Parameters e.g. Tahoe.

Tick Apply to Selected ObjectsOk


52/61

52

3 Server Configurations

Right Click on ServerEdit AttributesApplicationset Application Supported

Services and Click Ok.


53/61

53

4 Mobile Node Configurations

Select all Nodes Right Click on oneEdit Attributes. Click on Applications

Application: Destination Preferences and set some parameters.

Click on Application: Supported Profiles and set some valuesTick Apply to

Selected Objects. Click Ok.


54/61

54


55/61

55

5Click Topology menu Select Random Mobility set Mobility ProfileClick

ContinueClick Ok.

6 Mobility Configurations

Right Click on Mobility ConfigEdit AttributesClick Default Random Waypoint

Click Random Waypoint Parameters. Click OK.


56/61

56

7 Save the Project.

8 Click DES Menus

Click DES or Right Click on work area and select Individual Statistics.


57/61

57

Expand Node Statistics

a) Expand TCP Connection and Tick the Congestion Window Size (byte).Right Click on it then choose Change Collection Mode then Tick Advance and

Select All Values then Click Ok.

b) Expand Wireless Lan Tick Delay (sec) and Throughput (bit/sec) then Right Clickon both one by one and select Change Collection Mode then Tick Advance and

Select All Values then Click Ok.

9 Run the Project

Click on DES menu Select Configure/Run Discrete Event Simulation .ctrl+R.

Here you can set Time in Weeks, Days, Hours, Minuets, and Seconds. Then ClickRun.


58/61

58

When Simulation is Completed then Click Close.


59/61

59

10 Show the Result

Click on DES menu or Right Click on Work area then Select Results then View result


60/61

60

References:

[1]http://www.ietf.org/rfc/rfc1180.txt

[2]http://www.ietf.org/rfc/rfc2501.txt

[3] K. Leung and Vivtor O.K. Li, Transmission Control Protocol (TCP) in wirelessNetworks: issues, approaches and challenges, IEEE Communications Survey, Vol. 8 No.

4, pp. 64-79, 4th October 2006

[4] IEEE Standard 802.11, Wireless LAN Medium Access Control (MAC) and PhysicalLayer (PHY) Specifications, June 2007

[5] A. A. Hanbali, E. Altman, and P. Nain, A Survey of TCP over Ad Hoc Networks, IEEECommunication. Surveys and Tutorials, Vol. 7, No. 3, 3rd Quarter 2005,

[6] J. Postal,Transmission Control Protocol, Request for comments, RFC 793, September1981.

[7] V. Jacobson, Congestion avoidance and control, in: Proceedings of ACM SIGCOMM,pp. 314-29, 1988.

[8] Allman, V. Paxson, and W. Stevens, TCP Congestion Control, Request for Comments,RFC 2581, Network Working Group, Internet Engineering Task Force, April. 1999.

[9] Jany Hoe: S. Floyd and T. Henderson, The NewReno Modification to TCP's FastRecovery Algorithm, RFC 2582, Network Working Group, Internet Engineering Task

Force, April. 1999.

[10] D. Kliazovich, F. Granelli, C3TCP: Cross-layer congestion control in ad hoc wirelessnetworks, ELSEVIER Ad HocNetworks 4, pp. 687-708, 2006.

[11] J. Liu and S. Singh, ATCP: TCP for Mobile ad hoc Networks, IEEE Journal inSelected Areas in Communication (JSAC), Vol 19, No.7, pp. 1300-15, July 2001,.

[12] R. Cheng, H. Lin, A Cross-layer Design for TCP End-to-End PerformanceImprovement in multi-hop wireless networks, ELSEVIER, Computer communicationVol. 31, issue 14, pp. 3145-3152, Sep. 2008

[13] Routing and Multicasting Strategies in Wireless Mobile Ad hoc Networks by SungJuLee University of California, Los Angeles 2000.

[14] A. Zahary, A. Ayesh, Analytical study to detect threshold number of efficient routes inmultipath AODV extensions,proceedings of International Conference of Computer

Engineering & Systems, ICCES, 2007, pp. 95100[15] R. Misra and C.R. Mandal, Performance comparison of AODV/DSR on-demand

routing protocols for ad hoc networks in constrained situationICPWC International

Conference, IEEE, 2005, pp. 8689

[16] Lijuan Cao, K. Sharif, Yu Wang, T. Dahlberg, Adaptive Multiple Metrics RoutingProtocols for Heterogeneous Multi-Hop Wireless Networks,

[17] Changling Liu, Jrg Kaiser, A Survey of Mobile Ad Hoc network Routing Protocols*[email protected]

[18] Jin Mook Kim, In Sung Han, Jin Baek Kwon, Hwang Bin Ryou, A Novel Approach toSearch a Node in MANET,Information Science and Security, ICISS, 2008, pp. 4448
http://www.ietf.org/rfc/rfc1180.txthttp://www.ietf.org/rfc/rfc1180.txthttp://www.ietf.org/rfc/rfc1180.txthttp://www.ietf.org/rfc/rfc2501.txthttp://www.ietf.org/rfc/rfc2501.txthttp://www.ietf.org/rfc/rfc2501.txthttp://www.ietf.org/rfc/rfc2501.txthttp://www.ietf.org/rfc/rfc1180.txt


61/61

[19] R. Thorulp, Mobile Ad Hoc Networks and Routing Protocols,Implementing andEvaluating the DYMO Routing Protocol, Masters Thesis at the University of AARHUS,pp. 7- 20, 2007

[20] Jin Mook Kim, In Sung Han, Jin Baek Kwon, Hwang Bin Ryou, A Novel Approach to

Search a Node in MANET,Information Science and Security, ICISS, 2008, pp. 4448

[21] S. R. Chaudhry, A. Al-Khwildi, Y. Casey, H. Aldelou, A Performance Comparison ofMulti On- Demand Routing in Wireless Ad Hoc Networks, Wireless Network andCommunications Group (WNCG), School of Engineering & Design, Brunel University,

West London

[22] J. W. Creswell, Research Design: Qualitative, Quantitative and Mixed MethodsApproach, 2nd Ed, Sage Publications Inc, California, July 2002.

[23] http://www.cis.ufl.edu/~fishwick/introsim/node1.html

[24]http://www.opnet.com/
http://www.opnet.com/http://www.opnet.com/http://www.opnet.com/http://www.opnet.com/

ijaz master thesis 19march

Documents