版權(quán)說明:本文檔由用戶提供并上傳,收益歸屬內(nèi)容提供方,若內(nèi)容存在侵權(quán),請進行舉報或認領(lǐng)
文檔簡介
1
2
Abstract
Withincreasinglycomplexmobilecommunicationnetworksanddifferentiatedservicerequirementsofawidevarietyofindustries,themostimportantresearchdirectionandtrendforbeyond5G(B5G)/6GistodeeplyintegrateAI,bigdata,andcomputingintonetworks,anddevelopintelligentcommunicationnetworkstodeliverautonomousdrivingnetworkcapabilitiesandpreciselycustomizedservices.Embeddingdataandintelligenceintonetworksinordertodevelopanovelnetworkintelligencetechnologysystemfeaturingendogenousintelligenceaswellascomputeandcommunicationconvergenceraiseschallengesintermsofarchitecture,data,andAIalgorithms.Data,asthemostfundamentalandimportantpart,determinestheupperlimitofnetworkintelligenceperformance,whereasthearchitectureandalgorithmsarejustthemeanstoapproachthislimit.Therefore,weneedtoaddressthemostfundamentalissueofhowtoacquire,analyze,andleveragemobilebigdataandbuildacomprehensivedataacquisitionandanalysissysteminordertofuelthedevelopmentofthenovelB5G/6Gnetworkintelligencetechnologysystem.ThiswhitepaperpresentsadataacquisitionandanalysissystemofB5G/6Gnetworkintelligence.Thesystemrealizesdata-levelinformationintegration,information-levelknowledgeextractionandrepresentation,intelligence-levelknowledgecomputingandinference,andapplication-levelfeaturecustomization.Thesefunctionsareofferedthroughmobilenetworkdataacquisitionandstorage,translationofmobilecommunicationprinciplesandprotocols,networkdataknowledgegraphconstructionandknowledgegraph-basedparsing,andconstructionandevaluationoffeaturedatasetsfornetworkintelligence.Thepaperseekstocreateafeasibleapproachtodataacquisitionandanalysisofnetworkintelligence,exploringanoveldirectionandparadigmfornetworkintelligencedatatechnologies.
3
Contents
1VisionandChallengesofB5G/6GNetworkIntelligence 6
2DataAcquisitionandAnalysisSystemofB5G/6GNetworkIntelligence 9
2.1WirelessNetworkIntelligentOpenPlatform 9
2.2DataAcquisition 10
2.3KnowledgeGraph-basedRepresentationandAnalysis 11
2.4FeatureDataSetsConstruction 12
3ConstructionandApplicationsofNetworkDataKnowledgeGraphs 13
3.1IntelligentFeatureEngineeringofMobileBigData 13
3.2ConstructionofNetworkDataKnowledgeGraphs 14
3.2.1TechnicalArchitectureofKnowledgeGraphs 14
3.2.2ConstructionofNetworkDataKnowledgeGraphsBasedonCommunicationPrinciplesand
Protocols 15
3.2.3KnowledgeGraphCompletionMethod 15
3.2.4EdgeWeightsofKnowledgeGraphs 16
3.2.5NodeImportanceandEfficiencyofImpactBetweenNodesinKnowledgeGraphs 16
3.3ApplicationsofNetworkDataKnowledgeGraphs 16
4ConstructionandEvaluationofFeatureDataSetsforNetworkIntelligence 19
4.1DataModelsforNetworkDataKPIs 19
4.2PreliminaryKnowledgeGraph-basedFilteringofFeaturesThatAffectKPIsinSpecificScenarios
19
4.3ConstructionofFeatureDataSetsBasedonMachineLearning 20
4.4EvaluationSystemfortheConstructionProcessofFeatureDataSets 20
4.5EvaluationSystemforKPI-orientedFeatureDataSets 21
5FeatureDataSetConstructionandEvaluationCases 22
5.1ConstructionofaLocalKnowledgeGraphwiththeUplinkThroughputsatItsCore 22
5.2In-depthAnalysisoftheLocalKnowledgeGraphwiththeUplinkThroughputsatItsCore 23
