5G网络架构设计白皮书-英文版

5G-Advanced网络技术演进白皮书（2021）——面向万物智联新时代从产业发展驱动角度看，键，全球的主要经济体均明确要求将5G作为长期产业发展的重要一环。

从业务上5G将要进入千行百业，从技术上5G需要进一步融合DOICT等技术。

因此本白皮书提出需要对5G 网络的后续演进—5G-Advanced进行持续研究, 并充分考虑架构演进及功能增强。

本白皮书首先分析了5G-Advanced的网络演进架构方向，包括云原生、边缘网络和网络即服务，同时阐述了5G-Advanced的技术发展方向包括智慧、融合与使能三个特征。

其中智慧代表网络智能化，包括充分利用机器学习、数字孪生、认知网络与意图网络等关键技术提升网络的智能运维运营能力，打造内生智能网络；融合包括行业网络融合、家庭网络融合、天地一体化网络融合等，实现5G与行业网协同组网、融合发展；使能则包括对5G交互式通信和确定性通信能力的增强，以及网络切片、定位等现有技术的增强，更好赋能行业数智化转型。

，华为，爱立信（中国），上海诺基亚贝尔，中兴，中国信科，三星，亚信，vivo，联想，IPLOOK，紫光展锐，OPPO，腾讯，小米（排名不分先后）1 产业进展概述 (01)1.1 5G产业发展现状 (01)1.2 5G网络演进驱动力 (01)1.2.1 产业发展驱动力 (01)1.2.2 网络技术驱动力 (02)2 5G-Advanced网络演进架构趋势和技术方向 (04)3 5G-Advanced关键技术 (06)3.1 网络智能化 (06)3.1.1 网络智能化关键技术 (06)3.1.2 智能网络应用场景 (08)3.2 行业网融合 (08)3.3 家庭网络融合 (09)3.4 天地一体化网络融合 (10)3.5 交互式通信能力增强 (11)3.6 确定性通信能力增强 (11)3.7 用户面演进 (12)3.8 网络切片增强 (12)3.9 定位测距与感知增强 (13)3.10 组播广播增强 (13)3.11 策略控制增强 (13)4 总结和展望 (14)5G网络的全球商用部署如火如荼。

IMT-2020(5G)推进组5G网络架构设计白皮书目录引言P15G网络：挑战与机遇P25G网络架构设计P45G网络代表性服务能力P85G网络标准化建议P15总结和展望P17主要贡献单位P18IMT-2020(5G)推进组于2013年2月由中国工业和信息化部、国家发展和改革委员会、科学技术部联合推动成立，组织架构基于原IMT-Advanced推进组，成员包括中国主要的运营商、制造商、高校和研究机构。

推进组是聚合中国产学研用力量、推动中国第五代移动通信技术研究和开展国际交流与合作的主要平台。

IMT-2020(5G)推进组5G网络架构设计白皮书2IMT-2020(5G)推进组5G网络架构设计白皮书随着5G研究的全面展开并逐步深入，业界就环境，为不同用户和垂直行业提供高度可定制化5G场景形成基本共识：面向增强的移动互联网应的网络服务，构建资源全共享、功能易编排、业用场景，5G提供更高体验速率和更大带宽的接入务紧耦合的综合信息化服务使能平台。

能力，支持解析度更高、体验更鲜活的多媒体内5G国际标准化工作现已全面展开，需要尽容；面向物联网设备互联场景，5G提供更高连接快细化5G网络架构设计方案并聚焦关键技术方密度时优化的信令控制能力，支持大规模、低成向，以指导后续产业发展。

本白皮书从逻辑功能本、低能耗IoT设备的高效接入和管理；面向车和平台部署的角度，以四维功能视图的方式呈现联网、应急通信、工业互联网等垂直行业应用场了新型5G网络架构设计，并提炼了网络切片、移景，5G提供低时延和高可靠的信息交互能力，支动边缘计算、按需重构的移动网络、以用户为中持互联实体间高度实时、高度精密和高度安全的心的无线接入网和能力开放等5G网络代表性服务业务协作。

能力。

白皮书最后提出了5G网络架构和技术标准面对5G极致的体验、效率和性能要求，以及化工作的推进建议。

“万物互联”的愿景，网络面临全新的挑战与机遇。

5G网络将遵循网络业务融合和按需服务提供的核心理念，引入更丰富的无线接入网拓扑，提供更灵活的无线控制、业务感知和协议栈定制能力；重构网络控制和转发机制，改变单一管道和固化的服务模式；利用友好开放的信息基础设施1IMT-2020(5G)推进组5G网络架构设计白皮书1.极致性能指标带来全面挑战首先，为了满足移动互联网用户极致的视频在高吞吐量和大连接场景下容易导致流量过载和及增强现实等业务体验需要，5G系统提出了随时信令拥塞。

