《電子與通信工程專業(yè)英語》課件第18章

Unit18IntegratingRFIDonEvent-basedHemisphericImagingforInternetofThingsAssistiveApplicationsNEWWORDSANDPHRASES

NOTES

EXERCISES

參考譯文

EXTENSIVETEXT

1．Introduction

Theremoteactivitymonitoringacrosslargeenvironments,suchasgovernmentfacilities,publicbuildingsorindustrialenvironmentsinrealtime,isaprerequisiteforvariousmonitoringapplications.Suchactivitiescanalsoproveausefuladditioninassistiveenvironmentsinsmallersizespaces,likehousesandoffices.Modernvideo-basedsurveillancesystems,whichemploypowerfulrealtimeanalysistechniques,arewidelydeployed.Additionally,theuseofmultiplecamerastoprovidesurveillancecoverageoverawiderareawhileensuringobjectvisibilityoveralargerangeofdepths,introducestheneedtocoordinatethecamerasinordertodetecteventsofinterest,somethingwhichincreasessystemcomplexity.

RadioFrequencyIdentification(RFID)technologyisarguablytheidealsolutionforobjectidentification.Ithassuccessfullybeenusedinalargevarietyofapplications,likeenterprisesupplychainmanagementforinventorying,trackingandofcourseobjectsidentification.RFIDmayalsoproveusefulforpervasivecomputing,forprovidingidentitytovirtuallyeverything.ItisnotaccidentalthatRFID,alongwithwirelesssensorandnanotechnologieshavebeencombinedtoformwhatisknownastheInternetofThings.

Fornearlyeverypervasivecomputingapplication,anothervitalrequirementisrealtimelocating,whichemanatesfromtheinherentneedforjust-in-timeactionableinformation.Overtheyears,manysystemshaveaddressedtheproblemofautomaticlocationsensingwithvarioustechniques.InthispaperweelaborateonthefeasibilityofdevelopingahybridsystembycombiningvideosurveillancefeedandRFID,inordertoprovideasolidsystemforautomaticidentificationandtrackingofobjectsintheoutputofavideocamera.Theuseofhemisphericimagingcamerasthatmaximizetheareacoverageofasurveillancesystemisexamined,inordertoeliminatetheneedformultiplecamerasandthefactorsthatinfluencethepurposeofsuchasystemareidentified[1].

2．Underlyingtechnologies

1)

RFIDoverview

OneofthepivotaltechnologiesofpervasiveandubiquitouscomputingisRFID.Groupedunderthebroadcategoryofautomaticidentificationtechnologies,RFIDisusedasagenerictermtodescribeasystemwheretheidentity(intheformofauniqueserialnumber)ofanobjectistransmittedwirelessly,usingradiowaves.AtypicalRFIDsystemiscomposedof:①theRFIDtag,whichcontainsadigitalnumberassociatedwiththephysicalobjectthatitisattachedtoand②theRFIDreader(alsoknownasinterrogator)whichisusuallyconnectedtoabackenddatabase.Thereaderisalsoequippedwithanantenna,atransceiverandaprocessorthatbroadcastsaradiosignalinordertoquerythetagandreaditscontents.

TwoarethemostimportantcharacteristicsofanRFIDsystemeachencapsulatingdifferencesinrange,datatransferandtransmissionundercertainenvironmentalconditions:

①theenergyresourcesandcomputationalcapabilitiesofRFIDtagsand②itsoperatingfrequency.Accordingtothefirstcharacteristic,RFIDtagsaredistinguishedintopassiveandactive,aswellastheircombinations.Activetagsincorporateabatteryandcantransmitsignalsautonomously,overalongoperationrangewithhighperformancebuttheyareexpensiveandusuallyhavealargesize.Ontheotherhand,passivetagsrequireanexternalsourcetoprovokesignaltransmission,whichisacquiredusingeitherinductivecouplingorelectromagneticcaptureandtheycommunicatewiththereaderbyutilizingloadmodulationorelectromagneticbackscatter.Passivetagsarewidelyusedandinmanycasestheyarepreferredovertheactiveduetotheirlowcost,smallsize,andpracticallyunlimitedlifetime.Additionally,RFIDsystemscanbecategorizedintofouroperatingfrequencies:Low,High,UltraHighandSuperHighFrequency(orMicrowave).Asthefrequencyincreases,rangeanddatatransferratesalsoincrease,butpenetrationthroughwaterandmaterialssuchasmetaldecreases.

