Investigating the Performance of Tcp Variants and Routing Protocols in Mobile Ad Hoc Networks

No Thumbnail Available



Journal Title

Journal ISSN

Volume Title


Addis Ababa University


Mobile Ad Hoc Network (MANET) is a collection of mobile nodes that can dynamically and randomly move and self organize to form network topology. MANETs have provided new challenges, which are the results of the unique characteristics of the wireless medium, the dynamic nature of the network topology, lacks of well secured boundaries, route failure due to frequent link breakages, wireless channel error, and using multi-path routes, which affects the end-to-end transmission of data. Transmission control protocol (TCP) performs poorly in such networks, since TCP’s congestion control mechanism cannot distinguish between congestion and non-congestion related packet losses. TCP was previously developed for wired networks with the assumption that packet loss is an indication of congestion. However, in MANET TCP performs congestion control action for several types of losses that are not related to congestion. Consequently, when a packet loss is detected either by timeout or three duplicated acknowledgments, TCP slows down the sending rate by adjusting its congestion window size (cwnd) and unnecessarily retransmit a packet, which leads to lower throughput. In this dissertation research, we have proposed a method for TCP to distinguish between packet losses due to congestion or route failure due to mobility of nodes. The proposed protocol is called TCP Packet Loss Detection and Response (TCP-PLDR). We have developed an analytical model of throughput of TCP with selective acknowledgment (TCP-SACK) and TCPPLDR protocols as a function of packet loss probability and round trip time (RTT), in the presence of congestion and route failure losses. The model captures the behavior of TCP’s congestion avoidance mechanism and its impact on throughput. Results have shown that TCP-PLDR is TCP friendly while it improves the throughput of TCP-SACK when there is a packet loss ii due to route failure/change. Simulation was conducted using network simulator (ns-2) and results have shown that TCP-PLDR improves TCPSACK’s performance in MANET. As an example, simulation experiment for route failure and congestion loss scenario shows that TCP-PLDR improves the throughput of TCP-SACK on average by 39%. Moreover, the proposed protocol (TCP-PLDR) was evaluated in the presence of wireless channel error, and multi-path routing protocol like temporally-ordered routing algorithm (TORA) by making use of throughput, end-to-end delay, and packet delivery ratio (PDR) performance metrics and results showed that TCP-PLDR performed better than TCP-SACK. As stated above, in MANET TCP is unable to distinguish packet losses due to congestion or route/link failures due to mobility of nodes. Hence the way how routing protocols respond to route failures and route recovery mechanism has an effect on the performance of TCP variants. In the second part of this dissertation paper, the effect of routing protocols; Ad hoc On Demand Distance Vector (AODV), Destination Sequenced Distance Vector (DSDV), Dynamic Source Routing (DSR), and Temporally Ordered Routing Algorithm (TORA), on the performance of TCP variants; TCP-Newreno, TCP-Reno, TCPSACK, and TCP-Tahoe, under different mobility pattern and node density were studied thoroughly. Simulation results showed that, for all variants of TCP, AODV achieved the highest throughput. From TCP variants, TCP Newreno performed better than the other variants over the stated routing protocols. Besides, the best performing combination of routing protocols and TCP variants were identified. It is also confirmed that the performance of TCP variants are highly dependent on the underlying routing protocols in MANET. This result was taken as a valuable input to design and study the first part of this dissertation research. Moreover, performance of TCP-PLDR is compared with other variants of TCP (TCP-Reno, TCP-Newreno, TCP-Vegas, ATCP, and TCP-Westwood) and routing protocols (AODV, DSDV, DSR, and TORA). From TCP variants, in terms of throughput, TCP-PLDR performed better than the iii other variants. Whereas, in terms of delay, TCP-Vegas outperformed the other variants. From routing protocols, TCP-PLDR can be coupled with AODV and it is confirmed that TCP-PLDR is more suitable for reactive routing protocol than proactive in MANET. In MANET, packet drop attack due to malicious nodes can affect normal operation of routing protocols, performance of TCP, and performance of the network at large. The third part of dissertation study investigates the impact of malicious packet drop attack on the performance of two reactive TCP variants (Newreno and Sack) and one proactive TCP (Vegas) and two reactive routing protocols (AODV and DSR) by making use of throughput, end-to-end delay, and PDR performance metrics. Simulation was conducted by adding different percentages of malicious nodes in the network. Results showed that from TCP variants, Vegas outperformed Newreno and Sack in the absence of malicious nodes (0% malicious nodes). However, as the percentage of malicious nodes added in the network increases from 5% to 50%, Newreno and Sack performed better than Vegas though all of them are affected by malicious nodes. It is also noted that, even though both AODV and DSR protocols are highly affected by malicious nodes, AODV is more robust to malicious packet drop attack than DSR. Finally, performance analysis of TCP-PLDR and TCP-SACK were evaluated exhaustively under security attack (malicious packet drop attack). Upon completion of exhaustive simulation, it is confirmed that TCP-PLDR is more robust to malicious packet drop attack than TCP-SACK. Keywords: MANET, TCP-PLDR, TCP-SACK, Tahoe, Reno, Newreno, Vegas, Westwood, AODV, DSDV,DSR,TORA, route failure/change, packet loss, outof- order packet, congestion, wireless channel error, multi-path route, malicious packet drop attack.



Manet, Tcp-Pldr, Tcp-Sack, Tahoe, Reno, Newreno, Vegas, Westwood, Aodv, Dsdv, Dsr, Tora, Route Failure/Change, Packet Loss, Outof- Order Packet, Congestion, Wireless Channel Error, Multi-Path Route, Malicious Packet Drop Attack