messagesmicromedia

UserParticipatorySensingforDisasterDetectionandMitigationinUrbanEnvironmentsShin'ichiKonomi1,KazukiWakasa2,MasakiIto3(B),andKaoruSezaki1,31CenterforSpatialInformationScience,TheUniversityofTokyo,5-1-5,Kashiwanoha,Kashiwa,Chiba277-8568,Japan2DepartmentofSocio-CulturalEnvironmentalStudies,TheUniversityofTokyo,5-1-5,Kashiwanoha,Kashiwa,Chiba277-8563,Japan3InstituteofIndustrialScience,TheUniversityofTokyo,4-6-1,Komaba,Meguro-ku,Tokyo153-8505,Japanmito@iis.
u-tokyo.
ac.
jpAbstract.
Pervasivecommunicationtechnologieshaveopeneduptheopportunitiesforcitizenstocopewithdisastersbyexploitingnetworkedmobiledevices.
However,existingapproachesoftenoverlookthebrittle-nessofthetechnologicalinfrastructuresandrelyheavilyonusers'manualinputs.
Inthispaper,weproposearobustandresilientsensingenviron-mentbyextendingandintegratingcooperativelocationinferenceandparticipatorysensingusingsmartphonesandIoTs.
Theproposedapp-roachencouragesproactiveengagementindisastermitigationbymeansofeverydaydatacollectionandend-userdeploymentofIoTsensors.
Keywords:Participatorysensing·Disastermitigation·Smartphones·IoT·Urbanenvironments·Civiccomputing1IntroductionOwingtotherapidgrowthofthecommunicationbandwidthandotherresources,citizenscienceorcrowdscienceisnowconsideredtobeapowerfultooltogatherandanalyzescienticdata.
Especially,rapidpenetrationofsensor-richsmartphonesandIoTsensorsmakeitpossibletoretrievereal-timedataabouturbanenvironments.
Theirsensordatacanbeusedinordinarytimes,however,theywillalsoplayacriticalroleindisastermonitoring[1].
Understandingwhat'sgoingonandanalyzingthedatainnegranularitycanbeachievedonlybyuserparticipatorysensingbecausewecannotdeployconventionalexpensivesensorswithsucientdensity.
SmartphonesandIoTsensorscanbeveryusefulformitigatingtheimpactofdisastersifwecaneectivelyhandlethehugeamountofdatatheyproduce.
Weneedtomaketheirdataeasiertohandlebyapplyingalgorithmicandstatisticalapproachessuchasaggregation,indexing,ltering,compression,datamining,andmachinelearning.
WealsoneedtomakethedatamoreusefulbyactivatingcSpringerInternationalPublishingSwitzerland2016N.
StreitzandP.
Markopoulos(Eds.
):DAPI2016,LNCS9749,pp.
459–469,2016.
DOI:10.
1007/978-3-319-39862-442460S.
Konomietal.
robusttechnologicalinfrastructuresforcollectingandcommunicatingaccuratecontextualdatareliably.
Inthispaper,weproposearobustandresilientsensingenvironmentbyextendingandintegratingcooperativelocationinferenceandparticipatorysens-ingusingsmartphonesandIoTs.
Firstly,itisveryimportanttoconservebatterylifeofmobiledevicesindisastersituationsaspeopleusethemtoaccessandsharecriticaldisaster-relatedinformationandcommunicatewithfamilymembersandfriends.
Itisthereforehighlydesirabletodeterminethelocationsofmobiledeviceswithminimumenergyconsumption.
Oneoftheenergyecientlocal-izationtechniquesformobiledevicesistousewirelesslocationreferencepointsandpedestriandeadreckoningratherthanGPS.
However,currentlythereisnorobustpervasiveinfrastructureoflocationreferencepoints.
WeuseIoTdevicestoactivatesuchaninfrastructure.
Inparticular,weproposeacooperativelocationinferencemechanismtoautomaticallydeterminethelocationsofIoTdevices,therebyturningthedevicesintoubiquitouslocationreferencepoints.
