网络中的数据管理的挑战外文翻译

WuxiProfessionalCollegeofScienceandTechnology英文资料翻译系别物联网技术学院专业物联网应用技术班级物联网学生姓名 学号 指导教师 ChallengesforDatabaseManagementin

theInternetofThingsThisarticlediscussesthechallengesforDatabaseManagementintheInternetofThings.WeprovidescenariostoillustratethenewworldthatwillbeproducedbytheInternetofThings,wherephysicalobjectsarefullyintegratedintotheinformationhighway.WediscussthedifferenttypesofdatathatwillbepartoftheInternetofThings.Theseincludeidentification,positional,environmental,historical,anddescriptivedata.Weconsiderthechallengesbroughtbytheneedtomanagevastquantitiesofdataacrossheterogeneoussystems.Inparticular,weconsidertheareasofquerying,indexing,processmodeling,transactionhandling,andintegrationofheterogeneoussystems.Werefertotheearlierworkthatmightprovidesolutionsforthesechallenges.FinallywediscussaroadmapfortheInternetofThingsandrespectivetechnicalpriorities.TheInternetofThings(IoT)isaself-styledtermtodescribeobjectsthatareabletocommunicateviatheInternet.Objectsrangefromsensorinputstoactua-torsthatcontrolphysicalobjectswithnewinteractionsrequiringadvancesinmachineandhumaninterfaces.ItiswidelyforecastthattheseobjectswillnumberinthetrillionsoverthenextfiveyearsofInternetdevelopment.Halleretal.haveprovidedthefollowingdefinition.“Aworldwherephysicalobjectsareseamlesslyintegratedintotheinformationnetwork,andwherethey,thephysicalobjects,canbecomeactiveparticipantsinbusinessprocesses.Servicesareavailabletointeractwiththese ‘smartobjectstheInternet,querytheirstateandanyinformationassociatedwiththem,takingintoaccountsecurityandprivacyissues. ”Historically,theIoTreferredmainlytoRadio-frequencyidentification(RFID)taggedobjectsthatusedtheInter-nettocommunicate.Itsoriginslieinthemanufacturingarea,forexample,theAuto-idproject.TheCambridgeAuto-idlaboratoryproducedanumberofwhitearticles,journals,andconferencearticlesontheproject.RFIDisnottheonlymeansofconnectiontotheloT.Wirelesssensornetworkswillprovidecontinuousstreamsofdataonvariousenvironmentalcharacteristics,whichmaybefedintotheIoT.OthermoresophisticatedbridgestotheIoTincludeidentificationofobjectsviasensingdevices,forexample,objectrecognitionviadigitalimaging.Evermorecapabledisplaypersonal.Takentoitsextreme,anyeverydayobjectmightbecomepartoftheIoTandbemadeintelligent:eachbookweread,everydeviceinourhomes,ourpets,everyfoodproduct,everyitemofclothing,andevenourselves.Ofcourseonecanimaginegoodandbadscenariosinthisvision.Itmightbeconvenienttoarrivehomeafteraperiodofabseneetoareceptionofourmostcomfortablehomeenvironmentintermsofheating,lighting,anddigitalentertainment.Thisenvironmentissetupandmaintainedeconomicallyandenergy-efficientlythroughsensorsthatcaptureeveryenvironmentaldetail,makingdecisionsbasedontheirinput,knowledgeofcommod-itytarifffactors,andourpreferences.Letusconsideranothercase.Wecannotfindthebookwearereadingsowepressabuttononthescreenembeddedintoourlivingroomwallandinaninstaneethelocationofthebookisrevealed.Orevenmorefuturisticallythebookdetectsourarrivalhome,knowswewanttoreadit,andthereforebleepstoindicateitspresenee.TheIoToffersawealthofpossibilities,butperhapsthereisadownside.Ourpossessions,movements,andactivitiesmightbedetectedbyroguereceiversoperatedbypeoplewithintentionstoputsuchinformationtoillegaluse.Evenoverzealouslegitimatedatacollectioncanseemaninva-sionofprivacy.Securitysystemsarenotfail-proof.Wouldwewishsurveillaneeagentstoknoweverythingaboutus,forinstaneeourreadinghabits,themusicwelistento,thefoodweeat,withwhomweassociate,andwhereweare?Suchmattersneedtobecarefullyeonsidered.Size,ScaleandIndexingThesizeandscaleofthedataintheIoTwillbevast.Datawillneedtobemanagedviaresponsiblelocalownership.Localownerswilldecidewhichdataandservicestomakeavailabletotheglobalnetwork.Thus,theIoTmayoperateonmorethanonelevel:privateandpublic.Usersmayjoingroupsforaccesstocertainprivatelyowneddataormay,ontheotherhand,accessdatapubliclyavailableoverthepublicInternet.Theremaybediffer-encesinqualityofdatadependingonownershipandlevelofcare.Graduallytrustandreputationsystemswillprovideinformationtousersonthequalityofthedata.Intheglobalspacetherewillbeaneedforacentralauthority,formanagingaddressesandidentifiers,asthereiswiththecurrentInternet.Indexingwillbeamajorchallenge.FindingaparticulariteminaworldwhereallphysicalobjectshaveanIPaddresswillnotbeeasy,uniesswecandevisesuitableindexingmethods.Workinthelibrarycatalogmanagementmightprovidesomepoint-eraonhowtodothis,buttheIoTwillencompassmanydifferenttypesofobjects.Creatingacatalogofeverythingintheworld,readableacrosscountriesandIanguages,isadauntingtask.Someobjectswillbepubliclyaccessible;somewillneedvariouslevelsofaccesscontrol,andsomemaybeprivatetotheowner.AtfirsttheIoTislikelytodevelopthroughlocalsystemsthatcanbeindexedcoher-entlywithinaboundeddomain.Aslocalsystemsmergewithglobalsystems,newindexingmethodswillneedtobedeveloped.Categoriesofthingswillneedtobedefinedtogetherwithsubcategories.Specializedsearchsitesmayprovideaccesstocertaincategories.Forinstanee,ifonewantstofindaparticularcarpart,onemaygotoasearchsitethatspecializesinthattypeofproduct,andfromtherebeguidedtoaspecificIPaddress.QueryLanguagesCurrentpopularqueryIanguagesindatabasesystemsrelyonstructureddata.StructuredQueryLanguage(SQL)isthemostprominentexample.Overthelastfewyears,however,therehavebeenproposalsforqueryIanguagesforsemi-structureddata,whichismoretypicalofthedataheldontheInternetthequantitiesofdataaresovastthatitwouldbeunrealistictoexpectanysortofuniformstructure,exceptperhapsthatoftheloosestvariety,tobeimposedontheIoT.ExtensibleMarkupLanguage(XML)offersameansofrepresentinglessstructuredaswellasstructureddata,togetherwithsomelevelofselfdescription.Itisawell-acceptedtechnologythatsupportsinteroperabilityatatechnicalratherthanasemanticlevel.XQueryhasbeendevelopedbytheWorldWideWebConsortium(W3CXQueryaIanguageforqueryingXML,cancombinedocu-ments,webpages,andlinkstorelationaldatabases.QueryIanguagesforsemi-structureddatausuallyadoptanunderlyingunidirectionalgraph.RequiredobjectsarespecifiedbyprovidingapathexpressioninaIanguagethatisusuallyquiteintuitive.Thereare,however,inherentproblemswithhierarchicaldatamodels,suchas,difficultyinrepresentingmany-to-manyrelationships.Inspiteofthis,thehierarchi-caldatamodelhasbeenembracedbythewebcommunityasauseful,intuitive,andpracticalstructure.TheIoTwillhavebrieflyvisitasitetopickupsomedataorinformation,expertusersthatknowexactlywhatdatatheyneedandwheretofindit,andusersthatliesomewhereinbetween.Infactindifferentcontextsthesamepersoncanbeanyoneofthesedifferenttypesofusers.Itthereforeseemsnecessarythatdifferenttypesofdataaccessfacilitiesbeavailable.CasualuserswillneedtoaccesstheIoTviaauser-friendlygraphicaluserinterface(GUI),withdetailedexplanationavailableonanyobject,andmoreflexible,powerful,andefficientaccessinterfaceswillbeneededforexpertusers.Servicescanbeusedtoprovidebothtypesofaccess.WorkinqueryIanguagesforsemi-structureddatawillberelevantforthesedevelopments.ProcessModelingandTransactionsItislikelythatmostprocesseswillbedevelopedandsuppliedasservicesontheIoT.ServiceOrientedArchitects-ture(SOA)is becoming an important meansof supportinginteroperabilityinweb-basedsystems.Thecentralideaisthatindependentoutfitsofferservicesinauniformmanner,whichotheruserscanthentakeup.Thusimple-mentationdetailsarehiddenfromtheusersoftheservices.Applicationprocesseswilltypicallybemadeupofanum-beroflowerleveltransactions.Transactionsinturnwillbemadeupoflowerleveloperationsorservices.Therefore,thequestionoftransactionprocessingintheIoTarises.InthetraditionaldatabasesystemsthematterofconcurrenttransactionprocessinghasbeenhandledthroughthemaintenanceofACIDpropertiesthroughtimestamping,locking,andatwo-phasecommit.ACIDpropertiesareatomicity,consistency,isolation,anddura-ability.Atransactionmustcompleteinitsentiretyornotatall,atransactionmustleavethedatabaseinaconsistentstate,transactionsshouldnotshowothertransactions,andintermediateresultsandchangesmadebyatransac-tionmustbepermanent.Indistributeddatabasesystemsatwo-phasecommitisusedtopreserveconsistency.Allparticipatingsitesmustconfirmtheirreadinesstocommitbeforethecommitcommandisissuedbythecoordinatingsiteandwrittentothedatabaselog.IthasbeenrecognizedthattheACIDpropertiesdonotfitwebtransactionprocessingwell.Thisisbecausetheindividualwebservicesareessentiallyautonomousandmustindependentlypreserveconsistency.Thisrequire-meantmightconflictwithaconsistencyrequirementofauser'globaltransaction.Forinstaneeauserseesbookingaholidayasatransactionconsistingoftwooperations:bookingaflightfromoneoperatorandbookingahotelroomfromanotheroperator.Asfarastheuseriscon-cernedthetransactionshouldcompleteinitsentiretyorfail.However,theoperatorsofthetwobookingsystemsmaybeindependentandonebookingmaybesuccessfulwhiletheotherisnot.Astheunderlyingsystemsinthisexampleareindependentitisnotfeasibletoimposeatwo-phasecommitprocedure.Newmethodsandmodelshavebeenproducedforweb-basedtransactionhandlingTheseincludetheuseofcompensatingtransactionsandtransactionsystemsthatrelaxvariousACIDprop-erties