5G Network Architecture-A High Level View5G Network Architecture 5G Network ArchitectureA High-Level Perspective5GNetwork Architecture-A High Level ViewC ontentsA Cloud-Native 5G Architecture is Keyto Enabling Diversified Service Requirements1.1 The Driving Force Behind Network Architecture Transformation1.2 The Service-Driven 5G ArchitectureIndustries Based on One Physical Infrastructure3.1 Multi-Connectivity Is Key to High Speed and Reliability3.2 MCE4.1 Control and User Plane Separation Simplifies the Core Network4.2 Flexible Network Components Satisfy Various Service Requirements4.3 Unified Database ManagementSelf-Service Agile OperationCloud-Native Architecture is the Foundation of 5G InnovationA Cloud-Native 5G Architecture is Key to Enabling Diversified Service RequirementsThrough persistent effort and determination Telecom operators are implementing a digital transformation to create a better digital world. To provide enterprises and individuals with a real time, on demand, all online, DIY, social (ROADS) experience requires an end-to-end (E2E) coordinated architecture featuring agile, automatic, and intelligent operation during each phase. The comprehensive cloud adaptation of networks, operation systems, and services is a prerequisite for this much anticipated digital transformation.The "All Cloud" strategy is an illuminated exploration into hardware resource pools, ᵒProvides logically independent network slicing on asingle network infrastructure to meet diversified servicerequirements and provides DC-based cloud architecture tosupport various application scenarios.ᵒUses CloudRAN to reconstruct radio access networks (RAN)to provide massive connections of multiple standards andimplement on-demand deployment of RA N functionsrequired by 5G.ᵒSimplifies core network architecture to implement on-demand configuration of network functions throughcontrol and user plane separation, component-basedfunctions, and unified database management.ᵒImplements automatic network slicing service generation,maintenance, and termination for various services toreduce operating expenses through agile network O&M.distributed software architecture, and automatic deployment.Operators transform networks using a network architecturebased on data center (DC) in which all functions and serviceapplications are running on the cloud DC, referred to as a Cloud-Native architecture.In the 5G era, a single network infrastructure can meetdiversified service requirements. A Cloud-Native E2E networkarchitecture has the following attributes:5G Will Enrich the Telecommunication Ecosystem In the new exciting era of 5G, new communication requirements pose challenges on existing networks in terms of technologies and business models. The next-generation mobile network must meet diversified demands. The International Telecommunication Union (ITU) has classified 5G mobile network services into three categories: Enhanced Mobile Broadband (eMBB), Ultra-reliable and Low-latency Communications (uRLLC), and Massive Machine Type Communications (mMTC). eMBB aims to meet the people's demand for an increasingly digital lifestyle, and focuses on services that have high requirements for bandwidth, such as high definition (HD) videos, virtual reality (VR), and augmented reality (AR). uRLLC aims to meet expectations for the demanding digital industry and focuses on latency-sensitive services, such as assisted and automated driving, and remote management. mMTC aims to meet demands for a further developed digital society and focuses on services that include high requirements for connection density, such as smart city and smart agriculture.The expansion of service scope for mobile networks enriches the telecom network ecosystem. A number of traditional industries, such as automotive, healthcare, energy, and municipal systems participate in the construction of this ecosystem. 5G is the beginning of the promotion of digitalization from personal entertainment to society interconnection. Digitalization creates tremendous opportunities for the mobile communication industry but poses strict challenges towards mobile communication technologies.3DWork and play in the cloudFuture IMTArchitecture TransformationThe existing mobile network architecturewas designed to meet requirements for voiceand conventional MBB services. However,this previous organization has proven to beinsufficiently flexible to support diversified5G services due to multiple 3GPP versionupgrades, a large number of NEs, complexinterfaces. The driving force behind thenetwork architecture transformation includesthe following aspects:·Complex networks incorporating multiple services, standards, and site types5G networks must be able to provide diversified services of different KPIs, support co-existent accesses of multiple standards (5G, LTE, and Wi-Fi), and coordinate different site types (macro, micro, and pico base stations). The design challenge to create a network architecture capable of supporting such flexibility whilst meeting differentiated access demands is a brave endeavor to satisfy.·Coordination of multi-connectivity technologies5G is expected to co-exist with LTE and Wi-Fi for an extended period of time incorporating multi-connectivity technologies and the new 5G air interface. Multi-connectivity technologies must be coordinated based on traffic and mobility requirements of user equipment to provide sufficient transmission throughput and mobile continuity.·On-demand deployment of service anchors5G network architecture will be designed based on access sites and three-layer DCs. According to different service requirements, fiber/optic cable availability and network resource allocations, RAN real time and non-real time resources can be deployed on the site or on the access cloud side. This further requires that the service gateway location may also be deployed on the access cloud or on the core network side.·Flexible orchestration of network functionsService requirements vary with different network functions. eMBB requires a large throughput for scheduling. uRLLC requires ultra-low latency and high reliability. Networks must flexibly orchestrate network capabilities considering service characteristics, which significantly simplify network functions and increase network efficiency.·Shorter period of service deploymentVarious services have expanded the mobile network ecosystem and increased network deployment complexity. Rapidly deploying new services requires an improved set of lifecycle management processes involving network design, service deployment, and O&M.The service-driven 5G network architecture aims to flexibly and efficiently meet diversified mobile service requirements. With software-defined networking (SDN) and Network Functions Virtualization (NFV) supporting the underlying physical infrastructure, 5G comprehensively cloudifies access, transport, and core networks. Cloud adoption allows for better support for diversified 5G services, and enables the key technologies of E2E network slicing, on-demand deployment of service anchors, and component-based network functions.CloudRAN consists of sites and mobile cloud engines. This facility coordinates multiple services, operating on different standards, in various site types for RAN real time resources that require a number of computing resources. Multi-connectivity is introduced to allow on-demand network deployment for RAN non-real time resources. Networks implement policy control using dynamic policy, semi-static user, and static network data stored in the unified database on the core network side. Component-based control planes and programmable user planes allow for network function orchestration to ensure that networks can select corresponding control-plane or user-plane functions according to different service requirements. The transport network consists of SDN controllers and underlying forwarding nodes. SDN controllers generate a series of specific data forwarding paths based on network topology and service requirements. The enabling plane abstracts and analyzes network capabilities to implement