GTI 5G無(wú)線網(wǎng)絡(luò)智能化技術(shù)需求白皮書(shū) GTI 5G Radio Network Intelligence Technical Requirements White Paper

GTI5GRadioNetwork

IntelligenceTechnical

RequirementsWhitePaper

GTI5GRadioNetwork

InteligenceTechnical

RequirementsWhitePaper

Version:

V1.2

DeliverableType

□ProceduralDocument

√WorkingDocument

ConfidentialLevel

√

√

□

OpentoGTIOperatorMembers

OpentoGTIPartners

OpentoPublic

Working

Group

networkWG

Task

IntegrationAcrossAreas-AutonomousNetwork

Source

members

CMCC,Huawei,ZTE,CICTMobile,Ericsson

Support

members

NokiaShanghaiBell,H3C,RuijieNetworks，Comba

Editor

CMCC:JinghuaKuang,SongheLu,LinZhu

Huawei:YuanxianTang,YangZheng,ZhijieXu,ChengyuZhang,JianfengChen,YanSun,XiaozeXu,LiWang,QinxinLiu,JingLiang,XianmingLi,JiamingZhang,ChuanMa,JunjieZhang,JianFan,YanliZhuang,YingXu,BaokunWang,KunlunWang,KekeQiu,JianShi,XiLuo,XiaoweiQin,XiHe.

ZTE:YongZhan,PanLi,HongtaoZhang,ChengGuo,KekeSu

CICTMobile:LemingFang,LeweiHu

Ericsson:VivianBai,RuiWan,XiaoqingMa

LastEdit

Date

02-09-2023

Approval

Date

DD-MM-2023

Confidentiality:ThisdocumentmaycontaininformationthatisconfidentialandaccesstothisdocumentisrestrictedtothepersonslistedintheConfidentialLevel.Thisdocumentmaynotbeused,disclosedorreproduced,inwholeorinpart,withoutthepriorwrittenauthorizationofGTI,andthosesoauthorizedmayonlyusethisdocumentforthepurposeconsistentwiththeauthorization.GTIdisclaimsanyliabilityfortheaccuracyorcompletenessortimelinessoftheinformationcontainedinthisdocument.Theinformationcontainedinthisdocumentmaybesubjecttochangewithoutpriornotice.

DocumentHistory

Date

Meeting#

Version#

RevisionContents

31-08-2023

V1.0

Initialdraft

02-09-2023

V1.1

AddingUConradionetworkO&M

04-09-2023

V1.2

EditorialChange

TableofContents

Abbreviations 6

1CurrentStateandVisionforWirelessNetworks 9

1.1NewChallenges 9

1.2ConceptandStatusQuoofAutonomousNetworks 13

2TechnicalRequirementsforRANIntelligence 19

2.1RANIntelligenceArchitecture 19

2.2TechnicalRequirementsforRANIntelligence 20

2.3TechnicalDirectionofRANIntelligence 27

2.4PotentialTechnologiesforRANIntelligence-LargeModels 40

2.5DataRequirementsforWirelessNEs(OMCandBaseStations) 42

3UseCasesofRANIntelligence 44

3.1IntelligentMulti-BandCoordination 44

3.2DesignatedApplicationExperienceGuarantee 45

3.3IntelligentUserOrchestration 46

3.4Intent-drivenServiceExperienceGuarantee 47

3.5IntelligentMacro-MicroCollaboration 49

3.6LiveStreamingServiceExperienceAssurance 50

3.7EnergySavingWhileMaintainingaStableNetworkPerformance 52

3.8NetworkFaultPreventionandPrediction 53

3.9Networkintelligentabnormalstatusmonitoring 55

Abbreviations

Abbreviation

Explanation

3GPP

3rdGenerationPartnershipProject

5G

5thGeneration

5QI

5GQoSIdentifier

ADN

autonomousdrivingnetwork

AGI

artificialgeneralintelligence

AI

artificialintelligence

ANP

accessnetworkprovider

AoA

angleofarrival

API

applicationprogramminginterface

CCSA

ChinaCommunicationsStandardsAssociation

CEU

celledgeuser

CHR

callhistoryrecord

CSI

channelstateinformation

CV

computervision

E2E

endtoend

ETSI

EuropeanTelecommunicationsStandardsInstitute

FDD

frequencydivisionduplex

FDR

flowdatarecord

GHz

GigaHertz(GHz)

