1、SICE-ICASEInternationalJointConference2006Oct.18-21,2006inBexco,Busan,KoreaDevelopmentofPIRsensorbasedindoorlocationdetectionsystemforsmarthomeKyoungNamHa1,KyungChangLee2,SukLee3DepartmentofMechanicalandIntelligentSystemsEngineering,PusanNationalUniversity,Busan,Korea(Tel:+82-51-510-3091;E-mail:Ovin
2、centgpusan.ac.kr)2DivisionofControlandAutomationEngineering,PukyungNationalUniversity,Busan,Korea(Tel:+81-51-620-1639;E-mail:gcleegpknu.ac.kr)3SchoolofMechanicalEngineering,PusanNationalUniversity,Busan,Korea(Tel:+82-51-510-2320;E-mail:sleegpusan.ac.kr)Abstract:Smarthomeisexpectedtooffervariousintel
3、ligentservicesbyrecognizingresidentsalongwiththeirlifestyleandfeelings.Oneofthekeyissuesforrealizingthesmarthomeishowtodetectthelocationsofresidents.Currently,theresearcheffortisfocusedontwoapproaches:terminal-basedandnon-terminal-basedmethod.Theterminal-basedmethodemploysatypeofdevicethatshouldbeca
4、rriedbytheresidentwhilethenon-terminal-basedmethodrequiresnosuchdevice.Thispaperpresentsanovelnon-terminal-basedapproachusinganarrayofpyroelectricinfraredsensors(PIRsensors)thatcandetectresidents.ThefeasibilityofthesystemisevaluatedexperimentallyonatestbedKeywords:smarthome,location-basedservice,pyr
5、oelectricinfraredsensor(PIRsensor),location-recognitionalgorithmI.INTRODUCTIONThereisagrowinginterestinsmarthomeasawaytoofferaconvenient,comfortable,andsaferesidentialenvironment12.Ingeneral,thesmarthomeaimstoofferappropriateintelligentservicestoactivelyassistintheresidentslifesuchashousework,amusem
6、ent,rest,andsleep.Hence,inordertoenhancetheresidentsconvenienceandsafety,devicessuchashomeappliances,multimediaappliances,andinternetappliancesshouldbeconnectedviaahomenetworksystem,asshowninFig.1,andtheyshouldbecontrolledormonitoredremotelyusingatelevision(TV)orpersonaldigitalassistant(PDA)34.Espec
7、ially,attentionhasbeenfocusedonlocation-basedservicesasawaytoofferhigh-qualityintelligentservices,whileconsideringhumanfactorssuchaspatternofliving,health,andthefeelingsoftheresident5-7.Thatis,ifthesmarthomecanrecognizetheresidentspatternoflivingorhealth,thenhomeappliancesshouldbeabletoanticipatethe
8、residentsneedsandofferappropriateintelligentservicemoreactively.Forexample,inapassiveserviceenvironment,theresidentcontrolstheoperationoftheHVAC(heating,ventilating,andairconditioning)system,whilethesmarthomewouldcontrolthetemperatureandhumidityofaroomaccordingtotheresidentscondition.Variousindoorlo
9、cation-awaresystemshavebeendevelopedtodetecttheresidentslocationinthesmarthomeorsmartoffice.Ingeneral,indoorlocation-awaresystemshavebeenclassifiedintothreetypesaccordingtothemeasurementtechnology:triangulation,sceneanalysis,andproximitymethods8.Thetriangulationmethodusesmultipledistancesfrommultipl
10、eknownpoints.ExamplesincludeActiveBadges9,ActiveBats10,andEasyLiving11,whichuseinfraredsensors,ultrasonicsensors,andvisionsensors,respectively.Thesceneanalysismethodexaminesaviewfromaparticularvantagepoint.RepresentativeexamplesofthesceneanalysismethodareMotionStar12,whichusesaDCmagnetictracker,andR
11、ADAR13,whichusesIEEE802.11wirelesslocalareanetwork(LAN).Finally,theproximitymethodmeasuresnearnesstoaknownsetofpoints.AnexampleoftheproximitymethodisSmartFloor,whichusespressuresensors.Alternatively,indoorlocation-awaresystemscanbeclassifiedaccordingtotheneedforaterminalthatshouldbecarriedbytheresid
12、ent.Terminal-basedmethods,suchasActiveBats,donotrecognizetheresidentslocationdirectly,butperceivethelocationofadevicecarriedbytheresident,suchasaninfraredtransceiverorradiofrequencyidentification(RFID)tag.Thereforeitisimpossibletorecognizetheresidentslocationifheorsheisnotcarryingthedevice.Incontras
13、tnon-terminalmethodssuchasEasyLivingandSmartFloorcanfindtheresidentslocationwithoutsuchdevices.HoweverEasyLivinginvadetheresidentsprivacycanberegardedtowhiletheSmartFloorhasdifficultywithextendibilityandmaintenance.Thispaperpresentsanon-terminalbased2nobilecomputerInternetJhomegateS.kcellularphone,P
