gps multipath pattern recognitionenvironment.docx
- 文档编号:25456396
- 上传时间:2023-06-08
- 格式:DOCX
- 页数:12
- 大小:237.42KB
gps multipath pattern recognitionenvironment.docx
《gps multipath pattern recognitionenvironment.docx》由会员分享,可在线阅读,更多相关《gps multipath pattern recognitionenvironment.docx(12页珍藏版)》请在冰豆网上搜索。
gpsmultipathpatternrecognitionenvironment
PatternRecognition一BasedEnvironmentIdentificationforRobustWirelessDevicesPositioning
NesreenI.Ziedan
ComputerandSystemsEngineeringDepartment
FacultyofEngineering,ZagazigUniversity
Zagazig,Egypt
Abstract--Therehasbeenacontinuousincreaseinthedemands
forGlobalNavigationSatelliteSystem(GNSS)receiversinawiderangeofapplications.Moreandmorewirelessandmobiledevicesareequippedwithbuilt-inGNSSreceivers;theirusers'mobilitybehaviorcanresultinchallengingsignalconditionsthathavedetrimentaleffectsonthereceivers'trackingandpositioningaccuracy.Amajorerrorsourceisthemultipathsignals,whicharesignalsthatarereflectedoffdifferentsurfacesandpropagatedtothereceiver'santennaviadifferentpaths.Analysisofthereceivedmultipathsignalsindicatedthattheircharacteristicsdependonthesurroundingenvironment.Thispaperintroducesamachine-learningpatternrecognitionalgorithmthatutilizestheaforementioneddependencytoclassifythemultipathsignals'characteristicsandidentifythesurroundingenvironment.TheidentifiedenvironmentisutilizedinanoveladaptivetrackingtechniquethatenablesaGNSSreceivertochangeitstrackingstrategytobestsuitthecurrentsignalcondition.Thiswillleadtoarobustpositioningunderchallengingsignalconditions.ThealgorithmisverifiedusingrealandsimulatedGlobalPositioningSystem(GPS)signalswithaccuratemultipathmodels.
Keywords-component;GPS;GNSS;machinelearning;patternrecognition;PCA;PNN;multipath.
I.INTRODuCTTON
AGlobalNavigationSatelliteSystem(GNSS)[l,2]isaradionavigationsystemthatemploysspreadspectrumtechniquestotransmitrangingsignalsandnavigationdata.TherangingsignalsareusedbyaGNSSreceivertoidentifythevisibleGNSSsatellitesandmeasurethedistancebetweenthevisibleGNSSsatellitesandtheGNSSreceiver.Themeasureddistancesareusedwiththenavigationdatatosolvethenavigationequationtodeterminetheuser's3-diemntionalpositionandvelocity.ExamplesofGNSSsystemsaretheUSGlobalPositioningSystem(GPS),theRussianGLONASS,and
theEuropeanGalileoNavigationSystem.GNSSreceiversperformthreemainfunctions:
signalacquisition,signaltracking,andnavigationmessagedecoding.Signalacquisitionidentifiesthevisiblesatellitesandprovides
roughestimatesoftheDopplerfrequency,fa,andtheranging
codedelay,i.Signaltrackingappliesclosedlooptracking
techniquestoprovidecontinuousaccurateestimatesofthe
carrierphase,theDopplerfrequency,theDopplerrate,andthe
codedelay.ThoseestimatesareusedtomeasurethedistancebetweentheGNSSsatelliteandthereceiver.
GNSSreceiverscangivepositioningaccuracyuptoafewmillimeterswhenthereceiverisstableandhasaclearviewofthesky,wheretheLine-of-Sight(LOS)signalisreceivedwithstrongpower.However,inenvironmentslikeurban,suburban,andindoor,thereceivedsignalssufferfromattenuationandmultipatherrorsbecauseofthesurroundingobjects[3].Inaddition,theuser'smobilitybehaviorcansubjectthereceivertochangingandunstablesignalsdynamics.Thisleadstodeteriorationinthetrackingandpositioningaccuracy.
Multipathsignalsareamajorerrorsource.TheyappearwhentheGNSSsatellitesignalsarereflectedoffdifferentsurfacesandpropagatedtothereceiver'santennaviadifferentpaths.Thisleadstothereceptionofseveralversionsofthesamesignal,whichcausestrackingerrors.Analysisofthereceivedsignalsindicatedthattheircharacteristicsdependonthesurroundingenvironment[4,5,6,7,8].Urban,sub-urbanandindoorenvironmentsgeneratedifferentcharacteristics,whichincludemultipathsignals'parameterslikethenumberanddurationofechoesandsignalspower,andLOSsignal's
parameterslikeamplitudefluctuation,Dopplershiftandrate.Differenttrackingstrategiesareneededforeachenvironmenttomitigatemultipatherrorsandmaximizethetrackingperformance.
Therehavebeennumeroustrackingstrategiesthatareoptimizedforspecificsignalconditionorenvironment.Forexample,conventionaltrackingtechniques[l,2]areusedwithstrongsignals.KalmanFilterbasedtechniquesareusedwithweaksignals[3,9].Tightly-coupledGNSSwithInertialNavigationSystem(INS)techniquesareusedwithweakinterruptedsignalsorblockedsignals[10].Open-loopbatchprocessing,andcombinedbatchandsequentialprocessing
techniquesareusedinhighdynamicapplications[1l].Particle
Filter-basedtechniquesareusedfortrackinginmultipathenvironments【12,13,14】.