network optimization or open network capabilities in the form of API. The top layer of the network architecture implements E2E automatic slicing and network resource management.End-to-End Network Slicing for Multiple Industries Based on One Physical InfrastructureE2E network slicing is a foundation to support diversified 5G services and is key to 5G network architecture evolution. Based on NFV and SDN, physical infrastructure of the future network architecture consists of sites and three-layer DCs. Sites support multiple modes (such as 5G, LTE, and Wi-Fi) in the form of macro, micro, and pico base stations to implement the RAN real time function. These functions have high requirements for computing capability and real time performance and require the inclusion of specific dedicated hardware. Three-layer cloud DC consists of computing and storage resources. The bottom layer is the central office DC, which is closest in relative proximity to the base station side. The second layer is the local DC, and the upper layer is the regional DC, with each layer of arranged DCs connected through transport networks.According to diversified service requirements, networks generate corresponding network topologies and a series of network function sets (network slices) for each corresponding service type using NFV on a unified physical infrastructure. Each network slice is derived from a unified physical network infrastructure, which greatly reduces subsequent operators' network construction costs. Network slices feature a logical arrangement and are separated as individual structures, which allows for heavily customizable service functions and independent O&M.As illustrated in the preceding figure, eMBB,uRLLC, and mMTC are independently supportedon a single physical infrastructure. eMBBslicing has high requirements for bandwidth todeploy cache in the mobile cloud engine of alocal DC, which provides high-speed serviceslocated in close proximity to users, reducingbandwidth requirements of backbone networks.uRLLC slicing has strict latency requirements inapplication scenarios of self-driving, assistantdriving, and remote management. RAN real Timeand non-Real Time processing function unitsmust be deployed on the site side providing Reconstructing the RAN with CloudDuring the course of an evolution towards RAN2020, CloudRAN architecture is used on the RAN side to implement RAN Real Time functions, on-demand deployment of non-real time resources, component-based functions, flexible coordination, and RAN slicing. With Mobile Cloud Engine (MCE), CloudRAN can implement flexible orchestration for RAN Real time and non-real Time functions based on different service requirements and transmissionresource configuration to perform cloudification of the RAN.a beneficial location preferably based in close proximity to users. V2X Server and service gateways must be deployed in the mobile cloud engine of the central office DC, with only control-plane functions deployed in the local and regional DCs. mMTC slicing involves a small amount of network data interaction and a low frequency of signaling interaction in most MTC scenarios. This consequently allows the mobile cloud engine to be deployed in the local DC, and other additional functions and application servers can be deployed in the regional DC, which releases central office resources and reduces operating expenses.whilst located in close proximity to services. The RAN non-real time functions include inter-cell handover, cell selection and reselection, user-plane encryption, and multiple connection convergence. These functions require minimal real-time performance, latency requirements to dozens of milliseconds and are suitable for centralized deployment. A universal processor can be deployed in a MCE or site according to vast service requirements.MCE can implement complex management while coordinating multiple processing capabilities based on regional time, frequency bands, and space. This upgraded management system allows CloudRAN to support 4G, 4.5G, 5G, and Wi-Fi, and implement coordination and scheduling of macro, micro, and pico site types. Network functions are deployed on radio, backbone, or core convergence nodes to maximize both network efficiency and additional capabilities.AAUA. Multi-Connectivity Is Key to High Speed and ReliabilityMulti-connectivity is gaining a reputation as an underlying fundamentalconstruct for the deployment of the future network architecture. CloudRAN can be seamlessly deployed in a unified network architecture. This is a huge leap in radio network deployment. In current fragmented networks, increasing speed and reducing latency can improve user experience. Reliable high-speed data cannot depend on a single frequency band or standard connections. In h e t e ro g e n e o u s n e t w o r k s , m u l t i -connectivity helps provide an optimal user experience based on LTE and 5G capabilities, such as high bandwidth and rates of high frequency, network coverage and reliable mobility of low frequency, and accessible Wi-Fi resources. In scenarios that require high bandwidth or continuity, a user requires multiple concurrent connections. For example, data aggregation from multiple subscriptions to 5G, LTE, and Wi-Fi is required to produce high bandwidth. An LTE network access is required to maintain continuity after a user has accessed a 5G high-frequency small cell.In scenarios that source multiple t e c h n o l o g i e s , C l o u d R A N s e r v e s as an anchor for data connection which noticeably reduces alternative t ra n s m i s s i o n. I n t h e t ra d i t i o n a l architecture integrating base stations as an anchor for data connection, LTE, 5G, and Wi-Fi data is aggregated into a non-real time processing module of a specific standard to be forwarded to each access point. In the CloudRA N architecture, non-real time processing function modules in access points of different modes are integrated into the MCE, which serves as an anchor for data connection. Data flows are transmitted to each access point over the MCE, which prevents alternative transmission and reduces transmission investment by 15%, and latency by 10 ms.MCE is the logical entity of central control and management for CloudRA N, incorporating RAN non-real time functions, Wi-Fi AC, distributed gateway, service-related application distribution entity (App), and Cache. RAN non-real time functions include a general control plane (cRRC) to facilitate multi-connectivity and new technology deployment, and a centralized resource management module (cRRM) to ensure the efficient coordination of resources in heterogeneous networks. Cloud-based SON (cSON) is introduced to improve network capacity, coverage, and transmission resources to encompass vast extended areas and ensure the successful implementation of slicing management.MCE can run on a dedicated platform and generalCOTS platform that are deployed above the Cloud OS andthe COTS-based cloud infrastructure. This is to providecarrier-grade disaster recovery capability, on-demanddeployment based on Cloud-Native architecture, flexiblescale-in and scale-out functionality, and independentfeature upgrades.Cloud-Native New Core ArchitectureExisting network gateways integrate parts of both user plane and control plane functions. In the 5G era, many services with high requirements for latency require gateways to be relocated by a downward shift towards the local or central office DCs. This requires that the number of gateway nodes must increase by a factor of 20 to 30 times the original amount. If operators still opt to use the existing gateway architecture, complex gateway service configuration will significantly increase CAPEX and OPEX. In addition, if the control plane has subscribed reports of location and RAT information, a large amount of signaling will be generated between the site, distributed gateway, and network control plane. A large number of distributed gateways will result in heavy interface link load and handover signaling load on centralized control plane NEs.A. Control and User Plane SeparationSimplifies