Secondly,wedevelopauserparticipatorysensingenvironmentformitigat-ingtheimpactsofdisastersbasedontheIoT-supportedlocationinfrastructure.
Theproposedenvironmenthasthreekeyadvantagescomparedtoexistingpar-ticipatorysensingenvironments:(1)itfacilitatescollectionofgeo-taggedsen-sordatafromsmartphonesandIoTsensorswithsmallerbatteryconsumption,(2)itallowscitizenstocollectdatabefore,duringandafteradisasterusingsmartphones,omnidirectionalcameras,andenvironmentalsensorstobuildanintegratedlarge-scaledatabase,and(3)itappliesalgorithmicandstatisticalapproachessuchasaggregation,indexing,ltering,compression,datamining,andmachinelearningtodeliverrelevantinformationsuchassafety-enhancingrouterecommendationsatcitizens'ngertips.
2RelatedWorksWenowreviewexistinguserparticipatoryenvironmentsfordisasterdetectionandmitigation.
Peopleusesocialmediatoolstorespondtonaturaldisasterssuchasearthquakes,oods,andhurricanes.
Theyareoftenusedasameanstocollect(or"sense")criticalinformationbyorganizingandcoordinatingvolunteers.
Suchaformofcrowdsourcingenablesswiftsharingofdisasterinformationalthoughithascertainlimitationsintermsofdataqualityaswellaseaseofcollaborationandcoordination[2].
Olteanuetal.
haveanalyzedTweetsfromvariousrecentcrisesandshowntheirsubstantialvariabilityacrosscrises[3].
Wecanexploitsocialbigdatainamoreinformedmanneraswedeepenourunderstandingaboutthekindsofinformationcrowdsgenerateinvariouscrisessituations.
Crowdsourceddisasterinformationisoftenlinkedtolocationinformationandcanbevisualizedonamap.
Forexample,volunteersmonitoredwildresinSantaBarbarabyshowingtextreports,photosandvideosonadigitalmap[4].
Crowdscangeneratesuchmapsmuchbeforeauthoritativeinformationbecomesavailable,whichisanimportantbenetthatcanoutweighthecostoferror-pronecrowdsourcingdata.
LikelyrelevanttothisdiscussionisthatnotonlygrassrootsUserParticipatorySensingforDisasterDetection461organizationsbutalsogovernmentalagenciesarenowexploitingcrowdsourcing.
Forexample,theFederalEmergencyManagementAgency(FEMA)intheU.
S.
recentlyintroducedacrowdsourcingfeatureintheirmobileapp[5].
Smartphonesareoftenusedassocialandparticipatoryplatformsforcol-lectingdisaster-relevantinformation.
Moreover,thereareanumberofexperi-mentalprojectsthatexploretheusesofubiquitoussensorsinsmartphonestoinfercriticalinformationsuchasshakes,infrastructuraldamages,andresinearthquakes.
Smartphones'accelerometerscanbeusedtomeasureandcommu-nicatethestrengthsofshakesquicklyandcheaplywithmuchhigherspatialresolutionthanprofessionallymanagedhigh-qualitysensorssuchasK-NETinJapan.
ExistingresearchbyNaitoetal.
hasshownthatsmartphones'accelerom-etersareparticularlyeectiveformonitoringshakeswiththeseismicintensityover2ontheJapaneseseven-stageseismicscale[6].
Monitoringstrongshakesinbuildingswithhighspatialresolutioncanbeextremelyusefulforanalyzingcumulativeimpactofshakesonbuildingsandevenfordesigningsaferphysicalstructures.
CommunitySenseandResponsesystem(CSR)exploitsaccelerome-tersinsmartphonesanddedicateddevicestomonitorshakescheaplyandinfercomplexspatialpatternsofshakesbasedonamachinelearningmechanism[7].
CitizenSeismologyProjectinterestinglysenseswebtraconapopularearth-quakewebsiteandTwittermessagestodetectearthquakesquickly[8,9].
Fires,whichcanbetriggeredbyearthquakes,oftencausesignicantdamagetoinhabitants.
Earlydetectionofthelocationsofresisveryimportantforpredictingthespreadoftheresandmakingappropriateevacuationplansintime.
