《全球6G技術大會：2024年6G網絡架構設計白皮書（英文版）（77頁）.pdf》由會員分享，可在線閱讀，更多相關《全球6G技術大會：2024年6G網絡架構設計白皮書（英文版）（77頁）.pdf（77頁珍藏版）》請在三個皮匠報告上搜索。

1、1/76Contents目錄1Vision.22Requirement.53Architecture.103.1 Architecture Design Principles.103.2 System Architecture.124.Key Technologies.214.1 Distributed Network Technology.214.2 Network Programmable Technology.254.3 Service.274.4 Flexible Frequency Domain Arrangement.334.5 Stack-Free User Interface.

2、364.6 Data Services.404.7 Computation-Network Integration.464.8 Native trustworthiness.534.9 Generalized QoS.574.10 Intelligent Endogenous.645.Summary and Outlook.736.References.747.Abbreviations.758.White Paper Contributors.762/761VisionThe advancement of information and communication technologies

3、isushering in the era of 6G,which promises to revolutionize our digitalexperiences.Expected to build upon the foundations laid by 5G,6G aimsto deliver richer,more immersive experiences,achieve ubiquitouscoverage,and foster new forms of collaboration.As a next-generationtechnology,6G is poised to ena

4、ble a plethora of novel applicationscenarios while enhancing existing functionalities.The International Telecommunication Union RadiocommunicationSector(lTU-R)has set forth a vision for the capabilities of IMT-2030,encompassing both the enhancement of current 5G capabilities and theintroduction of b

5、rand-new functionalities targeted specifically at 6G,These capabilities are categorized into two main groups:the enhancementof existing 5G capabilities and new capabilities targeted at 6G.AI Capabilities:Native AI in 6G networks means considering AIduring the design phase.AI capabilities empower the

6、 entire network,including access networks,core networks,and endpoints.Furthermore,6G network intelligence is inherently endowed with each networkfunction and the entire network,achieving autonomous operation ofnetwork services,business,and functions,truly becoming a high-levelautonomous network,and

7、serving as a pivot and foundation for enablingintelligent industries,without human intervention for self-monitoring,3/76self-organizing,self-optimizing,and self-repairing.Communication and Sensing Convergence Capability:The 6Gnetwork will incorporate new sensing capabilities to support integratedcom

8、munication and sensing scenarios.This convergence leverages theextensive reach of mobile communication networks to not only enhancecommunicationabilitiesbutalsoempowerthenetworkwithhigh-precision sensing.This dual functionality facilitates the intelligentconnection of all things,creating a synergy b

9、etween communication andsensing that was not possible with previous generations of networks.Data Service Capability:With the proliferation of intelligentapplication scenarios and the deepening integration of passive systems,6G networks will be home to massive amounts of heterogeneous data.Toeffectiv

10、ely manage this data deluge,it is imperative that data services beincorporated into the network architecture.The shift from centralized todistributed data services represents a significant evolution in how data isprocessed.By distributing data services closer to the source,6G networkscan offer real-

11、time processing,thereby providing faster response times.Additionally,the nativeAI capabilities of 6G require vast amounts of datafor model training,further underscoring the importance of efficient datamanagement.As such,6G networks must be capable of delivering secureand efficient data services both

12、 internally and externally to enhance dataflow efficiency and user experience.4/76Ubiquitous Computing Capability:The limitations of centralizedcloud computing are becoming increasingly apparent,especially inscenarios that demand low latency,high bandwidth,and real-timeprocessingsuch as immersive cl

13、oud XR,autonomous driving,and otheradvanced applications.To meet these diverse computing needs,6G willleverage distributed computility deployments.This approach ensures thatcomputing resources are available wherever and whenever they areneeded,providingatrulyubiquitouscomputingcapability.Suchadaptab

14、ility is essential for meeting the demands of pervasive intelligentapplications and catering to a variety of data service requirements.Ubiquitous Connectivity:A hallmark of 6G networks will be theirability to offer wide-area connectivity that spans various environments,including sky,earth,air,and se

15、a.This comprehensive reach aims tosatisfy the heterogeneous access scenarios and performance requirementsof diverse application areas.By maintaining consistent user experiencesacross different locations,6G networks will tackle challenges related toconnectivity,coverage,capacity,data rates,and termin

16、al mobility.Ubiquitous connectivity is fundamental for enabling users to seamlesslytransitionbetweenenvironmentswhileenjoyinghigh-quality,uninterrupted service.5/762RequirementIn 5G,the separation of software and hardware through virtualizationand containerization technologies,coupled with the intro

17、duction of aserviced architecture,allows the network to have a certain degree offlexibility and scalability,but there are still some problems that need to besolved due to the limitations of the 5G architecture.Centralized Network Model:The centralized nature of 5Gnetworks presents limitations in ter

18、ms of deployment andcustomization for future applications.It also poses a single pointof failure risk,where an issue at one central node could impactthe entire network.Rudimentary User Requirement Management:5G networksemploy network slicing to differentiate services but still have arelatively unifo

19、rm architecture and logical function divisionamong slices.This approach lacks the granularity needed toeffectively cater to diverse scenarios anticipated in 6G.Insufficient Network Function Sinking:While 5G enables localdata services through user-plane function sinking and privatenetwork establishme

20、nt for certain ToC(Technology ofConsumer)and ToB(Technology of Business)applications,itscentralized management is inadequate for a broader spectrum of6/76industrial uses.Additionally,the lack of interconnectivitybetween private networks hinders flexible resource allocation.Separate Development ofAri

21、thmetic Network and MobileCommunication Network:The current division betweenarithmetic(computational)networks and mobile communicationnetworks leads to a situation where computational resourcescannot be efficiently utilized due to lack of integration.Centralized cloud computing is not sufficient for

22、 low-latency,high-bandwidth,real-time applications such as immersive cloudXR and autonomous driving.This separation needs to beaddressed to improve collaboration and reduce latency.Chimney-Type Data Services:In 5G,data services are organizedin silos,which hinders data quality and efficiency.Thischim

23、ney-type structure becomes a bottleneck in data collectionand sharing.Additionally,the storage,interfaces,and datamanagement protocols vary across different generations ofcommunication networks,creating compatibility issues.Network Intelligence Plug-in and Strong Coupling:Thenetwork intelligence in

24、5G is often implemented as a plug-in,withAI functions provided by independent entities inside oroutside the network.This results in a high degree of couplingbetweenAI components.The fragmented nature of networkAI7/76application scenarios and research requires a more integratedand streamlined approac

25、h to facilitate broader use cases andsmoother operation.In order to meet the demand for 6G multi-scenario differentiation,6Gnetwork innovation is required.Network Flexibility:Emerging applications such as holographiccommunication,immersive XR(Extended Reality),sensoryinterconnection,Telematics,and a

26、utonomous driving havediverse performance requirements in terms of delay,reliability,throughput,etc.This necessitates that 6G networks must beflexible and scalable to rapidly adapt to these varied needs.Deeper Servitization:The concept of servitization,which refersto the transformation of products i

27、nto services,needs to godeeper in the network.This includes enhancing the servitizationof the core network control plane,extending it to the user plane,then to the RadioAccess Network(RAN),and finally to theUser Equipment(UE).The goal is to achieve end-to-endservitization of the entire network.Distr

28、ibutedArchitecture:Scenarios like cloud gaming andimmersive XR require temporary,large bandwidth,andlow-latency access services.Holographic communication andTelematics not only need high bandwidth and ultra-low latency8/76but also demand data services,AI services,and computationservices.Applications

29、 like federated learning must consider thelocation of data sources,computational resources,and latency,requiring dynamic configuration of network functions andresources.These needs push the network towards a flexible,end-to-end distributed architecture that can provide ubiquitousconnectivity and pro

30、ximity network services,along withreal-time or near-real-time quality of service monitoring andmanagement capabilities to satisfy diverse user demands.Multi-factor On-Demand Intelligent Orchestration:Theadvent of new applications such as holographic communication,immersive XR,sensory interconnection

31、,connected vehicles,andautonomous driving demand more than just connectivityservices from 6G networks.These scenarios require perceptionservices,intelligent services,data services,and arithmetic(computational)services.Consequently,6G networks must becapable of cross-layer and multi-layer intelligent

32、 organizationbased on user needs to provide a comprehensive range ofservices.End-Edge-Network-Cloud Collaboration:For computationalservices,6G networks are expected to evolve towards adistributed multi-tiered computing collaboration framework that9/76integrates cloud,network,edge,end,and industry.Th

33、isarchitecture will involve the cloud handling large-volume andcomplex computations along with algorithm training,the edgefocusing on agile access and local computation,and theendpoint being responsible for pervasive computing tasks likesensing and interaction.As business,data,andAI move towardsdist

34、ributed development,efficient collaboration between cloud,edge,and endpoint computing resources becomes essential.Data Collaboration and Data Sharing:Recognizing thenaturally distributed nature of data,there is a need for a unifieddata service framework that manages and schedules the lifecycleof dat

35、a services to enhance efficiency.Centered around userdemands,this involves a global perspective on data sourceselection,network element function configuration,networktopology design,and other data service units.These elementsmust be organized in an orderly manner,with real-time analysisconducted by

36、joint edge data services to generate services thatmeet the requirements.Endogenous Intelligence:Arithmetic and data resources areubiquitous,and integrating network and terminal arithmetic nearthe data source to realize intelligent capabilities and provideintelligent services can improve processing e

37、fficiency,reduce10/76data transmission costs,reduce latency,protect data privacy,andrealize green and low-carbon.3Architecture3.1Architecture Design PrinciplesThe architecture design for 6G networks must be approached from aholistic perspective,aligning with the key capability indexes proposed bythe