5.3ConstructionoftheUplinkThroughputFeatureDataSet 24
5.4EvaluationoftheUplinkThroughputFeatureDataSet 25
6Outlook 26
References 27
4
Preface
Withincreasinglycomplexmobilecommunicationnetworksanddifferentiatedservicerequirementsofawidevarietyofindustries,themostimportantresearchdirectionandtrendforbeyond5G(B5G)/6GistodeeplyintegrateAI,bigdata,andcomputingintonetworks,anddevelopintelligentcommunicationnetworkstodeliverautonomousdrivingnetworkcapabilitiesandpreciselycustomizedservices.Atpresent,globalresearchonthisdirectionisstillinitsinitialstage.Embeddingdataandintelligenceintonetworksinordertodevelopanovelnetworkintelligencetechnologysystemfeaturingendogenousintelligenceaswellascomputeandcommunicationconvergenceraiseschallengesintermsofarchitecture,data,andAIalgorithms.Data,asthemostfundamentalandimportantpart,determinestheupperlimitofnetworkintelligenceperformance,whereasthearchitectureandalgorithmsarejustthemeanstoapproachthislimit.Whenmobilecommunicationnetworksarerunning,tensofthousandsofdatafieldsandindicatorsaregeneratedthroughoutthenetworks,fromUEs,radioaccessnetworks(RANs),transmissionbearernetworks,tocorenetworks,involvingdifferentsoftware,hardware,functions,andprotocolstacks.Forthedevelopmentofnetworkintelligence,itisimperativetoestablishacomprehensivenetworkdataacquisitionandanalysissystemthateffectivelyacquires,classifies,analyzes,andleveragesdifferenttypesofdata.
Asnetworkstructures,UEtypes,UEbehaviors,dataservicerequirements,andsystemresourceconfigurationsofB5G/6Gwillbehighlydynamic,time-sensitive,andcoupled,dataacquisitionandanalysisonmobilenetworkswillfacemanychallenges.Amajordifficultyliesindataacquisitionbecausedataisscatteredondifferentdevicesatdifferentlayers.Alongwiththis,variousdatatypesandcomplexstructuresmakeitdifficulttoanalyzedata,andhighlyabstractdatahindersunderstanding.Beyondthis,coupleddatawithcomplexcorrelationsresultsindifficultiesinmining.Tocopewiththesechallengesrequiresstandarddataattributegeneralization,efficientlyclarifyingrelationshipsbetweendata,furtherexploringin-depthassociationsbetweenrelationships,andclearlyrepresentingdataandkeyinformationinherentinrelationships.
Networkintelligencereliesheavilyondata.Establishingacompletedataacquisitionandanalysissystemthatsolvesthebasicissuesofdataacquisition,openness,andutilizationwilladvancethedevelopmentofnovelB5G/6Gnetworkintelligencetechnologiesandbecomeakeymilestonetowardsrealizingnetworkintelligence.Thispaperattemptstocreateafeasibleapproachtodataacquisitionandanalysisofnetworkintelligence,seekinganoveldirectionandparadigmfornetworkintelligencedatatechnologies.Knowledgegraph,asanewmethodofnetworkdataknowledgerepresentationandanewtoolofknowledgemanagement,clarifiescomplexcorrelationsbetweendata,andrepresentstypesandattributesofdataaswellasthoseofrelationshipsbetweendata.Furthertothis,topologyandnodeinformationofgraphsenablesintelligentinferenceandintelligentfeatureengineeringpractices.Knowledgegraphisefficientintacklingtheaforementionedchallengesindataacquisitionandanalysisonmobilecommunicationnetworks.ItiskeytothedataacquisitionandanalysissystemofB5G/6Gnetworkintelligence.