However,thereisarelativescarcityofprojectsthatexploresmartphone-basedredetection.
Somerecenthigh-endsmartphonessuchasSamsungGalaxyS4areequippedwithtemperatureandhumiditysensorsthatcanbeusefulfordetectinghightemperatureandlowhumidityaswellastheirtemporalvariancesinproximitytores.
Amjad'srecentprojectexploitssuchhigh-endsmartphonestobuildFireDitectorthatinfersoccurrencesofresinindoorenvironmentsusingNaiveBayesClassierwiththedatafromsmartphones'temperature,humidity,pressureandlightsensors[10].
Althoughexistingliteraturereportsmanysuccesscasesofuserparticipatorysensingfordisasterdetectionandmitigation,mostoftheexistingsystemsuseenergy-hungrylocalizationmechanismssuchastheonesthatheavilyrelyonGPS.
Whenusingstationarysensors,someonewouldhavetospecifytheloca-tionsofthedevicesatthetimeofdeployment.
However,oftentimes,deploymentprocessesarenotclearlydened.
3CooperativeLocationInferencewithIoTsTherewillbeasmanyas26billionInternetofThings(IoTs)in5years[11].
Aswediscussedealier,IoTscanbeextremelyusefulforcollectingenvironmentalinformationbefore,duringandafterdisasters.
Moreover,theycancooperatewithpersonalandwearabledevicesthatcitizenscarryaround.
Forexample,IoTdevicescouldhelpsmartphonestodetecttheircontextmoreaccuratelybyprovidingusefulreferencedata.
462S.
Konomietal.
SmartphonescanuseIoTdevicesaslocationreferencepointsor"locationtags"iftheycanidentifynearbyIoTdevicesbyusingshort-rangeradio,visualrecognition,audiodetection,etc.
Ourproposedmechanismconsiderstwotypesoflocationtags:(T1)theonesthatalreadyknowtheiraccuratelocationsand(T2)theonesthatdon'tknowtheiraccuratelocations.
Inaddition,locationtagshaveonstageandostagestates:thesystemusesonstagetagstocomputelocationinformation,andtrainsostagetagsuntiltheyarereadyto"goonstage.
"WenowconsideraphysicalspaceinwhichonstageT1/T2tagsandostageT2tagscoexist.
LetLbethelocationestimateofanostagetag.
Oursystemcollectslocationinformationfromthesmartphonesthatareinproximitytothetag,andincrementallycomputesLasfollows:Li+1=(i·Li)+Si+1i+1ItobtainsnewlocationestimateLi+1fromsmartphonelocationSi+1andexistinglocationestimateLi(0≤i).
Thiscomputationalprocesscanbetriggeredperiodically,usingthebestsmartphonelocationSi+1ineachinterval.
Also,whentherearemultiplesmartphonesnearby,Si+1isaweightedsumoftheirlocationinformation.
NotethatoursystemcurrentlyusesRSSI(ReceivedSignalStrengthIndicator)toselectthebestSi+1withineachinterval,andtoassignaweighttoeachsmartphone.
Anostagetagisturnedintoanonstagetagwhenitserrorestimationbecomessmallerthanathresholdvalue.
Weestimatetheerrorbyusingmaxi-mumlikelihoodestimatorofacorrespondingcovariancematrix.
Wethenderiveanellipsethatcontainsthetag'sreallocationwith95%condence,andusetheareaoftheellipseasthetag'serrorestimation.
Therearemultiplebenetsgainedfromprovidingsuchalocalizationmecha-nism.
Firstofall,asitinferslocationsofIoTdevicesautomatically,peopledon'talwayshavetodenethelocationsofIoTdevicesatthetimeofdeployment.
IoTdevicescaneventuallybeassociatedwithcorrespondinglocationinforma-tionandthedatatheyproducewillbegeotaggedregardlessofwhethertheyarelocatedindoorsoroutdoors,whethertheyhaveGPSmodulesornot,andsoon.