38、 International Telecommunication Union Radiocommunication Sector(ITU-R).Thedesignshouldaccommodatenewscenariosandtechnologies while allowing for the seamless evolution from 5G to 6Gnetworks.The following design principles are crucial:Simplicity:The 6G network is envisioned as a converged air and spa

39、ce integratednetwork.As the scale of network access grows and demands becomemore diverse,simplicity in architecture and protocols is essential.Thissimplicity enables efficient data transmission and on-demand networkfunction deployment.It aims to provide a lightweight and flexibleclosed-loop network

40、that effectively reduces energy consumption andavoids unnecessary network redundancy.ConvergenceThe 6G network aims to move beyond traditional unidimensionalcommunication and transmission services.It will integrate capabilities11/76such as sensing and data processing to offer multi-dimensional inter

41、naland external information services.By fully leveraging the networkresource platform,it can support comprehensive life-cycle services forthe development of a fully informational society.Flexibility6Gaimstoimplementafull-service/programmablenetworkarchitecture that separates software from hardware r

42、esources.Thisseparationallowsforflexibleresourceallocation,enhancingthenetworks adaptability.It reduces the costs associated with networkupgrades and replacements by enabling a more modular approach.Thenetwork should provide plug-and-play capabilities for various emergingapplication scenarios and ve

43、rtical industries,allowing them to integrateseamlessly.EndogenousThe6Gnetworkarchitectureshouldincorporateendogenousintelligence,arithmetic,and security.This integration aims to achieveubiquitous networking,computation,and intelligence,penetrating everydomain,network,and unit throughout their entire

44、 lifecycle.By doing so,the 6G network architecture will inherently support a new paradigm thatextends beyond traditional connectivity and is driven by artificialintelligence(AI).Openness12/76The 6G network should serve as an open platform that facilitatesbusiness opportunities by connecting the netw

45、orks internal capabilitieswith external demands.It should establish robust interoperability andopenness,supporting all individuals and enterprises capable of interactingwith the network.By leveraging network resources,it can offer valuableassistance to further the development of industry informatiza

46、tion.3.2 SystemArchitectureBased on the design principles,a 6G network architecture isproposed,including a network resource and infrastructure layer,anetwork function layer,a service and capability open layer,managementorchestration,and endogenous enablement.Figure 3-1 6G network architectureNetwork

47、 Resources and Infrastructure LayerThislayerprovidesthenecessaryinfrastructureandmultidimensional resources required for the generation of networkfunctions.It encompasses a range of resources including ubiquitous13/76wireless,computing,storage,and networking capabilities.Resourcesinclude unified vir

48、tualized resources,abstractable physical resources,anddedicated high-performance hardware resources.These resources are theoperational foundation for the entire network and are managed by thenetwork functions of the upper layers.As a pass-sense integrated network,it offers data information collectio

49、n capabilities,providing support for AIapplications through big data.Network Function LayerBuilding on the pervasive infrastructure from the lower layer,thislayerinterconnectsdynamicallydistributedresources.Itenablescommunication,sensing,computation,intelligence,and digital securitynetwork functions

50、 through the unified and cooperative scheduling ofmultidimensional resources.The layer supports atomization functionssuch as splitting,combining,or expanding that users require in differentnetwork systems.It is represented by multiple network elements that offer full-servicesolutions,meeting the fra

51、gmented,diverse,and complex networkdemands of various industry applications.The layer transforms thetraditional application-adapted network into an application-definednetwork by providing on-demand intelligent,simple,and trustedfunctions and services.Through endogenous empowerment,the networkcontinu

52、ously integrates and absorbs advanced technologies,becoming an14/76evolutionary force that progresses to more advanced stages.The networkfunctions support the explosive growth of new industries and applications,offering a smart and streamlined platform.This layer contributes to thetransformation and

53、 upgrading of economic and social digitization,networking,and intelligence.Data PlaneIt is responsible for data management functions such as collection,processing,analysis and service of all data in the whole network.Drivenby the demand of Smart Connection of Everything,Digital Twin,itimproves the p

54、erformance of data services,reduces the overhead of datatransmissiononbandwidth,carriesoutunified andcollaborativemanagement of data in the whole network,supports 6G to provide safe,shared and trustworthy data services for the whole network,and createsmore value through intelligent processing and an

55、alysis of data.At thesame time,it provides secure and efficient data services both inside andoutside the network to enhance data flow efficiency and user experience.Control PlaneThe control plane is an enhanced functional plane.6G networkarchitecture will support traditional connectivity services,bu

56、t also addcontrol services related to intelligence,arithmetic,security and othercapabilities,realizing a major shift from mobile communications tomobile information services.The control plane will also expand from a15/76single connection service control to multiple business dimensions ofcommunicatio

57、n,computing,intelligence and digital control,support theunified and simple access of star-earth convergence,facilitate thein-depthintegrationofendogenouscapabilitiessuchasintelligence/security/calculation power with the new architecture,andrealize flexible,intelligent,and secure control of the netwo

58、rk based onthe endogenous capabilities conferred by endogenous empowerment.User PlaneThe user plane is an enhanced functional plane.As the only face inthe network architecture that provides user data traffic processing andforwarding functions,the user plane is the core of realizing the ultimateexper

59、ience of user services,and the window for realizing the user as thecenter.The user plane supports the data forwarding and transmission ofmulti-dimensionalnewservicesthroughexpansiontomeetthetransmission requirements of 6G communication,perception,arithmetic,intelligence,data,and secure converged net

60、work;under the trend ofend-to-end full-service provisioning,the user plane servicing will bedriven by scenarios to activate the networks application mode throughordering according to a single dish,combined with the servicing ofRAN.In the trend of end-to-end full-service provisioning,user planeservic

61、e provisioning will activate the application mode of networkthrough scene-driven la carte,which,combined with RAN service16/76provisioning,will bring the possibility of optimization by splitting andrestructuring the user plane functions of the access network and the corenetwork at the same time.high

62、-precision time synchronization;in orderto achieve high-performance forwarding and network autonomousinterconnection,the user plane needs to enhance or introduce newtransport protocols.Service and Capability Open LayerIt extracts,encapsulates and combines the network functions of thelower layers,and

63、 provides on-demand capabilities or services that can beopened for internal network services or external applications,coveringconnectivity as well as a variety of capabilities or services such as data,computing,intelligence and orchestration management,which is theexternal embodiment of network as a

64、 service.The service and capability open layer interoperates with theapplication layer in the north direction and connects with capabilities andservices in the south direction,supports the safe opening of networkcapabilities,and provides callable,friendly and rich API interfaces foratomic capabiliti

65、es.According to the business requirements of differentscenarios,it provides on-demand network services for ubiquitous usersand improves the effect of user experience.At the same time,it interactswith other layers/faces to collect and encapsulate the service andcapability open layer of other layers/f

66、aces.17/76Endogenous empowerment realizes full life cycle management ofendogenous capabilities and on-demand scheduling of multiple elementsof endogenous capabilities within the network,creates endogenousmulti-dimensional capabilities,and empowers the network with networkfunctions such as intelligen

67、ce,arithmetic power,security,etc.,whichserves as the foundation of the 6G network,and builds up an on-demand,flexible and efficient endogenous capability resource pool.Based onendogenous empowerment,the 6G network can optimize networkperformance and achieve network autonomy by using endogenouscapabi

68、lities internally without external intervention;externally,it can givemore support to new services for applications in various industries andimprove user experience.The current capabilities of endogenousempowerment include intelligence,arithmetic,and security,and thecapabilities of endogenous empowe

69、rment will continue to expand withthe emergence of new business scenarios and vertical industries.Endogenous SecurityThe introduction of new application scenarios and new keytechnologies in 6G brings new security problems.6G network securitycapability should be integrated into the network architectu

70、re,andendogenous security empowerment is oriented by endogenous security,based on distributed trustworthiness,breaking the traditional securityboundaries,anddeployingdatasecurityandprivacyprotection18/76technologies,which can provide endogenous security sensing,defenseand prevention functions for ea

71、ch function and resource of the 6Gnetworksystemandmakethesecuritycapabilityon-demandcustomizable and on-demand access.The security capabilities can becustomized on demand,guaranteeing network and user security internallyand realizing the openness of security capabilities externally.Ultimately,the 6G

72、 network architecture empowers the formation of automaticimmunity,trust and consensus,synergistic elasticity,intelligent andefficient security system,and realizes the effective operation of the entirenetwork with multiple applications and high reliability and high securityin parallel.EndogenousAIArt

73、ificial intelligence will become the first endogenous driving forceforthedevelopmentofhigh-performancenetworks.6G networkintelligence will deeply integrate and design AIs arithmetic,data,algorithms,and other elemental requirements in network functions,architecture,protocols,andprocesses,realizingthe

74、on-demandscheduling and flow ofAI data,algorithms,arithmetic,and other resourceelements at the network level,and more real-time and high-efficiency AItraining/reasoning,as well as the intelligent identification of AI scenariosand the smart generation of use cases,so as to improve networkperformance

75、and efficiency through AI.The AI network can be used to19/76improve the performance and efficiency of the network through AI andrealize a high level of autonomy of the network.Through the intelligentnative capabilities of the network,it can achieve business intelligence andserviceintelligence,andtru

76、lybecomeanend-to-endhigh-levelautonomous network,as well as a link and cornerstone for empoweringthousands of industries with universal intelligence.Endogenous ComputilityAs the emergence of new users,technologies,applications andscenarios has brought about exponential growth in data volume,variousi

77、ndustries have put forward a more urgent need for arithmetic power andnetwork.6G network will be deeply integrated with arithmetic network,and arithmetic power capability will be built into the architecture,whichon the one hand will enable the mobile communication network toefficiently utilize arith

78、metic resources and build an enhanced networkarchitecture with endogenous arithmetic power;and on the other hand,itcan provide arithmetic power services for external industrial applicationsthrough the mobile communication network to empower the diversifiedapplication needs of thousands of industries