the Core NetworkB. Flexible Network Components Satisfy Various Service RequirementsGateway control and user plane separation divides complex control logic functions for convergence into control planes, which reduces the costs of distributed gateway deployment, interface load, and number of alternative signaling routes. In addition, the control plane and user plane separation supports scaling of the forwarding and control planes, which further improves network architecture flexibility, facilitates centralized control logic functions, and ensures easy network slicing for diversified industry applications. This segregation technique also decouples the forwarding plane from the control plane, which prevents frequent forwarding plane upgrades caused by control plane evolution. Two tasks must be completed to implement control and user plane separation. First, an implementation of lightweight functions to divide complex control logic functions. Second, the construction of models for the reserved core functions with the definition of a generalized template model complete with object-oriented interface for the forwarding plane to ensure that the forwarding plane is both programmable and scalable.After the control and user planes are successfully separated, interfaces providing the associative link connections operate through the enhanced GTP protocol. Based on subscriber access types and subscription data, the control plane initiates an orchestration for service objects and atomic actions, and sends the request to the forwarding plane over the enhanced GTP interface. The forwarding plane then responds with a service-based event notification confirming receipt which is directed back to the control plane.In the 5G era, mobile networks will provide diversified services. eMBB, uRLLC, and mMTC demand different requirements for network control functions. Existing mobile networks cannot customize control functions for a specific service type and can only provide one set of logical control functions for diversified services. Tightly coupled control functions and complex interfaces result in increasingly difficult service deployment and network O&M. Flexible and customizable control function components are a basic core necessity of next-generation mobile networks.In the service-oriented 5G network architecture, logical control functions can be abstracted as independent functional components, which can be flexibly combined according to service requirements. Logically decoupled from other components, network function components support neutral interfaces and implement an identical network interface message to provide services for other network function subscribers. Multiple coupling interfaces are transformedto converge into a single interface. A Network function management framework provides network registration, identification, and management. Independent features ensurethat the addition of network functions and potential upgrades do not affect existing network services.Compared to tightly coupled network controlfunctions, the control plane componentarchitecture significantly simplifies thedevelopment and deployment of newservices through flexible orchestrationand plug-and-play deployment,and lays a solid foundationfor 5G E2E networkslicing.Rapid fault recovery is required fornetwork data status information (such asuser data and policy data shared acrossdata centers), to meet network reliabilityrequirements after the virtualization offunctions. The traditional disaster recoverymechanism based on N+1 backup relies onprivate signaling interaction to implementstatus information synchronization, which produces system inefficiency and complex interaction of cross-vendor products.With separated data and control logic, network status information can be centralized in a unified database. All network functions can access metadata models through standard interfaces and locally store dynamic user data. Thanks to the distributed database synchronization, network status information can implement real-time backup between data centers. With the help of the service management framework, the unified database simplifies the procedure for network information retrieval functions introduced by the component-based control plane to reduce the required signaling overhead for data synchronization.Self-Service Agile OperationOne of the targets and driving forces of network architecture evolution is to provide diversified services using mobile networks. E2E network slicing isa fundamental technology to achieve this target. In the 5G era, a network willcontain multiple logically separated network slices. Each slice has a specific network topology, network function, and resource allocation model. If manual configuration is still used for network planning and deployment, operators' O&M system will potentially face a huge number of significant challenges.5G networks will possess self-serving agile operation capabilities. Network slicing services can be automatically generated, maintained, or terminated according to services requirements, which significantly reduces subsequent operating expenses. Third-party vertical industries can input mobile network slicing requirements on an operation platform. The operator analyzes customer requirements based on current network status.After a service level agreement procedure is complete, the operator mapsvarious service requirements on network requirements, and selects multiple network function components to generate a network slice. According to service features and deployment of data centers, the operator determines logical network function deployment nodes and defines a connection relationship, namely software-defined topology (SDT). After the network slicing topology is defined, an E2E protocol is defined, namely, software-defined protocol (SDP). According to service requirements, network resources are allocated for logical connections in the logical topology, namely software-defined resource allocation (SDRA). SDT, SDP, and SDRA constitute a list of key functions required for Service Oriented Network Auto Creation (SONAC).Conclusion: Cloud-Native Architecture is the Foundation of 5G InnovationIn existing networks, operators havegradually used SDN and NFV to implement ICT network hardware virtualization, but retain a conventional operational model and software architecture. 5G networks require continuous innovation through cloud adoption to customize network functions and enable on-demand network definition and implementation and automatic O&M.Physical networks are constructed based on DCs to pool hardware resources (including part of RAN and core network devices), which maximizes resource utilization. In addition, E2E network slicing provides logically separated virtualized network slices for diversified services, which significantly simplifies network construction for dedicated services.CloudRA N is built based on MCE. Multi-connectivity helps aggregate access capabilities of multiple RATs, frequency bands, and site types to maximize network efficiency. Flexible deployment of network functions helps customize networks for various differentiated services. CloudRAN allows operators to address challenges and proof themselves against potential prospective uncertainties.Based on the control and user plane separation, 5G core networks using component-based control planes, programmable user planes, and unified database will simplify signaling interaction a n d a l l o w f o r t h e d e p l o y m e n t o f distributed gateways. Customized network functions can allow operators to generate increasingly flexible additional network slices to better serve subscribers needs.SONAC implements 5G automation using SDT, SDP, and SDRA to ensure automatic implementation of service deployment, resource scheduling, and fault recovery based strictly on a detailed and thorough network data analysis.5GNetwork Architecture-A High Level View5G Network Architecture-A High Level View2016HUAWEI TECHNOLOGIES CO., LTD.Bantian, Longgang DistrictShenzhen518129, P. R. ChinaTel:+86-755-2878080821。