Wecanthenaccumulatealotofgeoreferenceddatawhichcanbeusedtodetectpointsofcriticaleventssuchasoccurrencesofreorcollapse,andpossiblyguidereghtersquicklytothepeopleinneedofrescue,helpcitizenstoevacuatesuccessfully,andassessandpredictdamagesaccurately.
Moreover,location-taggedIoTdevicescanprovidenearbysmartphoneswithaccurateloca-tioninformation.
Thesmartphonescanusethereceivedlocationinformationtoimprovetheirlocationestimationwithoutconsumingalotofenergy.
AstheproposedmechanismdoesnotrelyonGPS,itisparticularlyusefulinbuildings,undergroundpassages,andurbancanyons.
4UserParticipatorySensingMakingparticipatorysensingusefulindisastersituationswouldrequirepracticalsolutionstofundamentalproblemssuchasenergyecientsensing,integrationofUserParticipatorySensingforDisasterDetection463mobileandstationarysensing,integrationofsensingineverydayandemergencysituations,andprivacypreservation.
Wedescribeourapproachestotackletheseissuesbasedonourexperiencesdevelopingrelevantprototypes.
4.
1EnergyEcientSensingSomecomputationalprocessingismoreenergyconsumingthanothers.
Thus,wecansaveenergybyturningoenergy-consumingfunctionsmostofthetime.
Ourapproachtoenergyconservingparticipatorysensingexploitsenergy-ecientsensorssuchasaccelerometerstodetecttheappropriatetimingforturningonandomoreenergy-hungrysensors,communicationmodules,andcomputationalprocesses.
OneofourongoingresearchesaimstorecorddailyinteractionofapersonbyutilizingBluetoothinasmartphoneasasensor[12].
AlthoughBluetoothissuperiortootherdirect-communicationmethodduetoitsusableidentier(MACaddress)andusefulcommunicationrangeofapproximately10m,energyconsumptionisstillaproblem.
WedevelopedamethodthatimprovesenergyconsumptionofBluetoothbeaconingleveraging3-axialaccelerometersequippedonsmartphones.
Also,themethodimprovesrobustnessofndingsociallinksthattendtofailduetocollisionusingthesimilarityofaccelerationandsetsofBluetoothMACaddresses.
Thedetailedmethodtondothersmartphonesconsideringenergyconsump-tionisillustratedinFig.
1.
Firstofall,themethodrecognizesifauseris"staying"ornotwithanaccelerometerbasedonthemethodproposedbyRavietal.
[13].
Second,themethodrecognizesifauseris"talking"ornotwithamicrophone.
Themethoddoesnotutilizespeech-recognition,bututilizesonlythevolumeofsound.
Finally,themethodsensesproximityusinginquirymodeoftheBluetooththatisnormallyusedtosearchunpaireddevices.
ThephonecollectstheMACaddressesofnearbyphonesinacertainseconds.
Theproposedmethodpredictsasociallinkinarobustmanneragainstfail-uresofndingininquiryofBluetooth.
Inthefollowingequation,sij(B,t)isthestrengthofthesociallinkbetweenthepersoniandthepersonjfromtimettot+TwhereBitandBjtrepresentsetsofcollectedMACaddresses.
EvenwhenasmartphonecannotndbytheBluetoothdirectory,theequationgivesanindicationhowmuchtwosmartphonesarelocatednearby.
sij(B,t)=1(Found)Bit∩BjtBit∪Bjt(Notfound)Wehaveshownthattheproposedapproachcanreduceenergyconsumptionthroughpreliminaryevaluationstudies.
Webelievethatthistechniqueshouldbeextendedandintegratedwithvariouskindsofmobilesensingandcommunicationtoolsfordisasterdetectionandmitigation.
464S.
Konomietal.
Fig.
1.
Flowchartofproposedsensingmethod4.
2IntegrationofMobileandStationarySensingWhendisastersoccur,wewouldbemostlikelytoseekwaystobestutilizeallthetoolsanddatasetsincomplementarymannersinordertominimizethenegativeimpactsofdisastersoncitizens.
Itisthenveryimportanttodevelopoptimalstrategiesandbestpracticestousevarioustechnologiesandresourcesincombination.