79、.On the other hand,it canprovide arithmetic services for external industry applications through themobile communication network,empowering the diversified applicationneeds of thousands of industries.Through the unified endogenousempowerment and penetration of arithmetic service capability,it can20/7

80、6provide arithmetic service for other layers/facets.Orchestration ManagementOrchestration management is given new meanings and capabilitiesin 6G to match the management requirements of on-demand services in6G full-scenario networks.Through intelligent sensing of user intent andservice demand,it real

81、izes intelligent cooperative scheduling anddynamic scheduling of intent policies across multiple services,multipledomains,and the entire life cycle,and realizes closed-loop assurance ofservice quality in heterogeneous environments.Through in-depth sensingand intelligent management of heterogeneous r

82、esources in the wholedomain,it builds a panoramic knowledge space,multi-dimensionally anduniformly arranges network resources,network functions and networkservice capabilities,completes the closed-loop intelligent managementand control of sensing,analysis,decision-making and evaluation,realizeson-de

83、mand selection of network capabilities and balanced application,and supports integrated on-demand scheduling and management ofstar-earth convergence to maximize the utilization rate of resources.Intelligent Endogenous greatly expands the management boundaries oforchestration management,and realizes

84、automatic closed-loop control ofnetwork planning-construction-optimization by establishingintelligent anomaly detection and self-healing mechanisms in complexscenarios of the whole network and deeply integrating the digital twin21/76network to control the real network with the virtual.It realizes au

85、tomaticclosed-loop control of network planning-construction-optimization andenhances network robustness.At the same time,the orchestration management surface supports thedeep fusion of digital twin networks to control the real with the virtual inorder to achieve continuous optimization of the physic

86、al networkoptimization and simulation verification.4.Key Technologies4.1 Distributed Network TechnologyThe evolution of network architectures from 4G to 5G has primarilyrevolved around a centralized model,with an emphasis on deploying userplane functions near the user for local data offloading.This

87、approach,however,is being reevaluated in light of the burgeoning 6G landscape,which demands a more nuanced and flexible network design to cater to awide array of emerging scenarios.It is increasingly clear that merelyoffloading the user plane function cannot satisfy the multifaceted needs ofusers in

88、 this new era.As such,there is a growing impetus for the controlplaneand even the management planeto be brought closer to theusers vicinity,enabling distributed deployment and facilitating localprocessing services.The advent of communication environments thatintegrate AI necessitates networks capabl

89、e of handling vast amounts of22/76data and performing intricate computations proximately.Furthermore,asthe variety and number of terminal devices proliferate,the vulnerabilityof centralized deployments to single points of failure becomes ever moreapparent,underscoring the need for a transition towar

90、ds an end-to-enddistributed network paradigm.In anticipation of these developments,the 6G network is expected tofeature a multitude of edge,private,and campus networks coexisting andinteroperating.Achieving adaptable interconnectivity and mutual trustbetween these networks will likely see the adopti

91、on of a hybrid modelthat combines centralized and distributed elements.Envisioned is anetwork topology centered around a central node,complemented byvarious specialized and customizable distributed nodes.Each distributednode would be empowered to operate autonomously,encompassingself-contained proce

92、ss loops and independent service provisioning,alongwith attributes such as self-management and self-optimization.In this 6Garchitectural blueprint,central nodes will coexist and interlink withprivate and campus networks,creating a rich ecosystem of connectedentities.Both central and distributed node

93、s will independently furnishserviceswhilealsobeingcapableofintelligent,demand-drivencollaboration.This synergy aims to fulfill the expansive connectivityrequirements of 6G use cases,significantly enhancing network coverageand ultimately delivering ubiquitous internet services and customized23/76busi

94、ness solutions to users anytime,anywhere.Figure 4-1 Distributed network organization diagramThe dynamic nature of user demands,coupled with the in both thetiming and location coupled with the unpredictable fluctuations in boththe timing and location of these needs,underscores the importance ofcollab

95、orative service provisioning between distributed networks andcentral nodes.The essence of this cooperation lies in the ability to offerdifferentiated yet coherent business services to customers across variousscenarios.Network deployment strategies often weigh multiple factors such ascost and average

96、 user demand.However,there are instances where asudden surge in user demand occurs,which existing resources maystruggle to accommodate.This is particularly true during peak timesassociated with events like sports matches,large-scale performances,andother special occasions.Additionally,when users mov

97、e to new locations,24/76the deployed distributed network nodes may not always align with theirbusinessneeds,potentiallydisruptingservicecontinuityandcompromising the user experience.In response to these challenges,itscrucialfordistributednetworknodestocollaborateeffectively,optimizing the use of ava

98、ilable network resources to reduce costs whilemaintaining a seamless and consistent network experience.Looking ahead,the capabilities of UE,including their demands onthe network,are becoming increasingly sophisticated.For instance,UEsare no longer solely used by toC but are evolving to require not j

99、ustconnection services but also computational power,AI services,andother advanced functionalities.Furthermore,UEs are transitioning frombeing mere consumers to service providers,offering external dataservices and intelligence-related capabilities such as algorithms andmodels.In light of these develo

100、pments,its essential to consider both thecapabilities of UEs and the networks capacity to accurately match thetwo.This matching mechanism is critical for realizing precise on-demandservices that cater to the specific requirements of each user or serviceprovider,ensuringahighlyadaptiveandresponsivene

101、tworkenvironment.25/764.2 Network Programmable TechnologyNetwork programmable technology can flexibly program thecompilable instructions of network devices based on different businessrequirements,and can formulate and change the logical algorithms ofcontrol plane and user plane.The network programma

102、ble technology willbreak the traditional black box architecture of tightly coupled networkequipment,operating system and network applications,and bring moreopen capabilities to the network,which will bring a new paradigm for 6Gand future networks.The 5G core network architecture based on SDN separat

103、es thecontrol function of the network from the forwarding function,programsthe matching forwarding rules of user data through the control plane,andintroduces the control plane programmable technology,so that the mobilenetwork takes shape as a programmable prototype.Facing the needs of6G communicatio

104、n,perception,arithmetic,intelligence,data,security andother diversified rules,the programmable capability in the currentnetwork can no longer meet the business needs.In order to meet thedemand for data transmission of diversified services and improve thescalability and extensibility of the network,6

105、G should introduce userplane programmable network technology to realize the programmabilityof the whole network in the control plane and user plane.26/76Figure 4-2 5G to 6G Programmable Capabilities EvolutionSRv6 is poised to be a pivotal technology for 6G user planeprogramming.As an evolutionary ex

106、pansion of the existing networkinfrastructure,SRv6 augments network programmability by embeddingforwarding rule commands into packets at the origination point of thepath.This guidance directs packet forwarding along the intended route.For 6G user planes,adopting SRv6 or its advanced derivatives in p

107、lace ofthe conventional GTP stack is under consideration.Leveraging SRv6striple-layer programming capability,one can flexibly define the userplane,achieving programmability of paths,behaviors,and data within theuser plane to better cater to diverse network needs.Moreover,an SRv6-based user plane pro

108、tocol stack enables a NativeIP end-to-end channel that directly connects users to data centers.Thisapproach simplifies the network layers,resulting in a more streamlined,controllable,and adaptable network architecture.The protocol-independent packet processing power of P4 offers27/76robust and versa

109、tile programming capabilities.By defining packet headerformats,parsers,matching action tables,and control programs,P4realizes the programming of device packet processing.When 6G userplanes employ P4-enabled devices,they can deftly craft user planealgorithmstailoredtoservicerequirements.Thisleadstoap

110、rotocol-agnostic programmable user plane that significantly boostsflexibility and manageability.It paves the way for an innovative 6G userplane that readily accommodates and evolves with new 6G services,enhancing the user planes adaptability and offering an effective solutionfor the integration and

111、evolution of emerging 6G services.Inresponsetonovel6Gdemands,networkprogrammabletechnologies promise a greener,more malleable mobile network upgradeand evolution.They prepare the network to swiftly and nimbly adapt tofreshandfluctuatingservicedemands,facilitatingrapidservicedeployment and dynamic ad

112、justment.4.3 ServiceThe transition from 4G to 5G mobile communication networks hasbrought significant changes to the core network architecture compared tothe changes in air ports.The 5G core network adopts a service-basedarchitecturethatenablestheindependentscaling,evolution,andon-demand deployment

113、of network functions.This architecture builds28/76upon the separation of the control plane and user plane and undergoesextensive optimization.Continuous improvements are being made toenhance service-based functions and frameworks further.In the forthcoming 6G era,characterized by virtual reality and

114、interconnectedness of all things,new business requirements,applicationscenarios,and key technologies will emerge.These will be built upon theconcept of three-dimensional coverage to ensure seamless switchingbetween various scenarios while maintaining business continuity.Thediversity of business and

115、application requirements is set to increase,encompassing not only interaction forms and content but also servicecustomization and personalization.An integration of communication,computing,and sensing technologies will enrich service types andbusiness scenarios even further.The 5G core networks contr

116、ol plane has already achieved a flexiblesceneadaptation capabilitythroughitsservice-based architecture.However,due to the ongoing development of user-plane service,RANservice,and UE service,the current network does not yet offer anend-to-end full-scene adaptation capability.To meet the 6G demand for

117、comprehensive scene adaptation,deeper network service is necessary.This involves expanding beyond just enhancing the core networkscontrol plane service to include user plane service,extending to RANservice,and ultimately reaching UE service.Such a transformation aims29/76to achieve end-to-end servic

118、e and full service.4.3.1 user plane serviceThe advancement to 5G has primarily focused on the service of thecontrol plane,leaving the user plane yet to be fully transformed.Theintroduction of the R18 UPF with event exposure services marks asignificant milestone in this transition,signifying the comm

