content oriented surveillance system based on information...
TRANSCRIPT
ContentOrientedSurveillanceSystemBasedonInformation-CentricNetwork
XinQiWaseda Univers i ty, Tokyo
IEEEGLOBECOM2016WORKSHOP,WASHINGTONDC
OutlineIntroduction
Architecture◦ SystemArchitecture◦ ApplicationArchitecture◦ ContentNamingStrategy
FieldExperiment
Evaluation
Conclusion
IEEEGLOBECOM2016WORKSHOP,WASHINGTONDC 1
IntroductionUrban surveillance systems are being applied in a rapid pace with mature butinefficient solutions. The inefficiency is revealed with two aspects:
ØToo concentrated bandwidth consumption
ØToo concentrated processing requirement
To solve this problem, we proposed a content oriented surveillance system basedon Information- Centric Network. Instead of streaming live video to the centraldata center and processing multiple data stream in the same time, we havedesigned the nodes to process the captured raw data and produce objectivecontents for the subscriber.
IEEEGLOBECOM2016WORKSHOP,WASHINGTONDC 2
NetworkArchitectureThisfigure showsthenetworkarchitectureofthemethod.ThecameranodeswereconnectedviaICNnetworkanddividedintoseveralsetscoveringcertainareas.Insideeachsettherewasanodewithcontrolfunction,thisnodecouldrespondtoacomplexrequestbymanipulatingothernodesintheset.
IEEEGLOBECOM2016WORKSHOP,WASHINGTONDC 3
NetworkArchitectureThefigureshowstheflowchartofafullrequestingprocedure.Theprocedurestartedfromthegenerationoftheinterest.Theinterestwasacombinationoftheobjective’spatternparameterandtheareacodeforrecognition.Aftersendingouttheinteresttothecameranodescoveringtargetarea.Thecameranodeswouldinstantlystarttoanalyzethefreshcapturedframewiththegivenpatternparameter,thengeneratedtheresultdatatobetransmittedback.
IEEEGLOBECOM2016WORKSHOP,WASHINGTONDC 4
ApplicationArchitectureTheuser,forexample,couldinstructtheconsumerapplicationtosendarequestforthehumancountinareaone,containing4nodes.Aninterestwasgeneratedbytheapplicationandsentout.Oncetheinteresthitamatchingcontrolnode,thecontrolnodewouldinstructallthenodesincontrol(includingitself)togatherandintegrateinformationwithdifferentinterestcombination,thensentthecontentback.
IEEEGLOBECOM2016WORKSHOP,WASHINGTONDC 5
ApplicationArchitectureWeappliedtwokindsofparameterstothepatternrecognitionapplicationusedintheexperiment.Theywerefacecounterandcloth-colorrecognitionbasedonopensourcesoftware,OpenCV.Duringtheanalyzeoftheapplication,firstitlocatedthefaceareasofpossiblehumanappearancesinthecapturedframe.Whilecountingthevalueofthefacenumber,itlocatedtheupperbodyclothareasrightbeloweveryfacearea.Bysummarizingtheareapixels’ colorvalue,outputthefacecountandcloth-colordatatogetherorseparately.Algorism1describesthepatternrecognitionprocedures.
IEEEGLOBECOM2016WORKSHOP,WASHINGTONDC 6
ApplicationArchitectureOntherightshowstheexpressionoftherecognitionareasappliedonaperson.Theredzonecoversthefacearea,andbelowthatisthebluezone,coveringupperbody’sclotharea.First,therecognitionapplicationcheckstheexecutionparametertomakesureitislegal.Thenfindsoutifitdesiresfacecountingandcloth-colorrecognizing.Iftheapplicationispermittedtoexecutethefacerecognitionfunction,itwouldquicklyrecognizeeveryfaceintheframeandoutputthecountvalue.Afterthecounting,theapplicationwouldcheckifthereisacloth-colorrecognitioncommand.Ifthereisone,theapplicationwouldlocatetheareaundereveryfaceandsummarizethecolordataandthenoutputthecolordata.Withtheoutputtingofthedesireddata,thesystemcouldreplythedesiredcontent.
