一個傳感器板一個聲學(xué)交通監(jiān)控系統(tǒng)設(shè)計

ProjectGoalToenablethestudyofanewtechniquefortrafficmonitoringusinganacousticarraysensornetworkby:definingthesystemarchitectureofthesensorarrayboardderivingthesystemspecificationsofthesensorarrayboarddesigningaPCBforthesensorarrayboardandtheinterfacetothewirelesslinkPresentationOutlineIntroductiontoTrafficSurveillanceNetworksBackgroundonPassiveAcousticSensorsSystemArchitectureSystemSpecificationsPartSelectionSchematicCapturePCBLayoutFutureGoalsTechnologiesusedforRuralTrafficSurveillanceNationalSurveyconductedin2004byITS1found:26statesuseInductiveLoopDetectors(ILD)21statesuseRadarDetectors11statesuseVideoImageDetectors(VIDS)5statesuseAcousticDetectorsVehicleinfocollectedbythesensornetworksinclude:TrafficVolumeVehicleSpeedVehicleClassificationTravelTimeIncidentOccurrence1ITSorIntelligentTransportationSystemisadivisionofRITAwhichispartoftheU.S.DepartmentofTransportation(DOT).InductiveLoopDetectors(ILD)Howitworks:1ormoreloopsofwireareembeddedundertheroad&connectedtoacontrolbox.Whenavehiclepassesoverorrestsontheloop,inductanceisreducedshowingavehicleispresent.BenefitsEstablishedTechnologyNotimpactedbyenvironmentalconditionsAccurateindetectingvehiclepresencePerformswellinbothhighandlowvolumetrafficDisadvantagesHighCost(upto$10kforinitialcosts)InvasiveInstallationPotentialpoorreliabilityduetoimproperinstallationNotviableforcertainlocationsandroadconditionsUnabletodirectlymeasurespeedordirectionVideoImageDetection(VIDS)Employsmachinevisiontechnologytoautomaticallyanalyzetrafficdatacollectedw/ClosedCircuitTelevisionSystems(CCTV)Benefits:RapidIncidentdetectionWideareadetectioncapabilities(multi-lane,multi-direction)VehicleclassificationEstimatestrafficqueuesandspeedInstallationdoesnotrequirelaneclosuresCanbeintegratedwithothersensornetworksDisadvantages:HighCost(initialcost>$50k)HigherPowerthanothersensornetworksLightconditionscanaffectsurveillanceRadar-basedRoadsideSensorHowitworks:TransmitsradarpulsesAportionoftheenergyisreflectedorscatteredfromthevehicleandroadwaybacktowardthesensorThisenergyisreceivedandinterpreted.BenefitsLowPowerMostaccuratetechnologyfordetectingspeedTrafficCountaccuracyEasyinstallationLowcostDisadvantagesAccuracycanbeaffectedbyweatherconditions(hail,snow,rain)DirectionalDetectionispoorInterferencecouldoccurwithotherRFdevicesPassiveAcousticSensorsPassiveRoad-sideSensorthatreceivessoundwavesfrompassingvehicles.Benefits:LowPowerLowCostEasytoInstallDirectionalandMulti-laneDetectionAccuratelymeasurestrafficcountDisadvantages:AccuracyaffectedbyenvironmentfactorsSpeedmeasurementsarenotasaccurateasothermethodsSummaryofTrafficSurveillanceTechnologiesEmployedToday:RadarandAcousticsensorsaretheleastexpensivetodeployandarethelowestpower.VIDScollectthemostinformationandthedataprocessingpossibiltiesareendless.Acousticsensorsaretheleastaccurate,butthetechnologyisrelativelynew.Acousticsensorsfortrafficmonitoringhasroomforimprovement:ImproveaccuracyAbilitytodetectidletrafficIntelligentlyprocessdata(VehicleClassification,IncidentDetection)TechnologyTrafficVolume(Moving)TrafficVolume(Stopped)VehicleSpeedVehicleClassificationIncidentDetectionPowerCostILDsAACDCBCRadarAAACCBAVIDsBBBAACCPassiveAcousticBCBBCAACode:A–Excellent;B–Fair;C-Poor;D–Nonexistent;U–UnknownAcousticArraySensorResearchhasshownthatusingarraysofacousticsensorsnarrowsthedetectionzoneforimprovedSNR&betteraccuracyForthisprojectthefollowingconfigurationwillbeevaluated:4microphonearrayspersensor:2arraysformapairparalleltotheroad2arraysforapairorthogonaltotheroadEacharraycontains12microphonesSensorstobedeployedroad-sideabout10mabovetheroad.SensorNetworkSystemDetailsSounddetectedbyeachelementinasinglearraywillbesummedtogetherandamplified.4analogoutputswillbedigitizedseparatelyandprocessed.Processedsignalswillbetransmittedtoagatewaynodewithinthewirelesssensornetwork.SignalProcessing:ParallelPairAvehiclemovesacrossthedetectionzone(D2toD1):Asavehicleapproaches,soundreachesD2earlierthanD1.Thetimedeltawillbenegative(τ<0).Whenavehicleisatthecenterofthedetectionzone,τ=0.Asavehicleexitsthedetectionzone,τwillbepositive.Therateofchangeorslopeacrossthedetectionzonecorrespondstothevehiclespeed.Toextractthetimedelayfromtheactualsignals,thecrosscorrelationofD1andD2iscalculated:SignalProcessing:OrthogonalPairAvehiclemovesacrossthedetectionzone:Asthevehicleapproaches,soundwillreachD1earlierthanD2.Thetimedifferencewillbepositiveandincreasing(τ<0).Whenthevehicleisatthecenterofthedetectionzone,τwillpeak.Asthevehicleexitsthedetectionzone,τwillbepositiveanddecreasing.Thesoundmapforvehiclesinlanesclosertothesensor(smallery)willhavesmallerpeaksthanthoseinlanesfurtherfromsensor.SystemArchitecture2Boardstoformcompletesystem:SensorBoardMicrophoneArraysSummingStageAmplificationStageSystemBoardAnalog-to-DigitalConverterProcessorRFTransceiver(ISMbandof2.4GHz)eZ430-RF2480DemoKitfromTIchosenforSystemBoardUSB-basedwirelessdemotoolMSP430F2274Mixed-SignalMicrocontrollerCC24802.4GHzZigBeenetworkprocessor2.4GHzAntennaSensorBoardArchitectureForinitialprototype,1sensorboardperarraytoallowforarrayspacingexperimentationMajorityofSystemGainimplementedonSensorboardtomaximizeSNR.ClippingCircuitandAnti-AliasingFiltermaybeneededtoconditionsignalforADC.SystemBoardArchitectureMSP430Mixed-SignalMicrocontrollerAUXOp-AmpscanprovideextragainifneededOn-chip10-bitADCcanmultiplexin4analogchannelsfordigitizingCPUcanbeprogrammedtocomputecross-correlationfunctionsCC2480–ZigBeeProcessortotransmitdatatogatewaynodetomaincontrolcenterover2.4GHzISMbandDeterminingSensorBoardSpecsPowerSupplyRequirementsMaximumVoltageOutputSwingElementSpacingSummingStageConfigurationSystemGainAnti-AliasingFilterPowerSupplyRequirements2Specificationsneedtobedetermined:SupplyVoltagesMaximumCurrentSensorBoardneedstobeportablesolutionLowpowerBatteryoperatedReuseBatteryBoardincludedwitheZ430-RF2480DemoKit:2AAABatteriesinseries:3VsupplyCapacityofAAAs:900-1155mA/hMaximumVoltageOutputSwingNeedstobelimitedtoInputRangeofADC.DependingofADCfront-endarchitecture,overvoltageoninputscancauseconversionerrorsandinsomecasesdamagetheADC.10-bitADCintegratedontheMSP430willbeused.ADCinputrangeisprogrammablefromVcctoVss.Forthisproject,ADCwillbeprogrammedtoacceptaninputrangeof3Vto0V.NoClippingCircuitneededonSensorBoardAmplifiersonSensorboardwillnotproduceavoltagehigherthanitssupplyvoltageADContheMSP430clipsanysignalsgreaterthantheprogrammableupperinputrangelimit.Benefitsfromusingthesamepowersourcesolution!ElementSpacingSpacingb/tarrayelementswillbechosentoachievethedesireddetectionangle.DesiredDetectionAngle:Sensorboardmounted10maboveroad(z=10m)Desireddetectionzoneis2.5minanydirectionatroadlevel.DetectionAnglecanbecalculated:ExperimenttoDetermine