119、encement ofthe user planes evolution towards service.In adherence to the establisheddesign doctrine of distinct control and forwarding roles,enhancementsare being integrated into the UPF to extend service capabilities from thecore networks control plane to encompass the user plane.This expansionincl

120、udes the transformation of interfaces engaged with the control planeand radio into service-based entities.With the completion of user plane service,traditional interfaces suchas N4,N3,and N9 will be rendered obsolete,replaced by an exclusivelyservice-based interface landscape for all interactions.Th

121、ese changesmove away from the classic OSI model,which relies on hierarchicallayering where each layer is dependent on the services of the layer belowand provides services to the layer above.When interacting betweenupper and lower layers,follow the interface agreement,and wheninteracting between the

122、same layer,follow the protocol agreement.Leveraging cutting-edge technologies like Server Mesh,Volcano,30/76and a diverse suite of micro-service technology stacks,servingmultipleuser planes becomes a reality through service reconfiguration.Thisoverhaul allows the network to dynamically adjust anchor

123、 settings,engage in multi-dimensional business processing,and cater to individualuser preferences through network personalization.Moreover,it paves theway for an optimized service operation environment,ensuring a seamlessand efficient user experience that can be finely tuned to meet theever-evolving

124、 demands of the network.Figure 4-4 User plane service4.3.2 RAN ServiceIn the traditional communication model,RAN establish a predefinedpoint-to-point N2 signaling interface for interactions with the corenetworks control plane network function AMF,and a distinct predefinedpoint-to-point N3 data inter

125、face for communications with the user planenetwork function UPF.The introduction of new functions necessitatesenhancements to existing network functions and the definition of newpoint-to-point interfaces between these novel functions and theircorresponding interacting network functions.31/76The adve

126、nt of the service-based architecture in the 5G core networksignifies a transformative shift,allowing for expansion and evolution ofservice-based architecture.This architectural progression extends to theN2 interface,bridging the core network with the gNB CU-CP(NextGeneration NodeB-Centralized Unit-C

127、ontrol Plane).Within thisframework,the gNB-CU-CP operates as a unified RAN control planeservice,interfacing with the core networks control plane services,whilethe gNB-CU-UP(Next Generation NodeB-Centralized Unit-User Plane)functions as a unified RAN user plane service,interacting with the corenetwor

128、ks user plane services.Advancing to the subsequent phase,the RAN control plane functionundergoes reconfiguration into multiple RAN Control Plane Services(CPS),and the RAN User Plane function is restructured into multipleRAN User Plane Services(UPS).This service of the RAN control anduser planes redu

129、ces superfluous interactions with the core networks AMF.Concurrently,the nature of interactions with the core network evolvesfrom a linear,sequential pattern to a more efficient,multi-party parallelexchange.This paradigm shift optimizes the signaling interaction process,enhancing overall network per

130、formance and flexibility.32/76Figure 4-5 RAN Service4.3.3 UE ServiceThe 6G era of virtual reality and interconnection of all things,based on the user perspective means that the number,type,intelligenceand complexity of terminals accessing the mobile communicationnetwork have increased dramatically,a

131、nd major changes have occurredfrom the original only cell phones needing access to the mobilecommunication network to the present time when drones,cars,smarthome appliances,industrial equipment,etc.will all have access needs.Atthe same time,with the re-emergence of the cloud cell phone and cloudcomp

132、uter market,UEs can also have the service capability to provide UEservices(UE Service,UES)such as computing,metrics,and UEinformation to operator networks,third-party applications,tenants,otherUEs,etc.UE services will be combined with network services andaccessed to each other through service-orient

133、ed interfaces to realize moreflexible and direct information interaction.33/76Figure 4-6 Terminal Service4.4 Flexible Frequency Domain ArrangementMobilecommunicationneedstofaceincreasingbusinessperformance requirements,and wireless access networks relying solelyon one or two frequency bands can no l

134、onger meet the performancerequirements of businesses.Therefore,it is necessary to integrate and usemultiple frequency bands to achieve higher performance requirements.The spectrum resources face increasingly large carrier bandwidth anddifferences in physical characteristics,which have a significant

135、impact onthe ability of wireless access networks to adapt to services.Mainlyincluding the following key technologies。4.4.1 Carrier CascadeThe key elements of carrier cascade include baseband carrier,radiofrequency carrier,and their mapping relationship(hereinafter referred toas mapping relationship)

136、.In a multi carrier scenario,a baseband carrier34/76in a cell maps multiple RF carriers,and physical parameters and resourcescheduling are configured based on the baseband carrier,which cancombine the processing of the underlying RF carriers.At the same time,in order to reduce energy consumption,the

137、 activation/deactivation of RFcarrier,baseband carrier,and mapping relationship together determine theoverall activation/deactivation status of the carrier.RF carrier andbaseband carrier can be activated separately,and the mapping relationshiprequires both to be activated in order to be activated.4.

138、4.2 Decoupling of uplink and downlinkIn uplink and downlink decoupling,the cell consists of an uplinkpool and a downlink pool,each consisting of multiple carriers.Thedecoupling between uplink and downlink pools is mainly reflected in thefollowing aspects:the mapping relationship between uplink and d

139、ownlinkcarriers is not fixed;There is no limit to the number of uplink anddownlink carriers;There are no scheduling or feedback constraintsbetween uplink and downlink carriers.Furthermore,uplink and downlinkTRPs can also be flexibly used,and this networking solution can achieveflexible deployment of

140、 uplink and downlink TRPs,customize linkcoverage as needed,and adapt to extreme scenarios.35/764.4.3 Physical Channel arrangementThe communication process of a mobile communication system is toestablishacommunicationchannelthroughsignalingbeforethetransmission of user equipment and external data can

141、 be achieved.However,this data transmission method is not suitable for all businesses.In order to meet the diverse needs of users,the physical channel setcorresponding to users can be different for different businesses and times.In high user density and sudden small business package scenarios,energy

142、-saving orchestration can be used to carry synchronous informationand lightweight data through SSB,eliminating the process of receivingsystem messages,establishing connections,and receiving PDCCH.Inhigh user density and periodic business package scenarios,cost savingorchestration can be used to achi

143、eve decoupling of system information,control information,and business channels,and centralized transmissionof system and control information.4.4.4 Transmission Channel VirtualizationA TB(transmission block)contains multiple CBs(encoding blocks),and any error in one CB will result in the entire TB be

144、ing unable to besubmitted.The transmission channel virtualization scheme divides theoriginal data stream,transmission channel,and TB into multiple data substreams,transmission channels,and TB.In this way,any TB error at the36/76receiving end will not hinder the submission of other TBs.In order toeff

145、iciently schedule multiple TBs mentioned above,the transmissionchannel virtualization scheme also introduces the concept of TB Group(TBG).TBG contains multiple TBs,which can reuse the same schedulingsignaling.4.4.5 SI PoolingThe pooling of SI refers to the aggregation of system message SIfrom a phys

146、ical resource pool,including multiple carrier links,to be senton a single carrier.This allows carriers that do not send SI to sleep for alonger period of time,reducing the energy consumption of base stationswhile also reducing the workload and cost(configuration,scheduling)ofSI publishing,simplifyin

147、g network operation and maintenance.Afterreceiving physical resource pool information through broadcasting,theterminal can obtain resource configurations for pooled multiple accesschannels(RACHs)and select the optimal RACH resources for access.The optimal choice of RACH can improve the success rate

148、of access andquickly build connections and services with the best coverage/energyefficiency frequency band.4.5 Stack-Free User InterfaceThe current wireless communication protocol stack is a layered37/76architecture,with each layer containing fixed functions and unchangedexecution order.This rigid a

149、nd tightly coupled protocol stack greatlylimits the network capability of 6G.The non stack based user interfacedesign ensures the ability of 6G user interfaces to adaptively match thecomplex and ever-changing business needs of future scenarios.It mainlyincludes three aspects:componentization of func

150、tions,parallelization ofdata packet vectors,and deep collaborative intelligence of components.4.5.1 Functional componentizationThe user interface architecture in traditional systems is very rigid,and changes in functionality can lead to changes in the entire protocol,and the same functionality at ea

151、ch protocol layer cannot be reused.Inorder to reduce the coupling and redundancy between various functionalmodules of the user interface,based on the principle of mutualindependence and complete exhaustion,the user interface functions aredecomposed into components.Independent components can quicklyi

152、terate and upgrade,and can maximize function reuse,minimize changes,independently develop,deploy,and maintain.On site customization iscarried out according to personalized needs.At the same time,thecomponentizationofuserinterfacefunctionsextractscommoncomponents and processes,reducing protocol compl

153、exity.In the new userinterface architecture,independent components are aggregated to form a38/76functional component library,which not only supports functionaldecoupling,but also facilitates the introduction of new components andtheir replacement and upgrading.By configuring,orchestrating,andmanagin

154、g components in the component library,the user interface canadapt to building a data transmission chain according to variouspersonalized needs.4.5.2 Parallelization of packet vectorsIn the IT field,in order to quickly build switching and routingfunctions,a scalable and high-performance packet proces

155、sing frameworkcalled VPP(Vector Packet Processing)has been introduced.The mainidea of VPP is to vectorize and batch process continuous data packetswith the same features.When users face continuous data packets withcommon set features and no need to stay in order,vectorization cansignificantly improv

156、e packet processing speed.The granularity of packetvectorization can vary flexibly,allowing for VPP of components as wellas VPP of internal steps within components.The user interface can alsoachieve parallel processing of multiple independent and statelesscomponents for a single data packet,achievin

157、g the effect of multiplecomponents simultaneously processing the same data packet.The parallelexecution of components also increases the dimension of componentorchestration,forming various combinations such as serial,parallel,and39/76mixed serial parallel.Users can choose the most suitable way toorc