AGNSSreceiverisusuallytunedtoonetrackingtechnique,andtherehavebeennomethodsthatenableareceivertochangeitstrackingstrategybasedonthesurroundingenvironment.Thispaperintroducesamachine-learningpatternrecognitionalgorithmtoidentifythesurroundingenvironment,andhenceenabletheimplementation
ofatackingstrategyselectorthatadaptivelychangesthetrackingstrategytobestsuitthecurrentsignalcondition.
TheLOSandmultipathsignals'patternsofeachpossibleenvironmentarerepresentedbyaclass.Theintroduced
algorithmisstructuredintoseveralchannels,eachofwhichis
tunedtooneoftheclasses.Thechannelsaretrainedonsetsof
patternsfromeachclass,andthentheyareusedtoclassifynew
unclassifiedpatterns.Theproposedpatternrecognition
algorithmperformstwomainfunctions,whicharefeature
extractionandpatternclassification.Featureextractionisthe
processoflearningthedistinctivecharacteristicsofthedata
andremovingredundantdata.Thisisdonetogetacompact
androbustrepresentationofthedistinctivefeaturesofeach
class,thusreducingtheprocessingoverheadandmemory
requirementswithoutdegradingtheclassificationperformance.
Featureextraction,whichisusedinimageandfacerecognition
[15,16,17,18,19,20,21,22,23],isperformedusinga
PrincipalComponentAnalysis(PCA)approach.Pattern
classificationistheprocessofbuildingneuronsthatcapturethe
dominantfeaturessharedbydifferentrealizationsofeachclass,
andthenclassifyingnewpatternsintooneoftheclasses.
Neuronsarecomputationalunitsthatareconnectedby
weightedlinks.Patternclassificationhasmanyapplications,
likeradardetectionandremotesensing[24,25,26];itis
performedhereusingamulti-layerfeedforwardProbabilistic
NeuralNetwork(PNN)approach[27,28,29,30,31,32].
Theproposedtrackingstrategyselectormoduleutilizes
bothclosedlooptrackingtechniquesandopenlooptracking
techniquestoaccommodatevariouspatterns.Openloop
trackingtechniquesareactivatedinunstableorrapidly
changingsignalconditions,whileclosedlooptracking
techniquesareactivatedinrelativelystableorlightmultipath
environments.Inaddition,basedontheenvironment
classification,theactivatedtechniqueadjustsitsparametersto
achievereliabletrackingperformance.
Therestofthispaperisorganizedasfollows.SectionII
presentssomesignalspatternsthatappearinurbanand
suburbanenvironments.SectionIIIpresentstheproposed
patternrecognitionalgorithm.SectionIVdiscussesthe
adaptivetrackingconcept.SectionVpresentssometestingand
resultstoverifythealgorithmperformance.SectionVI
concludesthepaper.
2、LOSANDMULTIPATHPATTERNS
ThereceivedGNSSsignalconsistsoftheLOSsignaland
(Nmp)multipathsignals.Thedown-convertedsampledreceived
GPSC/Asignalcanbeexpressedas
Where,Aisthesignalamplitude.disthenavigationCistherangingcode.f(IF)istheintermediatefrequency(IF).fdistheDopplershift.aistheDopplerrate.thetanisthephase.taoisthecodedelay.nisthemeasurementnoise.pozaimistheattenuationinthemultipathsignalamplituderelativetothe
LOSsignalamplitude.taomisthemultipathsignaldelay.faimis
themultipathphaseadvancerelativetotheLOSsignal.
Thereceivedsignalisprocessedbythereceivertogenerate
theintegratedsignal,whichhastheform
Where,Aisareferenceamplitudethatisusedtoexpressanyamplituderelativetoit,e.g.theLOSamplitudeisA=gama0*daov,gamam=gama0*pozaim.R(.)istheauto-correlationfunction.taocuisthecodedelayestimationerrorattimeinstanceu.feistheDopplershiftestimationerror.thetacuisthephaseestimationerror.△isacodedelayrelativetotheestimatedcodedelayoftheLOSsignal.Tiistheintegrationtime.
TheattenuationoftheLOSsignalandthenumberanddistributionofthemultipathsignalsdependonthesurroundingenvironment.Elaboratestudiesexploringtheeffectsofurbanandsuburbanenvironmentsonrealsignalswerepresentedin[4,5,6,7,8].IdentifyingthedistributionoftheLOSandmultipathsignalswillenableadjustingthetrackingstrategyorthetrackingparameterstoobtainthebestattainabletrackingperformanceundervarioussignalconditions.Thesoftwareprovidedin[33]isusedtogeneratereceivedsignalpatternsthattypicallyappearinurbanandsuburbanenvironments.SomeofthesepatternsareshowninFigs.1-3.Eachfigureshowstwoplots:
a3-dimensional(3-D)plotthatexpressesthe
patternintime,delay,andpower;anda2-Dplotthatisarotatedversionofthe3-Dplot.
Fig.1showsapatterngeneratedinasuburbanenvironmentwhenauseriswalkingataspeedof4miles/hour.ThepatternexhibitsastrongLOSsignalwithlightmultipathsignalsthathaveweakpowercomparedtotheLOSsignal.Fig.2showsapatterngeneratedinasuburbanenvironmentwhenacarismovingataspeedof20miles/hour.TheLOSsignalhereisweakerthantheLOSsignalshowninFig.1,andt
- 配套讲稿:
如PPT文件的首页显示word图标,表示该PPT已包含配套word讲稿。双击word图标可打开word文档。
- 特殊限制:
部分文档作品中含有的国旗、国徽等图片,仅作为作品整体效果示例展示,禁止商用。设计者仅对作品中独创性部分享有著作权。
- 关 键 词:
- gps multipath pattern recognitionenvironment