ThiswhitepaperpresentsadataacquisitionandanalysissystemofB5G/6Gnetworkintelligence.Theachievementsofthissystemaremadeinmobilenetworkdataacquisitionandstorage,knowledgerepresentation,knowledgeinference,andconstructionandevaluationoffeaturedatasets.
KnowledgerepresentationofB5G/6Gmobilecommunicationnetworksbyusingtoolssuchasknowledgegraph,changesthenatureofcomplexinternalrelationshipsofmobilecommunicationnetworksfrom"blackbox"to"whitebox."Itpresentshighlydynamic,time-sensitivemobilecommunicationnetworkswitheasilycoupledelementsinavisualized,hierarchical,andstructuredmannerbasedonmobilecommunicationprinciplesandprotocols.Onthebasisofknowledgerepresentation,knowledgeinferenceofB5G/6Gmobilecommunicationnetworkseffectivelyprocessesandintegratesmassivedataandcomplexconnectionsonnetworksforquickknowledgeresponseandinference.Knowledgeinferenceempowersmobilecommunicationnetworkswithperceptionandcognitioncapabilities.
5
Basedonthecorrelationanalysisbetweendatafieldsinknowledgegraphsofmobilecommunicationnetworksandservicerequirements,dataisfurtherclassifiedintodifferenttypesandfeatureextractionisperformedonthedatatoconstructfeaturedatasetsofB5G/6Gmobilecommunicationnetworks.Featuredatasetsrepresentkeyfeaturesthathavesignificantimpactonkeyperformanceindicators(KPIs)ofmobilecommunicationnetworks.Thesetsalsocontainthedatacollectedbyfeatures.Asacombinationofexpertknowledgeandintelligentalgorithms,thesetscanserveasdirectinputsforsubsequentintelligentoptimization.Constructingsuchfeaturedatasetshelpsobtainimportantfeaturescloselyrelatedtotargetindicatorsduringintelligentdiagnosis,locating,andoptimizationofmobilecommunicationnetworks.Thisguidesfurtherintelligentnetworkperception,prediction,optimization,andeventually,digitaltwins.
Organizations:PurpleMountainLaboratoriesforNetworkCommunicationandSecurity,SoutheastUniversity,BeijingUniversityofPostsandTelecommunications,CICTMobileCommunicationTechnologyCo.,Ltd.,HuaweiTechnologiesCo.,Ltd.,Ericsson(China)CommunicationsCo.,Ltd.,vivoMobileCommunicationCo.,Ltd.,andInstituteofChinaTelecomCorporationLimited
Contributors:XiaohuYou,YongmingHuang,HangZhan,ShiwenHe,YunshanYi,JianjieYou,WenjingLi,LeiFeng,HuiXu,WanfeiSun,MingyuZhao,XueqiangYan,LileiWang,RenqianZhao,YannanYuan,BingQian,LuGao,andYangLiu
1VisionandChallengesofB5G/6GNetworkIntelligence
6
Inthepursuitofbeyond5G/6G(hereinafterreferredtoasB5G/6G),mobilecommunicationnetworkswillempowerawholerangeofindustriesandthereforefacechallengesposedbydiversifieddemandsinnumerousapplicationscenarios.ThetrendforB5G/6GistodeeplyintegrateAIandbigdataintocommunicationfornetworkintelligence.Incontrastwith4Gnetworksthatfocusonmobilephonecommunication,5GandB5Gneedtosupportall-scenariofullservicesacrossindustries,promoteintelligentdigitaltransformationofawidevarietyofindustries,anddelivernewhigh-valueservicesandtheultimateserviceexperience.Tomeettheserequirements,inadditiontointegratingtraditionalterrestrialnetworks,6Gwirelesscommunicationnetworkswillintegratesatellitecommunication,unmannedaerialvehicle(UAV)communication,andothernon-terrestrialnetworksforspace-air-ground-seaintegratedaccess.Allspectrawillbefullyexploredtofurtherimprovedatatransmissionratesandconnectiondensity.Additionally,6GwilladoptAIandbigdatatechnologiestointroducearangeofsmartapplicationsinordertocopewithmassivedatageneratedbytheuseofextremelyheterogeneousnetworks,variouscommunicationscenarios,numerousantennas,highbandwidths,andnewservicerequirements.Clearly,6Gmobilenetworkswillbedeeplyintegratedwithcomputing,bigdata,andAItechnologiestoprovidefullapplicationservicesfrompersonalapplications,verticalapplications,togovernanceapplications.Itisexpectedthat6Gwilldelivernewparadigmsfeaturing"globalcoverage","allspectra","fullapplications"[1,2].
