the-ONE-simulation机会网络仿真工具教程及帮助

OpportunisticNetworkEnvironmentsimulatorAriKeranenari.keranen@tkk.HelsinkiUniversityofTechnologyDepartmentofCommunicationsandNetworkingSpecialassignmentSupervisor:ProfessorJorgOttMay29,2008Contents1Introduction12Background33TheOpportunisticNetworkEnvironmentsimulator63.1Mobilitymodeling........................73.2Routingsimulation........................93.3Externaleventandreportingframeworks...........113.4Runningsimulations.......................123.4.1GraphicalUserInterfacemode.............123.4.2Batchmode........................143.4.3Post-processing......................153.4.4Conguringscenarios..................163.5Softwarearchitecture.......................173.5.1Movementmodels....................183.5.2Routingmodules.....................193.5.3Extendingthesimulator.................213.6Limitations............................233.7Results...............................244Conclusions25ATheONEReadMeIIAcronymsAPIApplicationProgrammingInterfaceASCIIAmericanStandardCodeforInformationInterchangeDTNDelayTolerantNetworkingGISGeographicInformationSystemGPLGNUGeneralPublicLicenseGPSGlobalPositioningSystemGUIGraphicalUserInterfaceMACMediaAccessControlMANETMobileAd-hocNETworkONEOpportunisticNetworkEnvironmentsimulatorPOIPointOfInterestRAMRandomAccessMemoryRWPRandomWayPointTTLTimeToLiveWKTWellKnownTextWLANWirelessLocalAreaNetworkIIChapter1IntroductionTodaymanypersonalmobiledevicesincludecapabilitiestocommunicatewithinfrastructurenetworksbutalsowitheachother.Thelattercanbeusedtoformad-hocnetworkswherecommoninfrastructureisnolongerneededforcommunicationamonghostsparticipatinginthenetwork.Ad-hocnetworkscanalsohelpmobilenodestoreachinfrastructureifsomenodeinthenetworkisableandwillingtoactasagatewayandpossiblyothernodesasrelaysforthetrac.Networkscanbeformedthiswayaslongasthenodedensityislargeenoughsothatthereexistspossibleend-to-endpathsbetweenallnodeswantingtocommunicate.However,ifthenodedensitydecreasesortheconnectivitybreaksforsomeotherreason(e.g.,theradiosareswitchedooccasionally),traditionalnetworkcommunicationprotocolsarenolongerabletoprovidemeansformulti-hopcommunication.DelayTolerantNetworking(DTN)[10,4]isacommunicationnetworkingparadigmthatenablescommunicationinenvironmentswheretheremaybenoend-to-endpaths,communicationopportunitiescomeandgoandtheirintervalcanbeverylongandnotevenknownbeforehand.Routingmessagesinthiskindofenvironmentscanbequitedierentcomparedtotraditionalnetworks.Thishascreatedaneedtondnewroutingprotocolsthattakeecientlyintoaccountthedistinctnatureofthesenetworks.Dierentap-proachescanbetestedandevaluatedbysimulation,butthesimulationresultsarereallyusefulonlyiftheyarearesultofsomewhatcrediblesimu-lationscenarios.1ThegoalofthisstudyistoaddmorerealismtothesimulationsofDelayTolerantNetworks.Forthispurposewecreatedanewsimulationenviron-mentcalledOpportunisticNetworkEnvironmentsimulator(ONE).UnlikeotherDTNsimulators,whichusuallyfocusonlyonroutingsimulation,theONEcombinesmobilitymodeling,DTNroutingandvisualizationinonepackagethatiseasilyextensibleandprovidesarichsetofreportingandanalyzingmodules.2Chapter2BackgroundOnehindranceforresearchonDTNsisthelackofgoodsimulators.ManysimulatorsexistforMANETs(e.g.,ns2[19]andOMNeT++[27])andalsoforDTNrouting(e.g.,dtnsim[13]anddtnsim2[20])buttheformerlackgoodDTNsupportandthelatterconcentratesolelyonroutingsimulation.TheDTNroutingsimulatorshavealsoseenlittleevolutionsincetheirrstpublicreleaseandimplementationsofrecentroutingprotocolsdon'texistorarenotpubliclyavailable.Anotherproblemforasimulatorthatconsidersonlyroutingisthatitneedsinputdatathattellstheroutingprotocolswhenanetworklinkbe-tweentwoDTNnodesisupandwhenitisdown.Thisdatacanbegen-eratedbasedonsomerandomprocess(e.g.,justdrawingcontactdurationsandendpointsfromapseudorandomnumbergenerator),itcanbederivedfromreal-worldtrace(e.g.,usingCRAWDAD[5]data)oritcanbederivedfrommobilitysimulation.Theproblemwiththerstapproachisthatval-uesfromsomerandomdistributionarehardtoprovetohavesomevaliditybehindthem.Forexample,evenifhumaninteractionsmayseemsome-whatrandom,humansusuallydohavesomepurposefortheiractionsandformeetingcertainpeopleandvisitingsomeplacesmoreoftenthanothers.Capturingthisinasimple,orevencomplex,randomnumberdistributionwouldbechallenging,tosaytheleast.Thereal-worldtracesobviouslycapturerealhumanbehaviorwellbuttheyhaveotherkindsofproblems.Thenumberofusabletraceshasbeen3loweventhoughtheCRAWDADprojectisbringingsomereliefintothis.Unfortunately,theexistingtraceshavelowspatialandtemporalgranularity.Tosavebatterylifeofthemobiledevicesthatareusedtotrackpeople,theintervalthedevicesscanforothershasbeenkeptlow[7].Thisresultsinmissingpossiblecontactsbutalsointhatwedon'tknowhowlongtheactualcontactslasted|wejustknowhowmanyscanningintervalstheywereincontactwitheachother.Thisinformationcanbeofcourseusedformanysimulations,butifweareinterestedinutilizingeventheshortestcontacts,orwanttoexperimentwithdierentscanningintervals,thesetracesarenotthatusefulanymore.Ifwehadtracesofthenodes'exactlocations,wecouldderivecontactti