Thenecessityofestablishinguniformengineeringcriteria,methods,processandpracticesforthedefiningcharacteristicsofRFIDsystems,ledtotheproposalofvariousstandards,twoofwhicharemoreprominent.EPCglobaldefinesacombinedmethodofclassifyingtagsthatspecifiesuniqueidentificationnumbers(ElectronicProductCodes),frequencies,couplingmethods,typesofkeyingandmodulation,informationstoragecapacityandmodesofinteroperabilityamongothers.SimilarlyISO(jointlywithIEC)developedstandardsforidentification,communicationbetweenthereaderandthetag,dataprotocolsforthemiddlewareandtesting,complianceandsafety.

(1)

IdentificationandRFID.Afundamentalrequirementofpervasivesystemsingeneral,istheabilitytouniquelyidentifythingsand/orentities.RFIDsatisfiesthisrequirementbynature.Therearealsoseveralothertechnologiesthatservethispurpose,eachhavingadvantagesanddisadvantages,butsteadilybeingreplacedinmostapplicationareaswithRFID.

Thebarcode,stillthemostwidelyusedproducttrackingmethodinsupplychainmanagementandthecheapestidentificationsolution,isanoptical,machinereadablewrite-oncerepresentationofanobjectcategory.Themostimportantweaknessesofbarcodetechnologyaretheinabilitytoprovideextrainformationregardingasingleobject(two-dimensionalsymbologiesdealwiththisissuehowever)andtherequirementofthebar-codedobjecttobeinaline-of-sight(LOS).

Cardtechnologies,isanothercategoryforidentificationofobjectsandentitiesthatincludemagneticcards,smartcardsandopticalcards.Usuallyembeddedinacreditcard-sizedplasticcardtheyencompasseitheramagneticstripeoranintegratedcircuitandtheyhavegreaterstoragecapacities.Incasetheyincorporateamicroprocessor,theyalsohaveincreasedprocessingcapabilities,somethingwhichallowsthemtobeusedindemandingapplicationssuchassecurity(withgreatlimitationshowever).Onthedownside,mostcardseitherrequirecontactortobeinaveryclosedistancewiththereader.Thecostsalsoincreasesignificantlyinproportiontothefeaturesofthecardinuse.RFIDbalancesbetweenefficiencyandcost-effectiveness,whileitalleviatestheneedofthetagbeinginaLOS.ThisisthereasonwhyRFIDiswidelyusedinaccesscontrol,anti-counterfeitingandtrackingandtracingamongothers.

(2)

RealtimelocatingandRFID.RFIDisprimarilymadeforidentification,overthelastdecaderesearcheffortsfocusedontheuseofRFIDforrealtimelocating.Thereareplentyofcompetitiveand/orcomplementaryrealtimelocatingtechnologiestoRFID,eachofwhichdiffersinaccuracy,precision,complexityandcostamongotherfactors.

Wi-Fi(alsoreferredtoasIEEE802.11),thetechnologyusedforwirelessdeviceinterconnection,isprobablytheidealsolutionforlocatingdevicesequippedwithWi-Fitags,likelaptopsandPDAs,alreadyconnectedinawirelessnetwork.Wi-Firealtimelocatinghoweverishighlydependentonnetworkinfrastructure,ithasseriousscalabilityissuesanditmayintroducesomeburdenonthenetwork,letasidetheaffectionbyvariousenvironmentalconditionslikeobstacles,temperatureandhumidity,commonlymetinmostwirelesstechnologies.

AnotherrealtimelocatingtechnologysimilartoWi-FiisBluetooth,thewirelessnetworkingstandarddesignedforlowpowerconsumptionandcommunicationinapersonalareanetwork.Bluetoothisstandardized,widelyadopted,multipurposeandrelativelyaccurate.Nevertheless,therangeofBluetoothaccesspointsisrathershortandbecauseoftheinquiryprocess,thepositioningdelayisrelativelyhigh.AlongwithWi-Fi,Bluetoothtagsarenotsuitedforverysmallobjects.

UltraWideband(UWB)isanotherradiotechnologythatcanbeusedatlowenergylevelsforshort-rangehighbandwidthcommunications.UWBsystemsprovidehighaccuracythatcanbereducedtoafewcentimeters,howeverUWBsignalsinterferencethroughmetallicandliquidmaterialsisaconstantproblemandtheircostisprohibitiveatleastforsmallscaleapplications[2].