Future6Gnetworksareexpectedtomeetservicerequirementsforsupportingvirtualandphysicalfusion,holographic,scenario-based,personalized,andubiquitouscommunication,aswellasmeetnetworkrequirementsfortheintegrationofheterogeneousnetworktechnologiesandspace-air-ground-seaintegratednetworks.However,thecurrentnetworkoperationparadigmswithrule-basedalgorithmsarelimitedbyrigidpresetrules,anditisdifficulttodynamicallyadapttheparadigmstoever-changinguserrequirementsandnetworkenvironments.Moreover,networkoperationexperiencecannotbeeffectivelyaccumulated,whichlimitscontinuousimprovementofnetworkcapabilities.Thatistosay,networksdonothavetheself-evolutioncapabilityinthecurrentparadigms.Upgradesorimprovementsrequiretime-consumingresearchbyengineers,whichisunacceptablefor6Gnetworkswithunprecedentedscaleandoperationcomplexity.AddingendogenousintelligenceintonetworkstoimproveautomationandintelligenceofB5G/6Giskeytohandlingtheprecedingissues[1].Inparticular,dataisessentialforB5G/6Gnetworkintelligence.Onlybyleveragingmobilebigdata,canweoptimizenetworksintermsofserviceexperience,networkquality,networkefficiency,andnetworkcosts[4].
Majorchallengesaretoutilizemobilebigdataandembedintelligenceintonetworksinordertointroduceanovelnetworkintelligencetechnologysystemintermsofdata,architecture,andAIalgorithms.Toaddressthechallenges,acomprehensivenetworkdataacquisitionandanalysissystemisneeded.Basedondataacquisitionandanalysisresults,thesystemcanbettermeetdifferentiatedapplicationrequirements,enablingflexiblenetworkfunctionorchestration,morerefinedmobilecommunicationnetworkresourceallocation,andefficientnetworkcontrol.
Asnetworkstructures,UEtypes,UEbehaviors,dataservicerequirements,andsystemresourceconfigurationsofB5G/6Gwillbehighlydynamic,time-sensitive,andcoupled,acquisitionandanalysisofmobilebigdatawillfacemanychallenges.Amajordifficultyliesindataacquisitionbecausedataisscatteredondifferentdevicesatdifferentlayers.Alongwiththis,variousdatatypesandcomplexstructuresmakeitdifficulttoanalyzedata,andhighlyabstractdatahindersunderstanding.Beyondthis,coupleddatawithcomplexcorrelationsresultsindifficultiesinmining.Tocopewiththesechallengesrequiresstandarddataattributegeneralization,efficientlyclarifyingrelationshipsbetweendata,furtherexploringin-depthassociationsbetweenrelationships,andclearlyrepresentingdataandkeyinformationinherentinrelationships.Inaddition,tensofthousandsofdatafieldsandindicatorsaregeneratedthroughoutmobilecommunicationnetworks,fromUEs,accessnetworks,tocorenetworks,andinvolvedifferentsoftware,hardware,functions,andprotocolstacks.