158、hestrate according to their needs.The non stack user interface alsosupports cross domain/multi hop deployment,which is very suitable formulti-level deployment scenarios,multi connections,MESH networking,and other scenarios.By deploying components on demand,morefunctional partitioning methods have be

159、en added besides CU/DUseparation,increasing deployment flexibility.4.5.3 Component Deep Collaborative IntelligenceThe non stack user interface that can be arranged fundamentallybreaks the hierarchical concept of traditional protocol stacks,deepens thecollaboration between functional components,and p

160、lays an importantrole in building service-oriented on-site adaptive customized flexiblenetworks.By intelligently configuring,orchestrating,and managing userinterface components,the user interface architecture can introduce bigdata and AI algorithm engines,thereby improving the level of userinterface

161、intelligence.Throughdeepcollaborationofdifferentcomponents,the synergy,robustness,flexibility,and scalability of thecomponents can be improved.For example,tightly orchestrating andlinking components to achieve one-time read and batch processing of datacaching.40/764.6 Data Services4.6.1 Trends in 6G

162、 Data ServicesThe advent of smarter,more connected devices through socialnetworksandtheInternetofThings(IoT)meansthatmobilecommunication networks will be handling more than just voice and textdata.Streaming data and machine-to-machine interactions will becomeincreasingly common.New business scenario

163、s demand lower latency andhigher reliability in data transmission,with a heightened focus on datasovereignty,data security,and data privacy.Processing large volumes ofdata prior to transmission,enhancing data transmission performance,andensuring data privacy are critical considerations that necessit

164、ate furtheroptimization of data processing functions.With an integrated air-heaven network that enables ubiquitousconnectivity,various data types from different devicessuch as sensordata,AI data,and network behavior and status datamay converge on asingle device for processing.This requires the inter

165、connection andcollaborative transmission between 6G multi-domain networks,allowingdata to flow across domains.Utilizing statistics-and AI analysis,newknowledge can be derived from this data,creating value through datacollaboration and federation.Data can generate new knowledge throughstatistics and

166、AI analysis,and new value can be generated through data41/76collaboration and federation.Data collected at once can be utilized andobtained multiple times.The cross-domain data flow and shared services,adds another layer of diversity to 6G data services.To meet the evolving development trend of 6G d

167、ata services,adeparture from the conventional point-to-point pipeline transfer ofcommunication session data and centralized deploymentof computingpower and storage resources.The 6G network must facilitate high-fidelitycapture and retrieval of distributed data,enabling the integratedrepresentation an

168、d amalgamation of multimodal data.It should bolster thedistributed storage of vast data volumes and leverage distributed networkresources and computational capabilities to accommodate data transferacross any network structure,including on-the-move processing.In response to these requirements,the int

169、roduction of an autonomousdata plane within the 6G network architecture is advocated.Thisdedicated plane aims to holistically address the complexities of Tosystematically address the challenges of controlling and monetizingnetwork data,operations and maintenance data,AI data,perception data,and comp

170、uting power data.By doing so,it seeks to furnish secure andefficient data-centric services for both internal network operations andexternal users,enhancing data fluidity and elevating the user experience.The proposed 6G data plane embraces a uniform framework thatspans the entire data lifecyclecolle

171、ction,forwarding,processing,and42/76distributionfor diverse business entities.Its network functionalities arecategorized into three main areas:data service orchestration managementfunction,data plane execution function and data plane managementcontrol function.Data service orchestration management f

172、unction(DSM):DSM asthe conductor of the 6G networks data services,agilely managingresources and service orchestration.This function is pivotal forconfiguringthefunctionalentitieswithinthedataplane,maintaining data models,and overseeing data services andapplications.The DSMs role is critical in achie

173、ving on-demanddeploymentofdata-planefunctionsandenablingelasticreconfiguration.Through its meticulous management,the networkis poised to deliver premium quality data services that meet theevolving needs of users.Data Plane Execution Function(DPF):DPF takes charge of thedata operations gamut,from col

174、lection and preprocessing,storage,forwarding,analysis,and the eventual provision of data services.The DPF ensures that data flows seamlessly and undergoesnecessary processing in route,thereby facilitating a highlyresponsive and adaptive network environment.Data Management and Control Function(DPM):D

175、PM wieldsoverarching control over the data plane.The DPM rapidly43/76executes orchestration and control of data plane executionfunctions based on data service requests originating fromterminalsnetworksandapplications.Itsupportsthecomprehensive lifecycle management of data,including strategiccontrol

176、of data services,access control,registration and discoveryof data plane execution functions,data storage oversight,andenforcement of authentication and authorization protocols.TheDPMs role is crucial for maintaining a secure,efficient,andreliable data management system across the network.4.6.2 6G Da

177、ta PlaneArchitectureBuilding upon the foundation of 5Gs network architecture,avisionary 6G system architecture has been conceptualized.At its core is arefined data plane architecture designed to optimize performance acrossthe network side,access network side,and terminal side.Figure 4-7 6G Data Plan

178、eArchitecture44/76Within the network side architecture,an innovative data plane hasbeen integrated to augment the traditional control plane,user plane,andorchestrationmanagement.This strategicenhancementwithintheorchestration management elevates data service business and resourceorchestration manage

179、ment capabilities.The new data plane is empoweredto autonomously undertake tasks such as data service request resolution,access control,policy management,and data governance.It alsoorchestrates data bearer paths and oversees in-route data processing,among other management control functions.The data

180、plane collaborateswith the augmented control and user planes to streamline the entire dataservice workflow,Implementing collaborative services between dataservices and connection services.The control plane itself undergoes enhancements to strengthenexistingnetworkfunctionalities,includingaccesscontr

181、ol,policyenforcement,registration and discovery processes,and data management,all aimed at bolstering data plane management control.Meanwhile,theuser plane advances its role,adapting to perform on-demand dataforwarding,storage,and preprocessing tasks as part of the data planeexecution functions.As d

182、ata services deeply integrate with mobilecommunication networks,control plane elements like UDM and UDRcan be reimagined and incorporated into the data planes structure.On the terminal and access network sides,data plane functions(DPs)45/76are deployed as needed.For instance,the data plane managemen

183、t andcontrol function facilitates signaling interactions for registering andupdating data plane capabilities.Concurrently,the data plane executionfunction comes into play,handling operations such as the collection,preprocessing,and forwarding of data originating from both the terminaland base statio

184、n sides.Distributed data management techniques for data services involvedeploying a central data management center in central network nodes forglobal control and collaborative scheduling of data,a regional datamanagementcenterindistributednetworknodesfordistributedprocessing,control,and orchestratio

185、n of data within the nodes,and dataplane functions on demand in all network nodes with the configuration ofcorresponding resources and capabilities.This approach adopts a unifieddata service framework that mines the implicit correlation relationship ofdataandenablescross-domaininterconnectionandeffi

186、cientcollaboration.DHT-based distributed data storage technology is crucial for thestorage and search problem of 6G massive data.It involves storing andsearching data based on its unique key value using Distributed HashTable(DHT)technology,whichavoidsthesingle-point-of-failureproblem.Blockchain-Base

187、dDataDistributedManagementTechnology46/76leveragesblockchainsinherentdecentralization,immutability,andtraceability to provide a secure distributed data management method for6G.However,storing data directly on blockchain lacks storage andretrieval efficiency.Currently,there are solutions for off chai

188、n datastorage and on chain transaction information,the hash value on the chainserves as the verification standard for the actual stored data offline,thusavoiding the harm caused by attacks on offline data.Furthermore,searchable encryption allows users to search encrypted data files usingkeywords,red

189、ucing user communication and computation costs.It is anencrypted search mode that does not leak sensitive information tomalicious servers during the process of searching encrypted data.Usingthis technology in blockchain can reduce search complexity and achievesecure data sharing.Additionally,blockch

190、ain can be used for distributedkey management,greatly alleviating the computational pressure of asingle authority.4.7 Computation-Network Integration6G networks are set to transform the digital landscape byemphasizing the optimization of physical resources at the edge,ondevices,within the cloud,and

191、across the network itself.As thesenetworks evolve,they are converging towards a symbiotic integration ofcomputational and networking capabilities,internally implementing47/76computing endogeneity and providing computing services externally.Thisparadigmshiftisremoldingthetraditionalcommunicationframe

192、work,fostering a profound synergy between networking,computing,and storage.With intelligent scheduling at its core,the 6G architecturewill harness information and communication resources more efficientlythan ever before,dynamically allocating physical resources to supporthigh-level applications.Comb

193、ining endogenous computing prowess withadded functionalities,6G will orchestrate a collaborative effort toconstruct and deliver computing services that meet the evolving needs ofa data-driven era.The logic function framework of computation-networkintegration is shown below:Figure 4-8 Logic Function

194、Framework of Computation-NetworkIntegrationThe computing power service in 6G networks are predicated on the48/76synergistic interaction among various architectural layers,including theinfrastructure,network function,service and capability Exposure,dataplanes,and orchestration management.This synergy

195、 is achieved througha suite of functions that encompass computing service orchestration,computing power rout strategies,computing power route generation,andcomputing power route forwarding.These elements collectively empowerthe networks native computing power,enhancing internal processingwhile simul

196、taneously empowering external industry applications withrobust computing services.To ensure the seamless integration of 6Gs advanced capabilities,such as computation,AI,and data handling,the networks architecturemust be underpinned by key technologies.These include computingpower sensing,computing p

197、ower modeling,computing rout controling,and computing power service orchestration.Harnessing the combinedresources of the computing power and network,these technologies willadapt to diverse application scenarios,ensuring the optimal allocation,delivery,and flow of computational and storage resources

