2011版-ofdm for underwater acoustic communication

OFDMforUnderwaterAcousticCommunicationArjunThottappillyThesissubmittedtotheFacultyoftheVirginiaPolytechnicInstituteandStateUniversityinpartialfulfillmentoftherequirementsforthedegreeofMasterofScienceinElectricalEngineeringA.A.(Louis)Beex,ChairJeffreyH.ReedStevenW.EllingsonAugust17th,2011Blacksburg,VirginiaKeywords:OrthogonalFrequencyDivisionMultiplexing,UnderwaterAcousticCommunication,WidebandDopplerCorrection,TimeWarpOFDMforUnderwaterAcousticCommunicationArjunThottappillyABSTRACTCommunicatingwirelesslyunderwaterhasbeenanareaofinterestforresearchers,engineers,andpractitionersalike.Oneofthemainreasonsfortheslowrateofprogressinthisareaisthattheunderwateracousticchannelisingeneralmuchmorehostileintermsofmultipath,frequencyselectivity,noise,andtheDopplereffectthantheover-the-airradiofrequencychannel.Inthisworkatimewarpbasedtechniquewhichcanbeusedtomodeltime-varyingwidebandDopplershifts(asseeninanUWAchannel)inMATLABisproposed.Acorrespondingproceduretoestimatetheparametersfromobserveddata,requiredforinvertingtheeffectofthetimewarp,isalsoproposed.TwodifferentDopplercorrectionmethodsarecompared;bothcanbeusedtoundotheDopplereffectinmeasureddatafromanexperimentsubjecttothewidebandDopplereffect.ThetechniquespresentedcorrectforthewidebandDopplereffectasifitchangedthetimescaleofthereceivedsignal.Thefirstresamplingbasedtechniquecorrectsfortheaverageexpansion/contractionoverapacket,inherentlyassumingtherelativevelocitytobeconstantoverthedurationofthepacket.Thesecondtimewarpbasedtechniquemodelstime-varyingDopplershift.Sinusoids,addedtothebeginningandendofeachpacket,areusedtoestimatetheparametersrequiredtoinverttheeffectofthewarp.ThetimewarpbasedmethodsaredemonstratedusingOrthogonalFrequencyDivisionMultiplexing(OFDM)signals,butwillinprincipleworkforotherkindsofwidebandsignalsalso.ThepresentedresultsusingMATLABbasedsimulations,andover-the-airexperimentsperformedinsuchawayastointroducetheDopplereffectinthereceivedsignalsemphasizetheimprovementsthatcanbeattainedbyusingthetimewarpbasedDopplermodelingandcorrectionmethod.Thethesisconcludeswithsuggestionsforfuturework.iiiAcknowledgementsFirstandforemost,IwouldliketothankmyadvisorDr.A.A.(Louis)BeexforgivingmetheopportunitytoworkunderhimintheDSPResearchLaboratoryatVirginiaTech.ThetimeIspentworkingunderhisguidancehasbeenmostproductiveperiodofmyacademiclifesofar,andIindeedlearnedalotmorethanjustDSPfromhim.Iwouldalsoliketothankhimforallhisinsightfulinstructionsatvariouspointsofthiswork.Noneofthiswouldhavebeenpossiblewithouthishelp.IwouldliketoextendmygratitudetoDrs.StevenEllingsonandJeffreyReedforbeingapartofmycommitteeandreviewingthework.SpecialthankstomyfriendGauthamforallhisexcellentsuggestionswhilereviewingthework.IwouldliketothankmylabmatesAmy,AregandRoshin.WorkingintheDSPRLwouldnthavebeenasmuchfunwithoutthemaround.IwouldalsoliketothankKrishnan,BharatandallmyfriendswhomadeBlacksburghomeawayfromhome.Wordscannotexpressmyloveandgratitudetomyparentsfortheirunconditionalloveandsupport.Theirdedicationandinteresttowardsteachingandresearchhasalwaysinspiredmetoputinmybesteffortintoallmyendeavors.iv1TableofContentsIntroduction.11.11.2Background.1TheExperiment.42OrthogonalFrequencyDivisionMultiplexing.2.42.5Introduction.6GenerationofOrthogonalSubcarriers.8GuardInterval.9ChoiceofOFDMParameters.12SynchronizationinOFDMSystems...4EffectsofSTO.14EffectsofCFO.16MethodsforAchievingTimingSynchronization.17TechniquestoEstimateandCorrectforCFO.222.6PeaktoAveragePowerRatioinMulticarrierTransmission.252.6.1PAPRReductionTechniquesforMulticarrierTransmission.273UnderwaterAcousticCommunication.323.13.2Introduction.32ChannelModel.3..43.2.5PropagationDelay.34TransmissionLoss.35Noise.37Multipath.38DopplerEffect.393.3AdvancesinUnderwaterAcousticCommunication.4.2InitialEfforts.41IncoherentandCoherentModulation.423.4OFDMforanUnderwaterAcousticChannel.433.4.1NonUniformDopplerCompensationTechnique.45v3.4.2PracticalImplementation.484AcousticOFDMOvertheAirExperiment.5Introduction.51OFDMTransmitter.52OFDMReceiver.5..44.3.5ReceivedBurst.61PacketDetection.63SymbolSynchronization.66DopplerEstimationandCorrection.68OperationsPerformedonaSymbolbySymbolBasis.705TimeWarpbasedDopplerCorrectionTechnique.805.15.2Introduction.80WidebandDopplerModel.8.25.2.3UniformContractionorExpansion.85Expansion/ContractionwithWithin-PacketChangeinDopplershift.89Estimatingand.975.35.4WidebandDopplerModelonanOFDMBurst.98Over-the-airExperiment.10.2RadialMovement.109CircularMovement.12167ConclusionandFutureWork.125Bibliography.127viACIAUVBERBPSKCCDFCDFCFOCIRCPCSDFTDMTDSWCFDMFFTICIListofAbbreviationsAdjacentChannelInterferenceAutonomousUnderwaterVehicleBitErrorRateBinaryPhaseShiftKeyingComplementaryCumulativeDistributionFunctionCumulativeDistributionFunctionCarrierFrequencyOffsetChannelImpulseResponseCyclicPrefixCyclicSuffixDiscreteFourierTransformDiscreteMulti-ToneDoubleSlidingWindowCorrelationFrequencyDivisionMultiplexingFastFourierTransformInterCarrierInterferenceviiIFOISILFMMFOFDMPAPRPRCPTSRCSLMSNRSONARSSSTOTLTRUWAIntegerFrequencyOffsetInterSymbolInterferenceLinearFrequencyModulatedMatchedFilterOrthogonalFrequencyDivisionMultiplexingPeaktoAveragePowerRatioPeakReductionCarriersPartialTransmitSequenceRaisedCosineSelectedMappingSignaltoNoiseRatioSoundNavigationandRangingSphericalSpreadingSymbolTimingOffsetTransmissionLossToneReservationUnderwaterAcousticviiiUWBUWSNVCVLSIZPUltraWideBandUnderwaterWirelessSensorNetworkVirtualCarriersVeryLargeScaleIntegrationZeroPaddedixListofFiguresFigure2.1.SpectralefficiencyofOFDMcomparedtoFDM.6Figure2.2.Eachsubcarrierexperiencesarelativelyflatfade.7Figure2.3.BasicstructureofaMulticarrierTransceiver.8Figure2.4.CyclicPrefix.10Figure2.5.OFDMTransceiverDiagram.13Figure2.6.Fourdifferentpossiblesymboltimingestimates.15Figure2.7.Doubleslidingwindowpacketdetection.19Figure2.8.BlockdiagramofthePTStechnique.29Figure2.9.BlockdiagramoftheSLMtechnique.29Figure4.1.Theover-the-airexperimentsetup.51Figure4.2.OFDMBurststructure.53Figure4.3.Combtypepilotinsertion.54Figure4.4.Blocktypepilotinsertion.54Figure4.5.OFDMTransmitterStructure.56Figure4.6.Pre-amble/Post-ambleStructure.57Figure4.7.OFDMBurst.58Figure4.8.SpectrumofthetransmittedOFDMBurst.58Figure4.9.OFDMReceiverStructure.60Figure4.10.ReceivedBurst.61Figure4.11.Spectrumofthereceivedburst.61Figure4.12.FrequencyResponseoftheBandPassFilter.62Figure4.13.Spectrumoftheburstafterthebandpassfilter.63Figure4.14.EnergyDetection.64Figure4.15.Energydetectionoutputcorrespondingtopacket3.65Figure4.16.PacketdetectionusingtheDoubleSlidingWindowCorrelationtechnique.66Figure4.17.Outputofthematchedfilterforallfourpackets.68Figure4.18.EnergyinthenullsubcarriersforvariousCFOs(symbolnumber15).71xFigure4.19.EnergyinthenullsubcarriersforvariousCFOs(symbolnumber17).72Figure4.20.CarrierFrequencyOffsetestimate(Hz)persymbol.72Figure4.21.FrequencyResponseestimateofthechannelobtainedfromsymbolnumber3.73Figure4.22.FrequencyResponseestimateofthechannelobtainedfromsymbolnumber4.74Figure4.23.BPSKsymbolsextractedfrompacket1.75Figure4.24.BPSKsymbolsextractedfrompacket2.75Figure4.25.Scatterplotofthereceivedsymbols.76Figure4.26.BERperOFDMsymbol.76Figure4.27.BERperOFDMsymbolwithoutDopplercorrection.77Figure5.1.Modeltocalculatetherelativevelocityasafunctionofangle(ortime).81Figure5.2.Velocityofthemicrophonetowardsthespeakerforsemi-circularmovementinthecounterclockwisedirection.82Figure5.3.RelativeDopplershiftduetosemi-circularmovementinthecounterclockwisedirection.83Figure5.4.(a)Timedomainversionofthepacket.(b)Spectrogramofthereceiveds