7
Forthedevelopmentofintelligenceinmobilecommunicationnetworks,itiskeytoeffectivelycollect,classify,analyze,andleveragedifferenttypesofgenerateddata,fullyunleashnetworkservicepotential,andgivefullplaytothetechnicaladvantagesofnetworks.
Traditionaldataanalysismethodscanbeclassifiedintotwotypes.Thefirsttypeismodelinganalysisbasedontheoriesandexperienceaccumulatedinthecommunicationfield.Forexample,wehaveexpertsystemsforrootcauseanalysisofanomalies,anomalydetectionbasedonmultipleKPIsinthewirelessnetworkknowledgebase[3],andoptimizationmodelanalysisforradioresourcemanagementoptimization.Thesecondtypeisdata-drivenmachinelearningmodelsandalgorithms,whichfocusonprovidingatypeofefficientalgorithmsforcertainproblems.AsB5G/6Gnetworksareexpectedtobemorecomplexthanever,anddata-drivenAItechnologieshavelimitations,theselearningmodelsandalgorithmscannotsupporttheend-to-endprocessfromthedetectionofnetworksymptomstotheimplementationofsolutions.Thisproblemiswidespreadinmobilecommunicationswhose"blackbox"internalmechanismlacksexplainabilityandscalability.Forthisreason,thesingleknowledge/model-drivenordata-drivenmodecannotmeettherequirementsofbigdataanalysisinB5G/6Gmobilecommunications.Dataanalysisinmobilecommunicationsneedstotransformintoanintelligentmodedrivenbybothknowledge/modelsanddata.
AsAIiseffectiveatdataanalysisinmobilecommunications,itsdevelopmentisofprimesignificance.EarlyformsofAI,representedbyexpertsystems,emphasizethelawofcausalityandknowledgestructurerelatedtosymbolicinferenceatthecostofviolatingthelawofexcludedmiddle.SubsequentformsofAI,representedbystatisticalmachinelearninganddeeplearning,valuethelawofexcludedmiddleatthecostofviolatingthelawofcausality.AIdevelopmentrequiresacombinationofthesetwotypesofAI.Attheendof2018,thetheoreticalframeworkofthethird-generationAIwasofficiallyreleased:(1)BuildingexplainableandrobustAItheoriesandmethods(2)Developingsecure,reliable,trustworthy,andscalableAItechnologies.TheAIroadmaphighlightsAItheoriesandmethodsofdataandknowledgeintegration.Asknowledgegraphisrelativelymatureatpresent,itcanfacilitateimplementationofthethird-generationAI.
Brain-likedatagovernancesystemshavebeenappliedtomanyfieldsbasedontoolslikeknowledgegraph,formingadata-information-knowledge-intelligencetechnicalpath.Suchclosed-loopbrain-likesystemsareatrade-offsolutionbetweenthetwotypesofAI.Likewise,inthemobilecommunicationsfield,toembeddataandintelligenceinordertointroduceanovelnetworkintelligencetechnologysystemalsoneedstofollowthedata-information-knowledge-intelligencepath.Therefore,suchatechnologysystemwillrelyondataacquisitionandanalysis.Knowledgegraph,criticaltothetechnicalpathofnetworkintelligence,integratestheadvantagesoffirst-andsecond-generationAI,andwillbekeyinpromotingB5G/6Gnetworkintelligence.
Knowledgegrapheffectivelyclarifiesthecorrelationsbetweenmulti-sourceheterogeneousmobilecommunicationdatawithloosedatastructures,hierarchicalandflattenednetworks,andendogenousnetworkperformanceelements.Inthismanner,knowledgegraphhelpsrealizeknowledgeinterconnectivity,efficientresourcemanagement,andintelligentnetworkmaintenance.Theendogenouselementsandcorrelationsofmobilecommunicationnetworkscanbedynamicallypresentedandaccuratelypositionedinaneasy-to-understandmannerwithclearstructures.Knowledgegraph-baseddataanalysisiscrucialtomobilecommunicationknowledgeacquisition,sorting,andrepresentation.
