Performance Evaluation of DSDV, OLSR and DYMO using and p MAC-Protocols

Transcription

1 Performance Evaluation of DSDV, OLSR and DYMO using and 82.11p MAC-Protocols Final Year Project Report Presented By Saad Wasiq CIIT/FA8-BET-96/ISB Muhammad Waqar Arshad CIIT/FA8-BET-87/ISB In Partial Fulfillment of the Requirement for the Degree of Bachelor of Science in Electrical (Telecommunication) Engineering DEPARTMENT OF ELECTRICAL ENGINEERING COMSATS INSTITUTE OF INFORMATION TECHNOLOGY, ISLAMABAD JUNE 212

2 Performance Evaluation of DSDV, OLSR and DYMO using and 82.11p MAC-Protocols Final Year Project Report Presented by Saad Wasiq CIIT/FA8-BET-96/ISB Muhammad Waqar Arshad CIIT/FA8-BET-87/ISB In Partial Fulfillment of the Requirement for the Degree of Bachelor of Science in Electrical (Telecommunication) Engineering DEPARTMENT OF ELECTRICAL ENGINEERING COMSATS INSTITUTE OF INFORMATION TECHNOLOGY, ISLAMABAD JUNE 212

3 Declaration We, hereby declare that this project neither as a whole nor as a part there of has been copied out from any source. It is further declared that we have developed this project and the accompanied report entirely on the basis of our personal efforts made under the sincere guidance of our supervisor. No portion of the work presented in this report has been submitted in the support of any other degree or qualification of this or any other University or Institute of learning, if found we shall stand responsible. Signature: Saad Wasiq Khan Signature: Muhammad Waqar Arshad COMSATS INSTITUTE OF INFORMATION TECHNOLOGY, ISLAMABAD June 212

4 Performance Evaluation of DSDV, OLSR and DYMO using and 82.11p MAC-Protocols An Undergraduate Final Year Project Report submitted to the Department of ELECTRICAL ENGINEERING As a Partial Fulfillment for the award of Degree Bachelor of Science in Electrical (Telecommunication) Engineering by Name Saad Wasiq Khan Muhammad Waqar Arshad Registration Number CIIT/FA8-BET-96/ISB CIIT/FA8-BET-87/ISB Supervised by Dr. Nadeem Javaid Assistant Professor, Department Of Electrical Engineering CIIT Islamabad COMSATS INSTITUTE OF INFORMATION TECHNOLOGY, ISLAMABAD June 212

5 Final Approval Performance Evaluation of DSDV, OLSR and DYMO using and 82.11p MAC-Protocols Submitted for the Degree of Bachelor of Science in Electrical (Telecommunication Engineering Name Saad Wasiq Khan by Registration Number CIIT/FA8-BET-96/ISB Muhammad Waqar Arshad CIIT/FA8-BET-87/ISB has been approved for COMSATS INSTITUTE OF INFORMATION TECHNOLOGY, ISLAMABAD Supervisor Dr. Nadeem Javaid Assistant Professor Internal Examiner-1 Dr. Safdar Assistant Professor Internal Examiner-2 Mr. Sharjeel Riaz Lecturer External Examiner M. Tahir Naveed Baig Engineer WiMAX Diagnostics and Research Centre Head Department of Electrical Engineering

6 Table of Contents vii 1 Introduction 1 2 Performance Evaluation of DSDV, OLSR and DYMO using and 82.11p MAC-Protocols 2.1 Related Work and Motivation Simulations and Discussions Throughput E2ED NRL Modeling Network Connectivity for DSDV, OLSR, DYMO and AODV above and 82.11p 3.1 Related Work and Motivation Network Connectivity In VANETs Link Duration Simulations and Discussions Modeling Network Connectivity for FSR, DYMO, DSDV and AODV above and 82.11p 4.1 Related Work and Motivation Modeled Mathematical Framework Distribution of Node Population Size Analysis of network Connectivity Indirect Communication

7 4.2.4 Modeling Of Communication Time Simulations and Discussions Modeling Probability of Path Loss for AODV, DSR, DSDV, and DYMO above and 82.11p 5.1 Related Work and Motivation Mobility Model for VANETs Node/ Vehicle location Simulations and Discussions Conclusion 83 Bibliography 87

8 [Student Name] A senior thesis submitted to the faculty of Brigham Young University in partial fulfillment of the requirements for the degree of Bachelor of Science Dr. Nadeem Javaid, Advisor May 212 Copyright 212 [Student Name] All Rights Reserved

9

10 ABSTRACT In this thesis, we simulate the following mention routing protocols Destination Sequenced Distance Vector (DSDV), Optimized Link state Routing (OLSR), DYnamic MANET On Demand (DYMO), Dynamic Source Routing (DSR), Ad-hoc On-demand Distance Vector (AODV) and Fisheye State Routing (FSR) in NS-2 to evaluate and compare their performance using two Maclayer protocols and 82.11p. A novel approach of this work is modifications in existing parameters to achieve high efficiency. Comprehensive stimulation work is done for each routing protocol and by modifying their routing strategies, the performance metrics Throughput, End to End Delay (E2ED) and Normalized Routing Load (NRL) was analyzed in the scenarios of variable mobility and scalability. After extensive simulations, we observe that AODV and DSDV out performs with 82.11p while DYMO gives best performance with A framework to accurately estimate network characteristics that depict the dynamics of network connectivity in VANETs that is link duration. Also, a framework is presented for node distribution with respect to density, network connectivity and communication time. A path loss model along with framework for probability distribution function for VANETs is presented as well. Keywords: MANETs, VANETs, AODV, DSDV,DYMO, FSR, OLSR, Routing, Throughput, E2ED, NRL

11

12 ACKNOWLEDGMENTS By the Grace of ALLAH (S.W.T) we are presenting our thesis report for final year project. We would not have completed it without the support of our family who were always there for us whenever we need them, the encouragement they give to keep us going and their love to empower us that never fails all the time. Their support and encouragement always kept us motivated. We would also like to acknowledge Dr. Nadeem Javaid whose support was with us all the time. He supported and helped us in every possible way. He is a great Inspiration for us. To Madam Aysha Bibi who helped us doing research work with her experience and knowledge. Special thanks also to all our final year project team, involvement of the helpful literature and invaluable assistance. Not to forget our friends who were a real support and source of motivation for us.. Especial thanks to the COMSATS Lab members who keep the lab open late at night.

13

14 Contents Table of Contents vii 1 Introduction 1 2 Performance Evaluation of DSDV, OLSR and DYMO using and 82.11p MAC-Protocols Related Work and Motivation Simulations and Discussions Throughput E2ED NRL Modeling Network Connectivity for DSDV, OLSR, DYMO and AODV above and 82.11p Related Work and Motivation Network Connectivity In VANETs Link Duration SIMULATIONS AND DISCUSSIONS Modeling Network Connectivity for FSR, DYMO, DSDV and AODV above and 82.11p Related Work and Motivation Modeled Mathematical Framework Distribution of Node Population Size ANALYSIS OF NETWORK CONNECTIVITY Indirect Communication Modeling Of Communication Time SIMULATIONS AND DISCUSSIONS Modeling Probability of Path Loss for AODV, DSR, DSDV, and DYMO above and 82.11p Related Work and Motivation Mobility Model for VANETs Node/ Vehicle location SIMULATIONS AND DISCUSSIONS vii