Inourpreviousproject,wehavecombinedstationarywirelesssensornet-worksystemsanduserparticipatorysensingtocollectne-grainedenvironmen-talinformation,therebyenhancingthesafetyofcitizensinextremelyhoturbanenvironments[14].
Thesensorsystemsaredeployedinanurbanarea,witharangeabout600*600m2,neararailwaystationinTatebayashiCity,Japan.
Therearetwoindependentsensorsystems:awirelesssensornetwork(WSN)togathertemperatureandhumidityinformationandadistributedcamerasys-temtodetectthetracowsofpedestrians.
Thecombinedsensornodeswhichmeasuretheconditionsoftemperatureandhumidityhavebeeninstalledontheutilitypolesalongsidethestreets.
ThesensornodestransferdatatoasinknodeandthentoacentralserverbyusingIEEE802.
15.
4protocol.
Thereare40com-binedsensornodeswhichhavebeendeployedinthetargetarea.
Stereocamerashavebeeninstallednearthestreetssothattheycanconvenientlycapturethescenesofpedestriancrowds.
ThecapturedscenesaredeliveredtoalocalPConwhichadetectionprogramrunstorecognizethetracowsandvelocitiesofpedestrians.
Thenthesenseddataaretransferredtothecentralserverbyusingwirelesscommunication.
Sixstereocamerashavebeendeployedinthetargetarea.
Oneofthemostimportantissueinthistypeofintegratedsensingisthespatialandtemporalcoverageofsensordata.
Onemightoptforeliminatingredundancy,however,redundantmeasurementscanbeusefulforassuringthequalityofcrowdsenseddata.
Thishastobesupportedbythedatamanagementmechanismsonthecloud,whichwewilldiscussinSect.
5.
UserParticipatorySensingforDisasterDetection4654.
3IntegrationofSensinginEverydayandEmergencySituationsUserparticipatorysensinggenerallyrequirescitizenstointeractwithmobilesensingtools.
Theamountofworkthatusersareexpectedtoperformdiersindierentparticipatorysensingtools.
Opportunisticsensingtoolsonlyrequiresuserstoinstallandactivatethetoolsunlessuserswanttodeactivateandactivatethetoolsfromtimetotimetosaveenergy,memoryspace,orprotectprivacy.
Otherdatacollectiontoolsmayrequireuserstoentertext,numbers,selectitemsfrommenus,takephotos,recordsoundorvideoclips,andsoon.
However,itisaquestionhowmuchtimeandmentalspacecitizensmayhavetoperformsuchoperationsduringadevastatingcrisis.
Inordertoaddressthisissue,weargueforanapproachthatintegratesensingineverydayandemergencysituations.
Wehavesoughttoidentifythekindofusefuldatawhichcanbecollectedineverydaylifesituationsandusedtofacilitateparticipatorysensingduringdisasters.
Oneofsuchkindofdatacanbeomnidirectionalcameraimagesalongurbanstreets.
Ineverydaylifesituations,suchdatacanforexamplebeusedtorecommendpleasantgreenroutesfortakingawalk.
Thesamedatacouldbeusedtoassessdamagesandrecommendsaferrotesindisastersituations,potentiallycombinedwithcomplementaryparticipatorysensingduringdisasters.
InexpensiveomnidirectionalcamerassuchasRicohThetaandKodakPix-proareincreasingpopular,andpeoplecantake360-degreephotographsusingsmartphonesaswell.
Ifcitizensaremotivatedtocaptureandsharegeo-taggedomnidirectionalimagesofstreetsintheireverydaylives,theaccumulatedimagescanbeprocessedasframesofreferenceforassessingtheimpactofdisasters.
Wehavedevelopedasystemforcitizenstocaptureomnidirectionalimagesalongurbanstreetsandextracttheamountofvisiblegreentorecommendpleas-antwalkingroutes.
ThesystemrstprocessesomnidirectionalimagesbasedonLambertazimuthalequal-areaprojection.
AsshowninFig.
2,itthenappliesanedgedetectorandanalyzesfractaldimensiontondvegetationintheimages.