5G网络架构和关键技术曹诚【摘要】2014年，IMT-2020(5G)推进组发布了第一份白皮书，第五代移动通信系统被提上日程，有望在2020年完成整个网络与系统的部署。

未来的数据流量，网络连接设备总量将会发生爆炸式增长，业务需求将会发生颠覆性变化，物联网与移动互联网将会相互融合，为人们提供多元化服务，如车联网，智能家居，智慧医疗，工业监测系统，超高清3D时频等。

文章介绍实现以上高速率，靠可靠性，低时延，智能化，功能多元化的5G网络实现的网络架构以及关键技术，如大规模MIMO，软件定义网，自组网技术等。

%2014 IMT-2020 (5G) propulsion group released the ifrst white paper, the iffth generation mobile communication system is put on the agenda, is expected to complete the deployment of the entire network and the system in 2020. The data lfow in the future, network connected devices will total explosion growth, business demand will happen to subvert the changes, networking and mobile Internet will fuse with each other, for people to provide diversified services, such as car networking, intelligent home, medical wisdom, industrial monitoring and measurement system, Ultra HD 3D frequency. The achieve the above high rate, by reliability, low latency, intelligent, diversiifed functions of 5g network to achieve the network architecture and key technology, such as large-scale MIMO, software deifned network, ad hoc network technology.【期刊名称】《无线互联科技》【年(卷),期】2015(000)009【总页数】2页(P16-17)【关键词】5G;超密集异构网络;D2D;MIMO;SDN;网络架构;扁平化【作者】曹诚【作者单位】北京邮电大学电子工程学院，北京 100876【正文语种】中文5G网络的构建需要达到超高速率，大吞吐量，超高可靠性，超低延时等指标，来为用户提供最佳的体验。