15 viii CONTENTS 6 Conclusion 83 Bibliography 87

16 Chapter 1 Introduction Ad-HOC is a Network consisting of Mobile nodes. Each node Participates in forwarding data (acts as a router). Ad-HOC Networks can be further classified into four categories Mobile ad-hoc Networks (MANET), Vehicular ad-hoc Networks (VANET), Wireless Mesh Networks (WMN) and Wireless Sensor Networks (WSN) [1]. Mobile Ad-hoc Networks (MANETs) are self-configuring networks of mobile nodes connected by wireless links. Mobile Ad-hoc Network (MANETs) is collection of independent mobile users taken as mobile nodes that communicate through wireless links.in MANETs each node is free to move in any direction with capability of organize their selves into a network by changing there links to other nodes in frequently manner. There is no fixed infrastructure is present, like a base station; therefore, it can provide attractive environment for communication quickly and instinctively [2]. MANETs consist of mobile/semi mobile nodes with no existing pre-established infrastructure. They connect themselves in a decentralized, self-organizing manner and also establish multi hop routes. The nodes in the network topology are mobile so they change their path rapidly and unpredictably over time. There are various applications for MANETs, constituting from little, static network topologies that are confine by power sources, to large-scale, highly dynamic network topologies. The creation of network protocols for these network topologies is a com- 1

17 2 Chapter 1 Introduction plex issue. MANETs use streamlined distributed algorithms to determine network organization, link scheduling, and routing. If the mobile nodes are vehicles then this type of network is called VANET(vehicular ad-hocnetwork) [3]. Vehicular Ad-hoc Networks (VANETs) are distributed, Self-assembling communication networks that are made up of by multiple autonomous moving vehicles, and peculiarized by very high node mobility. The major purpose of VANETs is to provide protection and ease to the travelers. Vehicles are furnished with VANETs device which can become a node in the Ad-hoc network and can get and pass on messages through the wireless network among the nodes [4]. It creates a network consisting of moving cars and each car behaves like a node as well as router. It allows car within 1-3m (sometimes even higher) to connect and participate in communication which in turn creates a wide mobile network. One important property that distinguishes MANETs from VANETs is that nodes move with higher avg. speed and number of nodes is assumed to be very large. Vehicular networks consist of vehicles and Road Side Units (RSU)equipped with radios. Plummeting cost of electronic components and permanent willingness of manufacturers to increase road safety and to differentiate themselves from their competitors vehicles are becoming "Computer on Wheels" rather than "Computer N/W on Wheels" [5]. Routing protocols are responsible for efficient behavior in MANETs and VANETs. There is no single unique protocol that is convenient for all networks impeccably. The protocols have to be commensurate to networks unique characteristics, such as density, mobility of the nodes. Routing protocols are subdivided into table driven and on-demand based on route calculation. In table driven which also known as proactive protocols are based on periodic exchange of control messages and maintaining routing tables. These protocols maintain complete information about the network topology locally [6]. These kind of protocols usually use link-state routing algorithms for flooding of link information. Link-state algorithms maintain a full or partial copy of the network topology. On the other hand, reactive protocols try to discover a route only on-demand, when it is

18 3 necessary. These protocols usually take more time to and a route compared to a proactive protocols [7]. The reactive routing protocols create and maintain routes only if these are needed, on demand. They usually use distance-vector routing algorithms that keep only information about next hops to adjacent neighbors. Thus, link-state routing algorithms are more reliable, less bandwidthintensive, but also more complex and compute- and memory-intensive. On other hand, on-demand routing protocols have a fundamental requirement for connectivity is to discover routes to a node via flooding of request messages. Our stimulation work is based upon comparison of protocols in MANETs and VANETs named as DSDV [8], FSR [9], OLSR [1], AODV [11], DSR [12] and DYMO [13]. Destination Sequenced Distance Vector (DSDV) is proactive protocol. Its path calculation depends Distributive Bellman Ford (DBF) Algorithm. DBF Algorithm, Host node broadcast routing information which includes new sequence number, destination address and no. of hops. Its route maintenance are on its Trigger updates. Its packet forwarding done by hop by hop routing. Dynamic MANET On Demand (DYMO) is reactive protocol. Its route calculation done by Ring search Algorithm. RREQ is broadcasted by originating node towards targeted node, it send back RREP for rout confirmation. RERR is present for route maintenance. Its follows source routing for forwarding its packets. Optimized Link State Routing (OLSR) is proactive protocol. Its path calculation depends on Dijikstra s Algorithm. Its route discovery is done by Shortest path finding Algorithm. Maintains the information about next two hops towards destination (periodic updates of network topology). HELLO and TC messages are present for route maintenance. Packet forwarding is done by Hop by Hop routing. Multipoint Relays (MPRs) is special feature of this protocol. Ad-Hoc On Demand Distance Vector Routing Protocol (AODV) is the reactive protocol. Its path calculation depend upon Distributive Bellman Ford (DBF) Algorithm. Route Discovery is done by broadcast RREQ and Unicast RREPs. RREPs, HELLO, Less Bandwidth usage due to distance vector information.

19 4 Chapter 1 Introduction Fisheye scope Routing Protocol (FSR) is the proactive protocol. Its route discovery depends upon the Ring Search Algorithm. Its route maintenance is done by RERR. Source Routing is done for its packet forwarding.

20 Chapter 2 Performance Evaluation of DSDV, OLSR and DYMO using and 82.11p MAC-Protocols 2.1 Related Work and Motivation Several papers have been published regarding the comparison of routing protocols in different simulation scenarios e.g. varying mobility, node density and mobility models using any of MAC layer protocol such as MANETs or 82.11p VANENTs. A few studies regarding evaluation of performance protocols are discussed here. The study of DYMO is been done in comparison with other protocols [14]. The performance metrics such as jitter, throughput and delay is their work. Another study [15] involved the consistently varying network topology and compared the Dynamic Source Routing (DSR) Protocol to Ad Hoc On Demand Routing (AODV) Protocol. The comparison is carried out for MANETs in different scenarios and performance metrics to propose the best scenario for each routing protocol to maximize its efficiency. Different routing protocols 5

21 6 Chapter 2 Performance Evaluation of DSDV, OLSR and DYMO using and 82.11p MAC-Protocols are compared by evaluating their performance with MAC CSMA/CA by varying CBR data traffic. The comparison of routing protocols AODV, LAR and ZRP is discussed and then compared in scalable networks [16]. The comparison of DSR and DSDV is done with four different mobility models such as RandomWaypoint, Group Mobility, Freeway and Manhattan model.

22 2.1 Related Work and Motivation 7 Table 2.1 Protocols with their features in brief Feature DSDV DYMO OLSR Path Calcula- Distributive Flooding based Dijikstra s Altion Bellman Ford gorithm Algorithm Flooding Exchange Expanding Broadcast Control topology Ring search topology Mechanism Info with (ERS) control neighbours Algorithm messages only through MPRs DBF Algorithm ERS Algorithm Shortest path finding Algorithm Route Calcu- Host node RREQ is Maintains the lation broadcast broadcasted Information routing info by originating about next two that include node towards hops towards new sequence targeted node, destination number, it send back (periodic destination RREP for route updates address and no. confirmation of network of hops topology) Route Main- Trigger update RERR, Route HELLO tenance rediscovery (Fig.2.5), redundant TC messages (Fig.2.4)

23 8 Chapter 2 Performance Evaluation of DSDV, OLSR and DYMO using and 82.11p MAC-Protocols Table 2.2 Simulation Parameters for MANETs PARAMETERS VALUES NS-2 Version 2.34 DYMO Implementation DYMOUM [19] OLSR Implementation UM-OLSR [2] Number of nodes 25,5,75,1 Speed (m/s) 2,7,15,3 Data Type UDP, CBR Simulation Time (s) 9 Data Packet Size PHY Standard Radio Propagation Model 512 bytes IEEE 82.11/82.11p Nakagami