ZigBeeisalow-cost,low-power,wirelessmeshnetworkingproprietarystandard.Sinceitisstandardized,interoperabilityofequipmentfromdifferentmanufacturersisguaranteed.Alsoitishasexcellentperformanceinlowsignaltonoiseratioenvironmentsanditisfaulttolerant.Nonetheless,ZigBeeroutershaveshortrangeandtheirsignalhaslowpenetrationthroughwalls,andotherobstacles.

AsfarasRFID,bothpassiveandactivetagscanbeusedforrealtimelocating.Passivetagscanbeacquiredataverylowcostandcanbeattachedtoalmosteverything.Theyalsofacilitatehigherreadrates(approx.1500tagspersecond).Howevertheyhavealsolowtoleranceonharshenvironmentalconditionsandtheyrequirethepresenceofmultiplereadersandantennasinordertocoverwiderareas.Activetagsontheotherhand,improvetheaccuracyandtolerancebuttheyposeseriousmaintenancechallengessincetheyhavelimitedlifetime.Currently,thereisnobestrealtimelocationsensingtechnique.Eachtechnologyhasitsowndistinctcharacteristicswhenappliedinrealenvironmentsandthechoiceisclearlyamatteroftradeoffbetweenaccuracy,precision,systemcomplexityandsuitabilityinagivenenvironment[3].

2)

Videosurveillance

Videocameras,intheformofaclosedcircuittelevision(CCTV),havebeenwidelyusedinsurveillanceapplications,whereahumanoperatorevaluatesthecapturedeventstoprovidealert.Thesesystemsareinstalledmainlyinpublicspaces,wheresecurityisthemajorconcern.However,thehumanfactorinvolvedintheproceduregreatlyinfluencessurveillanceeffectiveness,duetofatigueorlackofconcentration.

IntelligentSurveillance(IS)systemshavebeendeveloped,toautomaticallydetectobjects,trackapersonandrecognizeaneventinordertoreactuponanyabnormalbehaviortakingplaceinascene.TheISsystemsarebeingextensivelyresearchedinliteratureandsolutionshavebeenproposedforanumberofapplications,especiallyinthecasewhereahumanoperatorcannotbeoffered.Recently,thesesystemshavebeenconsideredforpervasiveapplications,includingambienthome,e-healthande-caresystems.

However,ISsystemsstillpresentcertainlimitation.Severalcamerasarerequiredtocoveralargespace,eveniftheyareequippedwithwideanglelenses.Inthiscase,extensiveinstallationmodificationslimitacceptanceofvideosurveillancebyawideaudience,despitethefactthatinsomecasesinstallingcamerascanbeprovenuseful,suchasahome-assistingenvironmentsfortheelderly.

Furthermore,theimageprocessingmodulesofthesesystemsaresensitivetocomplexenvironmentalchangesandobjectocclusions,thatis,whenatargetedobjecthidesbehindanother.Additionally,videocamerasthatusuallyhavealowerresolutionthanstillcameras,cannotrobustlydetectsmallordistantobjects.

TheChargedCoupledDevice(CCD)thatvideocamerasareequippedwithalsopresentsnoise,bydesign,whichinturnhinderstheobjectlocationperformance.Finally,multicameraenvironmentsandthestoragecapabilitiestheyrequireleadtoextracostfortheacquisitionofahigh-endpcandstoragearrays.

Therefore,othersolutionsshouldbeconsideredthatwouldlimitthecostofthesesystemsandincreasetheireffectivenessinordertobecomemoreappealingforwideraudiences.

NEWWORDSANDPHRASES

prerequisite n. 先決條件

surveillance n. 監(jiān)視，監(jiān)督

arguably adv. 可論證地，正如可提出證據(jù)加以證明的那樣

inventory n. 詳細目錄，存貨，財產(chǎn)清冊，總量

pervasive adj. 普遍深入的

accidental adj. 意外的，非主要的，附屬的

nanotechnology n. 納米技術(shù)

emanate vi. 散發(fā)，發(fā)出，發(fā)源

actionable adj. 可控告的

elaborate vt. 詳細描述

hemispheric adj. 半球形狀的

pivotal adj. 樞軸的，中樞的，關(guān)鍵的

ubiquitous adj.到處存在的，(同時)普遍存在的

interrogator n. 盤問者，質(zhì)問者

backenddatabase 后端數(shù)據(jù)庫

inductive adj. 誘導的,感應的

backscatter n. 反向散射(背反射)

InternetofThings 物聯(lián)網(wǎng)

penetration n. 穿過，滲透，突破

interoperability n. 互用性，協(xié)同工作的能力

bar-code n. 條形碼

laptop n. 膝上型電腦

prohibitive adj. (費用等)高得負擔不起的,過分昂貴的

fatigue n. 疲乏,疲勞,累活

NOTES

[1]Theuseofhemisphericimagingcamerasthatmaximizetheareacoverageofasurveillancesystemisexamined,inordertoeliminatetheneedformultiplecamerasandthefactorsthatinfluencethepurposeofsuchasystemareidentified.