198、 to meet thestringent requirements of future networks.Computing Power Sensing:Computing Power sensing can be divided into two aspects:computing power service sensing and computing power resource sensing.Computing power service sensing:This facet of sensing enables49/76the 6G computing network conver

199、gence to discern and map userdemands to the requisite computing power resources.By detectingthe computing power requirements at the application layer,thenetwork captures the computing power needs of users or services.This insight facilitates network programmability and automaticservice adaptation,al

200、lowing for the on-demand provisioning andflexible scheduling of computing power services.The service andcapability exposure layer provides a foundation for subsequentrouting and policy generation by gathering sensing informationsuch as application type and demand.Application demandinformation includ

201、es both network requirements like bandwidth,latency,jitter,packet loss rate,and computation necessities suchas computation type and volume.Computing power resource sensing:The 6G network leveragescomputing power resource sensing technology to manage andcontrol computing power and network resources e

202、ffectively.Computing power nodes sense and report their status,eitherthrough registration processes or by actively and periodicallyreporting their computing power information.Additionally,theorchestration management layer can proactively interrogate thesenodes for relevant data.Once the data plane h

203、as collected thissensing information,it relays it to the orchestration management50/76layer,which then oversees the global coordination,management,and scheduling of computing power resources.Key sensing dataregarding computing power resources typically include loadconditions,deployment of computatio

204、nal services,and the usageof computational resources.Computing Power Modeling:In the face of various heterogeneous resources distributed in the 6Gnetwork,based on the unified computing power resource model,a unifiedcomputing power capacity system is constructed by constructing arelatively fixed comp

205、uting power capacity template for data plane sensingcomputing power resources in order to support the abstract representationof different types of computing power resources,shielding the differencesof the underlying heterogeneous computing power,and realizing theunified nano-pipeline of resources.In

206、 order to support the 6G network tobuild flexible and scalable computing power service,the computingpower capacity can be divided into the following three directions:Heterogeneous hardware computing power modeling:Toeffectively manage and utilize the various hardware componentswithin the 6G network,

207、it is essential to have a unified set ofcomputing power modeling.These modeling must be applicableto different chips,chip combinations,and other forms of hardware.The goal is to perform a normalized quantification of the51/76performance delivered by heterogeneous chips and hardware.Thisallows for a

208、comparative analysis that can guide the selection andintegration of hardware components based on their computingpower contributions to the network.Diversealgorithmscomputingpowermodeling:Thecomputational tasks performed within the 6G network are likelyto employ a wide array of algorithms,including n

209、eural networks,reinforcement learning,deep learning,and others.Each algorithmhas unique computational requirements,and understanding theserequirements is crucial for effective resource allocation andmanagement.By analyzing the computing power modelingassociated with different algorithms,the network

210、can optimize itsuse of computational resources to achieve the best possibleperformance outcomes for specific computing tasks.user computingpower requirementsmodeling:The6Gnetwork exists to serve user needs,which vary depending onfactors such as network latency,computation volume,type ofcomputing,and

211、 service requirements.By establishing a modelingsystem that maps these user requirements to the necessarycomputing power size,the network can better align its resourceswith user demands.This modeling system provides criticalsupport for computing power orchestration management,ensuring52/76that the n

212、etworks computing power are directed where they areneeded most,enhancing the overall user experience and networkefficiency.Computing Power Routing Control:：The 6G network architectures computing power routing controltechnology is a groundbreaking approach that aims to optimize both thenetwork and co

213、mputing resources in a highly synergistic manner.Thistechnology introduces computing power information into the routingdomain,allowing for more flexible service scheduling and better resourceallocation.The orchestration management plane plays a crucial role inthis process.It generates the computing

214、power topology by sensing userdemand,network,and computing power resource information.Thisinformation is then used to create the computing power routinginformation table,which enables the joint scheduling of network+computing through an analysis of user service demands.Through thenetworkfunctionlaye

215、r,thesystemcanrealizetheroutingofmulti-dimensional resource integration of computing and network.Thisincludes establishing network connections and scheduling services to thebest available service nodes.Computing Power Service Orchestration:：The 6G network architecture is equipped with a sophisticate

216、dcomputing power service orchestration management function that enables53/76the seamless coordination of computing,storage,and network resourcesacross the entire network.This function capitalizes on the deployment ofmulti-dimensional resources to achieve a high degree of resourceoptimization.It inte

217、lligently dispatches computing power services to themost suitable nodes,catering to the specific needs of each service.Furthermore,it oversees the entire lifecycle of these services,frominstantiationtotermination,includingupdates,expansions,andcontractions.The system proactively adjusts the deployme

218、nt strategy fornetwork function virtualization instances,leveraging real-time businessdemand forecasts and the status of network resources.This dynamicapproach ensures that services are scheduled on an as-needed basis,streamliningnetworkoperationandmaintenance.Withunifiedmanagement and global optimi

219、zation at its core,the computing powerservice scheduling function enhances efficiency,elevating the networkscapacity to support a broad spectrum of computing power services.4.8 Native trustworthinessTrustworthiness includes fields such as trust and security.In thefuture,network systems will embed tr

220、ustworthiness into the system andensure the trustworthiness of the entire network through system design,forming a trustworthy infrastructure.This will enable trusted interactionbetween individuals within the network system,including identity54/76recognition,data exchange,rights protection,and value

221、transactions.Thetrustworthy function can be continuously iterated and updated as neededto match the actual needs of 6G networks.Specifically,it includes thefollowing aspects.4.8.1 Multi-trust modeThe trust mechanism used in traditional networks is a centralizedguarantee approach,which provides trust

222、 credentials for interactingentities through a trust center.This approach cannot meet the autonomouscharacteristics of 6G multi network.On the one hand,the distributednetwork form puts networks in an equal position and cannot form anauthoritative entity that all parties trust;On the other hand,a lar

223、geamount of business occurs at the edge,and after using the central trustmechanism,all interactions between subjects need to go through the trustcenter,which requires the trust center to have strong processingcapabilities and the subject to trust center to have strong transmissioncapabilities.The tr

224、ust mechanism of 6G networks will be a multimodaltrust mechanism,including central trust and multi-party trust.Multi partytrust provides trust through multi-party negotiation and joint assurance,and ensures the credibility of interactions based on consensus mechanismalgorithms.Multi party trust also

225、 includes an automatically running smartcontract mechanism,which realizes that multi-party confirmation is the55/76fact and avoids the risk of unilateral breach.The combination of centraltrust and multi-party trust can ensure the trustworthiness of the networkin different scenarios and needs,and ach

226、ieve trust between any node inthe entire network.Asadistributeddatabasetechnology,blockchainhasthecharacteristics of transparency,immutability,traceability,and automaticexecution,and is a natural technology for achieving multi agenttrustworthiness.The key technologies of 6G native blockchain include

227、blockchainledgerstructure,consensusalgorithm,smartcontractmechanism,blockchainnetworkarchitecture,etc.Bycombiningblockchainwithtelecommunicationsnetworks,itempowerstelecommunications networks to achieve stronger capabilities and widerapplication scenarios.4.8.2 Inter network collaborationTelecom net

228、work security has always been achieved through passivepatching to fill past system vulnerabilities,and 6G needs to take proactivemeasures to create a secure interactive environment.By implementingdistributed collaborative design based on the different characteristics ofmultiple entities,a 1+12 effec

229、t can be achieved,forming a secureendogenous system architecture that can ensure system security even indifferent environments,ultimately forming an endogenous security of the56/76network.Meanwhile,the distribution and pooling of networks can helpcombat network fluctuations and enhance the overall r

230、esilience of thenetwork.Collaboration between networks also helps to form networksystem trustworthiness in various new types of interactions.4.8.2 Unified Digital IdentityHolding a valid digital identity is a prerequisite for trustworthyinformation exchange.Digital identity is the identification of

231、the subjectin the network,where the subject refers to all rights and interestsrelated parties such as individuals and enterprise/group users,networkentities and functions,applications,etc.In the future,there will be a hugenumber of users,devices,and services in 6G networks,carrying moretypes,quantit

232、ies,and levels of digital identities.It is necessary to unifythe digital identities of entities within the network.Unification has twolayers of meaning:first,a unified digital identity format that allowsdigital identities to pass through different entities;The second is a unifieddigital identity ide

233、ntifier,which allows the subject to hold a digitalidentity and use it in different interactions.This puts higher demands onthe convenient and efficient management of digital identities,includingthe generation,derivation,allocation,assembly,use,maintenance,updating,and deletion of digital identities.

