Performance Evaluation of Two Reactive Routing Protocols of MANET

Archana Patidar Student of Digital Communication Acropolis Institute of Technology And Research, Indore archu_9196@rediffmail.com

Abstract Mobile ad-hoc network is a collection of wireless mobile hosts forming a temporary network without the aid of any stand-alone infrastructure or centralized administration. Mobile ad-hoc network ha e the attributes such as wireless connection! continuously changing topology! distributed operation and ease of deployment. "n this paper we ha e compared the performance of two reacti e M#$%& routing protocol #'() and (*+ by using $*-,. -oth share similar 'n (emand beha ior! but the protocol.s internal mechanism leads to significant performance difference. /e ha e analyzed the performance of protocols by arying network load! mobility and type of traffic 01-+ and &123.# detailed simulation has been carried out in $*,. &he metrics used for performance analysis are 2acket (eli ery 4raction! # erage end-to-end (elay! +outing ' erhead and $ormalized +outing 5oad. "t has been obser ed that #'() gi es better performance in 1-+ traffic and real time deli ery of packet. /here as (*+ gi es better results in &12 traffic and under restricted bandwidth condition. Keywords6 MANET, TCP, CBR

1.Introduction Mobile networks can be classified into infrastructure networks and Mobile #d 7oc $etworks 0M#$%&3 according to their dependence on fi8ed infrastructures 9,:."n an infrastructure mobile network! mobile nodes has wired access points 0or base stations3 within their transmission range. "n contrast! Mobile #d 7oc networks are autonomously self-organized

networks without support of infrastructure. "n a Mobile #d 7oc $etwork! nodes mo e arbitrarily! therefore the network may e8perience rapid and unpredictable topology changes. +outing paths in M#$%&s potentially contain multiple hops! and e ery node in M#$%& has the responsibility to act as a router 9;:. +outing in M#$%& has been a challenging task e er since the wireless networks came into e8istence. &he ma<or reason for this is the constant change in network topology because of high degree of node mobility. # number of protocols ha e been de eloped to accomplish this task.*e eral performance e aluation of M#$%& routing protocols using 1-+ traffic ha e been done by considering arious parameters such as mobility! network load and pause time. -iradar! *. +. et. al. 91=: ha e analyzed the #'() and (*+ protocol using >roup Mobility Model and 1-+ traffic sources. -iradar! *. +. et. al. 91=: in estigated that (*+ performs better in high mobility and a erage delay is better in case of #'() for increased number of groups. #lso +athy! +.?. et. al. 91@: in estigated #'() and (*+ routing protocols under +andom /ay 2oint Mobility Model with &12 and 1-+ traffic sources. &hey concluded that #'() outperforms (*+ in high load andAor high mobility situations. "n this paper we ha e in estigated the performance of #'() and (*+ 'n(emand 0reacti e3 routing protocol for performance comparison in the scenario. &he purpose of this work is to understand there working mechanism and in estigate that which routing protocol gi es better performance in which situation. &he rest of the paper is organized as follows. "n section ,! we ha e gi en the brief introduction of #'() and (*+ routing protocol. *ection = and ; deals with the simulation setup and results obtained on the e8ecution of simulation. 4inally! conclusion is drawn in section B. 3. Descri tion of Routing Protocol 3.1 Ad!"oc on Demand Distance #ector $A%D#&. &he #d-hoc 'ndemand (istance )ector routing protocol 91!=!1;: enables multihop routing between the participating mobile nodes wishing to establish and maintain an ad-hoc network. #'() is a reacti e protocol based upon the distance ector algorithm. &he algorithm uses different types of messages to disco er and maintain links. /hene er a node wants to try and find a route to another node it broadcasts a +oute +eCuest 0++%D3 to all its neighbors. &he ++%D propagates through the network until it reaches the destination or the node with a fresh enough route to the destination. &hen the route is made