Finally,theamountofgreenineachimageisdeterminedbasedonacolor-basedlteringtechnique.
Inparticular,colorhistogramdataconstructedfromsam-pleimagesofvegetationareusedtocomputethepercentageofvegetationineachomnidirectionalimage.
"Greenroutes"canberecommendedbasedontheresultinggeoreferenceddata.
Althoughwehavefocusedongreenroutes,otherinformationcanbeextractedfromomnidirectionalimagesusingdierentimageprocessingandspatialanaly-sistechniques.
ByopeningupthepossibilitiesforsucheverydayapplicationsFig.
2.
Extractingtheamountofvegetationfromomnidirectionalimages466S.
Konomietal.
ofomnidirectionalstreetimages,weexpecttoincreaseusefullocationindexeddatasetsthatcanbequicklyretrievedandusedindisastersituations.
4.
4PrivacyPreservationIfthereisanyconcernonprivacypreservationinuserparticipatorysensing,peoplearediscouragedtojoinanyparticipatorysensingapplications.
Further,ifprivacypreservationmechanismcannotbeeasilyunderstoodbytheusers,itwillalsodiscouragethem.
Inlightoftheseissues,wehaveproposedaperturba-tiontechniquecalledNegativesurvey[15]andsomeofitsextensions.
Negativesurveyanditsextensioncanbeappliedtouserparticipatorysensingfordisas-tersituation.
Typicalexampleistheusageofprivacy-preservingsmartphonesasseismometerstocomplementtheexistinginfrastructuredeployedbyK-NET[16].
Earlyanddetailedredetectionaswellasdetectionofpeoplefollowindisastersituationiswithinourscope.
Wehavealsoproposedmechanismsforprotectinglocationprivacy[17],whichmakesitdiculttotracethetrajectoryofaspecicnode.
Sincethedegreeoflocationprivacyisnotyetwelldened,wearenowtacklingtheissueandtrytore-deneit[18].
5SystemArchitectureforProvidingIntegratedServicesTousethedatacollectedthroughuserparticipatorysensingeectively,webrieydescribemethodsto(1)buildtheenvironmentaldatawarehouse(EDW)whichworksasaninfrastructureprovidingcomprehensiveandpredictiveenvironmen-talinformation,and(2)integrateheterogeneousenvironmentalinformationfrommulti-modalsensorsintoanaggregatevaluewhichfacilitatesfurtherprocessing,and(3)determinetheoptimalpathplansinenvironmentswhicharevaryingcontinuously.
Figure3showstheoverallarchitecture.
Rawmulti-modalsensordataareinputintofacttablesoftheEDWwheremultidimensionaldatamodelanddatapredictionmethodareapplied.
Thedimensionalinformationofspaceandtimeisextractedandaggregatedintodimensiontables.
TheEDWcontainspredictivefunctionsthereforeitcanprovidehistorical,currentandfutureenvironmentalinformation.
Thewalkablespaceofpedestriansismodeledasastreetnetwork.
Theinter-sectionsaretreatedasnodesandthewalkablestreetsegmentsbetweenintersec-tionsaretreatedasedges.
Mapmatchingisappliedtoassociatesensordatatoproperstreetedges.
Inordertointegratethemulti-modalsensordataconsistentlyandexibly,anovelmulti-factorcost(MFC)modelisproposed.
TheaggregatecostratesforedgesarecalculatedoutbyapplyingtheMFCmodel.
ThecostvalueofanedgeaccessedbythePPengineistheproductofaggregatecostrateandthetraveltimeforthatedge.
Basedontheformertwosolutions,theoptimalpathplanning(PP)problemissolvedinatime-dependentnetworkbyapplyingadynamicprogrammingUserParticipatorySensingforDisasterDetection467Fig.
3.
Overallarchitectureoftheproposedmethodsmethod.
ThePPenginereceivespathqueriesthataresubmittedbypedestriansinrealtime.
WehavedevelopedtheprototypeclientapplicationrunningonanAndroidsmartphone.
Amapviewisdisplayedonthesmartphoneandthepedestriancanspecifyheroriginanddestinationbytouchingthescreen.