GTI 5G Network SlicingWhite Paper5G Network Slicing White PaperV 1.0Executive SummaryThe 5G network system can build on-demand network slices to satisfy different service requirements.Network slicing enables the operator to create networks customized to provide optimized solutions fordifferent market scenarios which demand diverse requirements, e.g. in the areas of functionality,performance and isolation. For each scenario, the 5G network can provide appropriate network functionalists, to achieve the goal of building on-demand network and get better performance. Multiple 5G slices canconcurrently operate on the same infrastructure.Table of Contents5G Network Slicing White Paper (1)Executive Summary (3)Table of Contents (4)1.Introduction (5)1.1.Main challenges in the network and problems to be solved (5)work Slicing overview (5)e cases and main requirements for 5G (6)3.Solutions for 5G network slicing (7)work Slicing Architecture (7)work Slice Instance Selection and Association (7)work Slicing resource management (8)work Slicing mobility support (9)3.4.1.Cell selection and reselection (9)work slicing roaming support (9)work slicing handover support (11)work Slicing QoS support (11)3.5.1.QOS for network slicing (11)3.5.2.Support for UE associating with multiple network slices simultaneously (12)4.Abbreviations (13)1.Introduction1.1.Main challenges in the network and problems to be solvedIncreasingly, more and more vertical services would be supported by 5G. There are identified three types of use cases: MBB, Massive MTC and URLLC. These use cases do not exclude any new types of service in future. The performance requirements for these use cases are quite different. For MBB (Mobile Broad band), the scalable control plane, high performance user plane and high speed of mobility is required. For Massive MTC, it needs optimized handling for small data transmission, and the control plan is relatively simplercompared to MBB. In addition, mMTC demands extreme power saving mechanism. For URLLC, it needs highest reliability on control plane and user plane, meanwhile its user plane is required to locate as closely as possible to the edge in order to achieve lowest latency.On the one hand, traditional mobile devices, such as smartphones and tablets, and related services require support for high mobility, high aggregate traffic capacities, and stringent delays. In contrast, MTC requires support for typically stationary devices with low per-device throughput, relatively loose delays and highgeographic device densities.Figure 1-1 main challengeswork Slicing overviewNetwork slice is composed of a set of network functions (e.g., potentially from different vendors) , theresources to run these network functions as well as policies and configurations and specific RAT settings that are combined together for the specific use case or business model.End-to-end network slicing, including Access, Core and transport network, with physical resources dedicated to either this slice or being shared with other slices. The slices shall be isolated against each other, butsharing a common infrastructure layer for efficiency reasons.The network slicing concept consists of 3 layers: 1) Service Instance Layer, 2) Network Slice Instance Layer, and 3) Resource layer.The Service Instance Layer represents the services (end-user service or business services) which are to be supported. Each service is represented by a Service Instance. Typically services can be provided by thenetwork operator or by 3rd parties. In line with this, a Service Instance can either represent an operatorservice or a 3rd party provided service.A network operator uses a Network Slice Blueprint to create a Network Slice Instance. A Network SliceInstance provides the network characteristics which are required by a Service Instance. A Network Slice Instance may also be shared across multiple Service Instances provided by the network operator.Figure 1-2 network slicing overviewe cases and main requirements for 5GThe services foreseen in the 5G era fall into three typical scenarios: enhanced Mobile Broadband (eMBB), Ultra-Reliable and Low Latency Communications (URLLC), and massive Machine Type Communications (mMTC). eMBB focuses on services characterized by high data rates, such as high definition (HD) videos, virtual reality (VR), augmented reality (AR), and fixed mobile convergence (FMC). URLLC focuses on latency-sensitive services, such as self-driving, remote surgery, or drone control. mMTC focuses on services that have high requirements for connection density, such as those typical for smart city and smart agriculture use cases. Each scenario requires a completely different network service and poses requirements that are radically different, sometimes even contradictory.For example, 5G networks will connect the factory of the future and help create a fully automated andflexible production system. In the healthcare industry, hospitals will be able to arrange remote roboticsurgeries as if the surgeon were physically present next to the patient. At the same time, 5G connectedhealthcare chips will constantly monitor vital signs, prevent conditions from becoming acute, and adaptmedication to meet changing conditions. While on the roads, self-driving cars and smart infrastructuresenabled by 5G networks will reduce accidents and save millions of lives every year.The 5G network is designed to support very diverse and extreme requirements for latency, throughput,capacity and availability. Network slicing offers a solution to meet the requirements of all use cases in a common network infrastructure. The same network infrastructure can support, for example, smartphones, tablets, virtual reality connections, personal health devices, critical remote control or automotiveconnectivity. With network slicing, different end-to-end logical networks with isolated properties areprovided and operated independently. These enable operators to support different use cases, with devices able to connect to multiple slices simultaneously.Figure 2-1 network slices for s a variety of use cases3.Solutions for 5G network slicingwork Slicing ArchitectureEnabling network slicing in 5G requires native support from the overall system architecture. Which contains access slices (both radio access and fixed access), core network (CN) slices and the selection function that connects these slices into a complete network slice comprised of both the access network and the CN. The selection function routes communications to an appropriate CN slice that is tailored to provide specificservices. The criteria of defining the access slices and CN slices include the need to meet differentservice/applications requirements and to meet different communication requirements. Each CN slice is built from a set of network functions (NFs). An important factor in slicing is that some NFs can be used across multiple slices, while other NFs are tailored to a specific slice.Figure 3-1 5G Network Architecture for Network Slicingwork Slice Instance Selection and AssociationSlice selection refers to the mechanisms or set of mechanisms to identify the NSIs (Network Slice Instance) of a UE. The subscription of the UE to NSIs can be determined via a slice ID (S-NSSAI) in the subscription database. The S-NSSAI is used to identify a slice and, thus, to assist the 5G network in selecting a particular NSI.When a UE registers with a PLMN, the UE shall provide the network in RRC and NAS layer a Requested NSSAI containing the S-NSSAI(s) corresponding to the slice(s) to which the UE wishes to register or requires, in addition to the 5G-S-TMSI if one was assigned to the UE.The RAN shall route the NAS signalling between this UE and an AMF selected using the Requested NSSAI obtained during RRC Connection Establishment. If the RAN is unable to select an AMF based on the Requested NSSAI, it routes the NAS signalling to an AMF from a set of default AMFs.The AMF verifies whether the S-NSSAI(s) in the Requested NSSAI are permitted based on the Subscribed S-NSSAIs and check whether it can serve the UE. If the AMF can’t serve the UE or if the UE context in the AMF does not yet include an Allowed NSSAI, the AMF queries the NSSF, with Requested NSSAI, the Subscribed S-NSSAIs, PLMN ID of the SUPI, location information, and possibly access technology being used by the UE. The NSSF selects the NSIs to serve the UE and determines the NRF(s) to be used to select NFs/services within the selected NSI(s).When establishing a PDU session associated to an S-NSSAI and a DNN, a UE that is registered in a PLMN and has obtained an Allowed NSSAI, shall indicate in the PDU Session Establishment procedure theS-NSSAI