8
Basedoncorrelationanalysisusingnetworkdataknowledgegraphs,featuresoftargetindicatorsarepreliminarilyfiltered.Aftercontinuousalgorithmoptimizationandmodeltraining,featuredatasetsfordifferentapplicationscenariosareconstructed.Thisentireprocessrealizesfrommobilecommunicationknowledgesortingtomobilecommunicationintelligenceapplication.
Figure1-1Bigdata-basedcognitiveintelligenceconstructioncommonlyusedinvariousindustries[13]
9
2DataAcquisitionandAnalysisSystemofB5G/6GNetworkIntelligence
Embeddingdataandintelligenceintonetworksinordertodevelopanoveltechnologysystemraiseschallengesintermsofarchitecture,data,andAIalgorithms.Tocopewiththesechallenges,acomprehensivenetworkdataacquisitionandanalysissystemisneeded.Thesystemconsistsoffourfunctionalmodules:wirelessnetworkintelligentopenplatform,dataacquisition,knowledgegraph-basedcorrelationanalysis,andfeaturedatasetsconstruction.Thesemodulesenabledata-levelinformationintegration,information-levelknowledgeextraction,intelligence-levelknowledgecomputing,andapplication-levelfeaturecustomization.
Thewirelessnetworkintelligentopenplatformmoduleandthedataacquisitionmodulefocusonlayer-by-layerdataacquisitioninB5G/6G.Theknowledgegraph-basedcorrelationanalysismoduleandthefeaturedatasetsconstructionmodulehelpcopewithchallengingdatarepresentation,dataanalysis,andfeatureextractioninB5G/6G.ThefourmodulesworktogethertolayadatafoundationforB5G/6Gnetworkintelligenceandexplorewaystorealizeit.
Figure2-1DataacquisitionandanalysissystemofB5G/6Gnetworkintelligence
2.1WirelessNetworkIntelligentOpenPlatform
Thewirelessnetworkintelligentopenplatform,basedonB5G/6Gnetworksanddatawarehouses,deeplyintegratesbigdataandAI.GuidedbythenotionofthedataacquisitionandanalysissystemofB5G/6Gnetworkintelligence,theplatformaimsatdataopenness,datasharing,dataanalysis,anddataapplication.
TheB5G/6Gwirelessnetworkintelligentopenplatformmoduledeliversavisualizedone-stopsolutionforreal-timemobilenetworkdataacquisition,analysis,andtracing.Itprovidesprofessionalnetworkdatasupportandcontributestoapremiumecosysteminordertorealizerealdataopennessintheindustry.Thearchitectureofthisintelligentopenplatformconsistsofthreelayers:dataacquisitionandpreprocessinglayer,dataanalysislayer,anddataapplicationlayer,asshowninthefollowingfigure.
10
Figure2-2B5G/6Gwirelessnetworkintelligentopenplatform
Theopenplatformaccessesmulti-dimensionaldiverseB5G/6GdatathroughAPIstoformtime-varying,highlyintegratedB5G/6Gdatawarehousesthatareorientedtoapplicationsandhaverelativelystablestructures.Basedonsuchdatawarehouses,anetworkdataacquisitionandanalysissystemisestablished,whichcoversthenetworkdataknowledgegraphconstruction,knowledgegraph-basedcorrelationanalysis,constructionoffeaturedatasets,and5Gnetworkbigdatadictionary.The5Gnetworkbigdatadictionaryisacompletedatasystemthatreflectsthecharacteristicsofmobilenetworkoperations.Itwilllayasolidfoundationforthedevelopmentofnetworkintelligence.Theoutputs(generatedfromthenetworkdataknowledgegraphconstruction,knowledgegraph-basedcorrelationanalysis,andfeaturedatasetconstruction)interactwiththe5Gnetworkbigdatadictionary.Inparticular,majorachievementsmadethroughinteractionwillbeincorporatedinthedictionary.Theseachievementsatthedataanalysislayerhelprealizeapplicationssuchasdigitaltwin,anomalydetection,sourcetracing,intelligentoptimization,intelligentdecision-making,andintelligentevaluation.