IEEEGLOBECOM2016WORKSHOP,WASHINGTONDC 7
ContentNamingStrategyWeusedCCNx ,torealizetheICNnetworkarchitecture.AccordingtotheprincipleofCCNxfunctionalities,theservingapplicationmustcontainacertainformatofnameprefix.Hereweconsiderthenameprefixastheidentifierofthedesiredcontent.Inordertoapplydifferentparameterstothepatternsrecognitionappliance,theparameterswerefittedintothenameprefixinacertainformat.Thisfigureshowsanexampleofthefullnameprefixinthemethod.
IEEEGLOBECOM2016WORKSHOP,WASHINGTONDC 8
ContentNamingStrategyInordertomanagethelargescaleofthecameranodes,apropercontentnamecombinationisorganized.Thesystemappliesnamestothecontentsandclassifiesthecontentsinthesametime,makingitefficientsortinghugeamountofcontentsprovidedbythenodes,especiallytohumanusers.Forexample,theusercouldeasilyinstructthespecifictargetcameranodebyfollowingthesystemclassmapinthefigure.
IEEEGLOBECOM2016WORKSHOP,WASHINGTONDC 9
FieldExperimentWehadperformedafieldexperimentinMarch,atMojiPort(Mojiko),Kyushu,whichusedtobeaninternationaltradingportsincethelate19thcentury.
Inordertofulfillthepurposeofthedesignedarchitecture.Everycameranodewasgivencertainlevelofprocessingpower.
Therewere20cameranodesintheexperiment,dividedinto5setsevenlyandsignedto5experimentlocationsreferringontheright.Werentedoneofthefacilities’roomastheofficethatoperatedourexperiment.1. Kanmon KaikyoMuseum2. OldDalianLineShed(Rentedasoffice)
3. Mojiko RetroObservationTower1F4. MojiCustomsBuilding5. OldOsakaMerchantShipBuilding
Powersupply DC5V/2A
Hardware IntelComputeStickSTCK1A8LFC
Capturedevice UVCwebcam
Networkdevice(sharedby4units) LTEmobileWi-Fiaccesspoint
IEEEGLOBECOM2016WORKSHOP,WASHINGTONDC 10
EvaluationThesefiguresshowtheexperimentdataofthe5locations.Thefiguresrepresentedthequantityandproportionofthecoloroftheareas,meanwhilethecoloredbarsrepresentedthehumancount.
IEEEGLOBECOM2016WORKSHOP,WASHINGTONDC 11
EvaluationThereweretwomajorconclusionsfromtheexperiment’sresultsbasedontherepresentationofthehumancountingandcloth-color.Firstwecouldeasilyfindoutthetourist’sdensitywasmainlyconcentratedinafternoontimeexceptlocation2,OldDalianLineShed.Thismightbecausethereweremanypeoplerentedthefacility’sroomsonthatday,includingourexperimentingteam.Ontheotherhand,thereappearedtohavemostofthecolorsingrey,blackandotherdarkcolors.Theresultstandsfortheappearancethatbigpartofthewinterclothchoicewasdarkcolorslikegreyandblack,alongwithotherbrightcolorsinasmallpercentage.
IEEEGLOBECOM2016WORKSHOP,WASHINGTONDC 12
EvaluationFortheevaluationofourproposedmethod,weanalyzedthethroughputfromlocation3.Figureshowsthethroughputcomparing.Fromthefigurewecansee2throughputlines,blueandred.Thebluelinerepresentsthedatarateofatypicalsurveillancesystemwhichusesvideostreamingastheirdatafeed.Theredlinerepresentsthedataratefromthemethodweproposed,thedataisgeneratedonlywhenthetargetcontentisbeingretrieved.Foratypicalhigh-definitionvideofeed,thenetworkbandwidthconsumptionisaround4Mbps.Fromthethroughputwecantellthat,inacontinuestime,ourproposedmethodconsumeslessnetworkbandwidththanthetraditionalsurveillancesystem.Thisisbecausetheonlycontentsbeingtransferredinthenetworkarethosematchesthetarget’sinterests.Whenthesystemisinidle,thewholenetworkconsumptiondropstoalmostzero.
Thecomparedresultshowsgreatnetworkbandwidthsaveinthefield.Utilizingthisbenefit,thesystemcouldperformveryvaluableinareaswithlownetworkbandwidth,alsoinsomedisasterscenarios.Meanwhilethepatternrecognitionappliancescouldworkinanti-terrorismscenarios.