a ailable by uncasing a ++%2 back to the source. &he algorithm uses hello messages 0a special ++%23 that are broadcasted periodically to the immediate neighbors. &hese hello messages are local ad ertisements for the continued presence of the node! and neighbors using routes through the broadcasting node will continue to mark the routes as alid. "f hello messages stop coming from a particular node! the neighbor can assume that the node has mo ed away and mark that link to the node as broken and notify the affected set of nodes by sending a link failure notification 0a special ++%23 to that set of nodes. 3.' D(namic )ource Routing $D)R& (*+ is a reacti e routing protocol i.e. determines the proper route only when packet needs to be forwarded 9;!9!11:. 4or restricting the bandwidth! the process to find a path is only e8ecuted when a path is reCuired by a node0'n-(emand +outing3. "n (*+ the sender 0source! initiator3 determines the whole path from the source to the destination node 0*ource-+outing3 and deposits the addresses of the intermediate nodes of the route in the packets. 1ompared to other reacti e routing protocols like #-+ or **#! (*+ is beacon-less which means that there are no hello-messages used between the nodes to notify their neighbors about their presence. (*+ was de eloped for M#$%&s with a small diameter between B and 1@ hops and the nodes should only mo e around at a moderate speed. (*+ is based on the 5ink*tate #lgorithms which mean that each node is capable to sa e the best way to a destination. #lso if a change appears in the network topology! then the whole network will get this information by flooding. &he (*+ protocol is composed of two main mechanisms that work together to allow disco ery and maintenance of source routes in M#$%&. Route Discover(* /hen a source node * wishes to send a packet to the destination node (! it obtains a route to (. &his is called +oute (isco ery. +oute (isco ery is used only when * attempts to send a packet to ( and has no information of a route to (. Route Maintenance* /hen there is a change in the network topology! the e8isting routes can no longer be used. "n such a scenario! the source * can use analternati e route to the destination (! if it knows one! or in oke +oute (isco ery. &his is called +oute Maintenance. +. )imulation )etu

,e -ave used Networ. )imulator $N)&!' in our e aluation. &he $*-, is a discrete e ent dri en simulator 9B!6: de eloped at E1 -erkeley. /e ha e used +ed 7at 5inu8 en ironment with ersion $*-,.=; of network simulator. $*-, is suitable for designing new protocols! comparing different protocols and traffic e aluations. "t is an ob<ect oriented simulation written in 1FF! with an '&cl interpreter as a front end. $* uses two languages because simulator got to deal with two things6 i3 detailed simulation of protocols which reCuire a system programming language which can efficiently manipulate bytes! packet headers and implement algorithms! ii3 research in ol ing slightly arying parameters or Cuickly e8ploring a number of scenarios. &he mo ement of nodes in the >roup Mobility model is generated by a software called Mobility >enerator which is based on a frame work called "mportant 0"mpact of Mobility 2atterns on +outing in #d-hoc $e&work! from Eni ersity of *outhern 1alifornia39G!1G!1H:."n the scenario we ha e considered four group with twel e node and one group leader in each. Table 1* )imulation Parameters
2arameters )alue

+outing 2rotocols #'() (*+ M#1 5ayer H@,.11 2acket *ize B1, bytes &errain *ize 1@@@m I 1@@@m $odes B@ Mobility Model >roup Mobility Model $o. of >roups ; (ata &raffic 1-+! &12 $o. of *ource 1@! ;@ *imulation &ime 9@@ sec. Ma8imum *peed @-6@ mAsec 0inter al of 1@3 /e ha e used four traffic patterns with arying number of sources for each type of traffic 0&12 and 1-+3. &he source-destination pair may be in same group or in different group. &he goal of our simulation is to e aluate the performance differences of these two on-demand routing protocols. &he type of traffic 01-+ and &123 and the ma8imum number of sources are generated by inbuilt tool of $*, 96:. &he parameters used for carrying out simulation are summarized in the table 1.

+.1 Performance Metrics +41,B@1 91,: describe a number of Cuantitati e metrics that can be used for e aluating the performance of M#$%& routing protocols. /e ha e used the following metrics for e aluating the performance of two on-demand reacti e routing protocols 0#'() J (*+36 +.1.1 2acket (eli ery 4raction
"t is the ratio of data packets deli ered to the destination to those generated by the sources. "t is calculated by di iding the number of packet recei ed by destination through the number packet originated from source. 2(4 K 02rA2s3I1@@

/here 2r is total 2acket recei ed J 2s is the total 2acket sent. +.1.' Routing %ver-ead "t is the total number of control or routing 0+&+3 packets generated by routing protocol during the simulation. #ll packets sent or forwarded at network layer is consider routing o erhead. ' erhead K $umber of +&+ packets +.1.3 Normali/ed Routing 0oad $umber of routing packets .transmitted. per data packet .deli ered. at destination. %ach hop-wise transmission of a routing is counted as one transmission. "t is the sum of all control packet sent by all node in network to disco er and maintain route. $+5 K +outing 2acketA+ecei ed 2ackets ;.1.; Average End!to!End Dela( $second& &his includes all possible delay caused by buffering during route disco ery latenc(1 2ueuing at t-e interface Cueue! retransmission delay at the M#1! propagation and transfer time. "t is defined as the time taken for a data packet to be transmitted across an M#$%& from source to destination. ( K 0&r .&s3