2.2DataAcquisition
ThedataacquisitionmoduleisafundamentalpartofthedataacquisitionandanalysissystemofB5G/6Gnetworkintelligence.Withoutthismodule,allsubsequentdataanalysisandintelligentdataapplicationscannotberealized.Traditionaldataacquisitioninvolvescreatingtables,filteringdata,collectingdata,andimportingdataintothedatabase.Itlackson-demanddataprocessing.Traditionaldataacquisitionmethodsincludedrivetests,hardware-basedsignalingdatacollection,andsoftware-basedsignalingdatacollection.Thecollecteddatacanbeclassifiedintoreal-timeandnon-real-timedata.Tomakedataacquisition,datatransmission,anddatainteractionmoreefficient,thedataacquisitionfunctionalmodulecanfurthercustomizetimeandspacedataacquisitionbasedondifferentintelligentoptimizationapplicationscenariosandnetworkdeploymentinmobilecommunications.Thecollecteddatacanbestoredinbasicdatawarehouses.
11
Thedatacollectionmodulecollectsvarioustypesofdataonpilotnetworksthroughhardware-andsoftware-baseddatacollectionaswellasdrivetests.Thedataincludeswirelessairinterfacedata(ontheUEsideandbasestationside),corenetworkdata,andnetworkmanagementdata.
Wirelessairinterfacedata(ontheUEsideandbasestationside)includesdataatthephysicallayer(L1),datalinklayer(L2),andnetworklayer(L3).Thephysicallayerincludesthedataofdownlinkanduplinksharedchannels,downlinkanduplinkcontrolchannels,randomaccesschannels,andsoundingreferencechannels.Thedatalinklayerincludesthedataofdownlinkscheduling,uplinkscheduling,downlinkcontrol,andchannelstateinformation.Thenetworklayerincludesthedataofbroadcast,paging,andaccessdata.Byextractingtheaforementioneddata,thebasestationsidesetsupcommunicationinterfaces,andsendsreal-timedatatoexternalplatformstoimplementdataacquisitiononthebasestationside.
Corenetworkdataincludescontrol-planeanduser-planedata.Control-planedatamainlyreferstothedatathatcontrolssignalingprotocoltypesinordertocontrolthesetup,maintenance,andreleaseofserviceprocesses.User-planedatareferstoservicedata,suchasvoice,packetservice,instantmessaging,email,andvideodata.
Networkmanagementdataincludesperformance,alarm,andconfigurationdata.Specifically,performancedatareferstotheperformanceinformationcollectedfromdifferentnetworkelements(NEs).AlarmdatareferstovarioustypesofalarmreportsgeneratedbyallNEsfordevicefaults,networkincidentsaswellasforfaultsrelatedtonetworksandservices.ConfigurationdatareferstothebasicinformationaboutNEs.Itismappedtoentitiesandmainlyusedfornetworktopologyinformationmanagement.
Datapreprocessinginvolvesperformingoperations,suchasdirtydataprocessing,time-basedpartitioning,structuredprocessingonunstructureddata,anddataimporttothedatabaseviaETL,onthecollectedrawmobilenetworkdata.
ETLisatooltoextractdatasuchasrelationshipdataandplanedatafilesfromdistributed,heterogeneousdatasourcestothetemporarymiddlelayerforcleaning,conversion,andintegration.Iteventuallyloadsthedatatodatawarehousesordatamartsforonlineanalyticalprocessinganddatamining.