IEEEGLOBECOM2016WORKSHOP,WASHINGTONDC 13
ConclusionAcontentorientedsurveillancesystemisproposedinthispaperforthepurposeofutilizingnamedcontentsinmodernsurveillancesystems.WiththeICNarchitectureoptimizingthesystemperformanceundermultipleclientssubscribing.Thismethodoptimizesthenetworkbandwidthusageandefficientlydistributesprocessingpowerneedstothenodes.Whenthereareenoughnodescoveringmonitoringareas,itisalsopossibletopredicttarget’smovingpatterntoforeseethenextappearinglocation.Thismethodcouldperformgreatlyinurbanscenariossuchasflowcontrolinpopularareasandanti-terrorismbyevaluatingpeople’sappearances.
Withthefoundationofourcontentorientednetworkingarchitecture,moreappliances,likemotiondetectingandsoundanalyzing,couldbedevelopedtoutilizetheICNnetwork.Seeingthefutureofbigdataanalyzingandprocessing,webelievethedistributedcomputingsolutioncouldalsobenefitinadvance.Meanwhiletheproposedmethodhasalreadyopenedupasolutionofapplyingsurveillancesystemoverlowbandwidthnetworkindisasterscenarios,anddistributingtargetacquirementinanti-terrorismscenarios.
IEEEGLOBECOM2016WORKSHOP,WASHINGTONDC 14
Q&A
IEEEGLOBECOM2016WORKSHOP,WASHINGTONDC 15
REFERENCES1. Stauffer,C.,Grimson,WEL.(2000).Learningpatternsofactivityusingreal-timetracking.IEEETransactionsOnPatternAnalysisAndMachine
Intelligence,Volume:22Issue:8Pages:747-757.DOI:10.1109/34.868677.
2. Maddalena,Lucia,Petrosino,Alfredo.(2008).Aself-organizingapproachtobackgroundsubtractionforvisualsurveillanceapplications.IEEETransactionsOnImageProcessing,Volume:17Issue:7Pages:1168-1177.DOI:10.1109/TIP.2008.924285.
3. Wen,Z.,Zhang,D.,Yu,K.,&Sato,T.(2015).Informationcentricnetworkingfordisasterinformationsharingservices.IEICETransactionsonFundamentalsofElectronics,CommunicationsandComputerSciences,E98A(8),1610-1617.DOI:10.1587/transfun.E98.A.1610.
4. Jacobson,V.Networkingnamedcontent.InProceedingsofthe5thInternationalConferenceonEmergingNetworkingExperimentsandTechnologies(NewYork,NY,USA,2009),CoNEXT ’09,ACM,pp.1–12.
5. Maas,C.,Schmalzl,J.(2013).Usingpatternrecognitiontoautomaticallylocalizereflectionhyperbolasindatafromgroundpenetratingradar.COMPUTERS&GEOSCIENCES,Volume:58Pages:116-125.DOI:10.1016/j.cageo.2013.04.012.
6. Satta,R.,Ciardulli,A.(2015).Sensorpatternnoiseandimagesimilarityforpicture-to-identitylinking.IETCOMPUTERVISION,Volume:9Issue:5Pages:711-722SpecialIssue:SI.DOI:10.1049/iet-cvi.2014.0320.
7. ProjectCCNxTM.http://www.ccnx.org,Sep.2009.
8. An,L.,Chen,XJ.,Yang,S.,Bhanu,B.(2016).Sparserepresentationmatchingforpersonre-identification.INFORMATIONSCIENCES,Volume:355Pages:74-89.DOI:10.1016/j.ins.2016.02.055.
9. Sharan,RV.,Moir,TJ.(2016).Anoverviewofapplicationsandadvancementsinautomaticsoundrecognition,NEUROCOMPUTING,Volume:200Pages:22-34.DOI:10.1016/j.neucom.2016.03.020.
10. Rathore,MM.,Ahmad,A.,Paul,A.,Rho,S.UrbanplanningandbuildingsmartcitiesbasedontheInternetofThingsusingBigDataanalytics,COMPUTERNETWORKS,Volume:101Pages:63-80.DOI:10.1016/j.comnet.2015.12.023.
IEEEGLOBECOM2016WORKSHOP,WASHINGTONDC 16