GPS

GlobalPositioningSystem

GPT

GenerativePre-trainedTransformer

IAB

integratedaccessandbackhaul

ICT

informationandcommunicationstechnology

CSC

communicationservicecustomer

CSP

communicationserviceprovider

NOP

networkoperator

IoE

InternetofEverything

IoV

InternetofVehicles

ISAC

IntegratedSensingandCommunications

KPI

keyperformanceindicator

LLM

largelanguagemodel

LOS

lineofsight

MAC

MediaAccessControl

MIIT

MinistryofIndustryandInformationTechnology

ML

machinelearning

MML

man-machinelanguage

MnS

managementservice

MR

measurementreport

SDO

standardsdevelopingorganization

NaaS

networkasaservice

NE

networkelement

NLOS

non-lineofsight

NLP

naturallanguageprocessing

NMS

networkmanagementsystem

NR

NewRadio

OMC

operationmaintenancecenter

OPEX

operationalexpenditure

PLC

programmablelogiccontroller

QoE

qualityofexperience

RAN

radioaccessnetwork

RAT

radioaccesstechnology

RF

radiofrequency

RRM

radioresourcemanagement

RSRP

referencesignalreceivedpower

SLA

ServiceLevelAgreement

TDD

timedivisionduplex

TDOA

timedifferenceofarrival

TOA

timeofarrival

TTI

transmissiontimeinterval

AV

unmannedaerialvehicle

UC

usecase

UE

userequipment

XR

extendedreality

1CURRENTSTATEANDVISIONFORWIRELESSNETWORKS

1.1NewChallenges

TheFourthIndustrialRevolutionisrapidlyusheringinaworldwhereeverythingisconnectedandintelligent.Allindustriesareactivelyexploringthepathtodigitaltransformation.Communicationsnetworkshaveshiftedtheirfocusfromservingindividualstoservingtheentiredigitalsociety.Theemergenceofthedigitaleconomyhasgraduallybecomethesecondevolutionarycurveforoperators,bringingbothnewopportunitiesandchallengesforICTinvestmentsinthetelecommunicationsindustry.

1.1.1GrowingWirelessTrafficandEnergyConsumption

Globaloperatorsareprioritizinggreenandlow-carbonstrategiesascriticaltotheirsustainabledevelopment.Todate,32operatorshavesuccessfullydevelopeddual-carbonactionplans.Energyoperatingrevenues(OPEX)continuetoriseasapercentageofoperators'revenues.OneEuropeanoperator,forexample,hasseena10%increaseinenergyconsumptionoverthepastthreeyearsandhasconsistentlyexceededits1%targetoverthepastfiveyears.Asaresult,costreductionhasbecomeakeyfocusforthem.InChina,operatorshaveintensifiedtheirevaluationofenergyconsumption.Forexample,oneChineseoperatorrequiresa1%to3%reductioninpowerconsumptionperlogicalsiteandincreasestheweightofoperationalefficiency(self-intelligence/cost/energyconsumption)from5%to20%.InEurope,theescalatingenergycrisishasledtoasharpriseinelectricitycosts,withsomecountriesandregionsfacingenergyshortages.OperatorsarenowfacingthepressureofrisingOPEXandpotentialservicedisruptionsduetoloadshedding.AnotherEuropeanoperatorhasoutlinedaclearstrategytoreducepowercostsfor35,000sitesonthelivenetworkthroughenergyconservation

measureswhileensuringservicecontinuity.