and, if available, the DNN the PDU Session is related to.SMF discovery and selection within the selected NSI is initiated by the AMF when a SM message to establish a PDU Session is received from the UE. The NRF is used to assist the discovery and selection tasks of the required network functions for the selected NSI.The AMF queries the NRF to select an SMF in a NSI based on S-NSSAI, DNN and other information e.g. UE subscription and local operator policies, when the UE triggers the establishment of a PDU Session. The selected SMF establishes a PDU Session based on S-NSSAI and DNN.NSSAIFigure 3-2 Network Slice selectionwork Slicing resource managementThe overall architecture supports the isolation of NSIs, including resource isolation, O&M isolation, and security isolation. NSIs can be either physically or logically isolated at different levels.Resource isolation enables specialized customization and avoids one slice affecting another slice. E.g. RAN needs to provide and enforce differentiation, and maintain isolation between slices where resources are constrained including RF resource, backhaul transport resource and computing resource.Hardware/software resource isolation is up to implementation. Each slice may be assigned with either shared or dedicated radio resource up to RRM implementation and SLA (Service Level Agreement). The logical isolation of resources may require resource multiplexing. The amount of allocated resources can be scaled up or down for higher utilization efficiency depending on the traffic load of each NSI. Co-existence of sharedand dedicated functions allows time- and frequency-domain resource isolation without sacrificing resource efficiency.To enable differentiated handling of traffic for network slices with different SLA:- NG-RAN is configured with a set of different configurations for different network slices;- To select the appropriate configuration for the traffic for each network slice, NG-RAN receives relevant information indicating which of the configurations applies for this specific network slice.To guarantee security for the network services provided by an NSI, it requires embedding the security mechanism and security provisioning entity (e.g. security anchors and security functions) into the logical network architecture of the NSI.A terminal should be authenticated and authorized to access a specific NSI. The communication between the terminal and the allocated NSIs should be protected against attacks. In addition, terminals may require different levels of security protection associated with different NSIs.work Slicing mobility support3.4.1.Cell selection and reselectionLike in LTE, UE behavior in IDLE state includes cell selection and re-selection. As in LTE, UE can prioritize a frequency based on service, Cell broadcasts (e.g. in minimum SI) the services supported by it, and network slices can use these technologies.Figure 3-3 cell selection and reselection for network slicingTherefore the issue is whether Slice based prioritization is also used for NR and what is the impact of this function on NR idle behavior.It was agreed that for intra-freq cell reselection the UE try to always camp on the best cell. Additional functionality for RACH resource isolation/differentiated treatment will not be supported for slicing forRel-15.work slicing roaming supportConsidering the support of network slicing in roaming scenarios, there are two scenarios:Home routed roaming case: The UE’s traffics are transferred via the UP functions in both VPLMN and HPLMN. The NFs in the VPLMN cooperate with the NFs in the HPLMN to provide the end-to-end services for the roaming UE.Figure 3-4 network slicing roaming architecture-home routed scenarioLocal breakout roaming case: The UE’s traffics are transferred via UP functions within VPLMN, and the HPLMN may provide policy control function for the roaming UE.Figure 3-5 network slicing roaming architecture-local breakout scenarioIf the VPLMN and HPLMN have an SLA to support non-standard S-NSSAI values in the VPLMN, the NSSF of the VPLMN maps the Subscribed S-NSSAIs values to the respective S-NSSAI values to be used in the VPLMN. The S-NSSAI values to be used in the VPLMN are determined by the NSSF of the VPLMN based on the SLA.Depending on operator's policy and the configuration in the AMF, the AMF may be allowed to decide theS-NSSAI values to be used in the VPLMN and the mapping to the Subscribed S-NSSAIs. The NSSF in the VPLMN determines the Allowed NSSAI without interacting with the HPLMN. When the UE constructs Requested NSSAI, it may also provide the mapping of each S-NSSAI of the Requested NSSAI to theS-NSSAIs of the Configured NSSAI for the HPLMN. Upon successful completion of a UE's Registration procedure, the UE obtains an Allowed NSSAI, which includes one or more S-NSSAIs, from the AMF, possibly associated with mapping of Allowed NSSAI to Configured NSSAI for the HPLMN.In a PDU Session Establishment procedures, the UE includes a Subscribed S-NSSAI based on the NSSP (HPLMN S-NSSAI), and the related S-NSSAI from the Allowed NSSAI (VPLMN S-NSSAI). For home routed case, the V-SMF send the PDU Session Establishment Request message to the H-SMF along with the HPLMN S-NSSAI. When a PDU Session is established, the CN provides to the AN the VPLMN S-NSSAI corresponding to this PDU Session.work slicing handover supportSome slices may be available only in part of the network. The RAN and the CN are responsible to handle a service request for a slice that may or may not be available in a given area. Admission or rejection of access to a slice may depend by factors such as support for the slice, availability of resources, support of the requested service by other slices.To make mobility slice-aware in case of Network Slicing, S-NSSAI is introduced as part of the PDU session information that is transferred during mobility signalling. This enables slice-aware admission and congestion control.Slice-aware supported in the cells of its neighbouring gNBs may be beneficial for mobility in connected mode. It is assumed that the slice configuration does not change within the UE’s registration area. Solutions for how slice availability may be handled during mobility include: Neighbours may exchange slice availability on the interface connecting two nodes, e.g. Xn interface between gNBs, and the core network could provide the RAN a mobility restriction list. This list may include those TAs which support or do not support the slices for the UE.work Slicing QoS support3.5.1.QOS for network slicingIn LTE, the bearer in CN and RB in RAN are of one to one mapping. One PDN connection can contain multiple EPS bearers, and one EPS bearer can contain multiple IP flows. In the RAN side, the EPS bearer is mapped to the radio bearer with one-to-one mapping relationship.The 5G QoS framework can provide the wide range of existing and future emerging use cases/services. The 5G QoS model is based on QoS Flows. NG-RAN and 5GC ensure quality of service (e.g. reliability and target delay) by mapping packets to appropriate QoS Flows and DRBs. The QoS Flow is the finest granularity of QoS differentiation in the PDU Session.Figure 3-6 QoS architectureThe PDU sessions from different CN slices are just different service flows with different QoS requirements and implicitly mapping the CN slice to RAN via QoS characteristic and session characteristic (e.g. which slice’s UP entity the session connects to). That is beneficial to realize some specific slices, such asV2X, which have diverse traffic type (entertainment, traffic safety) with different QoS requirements.Figure 3-7 QOS mapping for network slicingIn case of network slicing, S-NSSAI information is added per PDU session, so NG-RAN is enabled to apply policies at PDU session level according to the SLA represented by the network slice, while still being able to apply (for example) differentiated QoS within the slice. And network slicing related parameters are included in Initial Context Setup signaling and establishment/modification/release of a PDU session signaling.3.5.2.Support for UE associating with multiple network slices simultaneouslyA single UE can simultaneously be served by one or more Network Slice instances via a 5G-AN. A single UE may be served by at most eight Network Slices at a time. The AMF instance serving the UE logically belongs to each of the Network Slice instances serving the UE, i.e. this AMF instance is common to theNetwork Slice instances serving a UE. If network slices are isolated with each other, the UE should notinclude the isolated S-NSSAIs in the Request NSSAI and will not access such network slices simultaneously.4.Abbreviations。

