Packet Loss in Mobile Ad Hoc Networks(4)

Abstract — We investigate packet loss in mobile ad hoc networks via simulation. Ad-hoc on-demand distance vector (AODV) and destination-sequenced distance vector (DSDV) are chosen as representatives of the on-demand and proactive routing protocols respect

Thewirelesschannelissobusythatthetimesofbackoffexceedthelimit.

Thechannelisassociatedwithaqueuethatbuffersallthepacketswaitingtobesent.Whenthequeueisfull,anycomingpacketisdropped.

IV.EXPERIMENTS

Aseriesofexperimentshavebeenconductedtoin-vestigatemobility-relatedandcongestion-relatedpacketlossesindifferentnetworkcontexts.Thenetworkcon g-urationfortheexperimentsisa1000mx1000msquare eldwith30hosts.Thebuffersizeis64-packetforeachrouteandMAClayer.Eachdatapointintheresult guresrepresentsanaverageof5runswithidenticaltraf cbutdifferentmobilityscenarios,whicharerandomlygeneratedwiththesameparameters(i.e.,samemaximumspeedandpausetime).Everyexperimentrunsforatleast1000seconds.

A.VaryingMobilityandCommunicationRequestThepurposeofthe rstsetofexperimentsistostudytheimpactofhostmobility.Pausetimeisvariedovertherangeof{0,50,100,200,300,500}seconds.Zeropausetimeresultsinthehighestmobilitysincehostskeepmovingwithoutapause.Fortheseexperiments,10,20,and30connections,whichrepresentlight,moderate,andheavycommunicationrequestsrespectively2,areused.Thepacketsendingrateforeachconnectionis4packets/s.Theresultsareshownin gure1.

1)PacketLossforAODV:Totalpacketlossgrowsfromabout3000to8000withtheincreaseofpausetimefrom0to500secondsfor10connections,asshownin gure1a.Incasethereare20connections,totalpacketlossgraduallyincreasesby10%( gure1b).For30connections,itgraduallydecreasesby10%( gure1c).Asthecommunicationrequestgrowsfrom10to20,totalpacketlossincreasesby9timeswhenpausetimeis0seconds,andby3timeswhenpausetimeis500seconds.Theincreaseofcommunicationrequestfrom20to30resultsindoubledtotalpacketloss.

Thereisalmostnocongestion-relatedpacketlosswhenthecommunicationrequestis10.Intheothertwocases,packetlossgraduallydecreasesbyaboutahalfaspausetimeincreasesfrom0to500seconds.From10to20and30connections,withnopausetime,packetlossincreasesto5000and20000respectively.Thepercentagewithrespecttototallossincreasesaswell,to20%and30%respectively.

2

Traf cloadisrepresentedbythesendingrateinthispaper.Ithasdifferenteffectonpacketlosscomparedwithcommunicationrequest.

Fig.1.Packetlossfor4packets/sCBRconnections

Mobilityisalwaysthedominantcauseforpacketloss.However,themajoritydecreasesasthecommunicationrequestincreases.Whenpausetimeis0seconds,thepercentageofmobility-relatedlossdecreasesfromabout100%to70%and60%,given10,20,and30connec-tions.Theabsolutevalueandthepercentageofmobility-relatedpacketlossincreasewithpausetime.

2)PacketLossforDSDV:ThegrowthoftotalpacketlosswithpausetimeforDSDVfollowsasimilarpatternasthatforAODV.For10connections,totalpacketlossincreasedfromabout3000to10000aspausetimeincreasesfrom0to500seconds( gure1d).Itisnearlyunchangedwithpausetimefor20and30connectionsasshownin gure1eand1f(graduallyincreasesby5%for20connectionsanddecreasesby5%for30connections).Increasingcommunicationrequestfrom10to20resultsintotalpacketlossgrows10timesand4timesfor0and500secondspausetime,respectively.Theincreaseofcommunicationrequestfrom20to30,however,onlydoublesthetotalpacketloss.

Thepercentageofcongestion-relatedpacketlossin-creaseswithcommunicationrequest.Congestionbeginstobethedominantcauseforpacketlossaftercommuni-cationrequestreaches20(itresultsinapproximate50%and60%totalpacketlosswith20and30connections,respectively).Thelossisfairlystablewithpausetime,butjittersexist.

Mobility-relatedpacketlossincreaseswithcommuni-cationrequest,butslowerthancongestion-relatedpacketlossdoes.