Withtherapiddevelopmentofservices,usersarebecomingincreasinglyreliantonmobilenetworksintheirdailylives.Itisestimatedthatmobilenetworktrafficwillincrease100-foldby2030,resultingintheexpansionofsites,spectrumandchannels.Asaresult,theenergyconsumptionofwirelessnetworkswillincreaseproportionally.Operatorsneedtobalanceservicedevelopmentwithenergyconsumption.Theyarefacedwiththetaskofachievingoptimaluserexperienceandenergyefficiencywithincomplexnetworkandservicescenarios.

Tomeetthischallenge,operatorsmusttakeeffectivemeasurestoimproveenergyefficiency.

1)Deployefficientpowermodulesandintelligentcoolingsystemstoreducebasestationpowerconsumption.

2)Optimizenetworkarchitecturetoreducethenumberofbasestations.

3)EstablishagreenO&Msystemtoimproveenergymanagement.EstablishgreenO&Mstandardsandprocessestocontrolenergyconsumptionandachievegreenoperations.

1.1.2GrowingDemandforDeterministicExperienceasNew

ServicesAreIntroduced

Ascommercialnetworkdeploymentsacceleratearoundtheworld,wirelessnetworksaregraduallytransformingfromconsumer-centricmobilebroadbandnetworkstoInternetofEverything(IoE).IntheupcomingIoEera,bothpeopleandthingswillbeconnectedthroughwirelesscellularnetworks.Thediversificationofservicesbringsdifferentnetworkrequirements.Forexample,theimmersiveExtendedReality(XR)experiencerequiressymmetricuplinkanddownlinkbandwidth,10mslatency,andaconsistentexperienceanytime,anywhere;IntelligentInternetofVehicles(IoV)requiresultra-widecoverageandlow-latencycontrol;UnmannedAerialVehicle(UAV)servicesrequirehigh-bandwidthuplinkbackhaulandlow-latencydownlinkcontrol;IndustrialMachineVisionrelieson1Gbpsto10Gbpsuplinkbandwidthalongwithultra-lowlatencyof4millisecondsat99.999%reliabilityduringthecoreproductionphase.Intransportandlow-altitudedetectionscenarios,wirelessnetworksarerequiredtoprovideIntegratedSensingandCommunications(ISAC)capabilities.Servicediversificationalsointroducesdifferentnetwork

coveragerequirements.Forexample,UAVservicesrequirenetworkcoverageatlowaltitudes,IoVrequiresultra-widecoverage,andindustriesrequirecoverageinminingareas,oceans,andindustrialproductionenvironments.Inaddition,thedevelopmentofdiversifiedservicesplacesdifferentdemandsonnetworkservicelevelagreements(SLAs).Forexample,XRimmersiveexperiencesandindustrialprogrammablelogiccontrollers(PLCs)requiredeterministicassuranceofnetworklatency.

Therefore,5Gnetworksmustmeetthefollowingrequirementstosupporta

varietyofservices:

1)Multi-dimensionalnetworkcapabilities:Thenetworkshouldmovefrombeingprimarilyfocusedondownlinkservicestoprovidingdiversecapabilitiessuchasuplink,downlink,latency,positioning,reliability,andsensing.

2)Diversifiednetworkcoverage:Thenetworkshouldevolvefromagroundnetworkfocusedonhumanuserstoamulti-dimensionalnetworkthatprovidesultra-widegeographiccoverage,low-altitudecoverage,andsuper-distancecoverageinoceans,aswellascampusandgroundcoverage.

3)DifferentiatedSLAlevels:Thenetworkshouldprovidedifferentiated,deterministicassurance,ratherthanjustbest-effortservices.

1.1.3MoreComplexO&MDuetotheIntroductionofNewSites,

RATs,andFrequencies

Thenumberoffrequencybandsincreasessignificantlytomorethan15frequencybands.Thesebandsrangefromsub-1GHzandsub-3GHztoC-band,sub-10GHzandmmWave.WirelessnetworksareevolvingintoFDD+TDDconvergencenetworks.Tomeettherequirementsofvariousscenarios,thestructureofwirelessnetworkswillevolveintoheterogeneousnetworksthatmayconsistofmacrobasestations,mastsites,microbasestations,andintegratedaccessandbackhaul(IAB).

