VSR-多sink机会移动传感器网络的数据收集

ISSN10001239CN111777TPJournalofComputerResearchandDevelopment47(8):1450-1458,2010:2009-07-03;:2009-12-03:(60933011);(2008AA01Z217);(60673178,60873241)VSR:sink熊永平1,2孙利民3马建4牛建伟51(100190)2(100190)3(100190)4(100013)5(100191)([e-mail])VSR:DataGatheringinOpportunisticMobileSensorNetworkwithMultipleSinksXiongYongping1,2,SunLimin3,MaJian4,andNiuJianwei51(InstituteofComputingTechnology,ChineseAcademyofSciences,Beijing100190)2(GraduateUniversityofChineseAcademyofSciences,Beijing100190)3(InstituteofSoftware,ChineseAcademyofSciences,Beijing100190)4(NokiaResearchCenter,Beijing100013)5(SchoolofComputerScienceandEngineering,BeihangUniversity,Beijing100191)AbstractOpportunisticmobilesensornetworkscanbeappliedinsomescenariossuchaswildanimalmonitoringandurbansensingutilizingsensorsembeddedintohandhelddevicestocollecturbaninformation.Intheseapplications,thegathereddatausuallyneedtobetransmittedfromthesourcenodetooneofmultiplebasestations(sinknodes).TheauthorsproposetheVSR(virtualspacebasedrouting)schemeadaptingstorecarryforwardparadigmtotransmitthedatamessagetothebasestations.InVSR,eachsensorestimatestheexpecteddeliverydelaytoallbasestationsbasedonthemeetinghistorybetweenthenodeandthebasestations.Then,eachsensorismappedintoacoordinatepointofahighdimensionalspaceaccordingtoitsdeliverydelaystoallsinks.Allsinknodesarecorrespondingtotheoriginofthespace.TheforwardingmetricisdefinedastheEuclideandistanceofthenodetotheorigin.Whentwonodesencounter,thenodewiththehighersuchmetricforwardsthecarriedmessagestothepeerwiththelowermetric,untilthemessagesaredeliveredtoanysinknodes.VSRisrobusttodynamicnetworkbecauseofitsfinegrainedforwardingdecisionandisappropriateforsensornodeduetoitslowcomputingandstorageoverhead.ExperimentsundertworandomscenariosshowthatVSRoutperformsthehistorybasedroutingproposedinZebraNetandrandomforwardingscheme.Keywordsvirtualspace;datagathering;opportunisticforwarding;mobilesensornetwork;delaydisruptiontolerantnetwork机会移动传感器网络可应用在野生动物监控,或利用手持设备嵌入的传感器收集城市信息等场景,往往需要将数据从源节点传输到多个基站中的任一个.提出了一个基于虚拟空间的路由机制VSR(virtualspacebasedrouting),采用存储携带转发的传输模式实现数据收集.每个传感器节点根据与多个sink节点的期望传输延迟映射成高维空间中的一个坐标点,消息传输对应于从源节点移动到空间原点的过程.细粒度的转发决策特性,使VSR自适应于网络的动态变化,具有很好的鲁棒性.此外,VSR机制具有很低的计算和存储开销,非常适合资源受限的传感器节点.两种不同随机特性场景下的模拟实验验证了VSR机制比ZebraNet的基于历史的转发机制和随机转发机制的性能更好.虚拟空间;数据收集;机会转发;移动传感器网络;容迟容断网络TP393;TN915.410,ZebraNet[1]TurtleNet[2],(urbansensing)[3]Cartel[4],,,,,,1.,sink,,(opportunisticmobilesensornetworks,OMSN)[5].Fig.1OMSNillustration.1OMSN,sink.,OMSN,,,,sink.,.,sink,,,AP.sinkOMSN,(virtualspacebasedrouting,VSR)sink.sink,sink,.sink,sink0,,,sink.VSR,.VSRONE,,ZebraNet.11.1(delaydisruptiontolerantnetworks,DTN)[6],[78].