2.3KnowledgeGraph-basedRepresentationandAnalysis
Knowledgegraph,anewwaytorepresentandmanageknowledge,hasemergedinmoreandmoreverticalapplicationfieldsandisplayinganincreasinglyvitalrole[7–11].Networkdataknowledgegraphsconstructedbasedonthemobiledataserveasindustryknowledgegraphsinverticalfieldsofmobilecommunications.Theconstructionofknowledgegraphshashighrequirementsondepth,breadth,andaccuracyofknowledgeregardingmobilecommunications.
Theknowledgegraph-basedcorrelationanalysismoduledrawsonnetworkdataknowledgegraphsforknowledgerepresentation,correlationanalysis,andin-depthmining,providingeffectiveknowledgerulesandknowledgecomputingsupp
溫馨提示
- 1. 本站所有資源如無特殊說明,都需要本地電腦安裝OFFICE2007和PDF閱讀器。圖紙軟件為CAD,CAXA,PROE,UG,SolidWorks等.壓縮文件請下載最新的WinRAR軟件解壓。
- 2. 本站的文檔不包含任何第三方提供的附件圖紙等,如果需要附件,請聯(lián)系上傳者。文件的所有權(quán)益歸上傳用戶所有。
- 3. 本站RAR壓縮包中若帶圖紙,網(wǎng)頁內(nèi)容里面會有圖紙預(yù)覽,若沒有圖紙預(yù)覽就沒有圖紙。
- 4. 未經(jīng)權(quán)益所有人同意不得將文件中的內(nèi)容挪作商業(yè)或盈利用途。
- 5. 人人文庫網(wǎng)僅提供信息存儲空間,僅對用戶上傳內(nèi)容的表現(xiàn)方式做保護處理,對用戶上傳分享的文檔內(nèi)容本身不做任何修改或編輯,并不能對任何下載內(nèi)容負責。
- 6. 下載文件中如有侵權(quán)或不適當內(nèi)容,請與我們聯(lián)系,我們立即糾正。
- 7. 本站不保證下載資源的準確性、安全性和完整性, 同時也不承擔用戶因使用這些下載資源對自己和他人造成任何形式的傷害或損失。
最新文檔
- 2025至2030年中國磁性燈座數(shù)據(jù)監(jiān)測研究報告
- 2025至2030年中國水晶鉆石貼片數(shù)據(jù)監(jiān)測研究報告
- 2025至2030年中國剃刀數(shù)據(jù)監(jiān)測研究報告
- 2025年中國陶瓷連體座便器市場調(diào)查研究報告
- 2025年中國印花條絨面料市場調(diào)查研究報告
- 二零二五年度綠色家居產(chǎn)業(yè)民營中小企業(yè)戰(zhàn)略合作合同4篇
- 二零二五年度賓館客房租賃合同租賃權(quán)轉(zhuǎn)讓合同2篇
- 個人電商店鋪轉(zhuǎn)讓合同2024年版3篇
- 2025程力危險品廂式車廠家定制化生產(chǎn)與物流配送合同4篇
- 二零二四年度智能制造委托擔保合同匯編3篇
- 二零二五隱名股東合作協(xié)議書及公司股權(quán)代持及回購協(xié)議
- 四川省成都市武侯區(qū)2023-2024學(xué)年九年級上學(xué)期期末考試化學(xué)試題
- 環(huán)境衛(wèi)生學(xué)及消毒滅菌效果監(jiān)測
- 2024年共青團入團積極分子考試題庫(含答案)
- 碎屑巖油藏注水水質(zhì)指標及分析方法
- 【S洲際酒店婚禮策劃方案設(shè)計6800字(論文)】
- 鐵路項目征地拆遷工作體會課件
- 醫(yī)院死亡報告年終分析報告
- 中國教育史(第四版)全套教學(xué)課件
- 2023年11月英語二級筆譯真題及答案(筆譯實務(wù))
- 上海民辦楊浦實驗學(xué)校初一新生分班(摸底)語文考試模擬試卷(10套試卷帶答案解析)
評論
0/150
提交評論