Tomeetthischallenge,operatorsneedtoincorporateAItechnologiestoimproveO&Mefficiencyandprovideproactivepredictionandpreventionbasedontheconceptofachieving"zerofault"andmaintainingalways-onlinewirelessnetworks.

1)Formassivealarms,intelligentalarmidentificationanddiagnosisisrequiredtoaccuratelylocatetherootcauseofalarmsandhelpoperatorsaccuratelytroubleshoot.

2)Faultpredictionandpreventioncanhelpoperatorsproactivelyidentifynetworkrisksinadvance,enablingproactiveO&Mandfacilitatingtheestablishmentof"zerofault"wirelessnetworks.

1.2ConceptandStatusQuoofAutonomousNetworks

Autonomousnetworksenablenetworkstobecomemoreautomatedandintelligent.ByincorporatingautomationandintelligencetechnologiessuchasAI,thesenetworksaimtoachievepredictabilityandoperationalautonomy,enablingthedevelopmentofautomatedandintelligentO&Mcapabilitiesacrosstheentirecommunicationsnetworklifecycle.

1.2.1EvolutionofAutonomousNetworkingIndustryStandards

Withthediversificationandrapiddevelopmentofnewmobilenetworkapplicationservices,operatorshaveraisedtheirexpectationsforensuringservicequalityandimprovingnetworkefficiency.TheyhavetakenmeasurestopromotetheadoptionoftechnologiessuchasAIinthetelecomindustry.Inaddition,therapidgrowthofmobiletraffichasplacednewdemandsonspectralefficiencyandthedevelopmentandoptimizationofnewnetworktechnologies.Operatorsarefacingsignificantchallengesintheirdigitaltransformationjourney,includingmeetingdiverseservicedemands,improvingcustomerexperience,streamliningupstreamanddownstreamcollaboration,andimprovingcollaborationefficiencyinamulti-vendorenvironment.Theindustryhasreachedapreliminaryconsensusthatnetworkintelligenceandautomationarekeytechnicalmeanstoaddresstheseemergingrequirementsandchallenges.Astechnologyfrontrunners,standardizationandindustryorganizationsinandoutsideChina,suchasthe3rdGenerationPartnershipProject(3GPP),TMForum,EuropeanTelecommunicationsStandardsInstitute(ETSI),andChinaCommunicationsStandardsAssociation(CCSA),havealreadyinitiatedstandardizationeffortsintheirrespectivetechnicalareas.Theyareactivelyexchangingtechnologiesbetweenstandardsorganizationsthroughjointconferencesandcontactletters.Theparticipationofstandardsorganizationsinenterprisediscussionscanidentifytheoptimalpathforintelligenceevolution,helpoperatorscollaborateeffectivelywithpartnersindigitalnetworktransformation,streamlineend-to-endprocessesinmulti-vendorenvironments,andenhancecompetitivenessinrapidservicedeployment.

3GPPisaninternationalstandardsorganizationresponsiblefordevelopingstandardsformobilecommunicationstechnologies.Thestandardsandspecificationsdevelopedby3GPParemanagedthroughaseriesofreleases,typicallyformulatedeveryonetotwoyears.3GPPReleases18,19,and20representthesecondinnovationof5G,knownas5G-Advanced.Newfeaturesin3GPPRelease18includetheintegrationofAIandmachinelearningtechnologies,whichprovidedata-drivenandintelligentnetworksolutions.RANintelligencehasbecomeamajorresearchprojectwithin3GPPRelease18.Releases19and20arecurrentlyunderactivedevelopment,withafocuson

advancingautonomousnetworksintheindustry.