24 2.2 Simulations and Discussions Simulations and Discussions The simulation scenario consist of Highway model involv- ing Vehicle going in two directions in the same way as it happens in four-lane real Highway environment. The simula- tion is performed with two Mac layer protocols and 82.11p. DSDV, DYMO and OLSR are used as routing layer protocols with bothmac layer protocols. The comparison of all these protocols is done in varying mobilities and densities of vehicles. The performance metrics used are shown in Table Throughput It is the measure of data received per unit time. Its unit is byte per second (bps). As shown in fig.2.1.a MANETs throughput taken in scalability scenario with the increase in the number of nodes throughput decrease. At 25 nodes and 5 nodes itšs below 4Kbps while at 75nodes its 25Kbps and on 1 nodes it decrease to 13Kbps. As number of nodes increase the packet delivery also increases. But the chance of link break and network partitioning is more likely to happen with large network, so throughput decreases. As shown in fig.2.1.b MANETs throughput taken in mobility scenario shows that for high mobility the throughput value increase simultaneously. Due to increase in speed of node it is not easy to discover routes. In Fig.2.1.c VANETs throughput taken in scalability scenario shows that its increases with the number of 25 nodes and 5 nodes but its value become high at 1 nodes. Due to increase in number of nodes more broadcast is take place in the network and route table updates are coming immediately. So itšs easy for the nodes to establish the route to destination. As shown in fig.2.1.d VANETs throughput taken in mobility scenario depicts that with varying mobility the throughput value overall decreases for DSDV. The chance of link break can be possible for the increase in the speed of nodes so throughput decreases. Fig.2.1.a shows that the throughput of OLSR is decreased as the number of nodes increased from 25 nodes to 1. In Fig.2.1.a the scenario consisted of varying number of nodes moving at a constant speed

25 1 Chapter 2 Performance Evaluation of DSDV, OLSR and DYMO using and 82.11p MAC-Protocols 7 6 DSDV OLSR DYMO 7 6 Throughput [Kbps] Throughput [Kbps] Number of nodes Number of nodes (a) MANETs Throughput vs Scalability 6 5 (b) MANETs Throughput vs Mobility 6 5 Throughput [Kbps] Throughput [Kbps] Speed m/s Speed m/s (c) VANETs Throughput vs Scalability (d) VANETs Throughput vs Mobility Figure 2.1 Throughtput of DSDV, OLSR, DYMO, AODV of 15m/s. The nodes are moving like vehicles in a four lane Highway. The source and destination nodes consistently change some of their first hop neighbors. With more vehicles being processed using the number of intermediate nodes increased which results in increase of the transmission delay. Also the vehicles are consistently changing their locations and that delay allowed the nodes to lose the trail of destination which caused more dropped packets and less throughputs. When same scenario is implemented using 82.11p the packet drops dramatically as shown by fig.2.1.c. The observation shows that the OLSR performs better with and give more throughputs which show its stability to scalability. From fig.2.1.b when the mobility is varied from 2m/s to 3m/s the throughput decrease to a negligible extent. Here the Mac

26 2.2 Simulations and Discussions 11 protocol used is Fig.2.1.d shows that Mac protocol 82.11p again shows no stability for OLSR and gives very low throughput. In MANETs, throughput taken in scalability scenario fig.2.1.a shows that with the increase in number of nodes the throughput of DYMO increases as compared to other protocols. DYMO exhibited the highest throughput compared to DSDV and OLSR since more routing packets are successfully deliv- ered by DYMO than DSDV and OLSR. The throughput for each routing protocol continues to fluctuate as the time progresses; DYMO still produces the highest throughput. In fig.2.1.b throughput taken in mobility scenario shows that with the increasing speed of node the throughput finally decreases as compared to the other routing protocols. In VANETs, throughput taken in mobility scenario fig.2.1.c this shows that the DYMO throughput is decreases with the increasing number of nodes, while at 1 nodes is gone zero. The protocols DYMO tends to have a higher packet delivery fraction (ratio), but DYMO throughput decreases routing protocol requires robust route discovery and route maintenance to cope with the dynamic changing topology Moreover DYMO has more throughput as reactive routing protocol. In Fig.2.1.d when throughput taken in the mobility scenario of nodes shows little high throughput value as others, but less at low speed E2ED It is the time taken for a packet to be transmitted across a network from source to destination. In fig.2.2.a MANETs E2ED taken in scalability scenario shows that at less number of node like 25 nodes its highest value. Wile at 5 nodes and 1 nodes it have lowest value. As DSDV because it is a proactive type routing protocol and advantage of these type of protocol is there is less delay to find out the route from source to destination nodes because path is immediately available when source need to send a packet. As shown in fig.2.2.b MANETs E2ED taken in mobility scenario the performance of DSDV is decreasing due to the mobility of nodes the load of exchange of routing tables becomes high and the frequency of exchange also increases. From fig.2.2.c in VANETs E2ED is taken in scalability scenario shows that at varying number of nodes its very less value even at 1 nodes it has no effect. As far as delay is concerned, DSDV performs better

27 12 Chapter 2 Performance Evaluation of DSDV, OLSR and DYMO using and 82.11p MAC-Protocols than OLSR and DYMO with large number of nodes. Hence for real time traffic DSDV is preferred over OLSR and DYMO. In fig.2.2.d VANETs E2ED taken in mobility scenario shows that; with the increase in speed the delay has no effect its value remains same. When the source and destination nodes goes away from each other while having congestion channel and maximum traffic, delay increases for DSDV because of high node mobility, it performs better than other protocols. It is easily visible from fig.2.2.a that E2ED increases as the number of nodes increase. The change in E2ED is not dramatic which is because of proactive nature of OLSR. OLSR maintains a route for every node even before starting the transmission of data packet. That is why increase in intermediate nodes does not highly affect the E2ED. The limited increase occurred because of propagation delay. The E2ED is even lower when OLSR is used with 82.11p. E2ED remains constant for increasing number of nodes but for OLSR with 82.11p it is obvious as shown in Fiq.1.c and Fig.2.1.d that too many packets were dropped therefore OLSR is not reliable for VANETs. The fig.2.2.b and fig.2.2.d shows that E2ED for OLSR when simulated with and 82.11p respectively increase slightly and does not show any major fluctuation with the variation of number of vehicles and speed. The OLSR proves its stability as expected when it comes to transmission latency. OLSR is reliable if speedy communication is not needed. From fig.2.2.a MANETs, E2ED taken in scalability scenario shows that delay is very less of DYMO as compared to others. As routes break, nodes have to discover new routes which lead to longer E2ED (packets are buffered at the source during route discovery). DYMO performs slightly better compared to DSDV and OLSR. As shown in fig.2.2.b E2ED taken in mobility scenario shows that it is very less value with comparison to other protocols. This is significant for the fact that, as the variations of packet delay becomes more predictable, the routing mechanisms can factor in that delay to determine whether the packet is lost or not. Instead of a mobile node waiting for high time for other routing protocols packet, it needs to wait for less time to determine whether the packet is lost or not. Thus saving crucial time that can be utilize to initiate fresh route discovery operations. In fig.2.2.c VANETs, E2ED taken in scalability scenario shows that delay of DYMO in VANETs in-

28 2.2 Simulations and Discussions DSDV OLSR DYMO E2ED [s].5 E2ED [s] Number of nodes Number of nodes E2ED [s] (a) MANETs Throughput vs Scalability Speed m/s (c) VANETs Throughput vs Scalability E2ED [s] (b) MANETs Throughput vs Mobility Speed m/s (d) VANETs Throughput vs Mobility Figure 2.2 E2ED of DSDV, OLSR, DYMO, AODV creases with the increase in number of nodes and its value is very high as compared to DSDV and OLSR. For average E2ED, DYMO is behaving like DSDV and OLSR for the less number of nodes and on words, DSDV and OLSR perform better than DYMO, routing protocol in term of stable and low average E2ED. As routes break, nodes have to discover new routes which lead to longer E2ED (packets are buffered at the source during route discovery). DYMO performs slightly better compared to DSDV and OLSR. See in fig.2.2.d when E2ED taken in mobility scenario experiences greater delay for DYMO.