ElementSpacingSeveral3x4elementarraysbread-boardedw/summingstage:Array1:1.75cmx1.75cmArray2:2cmx2cmArray3:2.5cmx2.5cmDetectionAnglemeasuredbymovinga4kHzsoundsourceacrossthearrayfromafixeddistance.Vpk-pkmeasuredatsummeroutputwithOscilloscopeResultsofElementSpacingExperimentsSummingStageConfigurationOriginalDesignused1op-amptosumall12elementsNewDesignsumstheelementsin2stages:Firststagesumsthe3rowsof4elementsseparatelySecondstagesumsthe3outputsofthe1ststageNewDesignallowsformoregainwithoutlargeRImprovedRise/FallTimes(τ=RC)IncreasedBandwidth–lessGainperstage(f3dB=GBW/Gain)DeterminingSystemGainExperimentsconductedtoestimatetheamountsystemgainrequiredintherealtrafficmonitoringenvironment.Microphonearrayprototypeboardisnotportable,andoutputcannotbestored.Aportabledigitalsoundrecorderwasusedtocollectfieldsamples:2Omni-DirectionalElectretCondenserMicrophonesMicpositionscanbeconfiguredfrom45oto135o(90owaschosen)7Pre-AmpLevelsRecordedsoundclipssavedas*.WAVSystemGainExperimentStep1–CalibratethedigitalrecordertothemicrophonearrayinacontrolledenvironmentwithsamesoundsourceforeachPre-AmpLevel.MeasuredinLabw/4kHzsoundsourceArrayandRecorderinsamelocationforthemeasurementStep2–CollectFieldDataforeachapplicablePre-Amplevel5-10strafficsampleswererecordedfromaheightof25-30ftfromtheroadAverageof5vehiclespassedtherecorderduringeachsoundclipincludingmotorcycles,smallcars,andSUVs.Step3–Downloadandprocess*.WAVfilestodeterminepeaksoundlevelsMATLABcodewaswrittentoanalyzethesefilesThestartandendofeachfilewasremovedtoeliminateinterferenceThe2microphoneoutputsweresummedThemaxpeakandrmslevelswerecalculated.Step4–Relatethepeaksoundlevelsfromdigitalrecorderbacktothemicrophonearrayanddeterminetherequiredsystemgain.SystemGainResultsDigitalRecorderPeakandRMSLevelsfromTrafficSoundClipsCalculatingtheSoundSourceRatioEstimatedArrayOutputVpk-pkandRecommendedGainSettingsDigitalRecorderPeakandRMSLevelsfrom4kHzSignalinLabvs.Pre-AmpLevelArrayOutput=100mVpk-pkPreAmpLevel3optimalinfieldTocalculateVpk-pkatthearrayoutputthefollowingformulaswereused:1.2.3.TotalSystemGainNeeded=4x3.5=14Anti-AliasingFilterAnti-aliasingfiltershouldbeplacedbeforeADCPreventsharmonics,spursandbroadbandnoiseoutsideofNyquistfromaliasingbackin-bandImproperfilteringleadstoadecreaseinSNR,areduceddynamicrangeandanincreaseinunwantedspurs.ADCinputbandwidth/channel:clockrange:450kHz-1.5MHzNyquistBW:225–750kHz4chsmuxed,BWperch=56.25–187.5kHzFrequenciesofInterest<10kHzFilterDesignLow-passButterworth–flatpass-&stop-bandF3dB=20kHzforflatterphaseinpass-band1storderprovides~20dBattenuationatFADC/2.Sincemicrophoneshaveafrequencyroll-offresponse,1stordershouldbeadequate.SummaryofBoardSpecificationsParameterSpecificationPowerSupply3Vonly(2AAAbatteries)CurrentDissipationAslowaspossibleMicrophoneSpacing2.25cmx2.25cm(detectionangle~16o)SystemGain14or11dB(10*log(14))MaximumOutputSwing0-3VLowPassFilterf3dB=20kHz,Order=1Schematic-CircuitDesignArchitectureisfinalizedActiveComponentshavebeenselected:AnalogDevicesAD8544QuadRail-to-RailOpAmpSingleSupplyOperation:2.7Vto5.5VGBWP~1MHzLowsupplycurrent45uA/amplifierEmkayMD9745APZ-FOmni-DirectionalMicrophoneOperatingVoltage:2.0Vto10.0VFrequencyRange:100Hzto10kHzS/NRatio:>55dBAltiumDesigner6softwarechosenforPCBSchematicandLayoutSchematicPage1SchematicPage2PCBLayout4LayerBoard2SignalLayersGNDandPowerlayerBoardSizeis