TMForumlaunchedtheAutonomousNetworksProject(ANP)inMay2019.Theprojectaimstodefinefullyautomated,zero-wait,zero-touch,zero-troubleinnovativenetwork/ICTservicesforusersandconsumersinverticalindustries.Inaddition,TMForumhastakentheleadinorganizingmulti-SDOworkacrossmultiplestandardsorganizationstoachieveconsensusonconcepts,frameworks,andkeyviewpointsintheautonomousnetworksdomainwhilefosteringcross-organizationalcollaboration.Todate,TMForumhaspublishedseveralspecificationsrelatedtoautonomousnetworks,coveringarchitecture,evolutionlevels,intelligenceusecases(UCs),andintentopenness.ThroughitsAutonomousNetworksCollaborationProject,theorganizationactivelysupportsthedevelopmentofwirelessintelligenceandprovidesindustryguidance.

ETSIisanindependent,nonprofitstandardsorganizationfocusedonthedevelopmentofinformationandcommunicationsstandards.ETSIhasrecentlypublishedawhitepaperentitled"UnlockingDigitalTransformationwithAutonomousNetworks",whichhighlightstheimportanceofautonomousnetworksinachievingdigitaltransformation.ThiswhitepaperservesasaproactiveinitiativebyETSItopromotetheadvancementofwirelessintelligenceandprovideguidancetotheindustry.

OnSeptember23,2021,TMForumandCCSAco-hostedthe2021AutonomousNetworksIndustrySummitinBeijingtostrengthenconsensuswithintheautonomousnetworksindustryandsharethelatestprogressinstandardsdevelopmentandbusinessachievements.Duringtheevent,keyfiguresfromChina'sMinistryofIndustryandInformationTechnology(MIIT),CCSA,andTMForumjointlyannouncedthreeindustrypromotioninitiatives,includingclarifyingthegenerationalcharacteristicsofautonomousnetworksandguidingtheirdevelopmentdirection;improvingtheworkingmechanismandestablishingindustrystandards;andpromotinginitialpilotprojectsandacceleratingthedeploymentofautonomousnetworksinlivenetworks.CCSAisactivelypromotingthedevelopmentofRANintelligenceandprovidingguidancetotheindustrythroughtheSummit.

Insummary,aglobalconsensushasbeenreachedamongoperatorsonthedeploymentofautonomousnetworks,withL4autonomousnetworksasthephasetargetforoperators.Bytheyear2022,morethan10leadingoperatorsworldwidehadannouncedtheircommitmenttoachieveL4autonomousnetworksby2025.Thesenetworkshavebecomeanintegraltoolforoperatorstotransformtheirdataintelligenceandarecurrentlyundergoingrapiddevelopment.

1.2.2AutonomousNetworkingPractices

TheAutonomousNetworksFrameworkdividesnetworkautonomycapabilitiesintosixlevels:L0toL5.TheAutonomousNetworkslevelsfromL0toL5

indicatedifferentnetworkfeaturesandcapabilities.TheAutonomousNetworksFramework,inaccordancewithTMForum'sGuidingPrinciples,outlinesthe

characteristicsofeachAutonomousNetworkslevelasitrelatestothe

implementationofITsystems,takingintoaccountthepracticalneedsofnetworkoperationsandmaintenance.

ChinaMobilehasbeenattheforefrontoftheautonomousnetworksindustry,leadinginbothdevelopmentandimplementation.Withtheconceptofautonomousnetworksinmind,thecompanyplanstodigitallyandintelligentlytransformnetworkO&M,strengthenthebuildingofitsautomationandintelligencecapabilities,andsettheoverallgoalofachievingL4autonomousnetworksby2025.ChinaMobileisbuildingintelligentinfrastructure,agileoperationsandall-scenarioservicesthroughfullyautomatednetworksandICTtoprovideoptimalcustomerexperienceforverticalindustriesandconsumers.Inaddition,theoperatoractivelypromotesinnovationinthetheoryofautonomousnetworks,developscapabilities,andimplementslarge-scaleapplications.Withafocusonsupportingcustomerdevelopmentandconsolidatingqualityleadership,ChinaMobileiscontinuouslyoptimizingthreeclosed-loopprocessesintermsofcustomerrequirements,end-to-endservicesandprofessional

networks.Thecompanyisalsocommittedtobuildingandimprovingtheintelligentoperationandmaintenancecapabilityofnetworksandnetworkmanagementsystems.