29 14 Chapter 2 Performance Evaluation of DSDV, OLSR and DYMO using and 82.11p MAC-Protocols NRL It is number of control messages transmitted to receive one data packet. When NRL taken in scalability scenario in fig.2.3.a MANETs its value is increase with the number of nodes. At 25 nodes and 5 nodes its value is lower. While at 75 nodes and 1 nodes number of nodes it increases but no significant change. NRL taken in mobility scenario in fig.2.3.b MANETs shows that at low speed of nodes its value is high but at 15m/s and 3m/s its value decreases. As in DSDV nodes are continuously updating their routing table so with the increase in number of either nodes or mobility it is easy for their route maintenance so it shows less NRL value. NRL taken in scalability scenario fig.2.3.c VANETs shows that the NRL value remains constant overall but lower at 25 nodes. When NRL taken in mobility scenario fig.2.3.d in VANETs shows that its value has no significant change with the increase in speed of nodes. DSDV shows very less value as compared to the other routing protocols in VANETs. Generally OLSR is known for producing low routing load because of its proactive nature as well as it is unique MPRs approach. The routing overhead may increase with the in- creasing number of topological changes. The analysis of Fig. 2.3 will determine the behavior of OLSR with variable speed and number of nodes for both and 82.11p. In Fig. 2.3(a) OLSR is been simulated with From figure we can conclude that the NRL increased to a certain extent as number of nodes increased which is expected because for more nodes, more control packets are exchanged and number of MPRs also increases. Also the nodes are moving at a constant speed of 15m/s which forces topological changes and also causes routing overhead. When OLSR is simulated with 82.11p Fig.2.3 (c) shows that the routing overhead is way higher than what it is for The NRL is expected to increase with more number of nodes but the increase here is more than expected from OLSR. It again shows the incompatibility of OLSR with 82.11p. From Fig 2.3(b) and Fig 2.3(d) again it is obvious that the NRL increased with varying mobility as expected but the increase is much more when OLSR is simulated with 82.11p and the increase in NRL is negligible when OLSR is simulated with From all these observations we conclude that OLSR gives lower E2ED at the cost of Throughput and NRL when exe- cuted with 82.11p and gives adorable throughput E2ED and NRL in

30 2.2 Simulations and Discussions DSDV OLSR DYMO 2 15 NRL 1 NRL Number of nodes Number of nodes NRL (a) MANETs Throughput vs Scalability Speed m/s (c) VANETs Throughput vs Scalability NRL (b) MANETs Throughput vs Mobility Speed m/s (d) VANETs Throughput vs Mobility Figure 2.3 NRL of DSDV, OLSR, DYMO, AODV From Fig 2.3(a) and Fig 2.3(b) in MANETs, when NRL taken in scalability and mobility scenario its value increases with the increase in the number of node and had high value as compared to other protocols. DYMO reduces the control messages due to reactive nature so its value must be very less but in comparison with the proactive protocols its value is high. In VANETs, The NRL value increases with the increase in the number of node accordingly to other protocols the value is moderate. From Fig 2.3(c) and Fig 2.3(d) in VANETs, when NRL taken in scalability and mobility scenario its value is very less, has no significant effect due to mobility of the nodes for DYMO.

31 16 Chapter 2 Performance Evaluation of DSDV, OLSR and DYMO using and 82.11p MAC-Protocols Table 2.3 Performance trade-offs made by routing protocols Routing Trade-Offs Reasons Protocols MANETs VANETs DSDV Better E2ED Average Proactive nature (Fig 2.2(a),(b) at) throughput Preexisting the cost of (Fig 2.1(c),(d)) and routes reduces throughput better NRL transmission (Fig 2.1(a),(b)) and (Fig 2.3(c),(d)) at the delay NRL NRL (Fig 2.3(a),(b)) cost of E2ED is not very (Fig 2.2(c),(d)) high because of Incremental updates DYMO Low throughput Low throughput Reactive nature (Fig 2.1(a),(b)), Low (Fig 2.1(c),(d) at) More link E2ED (Fig 2.2(a),(b)) cost of high breakage causes at cost of slightly E2ED (Fig 2.2(c),(d)) high NRL and higher NRL and excessive increase in (Fig 2.3(a),b) NRL (Fig 2.3(c),(d)) dropped packets OLSR Average Low throughput Shortest throughput (Fig 2.1(c),(d)) high path finding (Fig 2.1(a),(b)) high NRL (Fig 2.3(c),(d)) Algorithm E2ED (Fig 2.2(a),(b)) at cost of lower and low E2ED (Fig 2.2(c),(d)) NRL (Fig 2.3(a),(b))

32 Chapter 3 Modeling Network Connectivity for DSDV, OLSR, DYMO and AODV above and 82.11p 3.1 Related Work and Motivation Several papers have been published regarding the comparison of routing protocols in different simulation scenarios. The comprehensive modeling for network connectivity VANETs is been done in Viriyasitavat, W. and Bai [17]. The comparison for AODV and OLSR is carried out in realistic urban scenario with varying Node mobility and Vehicle density to observe the behavior of both protocols [18]. In this study, modified version of OLSR has been discussed. In accordance to this work we made some modifications in all other routing protocols which is shown below. The paper also shows comparison of DSR and DSDV with four different mobility models i.e., Random Waypoint, Group Mobility, Freeway and Manhattan model. A few studies are carried out to evaluate the performance of different routing protocols in VANETs for some scenarios [5] e.g. varying mobility, node density and different mobility models using any of MAC layer protocol such as or 82.11p for VANETs or MANETs. A few studies regarding evaluating 17

33 Chapter 3 Modeling Network Connectivity for DSDV, OLSR, DYMO and AODV above and 82.11p performance of routing protocols are discussed here. The study of DYMO is done in comparison with other routing protocols [13]. This publication deals with the performance of metrics such as jitter, throughput and delay. The importance of VANETs cannot be denied. VANETs demand the efficient throughput and minimum E2ED without considering any energy or bandwidth constraints. 3.2 Network Connectivity In VANETs As in [25] the Manhattan-grid road structure comprises M evenly-spaced horizontal and vertical two-lane. Every street has two lanes with opposite directions. Streets are arranged in a L km L km space. Each lane is then subdivided into N square cells of width L/N km, intended to accommodate at most one vehicle per cell at a time. In this section, we develop framework to accurately estimate network characteristics that depict the dynamics of network connectivity in VANETs that is link duration Link Duration Link duration between two adjacent vehicles is the time interval during which both vehicles are within transmission range of each other. It is effected by presence of intersections, as this may cause the link to break in various ways. We will encounter the combination of following cases where link may be broken: Case 1: In this case two neighboring vehicles travel in directions away from each other. As the distance increases beyond the transmission range, the link breaks. Similarly, when two vehicles travel in the same direction. The rear vehicle gets red light while the front vehicle gets green light, the distance between the vehicles increases causing the communication link to break. In order to simplify the analysis, two vehicles are assumed to move at a constant speed. Vehicle has equal probability of seeing red or green light upon arriving at intersection. Each vehicle has the same transmission coverage. Let X and Y denote the vehicles whose link is of interest. 1. Vehicles moving on same street in same direction: The cause of link breakage is the light indication of intersections to which X and Y vehicle arrives. Only