“thatmaximizetheareacoverageofasurveillancesystemisexamined,”是后置定語從句，修飾前面的“cameras”。“inordertoeliminatetheneedformultiplecamerasandthefactorsthatinfluencethepurposeofsuchasystemareidentified.”是為了說明為什么討論使用半球形攝像機。本句可譯為：為了減少對多個攝像機的使用需求討論了半球形成像攝像機，該攝像機可以使監(jiān)控系統(tǒng)的覆蓋范圍最大化，并且明確了影響系統(tǒng)目的的一些因素。

[2]UWBsystemsprovidehighaccuracythatcanbereducedtoafewcentimeters,howeverUWBsignalsinterferencethroughmetallicandliquidmaterialsisaconstantproblemandtheircostisprohibitiveatleastforsmallscaleapplications.

“thatcanbereducedtoafewcentimeters.”是后置定語從句，用來修飾“highaccuracy”。這里的“prohibitive”的意思是“昂貴的，不可承受的”。本句可譯為：UWB系統(tǒng)能夠提供小到幾個厘米的高精確度，但是UWB信號通過金屬和液態(tài)物質(zhì)時的干擾一直是常態(tài)問題，并且其系統(tǒng)成本至少對小規(guī)模應用來說是過于昂貴的。

[3]Eachtechnologyhasitsowndistinctcharacteristicswhenappliedinrealenvironmentsandthechoiceisclearlyamatteroftradeoffbetweenaccuracy,precision,systemcomplexityandsuitabilityinagivenenvironment.本句是一個復雜的并列句。全句可譯為：當應用到真實的環(huán)境中時每種技術(shù)都有自己獨特的性能。(在這之中的)選擇顯然是在準確度、精確度、系統(tǒng)復雜度和對特定環(huán)境的適應能力之間的平衡折中。

EXERCISES

Ⅰ．Translatethefollowingwordsorphrases.

InternetofThings locatingtechnology RadioFrequencyIdentification

licenseplate line-of-sight ChargedCoupledDevice(CCD)

anti-counterfeiting 普適計算 視頻監(jiān)控RFID

Ⅱ．TranslatethefollowingparagraphsintoChinese.

(1)

RadioFrequencyIdentification(RFID)technologyisarguablytheidealsolutionforobjectidentification.Ithassuccessfullybeenusedinalargevarietyofapplications,likeenterprisesupplychainmanagementforinventorying,trackingandofcourseobjectsidentification.RFIDmayalsoproveusefulforpervasivecomputing,forprovidingidentitytovirtuallyeverything.ItisnotaccidentalthatRFID,alongwithwirelesssensorandnanotechnologieshavebeencombinedtoformwhatisknownastheInternetofThings.

(2)

Videocameras,intheformofaclosedcircuittelevision(CCTV),havebeenwidelyusedinsurveillanceapplications,whereahumanoperatorevaluatesthecapturedeventstoprovidealert.Thesesystemsareinstalledmainlyinpublicspaces,wheresecurityisthemajorconcern.However,thehumanfactorinvolvedintheproceduregreatlyinfluencessurveillanceeffectiveness,duetofatigueorlackofconcentration.

(3)

Videosurveillancehasbeensuccessfullyappliedonanumberofsituations.Firstofall,camerashavebeeninstalledtoserveasimagesensorsandactasanon-intrusivemeansofincreasedsecurity.Secondly,videoprocessingmethodshavebeenextensivelyusedforindustrialapplications.Theindustryinspectionsystemsinclude,amongothers,qualitytextileproduction,metalproductfinishing,glassmanufacturing,machineparts,printingproductsandmanyothers.AnotherfieldinwhichvideosurveillancehasbeensuccessfullyappliedisthedevelopmentofIntelligentTransportationSystems.Theseincludeautomaticlanefindingandlicenseplaterecognitionsystems.Videosurveillancehasfoundrecentlyincreasedusabilityinbehavioralanalysis.Finally,manye-healthande-careapplicationshavebeenbenefitedfromtheuseofvideocameras.