ToachievethegoalofachievingL4autonomousnetworksby2025,ChinaMobileisactivelyexploringtheecosystemcooperationsystem,industryenablementplatform,andnewO&Mservicemode.Inaddition,thecompanyiscontinuouslydeepeningindustrycooperationtopromotethedigitalandintelligenttransformationofO&M,ultimatelyachievingindustrydevelopmentandprosperitythroughjointefforts.Inthecomingyears,ChinaMobilewillcontinuetofocusonbuildingautonomousnetworks,acceleratingthedigitalandintelligenttransformationandupgradingofO&M,andimprovingnetworkquality.Theseeffortsarenotonlyessentialtopromotevalueoperationsonalargescale,butalsoimperativetofacilitatingthetransitionfromrelyingsolelyondemographicdividendstoembracingintelligencedividends.WiththeongoingdevelopmentandimplementationofAItechnologies,webelievethatautonomousnetworkswillbringusmoresurprisingexperiencesinthenearfuture.

WiththecontinuousadvancementofAItechnologies,moreandmoreoperatorsarerealizingtheimportanceofnetworkintelligenceinimprovingtheefficiencyofnetworkoperationsanddeliveringbetterserviceexperiencestocustomers.Asaresult,manyoperatorsareproactivelyinvestinginAItechnologiestoimprovetheircompetitiveedge.

By2025,70%ofoperatorsintheglobaltelecomindustryareexpectedtoinvestindevelopingAIsoftware,hardware,andservices,andbuildingintelligentnetworks,includingnetworkplanning,O&M,optimization,andcustomerservices.60%ofoperatorswillinvestinAItechnologiesto

implementnetworkintelligence,accountingformorethan20%oftheirtotalinvestment.WiththecontinuousdevelopmentandimplementationofAItechnologies,webelievethatnetworkintelligencewillbringusmoresurprisingexperiencesinthefuture.

1.2.3OutlookofAutonomousNetworks

Nativeintelligenceisoneofthevisionsofnetworkautonomy.ItmeansthatcommunicationsnetworkshavefullAIcapabilities,includingdata,computingpower,andalgorithmcapabilities.Thedevelopmentofthistechnologywillenablecommunicationsnetworkstobettermeetthefunctionalandlogicalrequirementsofintelligentapplicationsinthefuture.

Thewhitepaper6GVisionandCandidateTechnologiespublishedbytheChinaAcademyofInformationandCommunicationsTechnologyalsomentionsthatnewairinterfacesandnewnetworkarchitectureswithnativeintelligenceareamongthetencandidatetechnologiesof6G.Withthesetechnologies,networkperformancewillbegreatlyenhanced,meetingtherequirementsofnewservicesandscenariosinthefuture,servingtheintelligentsocietyandlife,andcontributingtotherealizationofthe6Gvisionsof"intelligentconnectionofeverythinganddigitaltwin."

Inthefuture,autonomousnetworktechnologieswillcontinuetoevolvetowardintelligentscenariosandnativeintelligence.Intelligentscenarios:Theclosed-loopintelligentsystemwithvisiblebenefits,perceptiblescenarios,plasticpolicies,andcontrollableeffectsisbuilttoimplementefficientandaccuratescenario-basedservicesandcontinuousevolution.Nativeintelligence:Theintegratedcomputing-networkarchitectureintegratesAIcomputingcapabilitiesintocommunicationprocesses,enhancingtheintelligentlearningcapabilityandscenarioadaptabilityforthephysicallayer,MAClayer,andnetworklayer,andenablingcontinuousevolutionofpersonalizedintelligentservicecapabilities.Italsointegratestheintelligencelayeranddatalayerintothenetworkarchitecturedesign,NEs,andinterfaceimplementation.Asaresult,thefutureintelligentnetworkcaninternallyimplementself-adaptation,self-learning,self-correction,andself-optimization.Inaword,nativeintelligence,asoneofthefuturevisionsofnetworkautonomy,willplayanimportantroleinthecommunicationsfield.Itwilldrivecontinuousinnovationincommunicationtechnologies,promoteindustrytransformationandupgrading,andcontributetothesustainabledevelopmentofhumansociety.