.IthasbeenfoundthatthemaintenanceofACIDpropertiesisnotrequiredbyallapplications.Sometimes,increasedthroughputoftransactionsmaybeagreaterprioritythanpreservingACIDproperties.HeterogeneityandIntegrationSection2hasoutlinedthemanydifferenttypesofdatathatwillneedtobehandledintheIoT.TheIoTwillfurthermoreconsistofbillionsofindependentnodes,whichwillhavetheirownsystemsforholdingthedata.Interoperabilitywillnotbeachievedwithoutastandardapproachatsomelevelofabstraction.Inthecontextofdatabasestheareasofheterogeneityandintegrationhavebeenresearchedsincethe1980s,onceitwasconsideredusefultoachieveinteroperabilityacrossheterogeneoussystems[40-44].Consideringthatonemighthaveapersonnelsystemstoredatonecompanyinarelationaldatabasesystem,andinanothercompanyasimilarsystemmightbeheldinanetworkdatabasesystemorevenadifferentrelationaldatabasesystem,questionsariseastohowtointegratesuchdata.Varioussolutionshavebeenoffered.Somepromisingsolutionssuggesttheuseofacanonicaldatamodel,forinstaneeafunctionalorbinarydatamodel.However,itseemsthatoftenthesolutionsoffereddonotwarranttheeffortsneededtoachievethem.Nowwiththeabundanceofdataanddifferentsystemsonthewebtheproblemsofheterogeneityandinteroperabilityariseanew.XMLhasplayeditspartinofferingasolutiontosomedegree.Ithasofferedatechnical,practical,andefficientmeanstopassdatafromonesystemtoanother.However,XMLdoesnotsolvethesemanticproblem.//Forinstanee,doesthedataitem“studentin”nesystemmeanexactlythesameasdataitem“studentindifferent-entsystem?Thisquestioncannotbeansweredaccuratelywithoutdomainknowledge.Effortstocapturedomainknowledgerevolvearoundtheconceptsofontologyandthesemanticweb.Ontologesdefineconceptsandtherelationshipsthatexistbetweenthem.CurrentworkhasrootsintheArtificialIntelligenee(Al)andknowledgerepresentationworkofpreviousdecades.OWL(WebOntologyLanguage)isafamilyofIanguagesforrep-resentingontologyontheweb(W3C2004).TheideaisthatcommunitieswillagreeoncommontechnologiesandrepresenttheseusinganOWLsystem,whichwillinturnprovidethenecessarysupportforsemanticinteroperable-ity.Inthefuture,agentswillplayaroleinusingsemanticinformationtosupporttheimproveduseoftheloT.Rellermeyeretal.considerthelackofascalablemodeltodevelopanddeployapplicationsatopaheterogenouscollectionofubiquitousdevicesasoneofthebiggestchallengesinmakingtheloTareality.Theyproposeamodelbasedonanextensionoftheideasalreadyinuseformodularsoftwaredevelopment.ItislikelythattheSOAwillplayanimportantroleinprovid-ingafabricintowhichtheheterogeneousapplicationsoftheloTcanbeweaved.TimeSeriesAggregationTimeseriesaggregationisaninterestingarea,whichhasbeennotedasraisingchallengesinvariousapplica-tiondomains.Ithasbeenrecognizedthatinappropriatetimeaggregationscangiverisetospuriouscausality.Theproblemrevolvesaroundtheabilitytoselecttheoptimalsamplingperiodforcontinuousdata.Trade-offsineludeprocessingtimeandstoragespaceagainstaccuracyandrealisticrepresentation.Inthedatabasefield,interestingworkhasbeencarriedoutinthestreamdatacapture.Ithasbeenrecognizedthatnewmodelswereneededfordatastreams.Anumberofarticlesonthetopicsofdatabasesystems,datastreams,streammining,classification,andsum-marizationhavebeenproduced.Ithasbeenrecog-nizedthattraditionalqueryIanguagessuchasSQLarenotsuitableforqueryingtimeseriesdata.Otherworkhasconsideredhowmissingdata,whichmightoccurthroughexceptionssuchaspowerbreaks,canbeestimated.ThesedevelopmentswillbeimportantforintelligentdatastreamingcapturesystemsintheIoT.IntheIoT,theoptimaltimesamplingperiodwilldependverymuchonthenatureofthedataandtheapplicationarea.Suitablequeryingfacilitieswillneedtobedefined.Thesearequestionsthatwillneedtobeaddressedbydataownerswhowilloffersamplingservicesontheircontinu-ousdata.WorkthathasbeenongoingindatastreamingwillcontributetothestreamingservicesoftheIoT.Aninterestingnewideaisdata-centricmiddlewareforcontext-awarepervasivecomputing,wherecontextualdatadrivesbothapplicationbehaviorandserviceadap-tationinsidethemiddlewaresystemwheresensorsaretreatedasdatastreampublishers.Adaptablesche-mascouldbecomepartofthesolution.ArchivingInrecenttimestherehasbeenmuchinterestinarchivingtheInternet.Anon-profitorganization,InternetArchive,wasfoundedin1996,tobuildanInternetlibrary,whichincludedanarchiveofallwebpages.Theaimoftheorganizationwastooffer“permanenaccessforresearchers,historians,andscholarstohistoricalcollectionsthatexistindigitalformatandtostopdigitalpublicationsdisappearing[71].Thewebarchivecanbeaccessedviathe‘Wayback'machine.AstheInternetissovast,archivingisoftendoneautomaticallywithcrawlers,whichtakecopiesofwebpagesatdefinedintervals.InternetArchiveissupportedbyAlexaInformation,theNationalScieneeFoundation,theLibraryofCongress,andotherinstitutions.DatabasearchivingisaIong-establishedtechniquethatinvoIvestakingcopiesofentiredatabasesatspecifiedintervalsandkeepingtheminasecurestore.InawebcontextarchiveddatabasescanbeconvertedtoXMLandbasicqueryingcanbepermitted.Processescanalsobearchived.Thisisachievedbyusingthecapturesoftwarethattakescopiesofeachrequestandresponseenteringaservice,bethatawebservice,adatabase,oranyothersoftwaresystem.SuchtechniqueswillbeapplicabletotheIoT.InterestingareasarisewhenweconsiderthenatureofsomeofthedataandthemagnitudeoftheIoT.Thesolutionislikelytobelocalmanagementofarchiveddatawithgoodindexinganddiscoveryfacilities.Inthearchiveddata,themainoperationwillberetrieval,updatewillonlybenecessaryinexceptionalcases.Thissimplifiestheproblemspace.Interestingareaswillrevolvearoundefficientstorage,querying,andperfor-mance.Workindatawarehousinganddataminingmayoffersomedirections.DataProtectionTheIoTislikelytoholdmuchmorepersonalinformationthanisheldonthepresentInternet.Furthermore,accesstosuchinformationtechnicallyisborderlessfromthepointofviewofnationalboundaries.ManycountriessuchastheUKhaveDataProtectionlaws.TheUKDataProtectionActrequiresallorganizationswhichhandlepersonalinformationtocomplywithanumberofimportantprinciplesregardingprivacyanddisclosure.Unlessexempted,allUKdatacontrollersofpersonalinformationhavetoregisterwiththeInformationCom-missioner'ffice.CountriesintheECoperatealongprinciplessimilartotheUKlaw,themainprincipleofwhichisthatdatamuchbeprocessedfairlyandlegiti-mately.Anotheristhatsuitablesafeguardsmustinplacetoprotectthedata.Thereareeightprinciplesaltogether.PersonaldataintheIoTwillbehardertoprotectbecauseofitssheerquantityandtheinterconnectednessofsys-tems.OutsidetheEC,othercountriesalsohavesimilarlaws.FortheIoTtobeasuccess,however,protectionmechanismsandalegalframeworkthatcanworkacrossnationalbordersisessential.Someworkhasbeencarriedoutindevelopingdataprivacytaxonomy.网络中的数据管理的挑战本文讨论了网络中数据管理的问题。我们提供了一些情况来说明新的世界将由物联网的产生，而造成物理对象完全集成到信息的高速公路。我们讨论的数据管理问题，将物联网的分成不同类型。这些类型包括标识，位置，环境，历史，和描述性数据。我们认为，需要管理大量的异构系统之间的数据所带来的挑战很大。特别是，我们考虑的范围查询，索引，过程建模，事务处理，和异构系统的集成。我们早期的工作可能提供对这些挑战的解决方案。 最后，我们讨论的事情和各自的技术优势的互联网地图物联网（IOT）是一个自封的术语来描述，可以通过互联网交流的对象。对象范围从传感器输入的实际因素，控制新的相互作用需要在机器和人类进步的物理对象的接口。人们普遍预测，这些对象将在互联网发展的下一个五年以亿计的数量。哈勒等人。提供了以下的定义。“一个世界里，物体的无缝集成到信息网络，并在他们，物理对象，可以成为企业积极参与的过程。提供服务的互动与这些“智能对象”在互联网上，查询其状态和任何相关的信息，考虑到安全和隐私问题。”从历史上看，物联网主要是指无线射频识别（RFID）标签的对象，使用跨网通信。它的起源是在制造业领域，例如，汽车项目。剑桥Auto-ID实验室产生了一些白色的文章，期刊，并在项目会议用品.RFID是没有连接到物联网的唯一手段。无线传感器网络将提供各种环境特性的连续数据流，这可能是美联储为物联网。其他更复杂的桥梁，物联网包括识别物体通过传感装置，例如，对象识别通过数字成像。永远的显示能力，个人。物联网提供了丰富的可能性，但或许也有一个缺点。我们的财产，运动和活动，可能会检测到流氓接收机由人们刻意把这种信息的非法使用。 甚至过分热心的合法的数据收集看上去都是在侵犯隐私。安全系统是不会失败的证明。我们将希望监视代理知道关于我们的一切，比如我们的阅读习惯，我们听音乐，我们吃的食物，与我们联系，我们在哪里？这种事需要仔细考虑。物联网中的数据问题1.大小，规模和索引在物联网中的数据的大小和规模将是巨大的。数据必须通过当地负责制。当地业主将决定哪些数据和服务可用的全球网络。因此，物联网可以操作多个水平：私人和公众。用户可以加入团体或访问某些私有数据，另一方面,在公共互联网访问公开数据。可能会有不同的所有权和水平取决于安全数据质量比较。 逐渐的信任和信誉系统将用户对数据的质量提供信息。全球空间中会有一个中央集权的需要为目前的互联网管理地址和标识。但物联网将包括许多不同类型的对象。索引将是一个重大的挑战。在一个世界里，所有的物理对象有一个 IP地址找到一个特定的项目将是不容易的，除非我们能够设计合适的索引方法。在图书馆的目录管理工作可能会提供一些方法帮助我们如何做到这一点。创造世界万物的目录，读过的国家和语言，是一项艰巨的任务。创造世界万物的目录，读过的国家和语言，是一项艰巨的任务。一些对象将被公开，有些人会需要不同级别的访问控制，以及一些可能的私人业主。在物联网是可能通过本地系统索引一个有界域是一致地.类的东西需要被定义与范畴。专业搜索网站可能提供访问特定类别。例如，如果一个想在网站搜索车部分的信息，可以去搜索网站，专业产品类型，并从那里被引导到特定的IP地址。32查询语言目前流行的查询语言，数据库系统依赖于结构化数据。结构化查询语言（SQL是最突出的例子。然而在过去的几年里，已有的半结构化数据查询语言的建议，其中比较典型的是互联网上的数据举行。数据量是如此巨大，它可能会在任何均匀结构不切实际，除了是在物联网的实施，可能的宽松的品种。可扩展的标记语言（XML提供较少的代表结构化以及结构化数据，一定程度的自我描述。这是一个很好接受的支持互操作性的技术而不是语义技术水平。 XQuery已由万维网联盟的XQuery开发，一种查询XML的语言，可以将文件，网页，和相关国家数据库链接。半结构化数据查询语言通常采用一个基本的层次数据模型， 例如一个单向图。需要做的在一个通常是很直观的语言提供了一个路径表达式是指定的对象。然而，困难的是固有层次数据模型的问题，例如，表示多对多关系的困难。在尽管如此，基础性的数据模型已被网络社区作为一个有用的， 直观的，实用的结构。物联网将有各种各样的用户：普通用户浏览它停止在需要的一些数据或者信息，专家用户那确切地知道他们需要什么样的数据和在哪儿能找到它， 和用户之间的区别。事实上，在不同的上下文中一个人可以对这些不同类型的任何一个用户。因此似乎必要的，不同类型的数据访问设施是可用的。普通用户将需要通过一个用户友好的图形用户界面（GUI）的访问，与物联网详细说明可在任何物体，和更多灵活的，强大的，高效的访问接口将为专家用户需要。服务可以被用来提供两种类型的访问。在查询语言的工作半结构化数据将与这些发展相关。3.3流程建模和事务这是最有可能将开发和提供对物联网服务的过程。面向服务的体系结构（SOA正成为网络性体系可以较好的支持能力的重要手段。中心思想是：独立组织提供统一的服务，其他用户可以接受。从而实现技术细节隐藏在用户的服务。应用程序通常是由一个数量，误码率的低级别的事务。交易会是由较低层次的操作或服务。因此，在物联网的事务处理出现的问题。在传统的数据库系统的并发事务处理问题已经通过酸性质通过时间戳，锁维修处理，和一个两阶段提交。 ACID属性是原子性、一致性、隔离性和硬脑膜性.一个事务必须在它的全部或没有在所有完成交易，必须使数据库处于一致的状态，交易不应该显示其他交易，和中间结果和变化的交易做了一定是永久性的。在分布式数据库系统中的两阶段提交用于保持一致性。所有参与站点必须确认他们准备提交之前提交命令的协调网站发布和写入数据库的日志。 人们已经认识到，酸性质不适合Web事务处理好。这是因为单个Web服务基本上自治，必须独立维护的一致性。例如在用户看到预订假期有交易的操作有两部