14、DA,etc.homenetworksystemhomeapplianceIlight/HVACcontrcontrolnetworkbackbonenetworkNayITnetworkITAVnetworkaaudio&videoFig.1Archtectureofthehomenetworksystemforsmarthome89-950038-5-598560/06/$102006ICASE216277-location-awaresystemthatusesanarrayofpyroelectricinfrared(PIR)sensors.ThePIRsensorsontheceil
15、ingdetectthepresenceofresidentandarelaidoutsothatdetectionareasofadjacentsensorsoverlap.BycombiningtheoutputsofmultiplePIRsensors,thesystemisabletolocatearesidentwithareasonabledegreeofaccuracy.Thissystemhasinherentadvantageofnon-terminalbasedmethodswhileavoidingprivacyandextendibility,maintenanceis
16、sues.Inordertodemonstrateitsefficiency,anexperimentaltestbedhasbeenconstructed,andevaluatedexperimentallyundervariousexperimentalconditions.Thispaperpresentsaresident-detectionmethodusingPIRsensorstosensetheresident,anddescribesalocation-recognitionalgorithmtodeterminetheresidentslocationbyintegrati
17、ngtheinformationcollectedfrommultiplePIRsensors.Inaddition,thelocationaccuracyofthePIRsensor-basedindoorlocation-awaresystemisevaluatedexperimentallyusinganexperimentaltestbed,andthefeasibilityofthesystemisdemonstratedundervariousexperimentalconditions.Thispaperisorganizedintofoursections,includingt
18、hisintroduction.SectionIIpresentsthearchitectureofthePIRsensor-basedindoorlocation-awaresystem,andthelocation-recognitionalgorithm.SectionIIIdescribesaresident-detectionmethodusingPIRsensors,andevaluatestheperformanceofthesystemundervariousconditionsusinganexperimentaltestbed.Finally,asummaryandthec
19、onclusionsarepresentedinSectionIV.II.ARCHITECTUREOFTHEPIRSENSOR-BASEDINDOORLOCATION-AWARESYSTEMA.FrameworkofthesmarthomeGiventheindoorenvironmentofthesmarthome,anindoorlocation-awaresystemmustsatisfythefollowingrequirements.First,thelocation-awaresystemshouldbeimplementedatarelativelylowcostbecausem
20、anysensorshavetobeinstalledinroomsofdifferentsizestodetecttheresidentinthesmarthome.Second,sensorinstallationmustbeflexiblebecausetheshapeofeachroomisdifferentandthereareobstaclessuchashomeappliancesandfurniture,whichpreventthenormaloperationofsensors.Thethirdrequirementisthatthesensorsforthelocatio
21、n-awaresystemhavetoberobusttonoise,andshouldnotbeaffectedbytheirsurroundings.ThisisbecausethesmarthomecanmakeuseofvariouswirelesscommunicationmethodssuchaswirelessLANorradio-frequency(RF)systems,whichproduceelectromagneticnoise,ortheremaybesignificantchangesinlightortemperaturethatcanaffectsensorper
22、formance.Finallyitisdesirablethatthesystemsaccuracyisadjustableaccordingtoroomtypes.Amongmanysystemsthatsatisfytherequirement,thePIRsensor-basedsystemhasnotattractedmuchattentioneventhoughthesystemhasseveraladvantages.ThePIRsensors,whichhavebeen needtoturnonalightwhenitdetectsthehumanmovement,areles
23、sexpensivethanmanyothersensors.Inaddition,becausePIRsensorsdetecttheinfraredwavelengthsemittedfromhumansat9.4-10.4ptm,theyarereasonablyrobusttotheirsurroundings,intermsoftemperature,humidity,andelectromagneticnoise.Moreover,itispossibletocontrolthelocationaccuracyofthesystembyadjustingthesensingradi
24、usofaPIRsensor,andPIRsensorsareeasilyinstalledontheceiling,wheretheyarenotaffectedbythestructureofaroomoranyobstacles.Figure2showstheframeworkforthePIRsensor-basedindoorlocation-awaresysteminasmarthomethatofferslocation-basedintelligentservicestoaresident.Withinthisframework,variousdevicesareconnect
25、edviaahomenetworksystem,includingPIRsensors,roomterminals,asmarthomeserver,andhomeappliances.Here,eachroomisregardedasacell,andtheappropriatenumberofPIRsensorsisinstalledontheceilingofeachcelltoprovidesufficientlocationaccuracyforthelocation-basedservices.EachPIRsensorattemptstodetecttheresidentatac
26、onstantperiod,andtransmitsitssensinginformationtoaroomterminalviathehomenetworksystem.Consequently,theroomterminalrecognizestheresidentslocationbyintegratingthesensorinformationreceivedfromallofthesensorsbelongingtoonecell,andtransmitstheresidentslocationtothesmarthomeserverthatcontrolsthehomeapplia