Thentheplannedpathcalculatedonaserverisdisplayedonthemapviewtonavigatethepedestriantoapproachherdestination.
Thisarchitecturehasbeenusedtointegratethedatafromawirelesssensornetwork(WSN)togathertemperatureandhumidityinformationandadis-tributedcamerasystemtodetectthetracowsofpedestrians[19],therebyrecommendingcomfortableandsafenavigationroutesinanextremelyhoturbanenvironments.
6ConclusionWehaveproposedarobustandresilientsensingenvironmentbyextendingandintegratingcooperativelocationinferenceanduserparticipatorysensing.
Theproposeduserparticipatorysensingenvironmentsupportsenergyecientsens-ing,integratedsensingineverydayandemergencysituationsusingmobileandstationarysensors,andprivacypreservation.
Inparticular,theproposedenvi-ronmentencouragesproactiveengagementindisastermitigationbymeansofeverydaydatacollection.
Theautomatedlocationinferencefacilitatesend-userdeploymentofIoTsensorsaswell.
Userparticipatorysensinghasimportantrolestoplayevenwhenhighqual-itysensorsandsimulationsystemsareinplace.
Oftentimesdisaster-monitoringinfrastructuresareofnationaland/orregionalconcerns.
Infrastructures,suchasJapaneseK-NET,aredeployedandmanagedunderdierentbudgetaryrestrictions,whichmayleadtocompromisedspatialresolutionsofsensors.
IntheJapanesecontext,itisparticularlyimportanttoconsidercomplementary468S.
Konomietal.
relationshipsbetweencheap,quickanddensecrowdsensingandreliableinfrastructuralsensors.
Moreover,aspeopleoftenfacescarcityofinformationindisastersituations,providingmoredatathroughcrowdsensingcanhelpreducefalsenegativeproblemsoffailingtoissuealarmsandwarnings.
Computer-basedsimulationsystemshelpusunderstandhowthingsbehaveindisastersituationswithoutactuallyexperiencingthemintherealworld.
Con-nectingsimulationstoreal-worldeventscouldeectivelynarrowdownthespaceforwhat-ifexplorationsforpertinentdecision-making.
Crowdsensingthencanplayasignicantroleinmakingsimulationsusefulintime-criticaldisastersitua-tionsasitprovidesawaytofeedreal-worldinformationquicklyintosimulations,muchbeforeauthoritativeinformationismadeavailable.
Also,microscopicsim-ulationsofshakesandresatabuildingscalerequirene-grainedfeedofreal-worlddatathatcrowdsensingcouldcaterwellfor.
Furthermore,simulationscouldbeusefulformakingcrowd-sensingsystemsincludingcrowdbehaviorsandcomputationalprocessingmechanismssmarter.
Forexample,simulationresultscouldbeusedtorequestsensingtasksecientlybyprioritizingdatacollectionbasedonthemostcriticalgoalssuchassavinglives.
Weexpectthatourcurrentresultswillbeextendedtobeasystemicyetex-ibleenvironmentratherthanacomplex,monolithicsystem.
Thus,ourproposedmechanismscouldbeadaptedeasilytodierentdisastersituationsanddierentexternalsystems.
Acknowledgments.
WeacknowledgeProf.
ToshihiroOsaragiforprovidingusthemobilitysimulationdatarightafteragreatearthquake.
ThisworkwassupportedbyCREST,JST.
References1.
Inoguchi,M.
,Tamura,K.
,Sudo,S.
,Hayashi,H.
:Implementationofprototypemobileapplicationoperatedonsmartphonesformicromediaservice.
J.
DisasterRes.
9(2),139–148(2014)2.
Gao,H.
,Barbier,G.
,Goolsby,R.
:Harnessingthecrowdsourcingpowerofsocialmediafordisasterrelief.
IEEEIntell.
Syst.
26,10–14(2011)3.
Olteanu,A.
,Vieweg,S.
,Castillo,C.
,Whattoexpectwhentheunexpectedhap-pens:socialmediacommunicationsacrosscrises.
In:Proceedingsofthe18thACMConferenceonComputerSupportedCooperativeWork&SocialComputing(CSCW2015),pp.