234、Future communication will present multidimensional characteristics,57/76includinghighlysensitiveinformationsuchaspersonalprivacy,perceptual mapping,AI data learning,etc.,and it is necessary to ensurethat information is not leaked.Distributed digital identity can achieveautonomous and controllable id

235、entity,generate real-time digital identitiescontaining different content according to different needs,and protectother identity information from being exposed.In the dimension ofinformation transmission,it is necessary to encrypt the identity of thetransmission receiver and the transmission content

236、through symmetricand asymmetric encryption technologies.In terms of informationprocessing,privacy calculations are needed to ensure that data processingdoes not leak information.4.9 Generalized QoSThe new paradigm of future QoS design,the 6G generalized QoSsystem,not only provides guarantees for com

237、munication services,butalso expands its multi service capabilities in fields such as sensing,computing,intelligence,and information technology.The 6G QoSarchitecture is not a simple patchwork of services in each domain,butrequires organic control of the entire domain to achieve comprehensiveservice

238、effects of network integration.Based on three new designparadigms:project-based,hierarchical,andmulti-agent,the6Ggeneralized QoS system can achieve integrated integration of QoS control58/76while considering the characteristics of each domain.4.9.1 Project orientedThe domains of communication,percep

239、tion,computing power,AI,and trust are very strong,with significant differences in their respectiveservice characteristics.The 6G QoS architecture should considerpreserving domain characteristics while integrating across domains.Thegeneralized QoS system takes the comprehensive service SLA as theover

240、all goal,covering the complete closed-loop and full process controlfrom demand to delivery settlement.The operating mechanism of thegeneralized QoS system is similar to project-based management,so it canalso be seen as a project-centered or project-based QoS system,and thecomprehensive service SLAca

241、n also be considered as a project level SLA.The generalized QoS system utilizes a project-based managementapproach,which not only focuses on overall network efficiency and value,59/76but also enables various fields to maintain high cohesion and lowcouplingrelationships,suchascommunicationtasksetsint

242、hecommunication domain,perception task sets in the perception domain,andAI task sets in theAI domain.In the generalized QoS system,specificscenario use cases of toC and toB are used as requirement imports,andintent,business,user,environment,andscenarioareusedascomprehensive service awareness engines

243、.Based on the comprehensiveservice SLA,task orchestration is carried out in communication,perception,computing power,AI,data,security,and other fields to form awork chain.The arrangement of the work chain specifically includes:decomposing the comprehensive service SLA into domain level QoS,determini

244、ng specific tasks and task execution relationships/rules for eachdomain based on domain level QoS(such as task execution sequencerules,task conflict resolution rules,task collaboration rules,etc.),managing the entire lifecycle of work tasks,and managing the entirework chain throughout its lifecycle.

245、The generalized QoS system uses a project-based design paradigmfor work chain orchestration,which has the following benefits.Full process and full value chain closed-loop.Project managementoperation can achieve a complete closed-loop and full process control ofproject requirements,delivery,and settl

246、ement.The generalized QoSsystem uses ToC and ToB specific scenario use cases as requirements60/76imports,and uses intent,business,user,environment,and scenario as thegeneralized perception engine to drive the orchestration of the work chainand the orderly execution of various tasks on the work chain

247、.The workchain orchestration is adjusted based on task feedback and project leveleffectiveness/value evaluation.Manage the entire lifecycle from project to task.The generalizedQoS system not only includes full lifecycle management of projects,butalso manages each task on the work chain around projec

248、t requirementsthroughout the entire lifecycle.While ensuring consistency in taskorientation,improve task execution efficiency and make rational use ofsystem resources through full lifecycle management of tasks.In the fulllifecycle management of tasks,all aspects of QoS control are clear andtranspare

249、nt,making it easy to quickly identify issues and bottlenecks withpoor QoS performance,and thus optimize a specific task in a targetedmanner.For tasks that occupy too much system resources,the taskpriority can be lowered or tasks can be terminated early according toproject needs.Support multiple type

250、s of QoS mechanisms.Task orchestrationenables flexible scaling of the work chain,and different scales of workchains can be generated for specific use cases of ToC and ToB.In thework chain orchestration,it is possible to configure and select taskexecution nodes,supporting single point and multi-point

251、 task execution.61/76Not only does it support end-to-end services,but it also supports singleend QoS control services such as single base station environmentperception.The 6G generalized QoS system is managed throughproject-based management,with different work chain arrangementscorresponding to spec

252、ific QoS mechanisms.Each QoS mechanism willnotaffecteach other,facilitatingtheintroductionofnew QoSmechanisms and the iterative evolution of each QoS mechanism.In theinitial stage of 6G,only QoS control for communication,perception,andcomputing power is carried out.After the maturity of intelligentt

253、echnology,it is easy to increase the QoS control of AI by modifying thework chain arrangement method.4.9.2 GradingGrading based on the scope of QoS control,a generalized QoSsystem forms a multi-level QoS architecture for projects,domains,tasks,and components.As the first level,the project conducts o

254、verall QoScontrol to achieve cross domain QoS management;As the second level,manage the task set of feature space in the domain;As the third level,thetask is responsible for the specific execution of QoS work.QoS processesunder different systems and standards can be arranged as specific tasks inthe

255、work chain;As the fourth level,components break down protocolbarriers and select components from the component library to arrange62/76them into tasks based on task requirements.After perceiving projectrequirements,the overall QoS indicators at the project level can beobtained.From top to bottom,proj

256、ect level QoS indicators are graduallydecomposed into domain level QoS indicators,task level QoS indicators,and component level QoS indicators;From bottom to top,tasks areselected based on component level QoS to arrange multiple componentsto achieve task level QoS indicators.Domain level QoS indicat

257、ors areselected based on task level QoS indicators to arrange multiple tasks toachieve domain level QoS indicators.The project ultimately achievesoverall QoS indicators based on domain level QoS.The step-by-stepdecomposition of QoS indicators and the step-by-step satisfaction of QoSimplementation ca

258、n provide more targeted and personalized QoSguarantees for specific scenarios.4.9.3 MultisubjectivityGranting certain QoS autonomy to different network element entitiescan expand the types of QoS control entities.The current 5G networkcontrol entity only has the core network,but if base stations and

259、 terminalshave partial QoS control capabilities,there will be three types of entityperspectives in the network:core network,base stations,and terminals.Considering the security control requirements of actual network operation,appropriate control rights can be granted to network nodes such as base63/

260、76stations and terminals through core network authorization.Taking QoSreporting,modification,and customization rights as examples,it can beseen that multiple control entities have the following advantages in termsof business and air interface adaptation.QoS information reporting rights.The base stat

261、ion has the right toactively report abnormal alarms,idle resources,wireless environment andother information before the start of business,which can enable thenetwork to conduct distributed QoS negotiation and allocate tasksreasonably in advance,avoiding QoS control failures caused by suddenfailures,

262、resource congestion,and harsh environments.In addition,theterminal has the right to proactively report the QoS requirements ofspecial services,which can enable the base station to provide specializedsupport for special services in a timely manner.QoS policy parameter modification rights.When the bas

263、e station hasthe right to modify,there is no need to wait for the core network toinitiate a new QoS configuration.The base station can directly modify theQoS parameter configuration.This short process of QoS controlminimizes the delay of QoS control,ensuring the full utilization oflimited resources

264、by real-time air ports.If the network can perceivebusiness characteristics in a timely manner,active modification of QoSby base stations can also achieve deep integration of network andbusiness.64/76Customization rights for QoS policy parameters.When the basestation has QoS customization rights,QoS

265、personalized information canbe inserted into the user side business data and transmitted along the wayto achieve flexible control of personalized services.In addition,basestations can define diverse granularity of QoS control,achieving preciseQoS control for different services at different protocol

266、layers.4.10 Intelligent EndogenousThe profound fusion of 6G with AI technologies is set to steer theevolution of 6G networks toward enhanced intelligence.This integrationis poised to elevate the productivity and operational efficacy of networkoperators while also offering AI services that span acros

267、s diverse sectors,from industrial applications to consumer-facing solutions.Such a blendempowers industries with advanced intelligent capabilities,acting as acatalyst for the emergence of an intelligent society.To realize thisambition,6G networks must be tailored to fulfill the requisite needs forAI

268、 functionality,performance,privacy,and personalization.Leveragingdistributed AI,task-centric AI capabilities,and innovative technologieslike intent networks and AI-as-a-service,these networks will strengthentheir innate intelligent prowess.This enhancement is crucial to delivering6G solutions that a

269、re not only more adaptable and efficient but also65/76omnipresent,thus shaping the future of smart connectivity.4.10.1 DistributedAIThe 5G intelligent solution based on NWDAF is centralized andplug-in,which causes a certain delay and consumption of datatransmission resources during data processing.I

270、n order to meet thelarge-scale,high-performance,and low-latency AI business requirementsof 6G,the sinking of AI capabilities has become an inevitable trend innetwork evolution.Therefore,the 6G network should support thedeploymentofdistributedAIcapabilities.AIdistributionnodesthroughout the network s

271、tore pre-trained commonly used AI models.Bysinking computing power and models and other intelligent elements,datacan be collected and processed nearby to achieve source analysis data,reducing resource consumption for data migration and integration.Since the model is deployed locally,the collected se

272、nsory datadoesnt need to be uploaded,and inference and decisions can beperformed directly locally,thereby providing high-real-time services.Atthe same time,through the distributed implementation of algorithms,modelsanddata,combinedwithfederatedlearning,multi-taskparallelism and collaborative process

273、ing,complex AI tasks can bedecomposed,processed and aggregated to solve the problems ofcentralized AI in cost,energy consumption,and efficiency,and making66/76efficient use of network intelligent resources.In distributed AI networks,large-scale AI tasks can be dividedamong different AI nodes using s

274、plit learning technology.The centralcontrol node divides training data into multiple subsets and splits themodel into several parts,then sends learning tasks to distributed AI nodesthrough 6G transmission.Distributed AI nodes train submodels locally and cooperate data andmodel parameters through int

275、er-node communication,returning AI modelparameter or gradient updates.Then the central node aggregates a globalmodel,with AI tasks distributed among nodes and knowledge sharedamong nodes.Obviously,distributed AI relies heavily on communication betweennodes.Therefore,6G networks should provide high-p

276、erformance andhighly robust communication.The central AI nodes should have thecapability to coordinate node actions and communications and have largestorage resources to store a variety ofAI models and large datasets.4.10.2 Task-centeredAITraditional communication system design is connection-centere

277、d,mainly providing connections for data transmission and ensuringtransmission quality.But AI business processing is more complex,involving not just connections but also computing,data,algorithms,etc.67/76In order to have endogenous intelligence capabilities,the 6G networkneeds to introduce new resou

278、rce dimensions and design correspondingmanagement and control mechanisms.In addition,in order to adapt to the differences between AI andtraditional connection services,it is also necessary to change from aconnection-centered design idea to a task-centered design idea.Taskrefers to the collaboration