參考譯文

第十八單元基于事件的半球形成像RFID一體化物聯(lián)網(wǎng)輔助應用

1．簡介

對大范圍遠程行為，比如政府設(shè)備、公共建筑物或工業(yè)環(huán)境的實時監(jiān)控是許多監(jiān)控應用的先決條件。這些(監(jiān)控)行為被證實在較小范圍空間內(nèi)比如家庭和辦公室，也是一個輔助性的、有用的補充。現(xiàn)代基于視頻的監(jiān)控系統(tǒng)使用功能強大的實時分析技術(shù)，應用廣泛。此外，為了提供大范圍的監(jiān)控而使用了多個攝像機，且為了保證監(jiān)控對象的清晰度而對監(jiān)控攝像機之間的協(xié)作提出了需求，這種協(xié)作主要是為了探測感興趣的事件，這些事件增加了系統(tǒng)的復雜度。

射頻識別(RFID)技術(shù)是經(jīng)過證明的對象識別的理想方法。它已經(jīng)被成功地應用在許多地方，比如企業(yè)存貨的供應鏈管理和跟蹤，當然還有對象識別。經(jīng)證明，RFID對普適計算和任何東西的識別也是有用的。RFID和無線傳感器以及納米技術(shù)已經(jīng)結(jié)合在一起形成所謂的物聯(lián)網(wǎng)。對幾乎每一個普適計算應用來說，另一個必需的要求是實時定位，這來自于對即時可控信息的內(nèi)在需求。以前，許多系統(tǒng)使用不同的技術(shù)解決了自動定位感應問題。本文中我們詳細討論發(fā)展視頻監(jiān)控和RFID結(jié)合在一起的混合系統(tǒng)的可行性，目的是能在視頻攝像機的輸出端為進行自動識別和目標追蹤提供一個可靠系統(tǒng)。為了減少對多個攝像機的使用需求討論了半球形成像攝像機，該攝像機可以使監(jiān)控系統(tǒng)的覆蓋范圍最大化，并且明確了影響系統(tǒng)目的的一些因素。

2．潛在的技術(shù)問題

1)

RFID概述

普適計算中的一個關(guān)鍵技術(shù)是RFID技術(shù)。RFID被劃分為寬泛的自動識別技術(shù)，它是一個用來描述通過無線形式用無線電波來傳送對象識別(以獨一無二的序列號的形式)系統(tǒng)的通用詞匯。一個典型的RFID系統(tǒng)由(以下)幾部分組成：①RFID標簽，包含了一個與它連接的物理對象相關(guān)的數(shù)字號碼；②RFID讀卡機(也可稱為詢問應答機)，通常和后端數(shù)據(jù)庫相連。讀卡機還安裝了天線、轉(zhuǎn)發(fā)器和處理器，處理器為了詢問標簽并讀懂內(nèi)容而廣播無線信號。

RFID系統(tǒng)最重要的兩個特征在某些環(huán)境下的范圍、數(shù)據(jù)轉(zhuǎn)發(fā)傳輸上有不同之處：①能量來源和RFID標簽的計算能力；②操作頻率。按照第一個特征，RFID標簽可分為有源和無源的或二者的結(jié)合物。有源標簽使用電池且能夠在大的作業(yè)范圍內(nèi)高質(zhì)量地自主傳輸信號，但是價格昂貴且通常體積較大。另一方面，無源標簽需要外部能源來激勵信號的傳輸，這個過程需要利用感應耦合或電磁捕獲技術(shù)，它們利用輸入模塊和讀卡機通信。由于無源標簽技術(shù)的成本低、體積小以及無限的實際使用壽命，所以它們應用得比有源標簽廣泛，在很多情況下更受歡迎。此外，RFID系統(tǒng)可分為四種操作頻率：低頻、高頻、特高頻和超高頻(或微波)。隨著頻率的增加，傳輸范圍和傳輸速率隨之增加，但是對水和物質(zhì)(比如金屬)的穿透力卻隨之降低。

建立統(tǒng)一的工程標準、方法、過程和定義RFID系統(tǒng)特征的實踐的需求引發(fā)了不同標準(的建立)，其中兩個標準尤其突出。EPC全球定義了一個劃分標簽的復合方法，詳細說明了唯一的識別數(shù)字(電子產(chǎn)品編碼)、頻率、耦合方法、電鍵調(diào)制類型、信息存儲能力和彼此之間的互操作模式。類似地，ISO(與IEC組織聯(lián)合)開發(fā)了識別、讀卡機和標簽之間通信、中間件協(xié)議和測試、兼容和安全的標準。