[9],.ZebraNet[1]1451:VSR:sink:(),,,.,.Haas[10]flooding,,sink,sink.Wang[11]DFTMSN(delayfaulttolerantmobilesensornetwork),ZebraNet,,..Mascolo[12]SCAR(sensorcontextawarerouting).SCAR(),,,.Shah[13]DataMule,Muleagent,,.,.Su[14],(ondemandminimumlatencyrouting,ODML),,.ODMLoptimalPML,ODMLquickPML,.1.2Ratnasamy[15]P2PCAN,,,.Leguay[16]DTNMobySpace,,,,,Canberra..[17]MobySpace,,.2VSR,,,,VSR,VSR.2.1msink,,,,.sink,iMi,,,sink.,,.,,,,,.,i,Ni,()x!{i}∀Nih,R.,.h=argmaxx!{i}∀NiR(x).(1)2.2sink.sink(),0,.,Li(j)isinkj,t2,,DC,i14522010,47(8)sinkjW(t)3.,(2),nt.Li(j)=#t01tW(t)dt=1t(W(t))=∃nk=112DC2kt=12t∃nk=1DC2k.(2),sink.,.w,,Lwi(j),,,Lwi(j)[Li(j)],![0,1].Li(j)=%[Li(j)]+(1-)%Lwi(j).(3)2.3,sink,jsinkj,sink.m{L1,L2,&,Lm},sink0,.,,{Li(1),Li(2),&,Li(m)}i,Mi:Mi=∃mk=1L2i(k),(4),sink,sink,.,,sink.,sink.,3sink4,,sink,1,1,3,2,4sink.Fig.4VirtualspaceinanOMSNwith3sinks.43sinkOMSNsink,,Mi=min{Lk(k)|1∋k∋m}.sink,,Mi,.AsinkB,AB,,Asink,,AB,,,,.sink,,1453:VSR:sinksink,,,sink,sink,sink,.2.4,VSR(virtualspacebasedrouting).,MACbeaconbeacon.,i,j,iVSR,5:Fig.5VSRalgorithm.5VSRsinkj,iDCDC,sink;,,sinksink,Mi,,.FIFO,,,,.VSR,sink,O(m),msink,,O(1)..3VSR,,,.3.19010sink3000m%3000m.trace,ZebraNettrace,.LocalizedRandomWalk(LRW).200m%200m,home,home.3ms([0,6ms]).,,,.xP(x)=e-dx2,dxxhome,,,,home,,,.3.2ONE(opportunisticnetworkenvironment)[18]VSR,ZebraNet(Zebra),Zebra,,.0,sink1,Dsink1,sink,;,.,5050B,100.20000s,150m.,14542010,47(8)2000s,0.4.,5,.3.3VSR,.3.3.1LRW100,,sink.67.,,sink,3,sink,.,VSRZebra,VSRsink,,VSRZebra.VSRZebra,LRW100,Zebrasink.(),89.8,3,VSR.,,,,,.9,,,,,,Random,VSRZebra.1455:VSR:sink3.3.2LRW1000,sink.10,,sink,200m,390%.,,165m,ZebraRandom.VSRsink,,VSR.11,3,,RandomVSRZebra.12.VSRsink,,.,Zebra,110,Random.13,,,,VSR.,VSR,ZebraRandom.4,sinkVSR.,.VSR,sinksink,sink,,sink.,VSR.14562010,47(8)VSRZebraNet,VSR,.VSR,.trace,.[1]JuangP,OkiH,WangY,etal.Energyefficientcomputingforwildlifetracking:DesigntradeoffsandearlyexperienceswithZebraNet[J].SIGARCHComputerArchitectureNews,2002,30(5):96-107[2]CornerMD,BergerED.TurtleNet.(20090302)[20090821].http:prisms.cs.umass.edudometurtlenet[3]CampbellA,EisenmanS,LaneN,etal.Peoplecentricurbansensing[C]Procofthe2ndAnnualIntWirelessInternetConf(WICON).LosAlamitos,CA:IEEEComputerSociety,2006:2-5[4]HullB,BychkovskyV,ZhangY,etal.CarTel:Adistributedmobilesensorcomputingsystem[C]Procofthe4thIntConfonEmbeddedNetworkedSensorSystems.NewYork:ACM,2006:125-138[5]P