279、and deployment of connection,computing,dataand algorithm resources in multi-node scenarios involved in networkintelligence capabilities to jointly accomplish a specific goal,such as AIreasoning,AI training,computing and other task types.Bydesigningandorchestratingthecorrespondingintelligentmanagemen

280、t and control framework,6G realizes the collaboration ofintelligent elementsandmulti-level and multi-nodecollaboration,completes intelligent business processing at the granularity of tasks,andprovides task-level QoS guarantees to achieve universal benefits andendogenous intelligence requirements for

281、 6G.4.10.3 Intent-based networkIntent-based network is a network management and operationconcept that can provide more intelligent and automated network controland optimization.Compared with traditional networks,6G networks willbe larger and more complex.Intent-based 6G networks can achieve68/76netw

282、ork autonomy and optimization.6G intent-based networks can beimplemented through intent modeling and description,intent perceptionand intelligent analysis,automated configuration and optimization,andintelligent decision-making and adaptation.6G intent-based network canbe implemented in the following

283、 ways:1Intent modeling and descriptionNetwork administrators define and describe user intentions andneeds,translating them into an understandable form.This can involveusing natural language,graphical interface,or other methods to expressuserintent,forexample,providelow-latencyreal-timevideotransmiss

284、ion or optimize energy consumption and network coveragerange.2Intent sense and intelligent analysisThe network has sensing and analysis capabilities and can monitorand collect various data in the network,such as network status,trafficload,user needs,etc.Through intelligent analysis and machine learn

285、ingtechnology,the network can extract intent signals from these data andunderstand the users intent.3Automated configuration and optimizationBased on understood intentions,the network can automaticallyconfigure and optimize network resources,automatically adjust networktopology,routing policies,band

286、width allocation,etc.to meet user69/76intentions and needs,while reducing manual intervention and improvingnetwork efficiency and performance.4Intelligent decision-making and adaptabilityIntent-based networks can make intelligent decisions and adaptivelyadjust network behavior based on changes in th

287、e network environmentand user needs.It can dynamically adapt to different application scenariosand service requirements to provide a better user experience.4.10.4AI as a ServiceAI-as-a-Serviceharnessesthenearreal-timecapabilitiesandubiquitous connectivity of 6G networks to deliver pervasive AI servi

288、ces.These services extend beyond the confines of the network,reachingmobile users and third parties alike.Offering flexibility across a broadspectrum of applications,AI-as-a-Service accelerates the evolution of AIintelligence and its widespread benefits.Historically,enterprises lookingto leverage AI

289、 for enhanced production and management efficiency facedsignificant investments in time and cost for AI infrastructure deploymentand technical teams to construct and train models,establishing a highbarrier toAI service enjoyment.Leveraging the comprehensive resources of 6G networksincludingcore netw

290、orks,radio access networks,terminals,as well as computationaland storage assetsAI-as-a-Service integrates connectivity,computation,70/76algorithms,and storage to provide real-time,omnipresent AI services.Itcoordinates these four elements to offer AI model development,training,reasoning,management,an

291、d more to organizations or users,substantiallylowering the threshold typically associated withAI.By offering AI training services,AI-as-a-Service furnishes userswith computational and storage resources,guiding them through the AIgeneration cycle step by step at the nearest AI service node.This proce

292、ssrequires substantial datasets that align with model needs;users can drawfrom their own data or find suitable datasets within the cloud datamarketplace provided by AI-as-a-Service.Additionally,preprocessingtools are offered to facilitate understanding and manipulation of data,aiding in extraction,t

293、ransformation,and loading processes.For instance,during machine vision model development,users are supplied withlabeled training data and guided through data cleaning,cropping,andnormalization interactively at network AIservicenodes,withoutincurring significant storage and computational resource ove

294、rheads.Furthermore,AI-as-a-Serviceoptimizesuser-builtmodelarchitectures and hyperparameters,fine-tuning model parameters withconfigured optimization algorithms.Once training is complete,results aredeliveredthroughthe6Gnetwork,allowinguserstodownloadpre-trained models or deploy them to proximal netwo

295、rkAI service nodes.In providing AI reasoning services,6G network operators become71/76capableAI-as-a-Service providers,offering a rich repository ofAI modelsand application scenarios.Through 6G networks,pre-trained AI modelsregistered to the network become accessible,enabling organizations orusers t

296、o select required AI services based on need,and experienceintelligent perception,analysis,and decision-making capabilities.Forexample,a product inspection system developer can simply capture aproduct image with a camera and send it via 6G to the nearest AI servicenode,receiving computer vision AI se

297、rvices to detect defects.Withoutspending time on underlying model construction and implementationdetails,developers benefit from AI-as-a-Services demand-matchingmodelsandtimelyinferenceresultsdelivery.Moreover,variousevaluation metrics are provided,allowing users to gauge modelperformance effectivel

298、y.4.10.5 Intelligent Sensing NetworkThe 6G Intelligent Sensing Network is at the forefront of innovation,integrating artificial intelligence,big data analysis,and the Internet ofThings to analyze network Sensing data continuously.This networkfocuses on user AI service demands and experiences,achievi

299、ng aclosed-loop,dynamic supply-demand match that is unparalleled in itsprecision and responsiveness.With its intelligent Sensing capabilities,the 6G network can discern72/76the AI service requirements of various users,allocating computationalresourceson-the-flyforaccurate,on-demandempowerment.Thisau

300、tomated process not only matches AI services to users but alsoproactively pushes these services,allowing users to enjoy AI-enhancedexperiences seamlessly.Central to this system is the decision-making and execution processof intelligent Sensing,which encompassesAI model selection and storagefunctions

301、.The AI Evaluation and Selection Function(AIESF)plays apivotal role,matching pre-trained models to user needs based on factorslike data type and size.It evaluates computational and storage overheads,as well as time consumption,striking a balance between efficiency andaccuracy.This function also take

302、s into account user preferences and AIpolicies to make intelligent model choices.In tandem,the AI ModelStorage Function(AIMSF)maintains a repository of AI models suitableforvariousproblems,alongwithhyperparameters,andsupportsoperations for model addition,deletion,and iterative updating,ensuringthe n

303、etwork stays current with the most effective solutions.Furthermore,the 6G networks intelligent sensing extends toongoing user experience monitoring,regulating channel resources,andtransmissionpowertoguaranteeservicequality.Leveragingreinforcement learning,the network deciphers the relationship betwe

304、enQoS and user experience,analyzing real-time network performance data.73/76By using performance outcomes and user feedback,the networkfine-tunes QoS policies,enhancing service quality through adaptive,differentiated assurances tailored to each users needs.5Summary and OutlookFocusing on the beautif

305、ul vision of full coverage,intelligent sceneconnection,thiswhitepaperproposesthe6Garchitecture.Thearchitecture is built upon a foundation of ubiquitous infrastructure,integrating native capabilities such as intelligence,computing,andsecurity.It encompasses a range of network functions including data

306、,control,forwarding,and orchestration management.This comprehensiveframework aims to deliver the necessary capabilities and services bothwithin and beyond the network,empowering all sectors across society.The trajectory towards 6G is marked by a growing consensus withinthe industry on the developmen

307、t direction of mobile communicationnetwork architectures and key technologies.Promising architecturalcategories such as distributed,intelligent endogenous,endogenous trust,andcomputing-networkconvergencearegainingtraction.Thesecategories,alongwithpivotaltechnologieslikeprogrammability,blockchain,and

308、 service,are set to be synergistically integrated to formthe overarching structure of the 6G networkThere is a strong commitment to delve deeper into the network74/76architectures and key technologies that have begun to find commonground within the industry.The proactive advancement of innovative 6G

309、network architectures is underway,driven by the ambition to build aninclusive information society and support the United Nations sustainabledevelopment goals.The journey towards 6G is not just a technologicalendeavor;its a collaborative mission that calls for partnership andbreakthroughs across the

310、industry.6.References1 6G Technologies,Next GAlliance Report,2022.072 Liu G,Li N,Deng J,et al.The SOLIDS 6G Mobile NetworkArchitecture:DrivingForces,Features,andFunctionalTopologyJ.Engineering,2022,8:42-59.3 M.A.Uusitalo et al.,6G Vision,Value,Use Cases and TechnologiesFrom European 6G Flagship Proj

311、ect Hexa-X,in IEEE Access,vol.9,pp.160004-160020,2021,doi:10.1109/ACCESS.2021.3130030.4 Q.Li et al.,6G Cloud-Native System:Vision,Challenges,Architecture Framework and Enabling Technologies,in IEEE Access,vol.10,pp.96602-96625,2022,doi:10.1109/ACCESS.2022.3205341.5.ITU-R，Framework and overall object

312、ives of the future developmentof IMT for 2030 and beyond.Jun.2023.6.Report ITU-R M.2516，Future technology trends of terrestrial IMT75/76systems towards 2030 and beyond.Nov.2022.7.AbbreviationsAbbreviationFull name in English4G4th Generation Mobile Networks5G5th Generation Mobile Networks6G6th Genera

313、tion Mobile NetworksAFApplication FunctionAIArtificial IntelligenceQoSQuality of ServiceRANRadioAccess NetworkAMFAccess and Mobility ManagementFunctionUPFUser Plane FunctionToBTo BusinessToCTo ConsumerTRPTransmit/Receive PointSMFSession Management FunctionSDNSoftware Defined NetworkSRv6Segment Routi

314、ng IPv6DSMData serving managementDPFData plane function76/76DPMData plane managementUDMUnified Data ManagementDHTDistributed Hash TableNWDAFNetwork Data Analytics FunctionCPSControl Plane ServiceUPSUser Plane ServiceUESUE ServiceAIESFAI Estimate and selection functionAIMSFAI model and storage funciton8.White Paper ContributorsNumberMain contributors1CICT Mobile2China Unicom3ZTE4China Telecom