34 3.2 Network Connectivity In VANETs 19 one out of 4 (1/4) light combinations can cause link to break. N denote the number of intersections that each vehicle passes before link breakage. As a result, N follows a geometric distribution. This is an event whose probability is(< p<1) P(N = k)= p (1 p) k k=,1,2,3,... (3.1) The vehicles pass k number of intersections before link breakage. The expected value E(X) is given by: E(X)= k= k p (1 p) k (3.2) E(X)= p k= k (1 p) k (3.3) Multiplying both side of eq.3 by (1 - p): E(X)= p k=1 k (1 p) k (3.4) (1 p) E(X)= p k= k (1 p) k+1 (3.5) Now subtract eq.6 from eq.4: (1 p) E(X)= p k=1 (k 1) (1 p) k (3.6) E(X) (1 p) E(X)= p k=1 [(k) (k 1)] (1 p) k (3.7) p E(X)= p k=1 (1 p) k (3.8)

35 Chapter 3 Modeling Network Connectivity for DSDV, OLSR, DYMO and AODV above and 82.11p E(X)= k=1 (1 p) k (3.9) The infinite sum in eq.9 is a geometric series with term ratio (1 - p), so that it may be written: E(X)= 1 p 1 (1 p) (3.1) E(X)= 1 p p (3.11) Incidentally, E(X) of eq.1 corresponds to the following form of the geometric distribution: P(N = k)= p (1 p) k 1 k=1,2,3,... (3.12) As mentioned earlier 1/4 light combination cause link to break. eq.12 can be written as P(N = k)= 1 4 (1 1 4 )k 1 (3.13) In this case both vehicles see red light, the distance between two vehicles remains constant and the link is establish during the stay time for t Stay seconds. The link also remain establish for d inter v which is equal to the time a vehicle takes to reach the next intersection. Where d inter is distance between two adjacent intersections and v is speed of vehicle. Thus, the link duration can be expressed as follows: LD= M k 1 k=1 j= [( jt Stay +(k 1) d inter v ) P(J = j N = k) P(N = k)] (3.14) Vehicle pass k intersections and sees red lights at j intersection out of k intersections that is time period during which the link remains establish. In eq.14 M is number of streets which is equal to square root of total number of intersection. Now placing the corresponding values in the eq.14 from eq.13. LD= M k 1 k=1 j= [( jt Stay +(k 1) d ( ) inter k 1 v ) j (1 1 3 )k j 1 ( 1 3 ) j (1 1 4 )k ] (3.15)

36 3.2 Network Connectivity In VANETs 21 Figure 3.1 Probability of Communication Now simplified form of equation eq.15 can be expressed as LD= M k k= j= [( jt Stay + kd ( ) inter k ) ( 1 v j 2 )k+ j+2 ] (3.16) The expression ( jt Stay + kd inter v ) in the eq.16 corresponds to the time that a vehicle takes to pass k intersections and sees red lights at j intersection out of k intersections. The remaining part of this equation is the probability that a vehicle passes k intersections before the link breaks. As in this case vehicles X and Y are traveling on the same street in the same direction, then the average link duration, would be LD(d o ) avg,

37 Chapter 3 Modeling Network Connectivity for DSDV, OLSR, DYMO and AODV above and 82.11p where d o is distance between two vehicle at time t o M k LD(d o ) avg = d inter d o + [( jt Stay + kd inter ) v k= j= v ( k )( 12 (3.17) j )k+ j+2 ] d o < d inter M k LD(d o ) avg = 2d inter d o + [( jt Stay + kd inter ) v k= j= v ( k )( 12 (3.18) j )k+ j+2 ] d inter < d o < 2d inter 2. Vehicles moving on same street but in opposite direction: Assuming that both vehicles initially travel away from each other. It is clear that the distance both vehicles have to travel combined, before the link breaks, is T R LOS d o meters. Where T R LOS is transmission range for line-of-sight communication. Also, the maximum number of intersections both vehicles pass before link break must be less than TR LOS d inter. This implies that both vehicles X and Y pass one intersection each before the link breakage. Let L X and L Y denote the light indication at the intersections respectively. When one of the vehicles encounters red light and the other does not, the distance between the two vehicles increases by t Stay v meters. Followings are link durations in four possible light combinations summarized in following two points: 1. LD 1 = TR LOS d o 2v 2. LD 2 = t Stay + TR LOS d o vt Stay 2v when L X and L Y both Green/Red. when L X Green/Red and L Y Red/Green. Since each case has equally probability, then average link duration will be. LD away (d o ) avg = [LD 1+ LD 2 ] 2 LD away (d o ) avg = 1 2 t Stay+ T R LOS d o 2v When vehicles traveling towards each other then average link duration will be. (3.19)

38 3.2 Network Connectivity In VANETs 23 LD(d o ) avg = LD away ()+ d o 2v LD(d o ) avg = 1 2 t Stay+ T R LOS+ d o 2v (3.2) Case 2: In addition to the first case, communication link may break when the direction of one or both vehicles changes as they turn on intersections of roads. Depending on vehicle directions, the two vehicles, X and Y may be in one of the following subcases: 1. Vehicles approaching toward the intersection of roads In this approaching of vehicles case, X and Y travel in the perpendicular directions but toward the same intersection denoted as inter. Both vehicles take at least T to Inter = min(d X,d Y ) u seconds before arriving at the intersection.where d X and d Y is the minimum distance of the vehicles from the intersection and T R NLOS is the Transmission range for non-line-of-sight communication. L X and L Y denote the light indications at intersections. Based on the light indication at the intersection, the following cases should be considered. 1. LD 1 = T to Inter+ TR NLOS 2v, when L X and L Y both Green/Red. 2. LD 2 = T to Inter+t Stay + TR NLOS vt 2v Stay, when L X Green/Red and L Y Red/Green. Once both vehicles pass the intersection, they begin to travel away from each other. In this case where one of the two vehicles, X stops at an intersection L X red and L Y green, the distance between X and Y is equal to vt Stay when vehicle X starts to move again. After the light turns green for X, both vehicles Amove and hence, the distance between them increases at a rate 2v m/s. Therefore the link continues to be establish for additional TR NLOS vt 2v Stay seconds. Since each case above occurs with equal probability, LD avg (d X,d Y )= [LD 1+ LD 2 ] 2 (3.21) LD avg (d X,d Y )= min(d X,d Y ) v +( )t Stay+ T R NLOS 2v (3.22)

39 Chapter 3 Modeling Network Connectivity for DSDV, OLSR, DYMO and AODV above and 82.11p The first and second terms in eq.22 correspond to the time a vehicle takes to reach the intersection and the time, after it reaches the intersection, before the link breaks, respectively. The last term takes into account the events that one or both vehicles see the red light at intersections. 2. Vehicles moving away from the intersection of roads If X and Y are initially traveling away from each other, then By using basic geometry we can deduce the formula. When both vehicles travel away from each other in the perpendicular directions, the distance increases by 2v m/s. LD away (d o )= T R NLOS do 2v (3.23) 2. Vehicles moving in opposite direction from intersection of roads In this case, vehicle X travels toward the intersection while the other, Y travels away from the intersection. Hence, link duration in this case depends on the locations of vehicles X and Y at the time they first connect. Specifically, depending on the distance from X to the next intersection, one of the following three cases is possible. 1. Locations of X and Y are shown on in Figure (). Based on the transmission range, when vehicle X is in the middle of the road, in order for vehicle Y to connect with vehicle X, d Y < d inter. Due to the slow down of vehicle Y as it approaches an intersection, the link between X and Y remains establish until either vehicle X moves beyond the middle of the road or vehicle Y reaches the next intersection. If <d X < d inter /2 d inter 2 d X LD(d X )=min( v T R 2, NLOS dx 2 d inter ) (3.24) v 2. In this case, the link always breaks at the next time step unless vehicle X stops at intersection. This is due to d(x,y) is increasing when dx > dy. Consider that X stops at the intersection, the link between X and Y continues to establish and later breaks when X starts to move again. By the