(1)識別和RFID。通常，普適系統(tǒng)的一個十分重要的要求是能夠唯一地識別物品和實體，而RFID與生俱來就滿足這種要求。還有其他幾種技術(shù)適合這個目的，每種都有優(yōu)點和缺點，但是在大多數(shù)應用領(lǐng)域都被RFID強勢取代了。

條形碼是光學的、機器可讀的、一次寫的物品分類表示，它依然是供應鏈管理中應用最廣泛的跟蹤產(chǎn)品，也是最便宜的識別方法。條形碼技術(shù)的最大缺陷是不能提供單個物品的額外信息(但是涉及這個問題的二維象征學)和對條形碼物品需在瞄準線以內(nèi)(LOS)的要求?？ㄆ夹g(shù)是另一類物體識別技術(shù)和相應的實體，它包括磁卡、智能卡和光學卡。這些卡片通常都內(nèi)嵌有信用卡體積大小的塑料卡片，它們包含了磁條或集成電路，并且具有較大的存儲能力。一旦它們包含了微處理器，就具有增強的處理能力，這樣這些卡片就可被運用在苛刻的應用中，比如可靠性應用中(但是具有較大的限制)。不過，大部分卡片都要求緊貼或非?？拷x卡器。在應用中，卡的成本也隨著卡的特性(的上升)而顯著上升。RFID技術(shù)在功效和成本效率之間達到了平衡，同時降低了對瞄準線以內(nèi)的需求。這就是為什么RFID能夠在接入控制、反偽造和跟蹤追查中能夠廣泛應用的原因。

(2)實時定位和RFID。RFID主要用作識別，過去十年來(對RFID)的研究成就集中在RFID用作實時定位上。還有許多RFID的競爭性或補充性的實時定位技術(shù)，它們的差別在于精確度、精密度、復雜度和成本。

用作無線設(shè)備互連的Wi-Fi(也稱為IEEE802.11)技術(shù)大概是配置有Wi-Fi標簽的定位裝置(用來定位)的完美方法，比如膝上型電腦和掌上電腦已經(jīng)能(使用Wi-Fi技術(shù))連接到無線網(wǎng)絡(luò)中。但是，Wi-Fi實時定位技術(shù)對網(wǎng)絡(luò)基礎(chǔ)設(shè)施依賴性較高，它具有嚴重的擴展性問題，引入了相當?shù)木W(wǎng)絡(luò)負擔，容易受到大多數(shù)無線技術(shù)都能碰到的惡劣環(huán)境條件下問題的影響，比如障礙物、溫度和濕度。藍牙(Bluetooth)是與Wi-Fi相似的另一種實時定位技術(shù)，它是為低功耗和單人范圍內(nèi)網(wǎng)絡(luò)通信而設(shè)計的無線組網(wǎng)標準。藍牙是標準化的、被廣泛采用的、多用途的且相對精確的(定位技術(shù))。然而，藍牙的接入點范圍相當短，并且由于它的詢問過程(較長)，因此配置時延相對較長。和Wi-Fi一樣，藍牙標簽不適合非常小的物體目標。

超寬帶(UWB)是另一種可用在低功耗、短距離、高帶寬通信環(huán)境的無線技術(shù)。UWB系統(tǒng)能夠提供小到幾個厘米的高精確度，但是UWB信號通過金屬和液態(tài)物質(zhì)時的干擾一直是常態(tài)問題，并且其系統(tǒng)成本至少對小規(guī)模應用來說是過于昂貴的。

ZigBee技術(shù)是一種低成本、低功耗的無線網(wǎng)狀網(wǎng)組網(wǎng)私有標準。因為它已經(jīng)標準化過了，所以對來自不同廠商的設(shè)備的互操作能力是可以保證的。它在低信噪比環(huán)境下也具有優(yōu)越的性能，且具有容錯能力。然而，ZigBee路由器的工作范圍短，信號對墻壁和其他障礙物的穿透力較弱(衰減大)。

至于RFID，無源和有源標簽都可以用作實時定位。無源標簽成本非常低，可以安裝連接到幾乎任何物體上。它們也適合較高的閱讀速率(大約為1500個標簽/秒)。但是無源標簽對惡劣環(huán)境的容忍性較差，且為了達到對較寬區(qū)域的覆蓋，它們需要許多個讀卡器和天線。另一方面，有源標簽提高了精確度和容忍能力，但是因為有源標簽的使用壽命有限，它們對系統(tǒng)維護提出了嚴重的挑戰(zhàn)。當前，沒有最好的實時定位感應技術(shù)。當應用到真實的環(huán)境中時每種技術(shù)都有自己獨特的性能。(在這之中的)選擇顯然是在準確度、精確度、系統(tǒng)復雜度和對特定環(huán)境的適應能力之間的平衡折中。