/here &r is recei e &ime and &s is sent &ime. +.1.3 Pac.et deliver( ratio* "n case of 1-+ traffic both protocols deli ers almost all originated data packets 0around 99-1@@L3 when mobility is low and number of sources is also low 01@3. -ut the packet deli ery fraction starts degrading gradually when there is increase in number of sources 0;@3 and with the increase in speed of nodes. (*+ perform better when number of sources is low! but when network load increases! packet deli ery ratio decreasing. #'() perform eCually under all assumed load condition in 1-+ traffic 0fig. ,3. -ut in case of &12 traffic! (*+ performs better irrespecti e of network load and speed 0fig. =3. ;.1.6+outing ' erhead6 4or 1-+ traffic! (*+ protocol ha e significantly low routing o erhead than #'() when the mobility is increased. /e ha e in estigated that! when number of sources is low 01@3! the performance of (*+ and #'() is similar regardless of mobility. 6. 1onclusions 4rom the figure , to 9! we conclude that in >roup mobility model with 1-+ traffic sources #'() perform better. -ut in case of &12 traffic! (*+ perform better in stressful situation 0high load or high mobility3. (*+ routing load is always less than #'() in all type of traffic. # erage end toend delay of #'() is less than (*+ in both type of traffic. ' er all the performance of #'() is better than (*+ in 1-+ traffic and real time deli ery of data. -ut (*+ perform better in &12 traffic under restriction of bandwidth condition. "n this paper! two routing protocol are used and their performance ha e been analyzed under >roup mobility model with respect to four performance metrics. &his paper can be enhanced by analyzing the other M#$%& routing protocols under different mobility model and different type of traffic sources with respect to other performance metrics.

References
91: *. (as! 1. %. 2erkins! %. +oyer! .#d 7oc 'n (emand (istance )ector 0#'()3 +outing.! "%&4 (raft! Mune ,@@,. 9,: 1-? &oh .#d 7oc Mobile /ireless $etworks 2rotocols and *ystems.! 4irst %dition! 2rentice 7all "nc! E*#! ,@@,. 9=: 1.%. 2erkins and %.M.+oyer! .#d-7oc 'n (emand (istance )ector +outing.! 2roceedings of the ,nd "%%% /orkshop on Mobile 1omputing *ystems and #pplications! $ew 'rleans! 5#! E*#! pages 9@-1@@! 4ebruary 1999. 9;: %lizabeth M. +oyer and 1hai-?eong &oh! .# +e iew of 1urrent +outing 2rotocols for #d 7oc Mobile /ireless $etworks.! "%%% 2ersonal 1ommunications! pages ;6-BB! #pril 1999. 9B: E1-A5-$5A)"$& $etwork *imulator! http6AAwwwmash. cs.berkeley.eduAnsA referred on March ,@1@. 96: .&he $etwork *imulator - ns-,!. a ailable at http6AAwww.isi.eduAnsnamAnsA referred on March ,@1@. 9G: 4an -ai! #hmed 7elmy .# 4ramework to systematically analyze the "mpact of Mobility on 2erformance of +outing 2rotocols for #d hoc $etworks.! "%%% "$4'1'M ,@@=. 45: (. Mohnson! (a e Maltz! N 7u! Mor<eta Metche a! .&he (ynamic *ource +outing 2rotocol for Mobile #d 7oc $etworks.! "nternet (raft! 4ebruary ,@@,. 99: *uresh ?umar! +.?. +athy and (iwakar 2andey! .&raffic 2attern -ased 2erformance 1omparison of &wo +eacti e +outing 2rotocols for #d-hoc $etworks using $*,.! ,nd "%%% "nternational 1onference on 1omputer *cience and "nformation &echnology! ,@@9. 416: (. Mohnson! N. 7u! and (. Maltz! .&he (ynamic *ource +outing 2rotocol 0(*+3 for Mobile.! +41 ;G,H! 4eb ,@@G.

4117 ). R. 8iradar1 "iren " D )-arma1 9al ana )-rama and )ubir 9umar )ar.ar1 :Performance ;om arison of Reactive Routing Protocols of MANETs using <rou Mobilit( Model:1 IEEE International ;onference on )ignal Processing )(stems1 ages 1='!1=3 '66=. 41'7 ;. Per.ins1 E. 8elding!Ro(er1 ). Das1 2uet1 :Ad -oc %n!Demand Distance #ector $A%D#& Routing:1 R>; 33?11 @ul( '663. 4137 A. "ong1 M. <erla1 <. Pei1 and ;.!;. ;-iang1 : A grou mobilit( model for ad -oc wireless networ.s1: in ACM/IEEE MSWiM1 August 1===. 41+7 -tt *BBwww!scf.usc.eduBCfbaiBim ortantB1 referred on >ebruar( '616.

4137 -tt *BBnile.usc.eduBim ortantB1 referred on >ebruar( '616.