基于多数据源的室内定位方法分析-软件工程专业论文.docx

AbstractNowadays,withthequickdevelopmentindataservicesandmultimediaservices,thedemand fornavigationandlocalizationisbecomingmoreandmore.Specificallyinsomecomplexindoor room-levelenvironments,suchasthecompany,hospital,tearoom,supermarket,market, museum,cinemaandanyotherenvironments,peopleoftenneedtoabtainthelocation information of the mobile terminalor their holders, facilities or thethingsintheroom.Dependingontheindoorandoutdoordifferentpositioningenvironments,nowadayswireless locationtechnologycanbedividedintoindoorpositioningtechnologyandoutdoorpositioning technology.Indoorpositioningtechnologynamesuggests,referstoawaytogettheobjectsin indoorlocationtechnology.Ithasabroadapplicationfutureinmilitaryandcivildomains.The knownlocalizationtechnologymainincludewirelesslocalareanetworks,lighttrackingsenser networkpositioningtechnology,infraredraypositioningtechnology,radiofrequency identificationpositioningtechnology(RFID).Therefore,sensortechnologyandwirelesslocal areanetworkstechnologywiththeirownadvantagescanbeappliedtoindoorpositioning method.Thethesiscombinessensornetworksandwirelesslocalareanetwork(WLAN)tolocalize andtrackthesingletargetindoor.First,thethesisdescribesthebackgroundofusingsensor networksandWLANtechnologyforindoorlocation,andthendiscussesthereceivedsignal strengthdatafortheconductofactivities,usinglightintensitysensorandwirelessLANinthe indoor environment. Specifically, we use the light intensity sensor to abtain the initial positioning ofthepersonintheroom,andthenusethewirelessLANreceivedsignalstrengthdataforthe precisepositioning.Amongmanymachinelearningalgorithms,theneuralnetworkhas advantagessuchasbeingadeptinresolvingnonlinearmappingproblemsandprecision adjustableThisthesisusesthemethodforpersonstoobtainlocationinformationbasedonthe receivedsignalstrengthandtheBPneuralnetwork.Weproposeanindoorlocalizationmodel fromthree-layerBPneuralnetworkwhichhasoneinputlayeronehiddenlayerandoneoutput layer.TheBPneuralnetworkusesignalintensityasitsinputvectoran