40 3.2 Network Connectivity In VANETs 25 time the light turns green for X, vehicle Y moves closer to X by vt Stay meters. Let t denote the time after X gets green light until the link breaks and t(dx) denote the minimum value of t at which T R NLOS = d X (t) 2 + d Y (t) 2 (3.25) where d X (t)= d inter + vt (3.26) d Y (t)= d Y () vt Stay vt d Y (t)= T R 2 NLOS d2 X vt Stay vt (3.27) Therefore, If d inter /2<d X < d inter, then LD dx ={ 1 if X is not at intersection 1 2 (1+t Stay +td X ) elsewhere Therefore, in the case X stops at intersection, the link duration is equal to t Stay + t(d X ). Since vehicle X has an equal probability of seeing red/green light upon arriving at the intersection, the average link duration of vehicle X is equal to 1/2[1+(t Stay +t(d X ))]. 3. Observe from the fig that the only intersection that affects the link between X and Y is the intersection at which Y is about to arrive. The link between X and Y always breaks unless vehicle Y stops at the intersection. In the case where Y encounters red light and stops at the intersection, both X and Y begin to connect through T R LOS communication, the link may break in one of the following two situations: X moves out of the transmission range of Y. In this, the link duration is

41 Chapter 3 Modeling Network Connectivity for DSDV, OLSR, DYMO and AODV above and 82.11p equal to (T R LOS d X )/v. Y gets green light and moves out of X transmission range, then the link duration is t Stay. If d X > d inter LD(d X )={ 1 if Y is not at intersection 1 2 (1+min(t Stay, T R LOS d X v )) elsewhere Thus, the link duration in the case that X first encounters red light at the intersection is equal to min(t Stay, T R LOS d X ) v. In addition to this the turning decision of vehicles is one of the dominant factors that determine duration of the link, it is worth noting that we can predict the link duration of two vehicles more accurately if the route itinerary of the vehicles that is destinations and turning decisions at each intersection are known. 3.3 SIMULATIONS AND DISCUSSIONS The simulation scenario consist of Highway model involving Vehicles moving in two directions in the same way as it happens in four-lane real Highway environment. The simulation are performed with two Mac layer protocols and 82.11p. DSDV, DYMO, OLSR and AODV are used as routing layer protocols with both Mac layer protocols. The comparison of all these protocols is done by varying the mobility and density of Vehicles. The performance metrics used are shown in Table 3.1 Throughput is the measure of data received per unit time. Its unit is bytes per second (bps). In Fig.3. 1.a for the scenario in which node density is considerably low and the MAC layer protocol is used; maximum throughput is generated by DYMO while MOD OLSR, and DSDV are also producing good throughput. DYMO is performing well. DYMO is a reactive protocol it

42 3.3 SIMULATIONS AND DISCUSSIONS 27 Table 3.1 Modifications in Protocols Protocols DSDV Modified Features Minimum time b/w trigger update 15 to 3perup (periodic update interval) Weighted settling time=7 OLSR HELLOmessages or TCmessages time interval is decrease can only transmits or receives the routing packets when data arrives. fig.3.1.a having scalability scenario DYMO is on the top once again while the other protocols shows decrease throughput. (MOD OLSR, OLSR, MOD DSDV, and DSDV). Nodes are mobile at a constant speed of 15m/s for all scalabilities scenarios. In wireless networks, generally the link quality considerably varies in different periods of time. The reasons may be; some mobile nodes are moving randomly, some go-out of range, some intentionally cutoff the ongoing communication, some die-out due to battery and so on. The respective routing protocols must be able to dynamically cop with the situation. MOD OLSR and DSDV both produce good amount of throughput because of their proactive nature and ability of decision made by each node; each node decides route up to next two hops. MOD OLSR performed slightly better because it s NRL is increased by increasing the number of control packets. Increase in the number of control packets is achieved by reducing the time interval for HELLO and TCmessages. DSDV did not produce enough throughput. ; in fact it has the lowest among all discussed protocols. It happens because the topology is varying quickly and node find a route too often to send the packets. But when we increased the time between trigger updates, periodic updates and periodic update interval these issue was resolved and more packets are sent to achieve higher throughput.

43 Chapter 3 Modeling Network Connectivity for DSDV, OLSR, DYMO and AODV above and 82.11p Table 3.2 Simulation Parameters Parameters Values Simulator NS-2(Version 2.34) Channel type Radio-propagation model Network interface type MAC Type Interface queue Type Bandwidth Packet size Wireless Nakagami Phy/WirelessPhy, Phy/WirelessPhyExt Mac /82.11, Mac/82.11p Queue/DropTail/PriQueue 2Mb 512B Packet interval.3s Number of mobile node Speed Traffic Type Simulation Time Routing Protocols 25 nodes, 5 nodes, 75 nodes,1 nodes 2 m/s,7 m/s,15 m/s,3 m/s UDP, CBR 9 s DSDV, DYMO, OLSR, AODV, MOD DSDV, MOD OLSR In Fig.3. 3.b for low mobility DYMO again produce the maximum throughput while OLSR, DSDV, MOD DSDV and OLSR MOD produced in descending order. For high mobility the maximum throughput is generated by the two of proactive protocols DSDV and OLSR. MOD OLSR produces the highest amount of throughput while DSDV, OLSR and MOD DSDV also produces good amount of throughput. Due to their proactive nature and the trigger updates by DSDV and MOD DSDV and HELLO plus TCmessages allows the DSDV and OLSR to know the state of route. Due to which the packets are more likely to

44 3.3 SIMULATIONS AND DISCUSSIONS 29 reach their destination. DYMO on the other hand has no procedure for link maintenance except RERR therefore it gives low throughput. In high mobility higher throughput for DYMO because of lower interval of ROU T E_T IMEOU T which with changing topology new routes are found. The DYMO depends on link layer s feedback for activation and deactivation of routes. Since 82.11p is better than therefore MOD DYMO produces better throughput. MOD DSDV is also showing good throughput because of proactive nature (which suits low scalability and mobility). DYMO is also performing well while MOD OLSR and OLSR are at last. MOD OLSR slightly better than OLSR because of more proactiveness. From Fig.3. 3.c we can observe that after getting some improvements in MOD DSDV and in throughput of MOD OLSR. DYMO has been producing the minimum throughput because of high route timeout which means a useless route is stored for long period of time. MOD DSDV improved the performance of DSDV because of decreased robustness and proactiveness. In MANETs, AODV improves its efficiency but also outperforms among all other protocols. There is a less expanding ring values in initial default ERS values up to T T L_T HRESHOLD. We can expand these rings by increment the T T L_INCREMENT value from 2 to 4.It lessens routing delay and increase communication probability as from equation. Also this results, not only low routing load but also lowers the routing latency. Ultimately, the throughput value increased. Overall, AODV have the highest throughput values, comparing to all other protocols. LLR is a distinguish feature of AODV to make this protocol more scalable along with gratuitous RREPs. Although DYMO is been acting decently but in the given scenario this proved its worth due to its smaller route timeout interval and decreased in RREQ wait time. Due to smaller route timeout the need of finding a new route is increased which means each route is valid (possibly). Also wait time decreases means more number of RREQs that leads to increased possibility of finding a new route. DSDV is working well for low mobility because of proactive nature while MOD DSDV works reasonable but not better than DSDV. OLSR also works well but MOD OLSR is working better due to more number of HELLO and TC messages. MOD DSDV produced second highest throughput while DSDV is the third. OLSR and MOD OLSR both are underachievers due to very proactive nature. Mostly pre calculated routes are timed out in the case of high mobility.