27、ncestoofferlocation-basedintelligentservicestotheresident.Withinthisframework,thesmarthomeserverhasthefollowingfunctions.1)Thevirtualmapgeneratormakesavirtualmapofthesmarthome(generatingavirtualmap),andwritesthelocationinformationoftheresident,whichisreceivedfromaroomterminal,onthevirtualmap(writing
28、theresidentslocation).Then,itmakesamovingtrajectoryoftheresidentbyconnectingthesuccessivelocationsoftheresident(trackingtheresidentsmovement).2)Thehomeappliancecontrollertransmitscontrolcommandstohomeappliancesviathehomenetworksystemtoprovideintelligentservicestotheresident.3)Themovingpatternpredict
29、orsavesthecurrentmovementtrajectoryoftheresident,thecurrentactionofhomeappliances,andparametersreflectingtheSmartHomeServerirtamapgenerator.|Generatingavirtualmap.|Writingtheresidentslocation|rackingtheresidentsmovement|MovingpatternpredictorStHPtnHomeappliancecontrollerIntelligetFig.2Frameworkofsma
30、rthomeforPIRsensor-basedindoorlocation-awaresystem.2163Home.currenthomeenvironment,suchasthetimetemperature,humidity,andillumination.Afterstoringsufficientinformation,itispossibletoofferhuman-orientedintelligentservicesinwhichthehomeappliancesspontaneouslyprovideservicesintendedtosatisfyhumanneeds.F
31、orexample,ifthepredictorknowsthattheresidentalwayswakesupat7:00A.M.andtakesashower,itispossibletoopenwindowsatanappropriatetimetoensurethattheroomiswellventilatedautomatically.Inaddition,thetemperatureoftheshowerwaterissetautomaticallyfortheresident.B.Location-recognitionalgorithmInordertorecognizet
32、helocationoftheresidentwithinaroom,anarrayofPIRsensorsareusedasshowninFig.3.Inthefigure,thesensingareaofeachPIRsensorisshownasacircle,andthesensingareasoftwoormoresensorsoverlap.Consequently,whenaresidententersoneofthesensingareas,thesystemdecideswhetherhe/shebelongstoanysensingareabyintegratingthes
33、ensinginformationcollectedfromallofthePIRsensorsinoneroom.Forexample,whenaresidententersthesensingareaB,sensorsaandboutputONsignals,whilesensorcoutputsOFFsignal.Aftercollectingoutputs,thealgorithmcaninferthattheresidentbelongstothesensingareaB.Accordingtothenumberofsensorsandthearrangementofthesenso
34、rssignalingON,theresidentslocationisdeterminedinthefollowingmanner.First,ifonlyonesensoroutputsONsignal,theresidentisdeemedtobeatthecenterofthesensingareaofthecorrespondingsensor.Next,iftheoutputsoftwoadjacentsensorsareON,theresidentislocatedatthepointmidwaybetweenthetwosensors.Finally,ifthreeormore
35、sensorssignalON,theresidentislocatedatthecentroidofthecentersofthecorrespondingsensors.Forexample,inthefigure,theresidentislocatedatpoint1whenonlysensorasignalsON,whiletheresidentislocatedatpoint2whensensorsaandbbothoutputONsignals.Inthissystem,locationaccuracycanberegardedasthemaximumvaluefortherad
36、iiofthecircumscribedcircles,includingthecentersofsensorsthatoutputONsignalwhenaresidentstaysataspecificpoint.Forexample,whenaresidententerssensingareaA,theresidentisdeemedtobeatpoint1andtheradiusofthecircumscribedcircleisequivalenttotheradiusofPIRsensorabcsensingareaA.Ontheassumptionthataresidentcan
37、berepresentedbyapointandtheradiusofthesensingareaofaPIRsensoris1m,weknowthatthelocationaccuracywithinsensingareaAis1m.Alternatively,whentheresidentislocatedinsensingareaB,theresidentisdeemedtobeatpoint2,andthenthelocationaccuracyinsensingareaBis/2m,whichisthedistancebetweentheresidentslocation(point
38、2)andthefarthestpointfromtheresidentslocation(point3).Therefore,thelocationaccuracyofthetotalsystemcanberegardedas1m,whichisthemaximumvalueofthelocationaccuracyofeacharea,forthesensorarrangementshowninFig.3.SincethenumberofsensorsandthesizeoftheirsensingareasdeterminethelocationaccuracyofthePIRsenso
39、r-basedindoorlocation-awaresystem,itisnecessarytoarrangethePIRsensorsproperlytoguaranteethespecifiedsystemaccuracy.Inaddition,becausethesystemrecognizestheresidentslocationbyintegratingsensinginformationcollectedfromallPIRsensorsperiodically,afaultsuchasthefailureormalfunctionofasensorcanbeovercome.
40、Forexample,whentheresidentremainswithinsensingareaAinthefigure,thenextlocationoftheresidenthastobenearsensingareaB.However,ifsensorc,whichbelongstosensingareaCoutsidesensingareasAandB,isactivated,thesystemregardsthisasanerrorofsensorcandignoresitsoutput.Inordertodeterminetheresidentslocationprecisel
41、yandincreasetheaccuracyofthesystem,itisnecessarytoincreasethenumberofsensingareas,asdeterminedbythenumberofavailablesensors,andtoequalizethesensingareas.Forexample,Fig.4showsthelocationaccuracybasedonthesensorarrangementmethod.Fig.4(a)and4(b)showarrangementsinvolvingninesensorsthatinvolve40and21sens
42、ingareas,respectively.Bycontrast,twelvesensorsinFig.4(c)giveriseto28(a)40sensingareas(b)21sensingareassensingareA2precisionawhs-n/(c)28sensingareaswithtwelvesensorsFig.3.Thelocation-recognitionalgorithmforPIRsensors.Fig.4.LocationaccuracyaccordingtothesensorarrangementofPIRsensors.2164sensingareas.A
43、stherearemoresensingareasinFig.4(a)thaninFig.4(b),buttheoutersensingareainFig.4(a)islargerthanthatinFig.4(b),thelocationaccuracyinFig.4(a)ispoorerthanthatinFig.4(b)fortheoutersensingareas.Bycontrast,theareasofFig.4(b)andFig4(c)aresimilar,butthelocationaccuracyinFig.4(c)isbetterthanthatinFig.4(b)duet
44、othenumberofsensingareas.WhenPIRsensorsareinstalledaroundtheedgeofaroom,asshowninFig.4(c),itispossibletoincreasethelocationaccuracybymovingtheresidentslocationfromthecenteroftheoutersensortothecenterofgravityoftheoutersensingarea.Forexample,whentheoriginalalgorithmisused,asshowninFig.5(a),theresiden
45、tslocationsarepointsaandb,whicharethecentersofsensors1and2,respectively.Conversely,afterapplyingthecompensationalgorithmforthecenterpointoftheoutersensors,theresidentslocationsareadjustedtopointsaandb,whicharethecentersofgravityofthecorrespondingareas,asshowninFig.5(b).III.PERFORMANCEEVALUATIONOFTHE
46、PIRSENSOR-BASEDINDOORLOCATION-AWARESYSTEMA.Resident-detectionmethodusingPIRsensorsSincethePIRsensor-basedindoorlocation-awaresystemrecognizestheresidentslocationbycombiningoutputsfromallthesensorsbelongingtoonecell,determiningwhetherasinglesensorisONorOFFdirectlyinfluenceslocationaccuracy.Ingeneral,
47、becausetheON/OFFvaluescanbedeterminedbycomparingapredefinedthresholdandthedigitizedsensoroutputacquiredaftersamplingtheanalogsignalfromaPIRsensor,itisnecessarytochooseanappropriatesignallevelforthethreshold.Forexample,SmartFloor,whichisanothernon-terminalmethod,canrecognizearesidentslocationexactlyb
48、ycomparingtheappropriatethresholdandasensorvalue,becauseapressuresensoroutputsaconstantvoltagebasedontheresidentsweightwhenheremainsataspecificpoint.However,becauseaPIRsensormeasuresthevariationintheinfraredsignalproducedbyamovinghumanbody,itsoutputisinanalogform,asshowninFig.6.Thatis,asthevariation
49、inthehumaninfraredincreaseswhenaresidententersasensingarea,thePIRsensoroutputsanincreasingvoltage.Conversely,thevoltagedecreasesastheresidentleavesthesensingarea.Iftheresidentdoesnotmovewithinthesensingarea,theSensor1Sensor2ar4abab(a)Beforecompensating(b)Aftercompensatingfortheoutersensorsfortheoute
50、rsensorsFig.5.Theeffectofcompensatingforthecenterpointoftheoutersensors.variationinthehumaninfraredsignaldoesnotexistandthePIRsensoroutputszerovoltage.Therefore,itisverydifficulttodeterminewhenaresidentisstayingwithinaspecificsensingareausingonlythevoltageorcurrentthresholdofaPIRsensor.Inordertoguar