2TECHNICALREQUIREMENTSFORRANINTELLIGENCE

2.1RANIntelligenceArchitecture

TheRANintelligencearchitectureconsistsofthreelayers.Itaddressesthe

challengesofthenetworkfabric.Thenetworkmanagementsystem(NMS)

platformisresponsibleformanagingintelligenceacrossdomainsandvendors.Thesingle-domainoperationandmaintenancecenter(OMC)platformis

responsibleforsingle-domainwirelessnetworkintelligence.Thebasestationlayerprovidessingle-siteNEintelligence.Inthisway,layeredprocessingcanmaketheentirenetworkmoreefficientandintelligent.

Cross-DomainIntelligence:Cross-domainandcross-vendornetwork-levelintelligencecangloballymonitorandpredictthehealthoftheentirenetworkbyaggregatingandanalyzingdatafromdifferentdomainsandvendors.Inaddition,collaborativealgorithmscanbeusedtooptimizeresourceallocationacrossdomainsandvendors,improvingoverallsystemperformanceandefficiency.

NetworkIntelligence:Networkstatuscanbemonitoredandpredictedthroughcentralizedmanagementanddataanalysisofwirelessbasestations.Inaddition,intelligentalgorithmscanbeusedtooptimizeresourceallocationatthenetworklevel,improvingnetworkperformanceandefficiency.

NEIntelligence:Basestationscanuseembeddedintelligenttechnologiestoimplementreal-timeperceiving,modeling,prediction,andmulti-dimensionaldecisionmaking.Intelligentalgorithmscanbeusedtoimplementon-demandresourceconfiguration,providebestexperience,andachieveoptimalcapacitytoachieveoptimalperformanceandenergysavings.

Inaword,thearchitectureofRANintelligenceisevolvedaccordingtonetworkchallenges.Thegoalistoimplementmoreefficientintelligence

throughlayeredprocessing.

2.2TechnicalRequirementsforRANIntelligence

Aswirelessnetworksevolve,therearetechnicalbottlenecksintheoperationandoptimizationofdifferentnetworklayers.Asaresult,networksandNEsfaceenormouschallengesintermsofperceiving,analysisanddecisionmaking,intent,andexperience.Tomeetthesechallenges,operatorsandvendorsmustcontinuetoinnovateandupgradetechnologiestoimprove

networkintelligence.

Intelligence

Capability

Technical

Requirements

TechnicalDirection

Intelligent

Data

perceiving

Digital

Twin

Intent

Opennes

s

IntelligentAirInterfa

ce

Network

intelligence

(planning,

building,

maintainin

g,and

optimizing)

Network

perceivin

g

PreciseUE

positioning

Precisedevice

engineering

parameters

Precise

environment

modeling

Preciseradio

signal

measurement

Y

Analysis

and

decision-

making

Multi-objective

optimization

Multi-method

paralleloptimizationOne-shot

optimization

Y

Human

machine

interactio

n

Naturallanguageinteraction

Simplifiedinterfaceinvocation

Y

NE

intelligence

(performan

ce

experience

)

Wireless

perceivin

g

EfficientL3

multi-frequency

measurement

Y

Air

interface

interactio

n

EfficientL1channelmeasurement

Y

Intermsofperceiving,operatorsneedtoaccuratelymonitornetworkstatusinrealtime,collectandanalyzelargeamountsofdata,anduseefficientdataanalysisalgorithmsandpowerfuldataprocessingcapabilitiestoaccuratelymodeltheenvironment.

In