45 Chapter 3 Modeling Network Connectivity for DSDV, OLSR, DYMO and AODV above and 82.11p Throughput [Kbps] DSDV OLSR DYMO AODV Throughput [Kbps] Throughput [Kbps] Number of nodes (a) MANETs Throughput vs Scalability Number of nodes (c) VANETs Throughput vs Scalability Throughput [Kbps] Speed m/s (b) MANETs Throughput vs Mobility Speed m/s (d) VANETs Throughput vs Mobility Figure 3.2 Throughtput of DSDV, OLSR, DYMO, AODV E2ED It is the time required for a packet to reach its destination. When the scenario is for MANET and the node density is low, DSDV produces high E2ED depicted in Fig.3. 3.a because it is proactive protocol therefore it is not well suited for mobility. ( all nodes are mobile at a constant speed 15m/s It s trigger and periodic updates means that it has a preexisting path; which does not hold for mobile scenarios. In Mobile scenarios the paths are changing frequently. For MOD DSDV the E2ED decreases considerably. The reason is that on a preexisting path the packets are sent. Then there are only two possibilities packet is dropped or reaches its destination quickly. OLSR and MOD OLSR produced quite impressive average

46 3.3 SIMULATIONS AND DISCUSSIONS 31 Throughput [Kbps] MOD DSDV MOD OLSR Throughput [Kbps] Throughput [Kbps] Number of nodes (a) MANETs Throughput vs Scalability Number of nodes (c) VANETs Throughput vs Scalability Throughput [Kbps] Speed m/s (b) MANETs Throughput vs Mobility Speed m/s (d) VANETs Throughput vs Mobility Figure 3.3 Throughtput of MOD DSDV, MOD OLSR E2ED. It is due to the fact that it maintains a view of neighbors which stay as its neighbors for a long time or the neighbors are more stable, therefore it produces a low E2ED. DYMO has produced very impressive E2ED in the current situation. Because of the fact that it holds a route for long time and also number of control packets sent are more opposite of above conditions. For the scenario of high scalability view from Fig.3. 4.a DSDV has produced a high E2ED because the routes are frequently changing and waited settling time is a problem because the routes have changed too many times. Also the less interval of trigger updates means the route can change before is reached. When the trigger interval is increased then the stable routes are found more often.

47 Chapter 3 Modeling Network Connectivity for DSDV, OLSR, DYMO and AODV above and 82.11p OLSR gives a high E2ED because of less proactive nature while MOD OLSR gives less delay because of more proactiveness. Overall OLSR has produce more E2ED than other routing protocols in the scalability. DYMO gives a average E2ED because of reactive nature and source routing. shown in fig.3.3.a. When the low mobility is considered the DSDV has high E2ED and the MOD DSDV is again performing well due to the face that it founds more stable routes than DSDV. The Fig.3. 3.b reveals that for OLSR the E2ED is lower than MOD OLSR. (which has the highest E2ED for low mobility) DYMO is the best in E2ED for low mobility. For high mobility the DSDV and OLSR are not performing better than DYMO. The reason for high E2ED of DSDV and OLSR is Proactive nature, Stale routes and MPR s loss in OLSR. DYMO is being reactive protocol have reasonable E2ED nor higher and neither low. For the case of low scalability (when protocols is used is a special one for VANETs) Fig.3. 4.c shows that MOD OLSR is the best in case of E2ED while MOD DYMO and OLSR are also competing with it. OLSR, DSDV and MOD OLSR have such variation because of 82.11p. It is obvious that for high scalability the behavior of DSDV is same as in MANETs. The Fig.3. 4.d when simulations are performed in VANETs and variable mobility scenario then E2ED decreases by small value for OLSR and MOD OLSR but order remains the same. For DYMO, order is same only values have changed slightly. For low and high mobility behavior of all protocols is the same. While AODV store only one rout for one destination for small interval. Therefore AODV has to find a new route to destination more often. As AODV has ERS mechanism with LLR without caching, thus introduce more delay. NRL is number of control messages transmitted to receive one data packet. Fig.3. 5.a In MANETs NRL taken in low scalability scenario. The DYMO having highest NRL while MOD DSDV and DSDV is having the lowest. When taken in higher scalability the value of DYMO relatively increases as compared to other protocols and MOD DSDV shows the same lower behavior. NRL depends on number of routing messages

48 3.3 SIMULATIONS AND DISCUSSIONS 33 E2ED [s] DSDV OLSR DYMO AODV E2ED [s] Number of nodes Speed m/s (a) MANETs Throughput vs Scalability 2 (b) MANETs Throughput vs Mobility E2ED [s] 1 E2ED [s] Number of nodes Speed m/s (c) VANETs Throughput vs Scalability (d) VANETs Throughput vs Mobility Figure 3.4 E2ED of DSDV, OLSR, DYMO, AODV RERR, RREP, RREQ and HELLOperiodicmessages. DYMO reduces the messages so its value must be less but in comparison with the proactive protocols its value is very high. For larger networks the NRL is relatively lower. Rate limit is reduced which will eventually help in decreasing NRL of DYMO but E2ED may increase. MOD DSDV NRL was expected to decrease due to the modifications that have been involved minimum time between trigger update due to the fact that less frequent topology updates. RREPs generated by intermediate nodes in low scalabilities are more suitable, while these RREPs in high densities producing high routing load. AODV produces more routing load, as ERS value is more in AODV and make it more scalable. More frequent exchange of routing updates generates more NRL.

49 Chapter 3 Modeling Network Connectivity for DSDV, OLSR, DYMO and AODV above and 82.11p MOD DSDV MOD OLSR E2ED [s] 1.5 E2ED [s] Number of nodes (a) MANETs Throughput vs Scalability Speed m/s (b) MANETs Throughput vs Mobility 1.5 E2ED [s] E2ED [s] Number of nodes Speed m/s (c) VANETs Throughput vs Scalability (d) VANETs Throughput vs Mobility Figure 3.5 E2ED of MOD DSDV, MOD OLSR In Fig.3. 5.b in MANETs, NRL is taken in mobility scenario. At low speed of nodes its value is high but at 15& 3m/s its value gone decreases. As in DSDV nodes are continuously updating their routing table with the increase in number of nodes or mobility it s easy for route maintenance to cause less NRL value. The modifications that have been made in MOD DSDV, NRL was expected to decrease due to the modifications that have been made involved minimum time between trigger update due to the fact that less frequent topology updates. Rate limit is reduced which will eventually help in decreasing NRL of DYMO. In Fig.3. 5.c in VANETs when NRL is taken in low scalability shows that DSDV, MOD DSVD and DYMO shows relatively lesser value as compared to OLSR and MOD OLSR. When NRL taken in high

50 3.3 SIMULATIONS AND DISCUSSIONS 35 scalability in the same scenario the MOD OLSR shows the highest value while MOD DSDV shows the very less value. Generally OLSR is known for producing low routing overhead because of its proactive nature as well as its unique Multipoint relays approach. The routing overhead may increase with the increasing number of topological changes. From Fig.3. 5.c we can conclude that the normalized routing load increases to a certain extent as number of nodes increased which is expected because for more nodes more control packets are exchanged and number of MPR sets also increases. Also the nodes are moving at a constant speed of 15m/s which forces topological changes and also causes routing overhead. When OLSR is simulated with 82.11p. When we talk about the MOD OLSR modifications same response has been shown. HELLO messages or TC messages time interval has been decrease so that the NRL value become greater so that in less time more routing packets are flowing in the network topology. So that it make good view of whole topology and more and more packets will reach to its destination. In Fig.3. 5.d VANETs NRL is taken in mobility scenario shows that its value has no significant change with the increase in speed of nodes. DSDV shows very less value as compared to the other routing protocols in VANETs. The MOD OLSR value is very high, at low speed of nodes its value is high but at 15 and 3m/s its value gone decrease. As in DSDV nodes are continuously updating their routing table so with the increase in number of nodes either or mobility it s easy for their route maintenance so it shows less NRL value.