994–1009(2015)4.
Goodchild,M.
F.
,Glennon,J.
A.
:Crowdsourcinggeographicinformationfordisas-terresponse:aresearchfrontier.
Int.
J.
DigitalEarth3(3),231–241(2010)5.
DisasterReporter.
http://www.
fema.
gov/disaster-reporter.
Accessed1Jan20156.
Naito,S.
,Azuma,H.
,Senna,S.
,Yoshizawa,M.
,Nakamura,H.
,Hao,K.
,Fujiwara,H.
,Hirayama,Y.
,Yuki,N.
,Yoshida,M.
:Developmentandtestingofamobileapplicationforrecordingandanalyzingseismicdata.
J.
DisasterRes.
8(5),990–1000(2013)7.
Faulkner,M.
,Clayton,R.
,Heaton,T.
,Chandy,K.
M.
,Kohler,M.
,Bunn,J.
,Guy,R.
,Liu,A.
,Olson,M.
,Cheng,M.
,Krause,A.
:Communitysense,responsesystems:yourphoneasquakedetector.
CACM57(7),66–75(2014)UserParticipatorySensingforDisasterDetection4698.
EMSC,CitizenSeismology.
http://www.
citizenseismology.
eu/.
Accessed1Jan20159.
Meyer,P.
:Usingashcrowdstoautomaticallydetectearthquakesandimpactbeforeanyoneelse.
http://irevolution.
net/2014/10/27/using-ashsourcing-to-automatically-detect-earthquakes/.
Accessed1Jan201510.
Amjad,M.
M.
M.
:Naivebayesclassier-basedredetectionusingsmartphonesen-sors,Master'sThesis,UniversityofAgder11.
Gartner,Inc,Says,Gartner,4.
9BillionConnected"Things"WillBeinUsein(2015).
http://www.
gartner.
com/newsroom/id/2905717.
Accessed1Oct201512.
Shimizu,K.
,Iwai,M.
,Sezaki,K.
:Sociallinkanalysisusingwirelessbeaconingandaccelerometer.
In:IEEE27thInternationalConferenceonAdvancedInformationNetworkingandApplicationsWorkshops(WAINA),pp.
33–38(2013)13.
Ravi,N.
,Dandekar,N.
,Mysore,P.
,Littman,M.
:Activityrecognitionfromaccelerometerdata.
In:ProceedingsoftheNationalConferenceonArticialIntel-ligence,vol.
20,no.
3,p.
1541(2005)14.
Dang,C.
,Iwai,M.
,Umeda,K.
,Tobe,Y.
,Sezaki,K.
:NaviComf:navigatepedes-triansforcomfortusingmulti-modalenvironmentalsensors.
In:IEEEPervasiveComputingandCommunication(Percom2012),Switzerland,March201215.
Esponda,E.
,Guerrero,V.
M.
:Surveyswithnegativequestionsforsenstiveitems.
StaticsProbab.
Lett.
79(24),2456–2461(2009)16.
Konomi,S.
,Kostakos,V.
,Sezaki,K.
,Shibasaki,R.
:Crowdsensingfordisasterresponseandpreparedness.
In:The77thNationalConcentionofIPSJ,pp.
449–451(2015)17.
Huang,L.
,Matsuura,K.
,Sezaki,K.
:Enhancingwierelesslocationprivacyusingsilentperiod.
WCNC2005,1187–1192(2005)18.
Matsuno,Y.
,Ito,M.
,Sezaki,K.
:Impactoftime-varyingpopulationdensityonlocationprivacypreservationlevel.
In:The5thIEEEInternationalWorkshopontheImpactofHumanMobilityinPervasiveSystemsandApplications(IEEEPerMoby).
Sydney,Australia(2016)19.
Umeda,K.
,Hashimoto,Y.
,Nakanishi,T.
,Irie,K.
,Terabayashi,K.
:Subtractionstereo:astereocamerasystemthatfocusesonmovingregions.
In:ProceedingsofSPIE7239,Three-DimensionalImagingMetrology,p.
723908(2009)