2)視頻監(jiān)控

閉路電視形式的視頻監(jiān)控(CCTV)已經(jīng)被廣泛應用在使用人類操作員的監(jiān)控應用中，人在此處估計事件，以此來提供警惕。這些系統(tǒng)主要安裝在安全受到主要關(guān)注的公共場合。但是，由于人會疲憊或注意力不集中，(所以)監(jiān)控過程涉及的人的因素極大地影響了監(jiān)控的有效性。

智能監(jiān)控(IS)系統(tǒng)已經(jīng)被開發(fā)出來進行目標的自動檢測，以對發(fā)生在現(xiàn)場的任何反常行為進行反應，從而跟蹤人跡及識別事件。很多文獻已經(jīng)對IS系統(tǒng)進行了廣泛研究，并有許多應用，尤其是在沒法使用人類操作員的情況下提出了解決辦法。最近，這些系統(tǒng)已經(jīng)被考慮應用在普適應用中，包括(應用在)家庭、電子健康和電子照看系統(tǒng)中。

但是，IS系統(tǒng)仍然有局限性。即使攝像機配備了廣角鏡頭，為了覆蓋大的空間范圍仍需安裝很多攝像機。在這種情況下，由于廣泛安裝視頻監(jiān)控不能被公眾廣泛接受，這也限制了IS系統(tǒng)的接受度。盡管在一些情況下安裝攝像機被證明是有用的，比如家庭有老人需要照看的情況下。此外，這些系統(tǒng)的圖像處理模塊對復雜環(huán)境的變化和對象掩蔽很敏感，對象掩蔽是由跟蹤對象隱藏在另一個對象后面引起的。此外，視頻相機(通常比靜態(tài)相機的分辨率低)不能有效監(jiān)測小的或遠的物體。

視頻攝像機配備的CCD設(shè)備的噪聲也影響了目標定位性能。最后，多攝像機環(huán)境和對存儲能力的要求導致了額外的成本，這些額外成本來自購買高端PC和存儲陣列。

因此，為了使設(shè)備對用戶有更大的吸引力，可以考慮其他的解決辦法來限制這些系統(tǒng)的成本，提高它們的有效性。

EXTENSIVETEXT

AnalyticsfortheInternetofThings

1．Introduction

Nowadays,thealmostubiquitousInternetisaccessednotonlybytraditionalcomputingmachinery,butalsobynewtypesofdevices,connectedproducts.Theseproductsofferacertain,oftenspecializedfunctionalitytousers,enabledbyaworld-wideconnectiontoseveralonlineservices.ThedevicesthatformtheInternetofThingsarediverseandmayrangefromRFIDtagequippedpackagingmaterialtocomplexelectronicproducts.Thelatteraimatentertainment,mobileapplications,orproductivity,andspecializeontasksthatcouldbecarriedouteasilybypersonalcomputers,butareplacedintonewcontexts,realizeabetteruserexperience,orpackagefunctionalityaspectsinanew,cheaper,ormorerobustway.Sincetheseconnecteddevicesimplementastrongrelationshiptotheusagecontext,evaluationhastobecarriedoutundersimilarconditions,dataacquisitionintheuser’shabitualenvironmentbecomescrucial.

However,atthesametime,methodsfordoingsuchanevaluationmustevolve,too.Duetohighmarketpressureandshorterdesignanddevelopmentcycles,userexperiencetestingwhichisdoneinthetraditionalwayseemstobetooslowtobeacceptableintermsoffeasibilityandefforttodesigners,engineers,qualityexperts,inshort,domainexperts,incharge.