51 Chapter 3 Modeling Network Connectivity for DSDV, OLSR, DYMO and AODV above and 82.11p NRL DSDV OLSR DYMO AODV Number of nodes (a) MANETs Throughput vs Scalability NRL Speed m/s (b) MANETs Throughput vs Mobility NRL 1 NRL Number of nodes Speed m/s (c) VANETs Throughput vs Scalability (d) VANETs Throughput vs Mobility Figure 3.6 NRL of DSDV, OLSR, DYMO, AODV

52 3.3 SIMULATIONS AND DISCUSSIONS MOD DSDV MOD OLSR 4 3 NRL 1 NRL Number of nodes Speed m/s (a) MANETs Throughput vs Scalability (b) MANETs Throughput vs Mobility NRL 1 NRL Number of nodes Speed m/s (c) VANETs Throughput vs Scalability (d) VANETs Throughput vs Mobility Figure 3.7 NRL of MOD DSDV, MOD OLSR

53 Chapter 3 Modeling Network Connectivity for DSDV, OLSR, DYMO and AODV above and 82.11p Table 3.3 Performance Trade Offs by Routing Protocols Routing MANETs VANETs Reasons Protocols DSDV Better E2ED Average Proactive at the cost of throughput nature. Preexisting throughput & routes reduces & NRL better NRL delay E2ED & NRL not very high due to incremental updates DYMO Low Low throughput Reactive nature. throughput, high E2ED More link Low E2ED and excessive breakage and slightly NRL & stale higher NRL routes causes high NRL & increase in dropped packets OLSR Average Low throughput Proactive nature throughput high NRL causes higher high E2ED and low E2ED NRL & lower NRL low E2ED. MPR reduces NRL

54 3.3 SIMULATIONS AND DISCUSSIONS 39 Table 3.4 Performance Trade Offs by Routing Protocols AODV High Throughput LLR, HELLO at the cost of high NRL messages & RREQ messages & E2ED MOD Much better Average Increased DSDV throughput throughput settling time, at the cost and the cost node of decreased E2ED waits for E2ED and NRL fresh routes. & NRL This increases throughput & decreases NRL MOD Average Average throughput HELLO messages OLSR throughput low E2ED or TC messages low E2ED increased time interval increased NRL NRL decreases greater routing packets so throughput increases & also NRL

55 Chapter 3 Modeling Network Connectivity for DSDV, OLSR, DYMO and AODV above and 82.11p

56 Chapter 4 Modeling Network Connectivity for FSR, DYMO, DSDV and AODV above and 82.11p 4.1 Related Work and Motivation In [19] [2], communication time between nodes is found when the nodes are moving in same and opposite direction with same or different speeds. In our work we improvement the work and calculate the probability of link establishment between nodes when they are moving in same and opposite direction with same or different speeds. In [21] the authors modified the OLSR protocols in their paper. We also do some modifications and evaluate AODV, DSR and FSR for both MANETs and VANETs. The study [14] involved the consistently varying network topology and comparison of DSR with AODV in MANETs for different scenarios and performance metrics to propose the best scenario for each routing protocol to maximize its efficiency. We can further extend our work on the communication of nodes by 41

57 Chapter 4 Modeling Network Connectivity for FSR, DYMO, DSDV and AODV above and 82.11p calculating the communication time for which the given set of nodes will have a direct communication within respective segments. The route breakage probability of nodes within the segments is modeled in [23], as the probability is independent of time. 4.2 Modeled Mathematical Framework Distribution of Node Population Size This work determine the steady-state distribution of the number of nodes within each segment. Let N k j is Poisson distribution with the parameter λ t k j B x(τ) (R k j )dτ. The corresponding steady state distribution can be found by determining the limit of the Poisson parameter as the time approaches to infinity. Let N i denote the node population within segment i. Defining k as the sub-strip that segment i is located, then k=max(1,... j...k) i j < i. N i has also Poisson distribution with parameter φ i. The probability distribution of the number of the nodes within segment i and its probability generating function (PGF) at the steady state is given by: Pr(Ñ i = n)=e φ φ n i i n! and P(z)=E[z N i ]=e φ i(1 z) (4.1) Fig 4.1 shows detail scenario of communication time between node with probability of link establishment. The above mention work can be extended and improvement can be made in the distribution of node population size. Probability of node population was carried out within a segment N i. The improvement has been made to find the node population size within the whole topology containing n segments. Let N ntotal denote the population within n segments. Defining s as the strip containing n segments, then the Poisson distribution of N ntotal with parameter φ ntotal is given by: φ ntotal = s S l=1l=s+1 φ ln (n total ) (4.2)

58 4.2 Modeled Mathematical Framework 43 where, φ ln (n total )=φ ln P ln (n total ) and; P ln (n total )= ntotal d R 1ntotal (n total 1)d R 1ntotal b xln (r)dr = emn totald e m(n total 1)d e mr 1n total (e mr ln 1) (4.3) Where d is node s transmission range, R is total length of the strip containing n segments. Also; m= µβ 2σ 2 +µ 2 The probability distribution of the number of the nodes within n segments and its PGF at the steady state are given by: Pr(Ñ ntotal = n)= P(z)=E[z N n total ]= n total T=1 e φ n total φ n n total n! n total e φ n total (1 z) T=1 (4.4) ANALYSIS OF NETWORK CONNECTIVITY In this section, we will determine the network connectivity of a new arriving node at the beginning of the service strip at the steady state. It is assumed that two nodes will be able to communicate directly if L<d where L is the distance between the nodes and d is the constant transmission range of a node. Clearly, all the nodes within a given segment are able to communicate directly as they are within each other s transmission range. Next, let us define the direct communication probability of two nodes at consecutive segments i and i+1 as; P i =Pr(L<d) two nodes are located at consecutive segments i and i+1). Moreover this probability of two nodes located within two consecutive segments has a constant value of 1/2. The direct communication probability of two nodes located at consecutive segments i and i + 1 respectively is given by, P i = Pr(x i+1 x i < d)= ri ri+1 o r i b xi (r i )b xi+1 (r i+1 )dr i dr i+1 (4.5)

59 Chapter 4 Modeling Network Connectivity for FSR, DYMO, DSDV and AODV above and 82.11p We can extend work mentioned in [37] and find the indirect communication between nodes in the segments. The indirect communication can take place between 3 4 for n segments. Figure 4.1 System Model Indirect Communication The probability of direct communication is given in eq.(5). For three segments, the probability of nodes having indirect communication between them is given by: = ri+1 ri+2 r i P i+1 = Pr(x i+2 x i+1 < d) r i+1 b xi+1 (r i+1 )b xi+2 (r i+2 )dr i+1 dr i+2 (4.6)