TheInternet,beingthetechnicalbasisforsuchdevicesandrelatedbusinesscases,canalsoserveasasettingthatenablescheapandlight-weightuserexperienceevaluationservices.WewouldliketodrawananalogytoGoogle’sAnalyticsservice(andalsotootherwebsitestatisticsproviders)thatprovideseasy,butpowerfulwebsiteusagestatisticstowebsiteownersaccessibletobothsmallhomepagesandhugeenterprise-gradewebportals.Thisserviceoffersaverysimpleintegrationmechanism(bymeansofJavaScriptsnippetsintegratedintotheHTMLsourceofthewebpage),andpowerfultoolstoleveragetheacquireddata.Verysimilarly,ourapproachaimsatreducingtheeffortneededtointegrateobservationintoproducts,whileprovidingextensivetoolsfor(creatively)workingwiththegainedobservationandanalysiscapabilities.Theapproachpresentedhereaimsatchangingtheevaluationprocesstowardsamoreparallelandstreamlinedactivity.Thisallowsmultipledomainexpertstocarryoutexperimentsatthesametime,aswellasitallowsthentochangeexperimentson-the-flytoderivemorerefineddata.

Intheremainderofthispaper,wefirstpointatrelatedwork,thenshowtheapproach.Thisisfollowedbyadescriptionofourimplementation.Finally,weconcludethepaperandgiveanoutlookonfuturesteps.

2．Relatedwork

Theuseofmonitoringsystemstoacquireusageinformationfromremotesystemsisnotnew.Suchapproacheshavebeenincorporatedbothformaintenancemonitoringandfault-detectionpurposesaswellaswaystocollectdatabeneficialforevaluationtasks.Theresearchpresentedhereandinsomedocumentdiffersinthreemainpointsfromsimilarapproaches:theuseofamoreuser-friendlyvisuallanguageforobservationspecification,thecapabilitytoredefineexperimentson-the-fly,andtheuseofanengineeringapproachthatrendersthetechnologyapplicabletoawidespectrumofconnectedproducts,broadeningthecircle.

3．Approach

Iterativeobservationisaprocessdividedbasicallyintofourparts:①conceptualization,②observationdefinition,③automaticdatacollection,and④dataanalysis.Withahighdegreeofautomationinstep3,humancapabilitiesandexperiencecanfocusonessentiallyhumantasks,suchasconceptualization,definitionanddissemination.Fig18.1showstheconnectionoftheseprocessparts.Fig18.1InterativeapproachtocollectmeaningfulandrelevantdatawiththeD’PUISframework

Intheconceptualizationphase,domainexpertsthinkaboutthesemanticstructureofthedatatobecollected.Oftenthisinvolvesalsosemanticallystructuringtheproduct’sfunctionalityornon-functionalaspectssuchasuserexperience,fault-tolerance,orperformancecharacteristics.Thisknowledgeshouldbeformalizedintermsofontologies(graphsofsemanticconcepts),tobeusableasinputfortheobservationframework.

Duringtheobservationdefinitionphase,moreorlessvaguequestionsaboutuserbehaviorarecastintomoreelaboratemetrics,e.g.elaborating“HowoftenisfunctionAused?”to“WhatistheaveragenumberoffunctionAusedtriggersperdayperuser?”Similartothefirstphase,thisinformalmappingtaskresultsinaformalspecificationofobservationwhichisadirectinputfortheobservationframework.Inthisphase,thedatacollectionisparameterizedonanabstractlevelusingasimplevisuallanguage.Thislanguageallowsforagraphicalconnectionbetweendatasources,processingnodes,semanticnodesanddatahandlingblocks,intuitivelyvisualizingaflowofinformation.

Subsequently,inthedatacollectionphase,observationspecificationsareautomaticallydistributedtoallconnectedproducts(that“participate”intheexperiment),andobservationmoduleswithintheproductsstarttocollectdataaccordingtothespecifications.Thedataitemsarefedbacktoacentraldatacollectionunitandremaininastoragespace.

Whilethedataiscollected,thecentraldatacollectionunitprovidesareal-timeviewofaggregateddataandalsoallowsforvisualizationsofincomingdata.Thisreal-timevisualizationcanoftenanswerfirstquestionssuchas,whethertherightandrelevantdataisbeingcollected.Inaddition,ithelpstohighlightinterestingpartstheoveralldataspacethatrequireddeeperdiggingand,thus,achangeofobservationspecifications.

However,theclosingofthecycleiscrucial.Collecteddatacanbeaccessedandanalyzedatanypointintimeduringtheexperiment,and,consequently,thedatacollectionprocessitselfcanbeadapteddynamicallyanytime.Moreover,differentdatacollectionmechanismscanworkinparallel,eache.g.usedbyadifferentdomainexperttoacquireadedicatedviewonproductusage.Thisapproachpresentstheexperimenter(oftenalsoagroupofexperimenters)withawaytomaximizetheuseofcostlyexperimenttimeandparticipants.

Fig18.2visualizestheoutcomeofsuchanapproach:duringafirst–explorative–