5 G 用例和要求（NOKIA）目录1. 5 G 是什么，以及为什么它会来------------------------------------------------------32. 用例--------------------------------------------------------------------------------6 2.1 移动宽带------------------------------------------------------------------------- 6 2.2 汽车-------------------------------------------------------------------------------8 2.3 智能社会--------------------------------------------------------------------------9 2.4 智能电网 -------------------------------------------------------------------------9 2.5 健康----------------------------------------------------------------------------- 10 2.6 工业----------------------------------------------------------------------------- 102.7 物流 / 货运跟踪----------------------------------------------------------------- 103.0 设计原则--------------------------------------------------------------------- 10 3.1 灵活性--------------------------------------------------------------------------- 103.2 可靠性--------------------------------------------------------------------------- 114. 要求 --------------------------------------------------------------------------------125. 摘要-------------------------------------------------------------------------------- 151.5 G 是什么，以及为什么它会来为更好的移动从订阅服务器需求的持续增长宽带的经验鼓励业界向前看如何准备好网络，以满足未来的极端容量和性能要求。

目录IMT-2020(5G)推进组于2013年2月由中国工业和信息化部、国家发展和改革委员会、科学技术部联合推动成立，组织架构基于原IMT-Advanced推进组，成员包括中国主要的运营商、制造商、高校和研究机构。

推进组是聚合中国产学研用力量、推动中国第五代移动通信技术研究和开展国际交流与合作的主要平台。

引言5G承载网络总体架构5G承载转发面架构与技术方案5G承载协同管控架构和关键技术5G同步网架构和关键技术我国5G承载产业发展趋势分析总结和展望主要贡献单位P1P2P4P21P25P29P34P35I M T -2020(5G )推进组5G承载网络架构和技术方案白皮书I M T-2020(5G)推进组5G承载网络架构和技术方案白皮书2I M T-2020(5G)推进组5G承载网络架构和技术方案白皮书引言随着3GPP 5G非独立（NSA）和独立（SA）组网标准的正式冻结，我国运营商同步启动规划和设计5G试点和预商用方案，5G迈向商用的步伐逐步加快。

相对4G网络，5G在业务特性、接入网、核心网等多个方面将发生显著变化，其中在业务特性方面，增强型移动宽带（eMBB）、超可靠低时延通信（uRLLC）、大规模机器类通信（mMTC）等典型业务场景将分阶段逐步引入；在无线接入网方面，将重塑网元功能、互联接口及组网结构；在核心网方面将趋向采用云化分布式部署架构，核心网信令网元将主要在省干和大区中心机房部署，数据面网元根据不同业务性能差异拟采用分层部署方案，随着物联网（IOT）等垂直行业的业务发展，5G控制平面也将呈现大区部署趋势。

5G新型特性变化为承载技术的新一轮快速发展提供了契机。

根据IMT-2020(5G)推进组5G承载工作组2018年6月发布的《5G承载需求分析》白皮书， 5G对承载网络主要带来三大性能需求和六类组网功能需求，也即在关键性能方面，“更大带宽、超低时延和高精度同步”等性能指标需求非常突出，在组网及功能方面，呈现出“多层级承载网络、灵活化连接调度、层次化网络切片、智能化协同管控、4G/5G混合承载以及低成本高速组网”等六大组网需求，如何满足和实现这些承载需求至关重要。

引言5G网络：挑战与机遇5G网络架构设计5G网络代表性服务能力5G网络标准化建议总结和展望主要贡献单位P1 P2 P4 P8 P15 P17 P18目录IMT-2020(5G)推进组于2013年2月由中国工业和信息化部、国家发展和改革委员会、科学技术部联合推动成立，组织架构基于原IMT-Advanced推进组，成员包括中国主要的运营商、制造商、高校和研究机构。

推进组是聚合中国产学研用力量、推动中国第五代移动通信技术研究和开展国际交流与合作的主要平台。

1随着5G研究的全面展开并逐步深入，业界就5G场景形成基本共识：面向增强的移动互联网应用场景，5G提供更高体验速率和更大带宽的接入能力，支持解析度更高、体验更鲜活的多媒体内容；面向物联网设备互联场景，5G提供更高连接密度时优化的信令控制能力，支持大规模、低成本、低能耗IoT设备的高效接入和管理；面向车联网、应急通信、工业互联网等垂直行业应用场景，5G提供低时延和高可靠的信息交互能力，支持互联实体间高度实时、高度精密和高度安全的业务协作。

面对5G极致的体验、效率和性能要求，以及“万物互联”的愿景，网络面临全新的挑战与机遇。

5G网络将遵循网络业务融合和按需服务提供的核心理念，引入更丰富的无线接入网拓扑，提供更灵活的无线控制、业务感知和协议栈定制能力；重构网络控制和转发机制，改变单一管道和固化的服务模式；利用友好开放的信息基础设施引言环境，为不同用户和垂直行业提供高度可定制化的网络服务，构建资源全共享、功能易编排、业务紧耦合的综合信息化服务使能平台。

5G国际标准化工作现已全面展开，需要尽快细化5G网络架构设计方案并聚焦关键技术方向，以指导后续产业发展。

本白皮书从逻辑功能和平台部署的角度，以四维功能视图的方式呈现了新型5G网络架构设计，并提炼了网络切片、移动边缘计算、按需重构的移动网络、以用户为中心的无线接入网和能力开放等5G网络代表性服务能力。

白皮书最后提出了5G网络架构和技术标准化工作的推进建议。

5G 网络架构顶层设计理念C ontents5G 原生云化架构 - 满足多样化商业需求的关键5G将强化电信生态系统1.1 网络架构演进的需求1.2 商业需求驱动5G架构变革端到端网络切片 - 统一物理设施支撑多种垂直行业利用云重构无线接入网络3.1 多连接技术是使能高速传输与高可靠性的关键3.2 移动云引擎原生云化新核心网架构4.1 用户面/控制面分离简化核心网络4.2 灵活的组件化网络使能多业务需求4.3 统一数据库管理自服务敏捷运维结论：原生云化架构是5G创新的基础040508091316185G 原生云化架构-满足多样化商业需求的关键为了更好的面向数字化的世界，服务数字化的社会，全球范围内的运营商都在进行数字化转型。

运营商数字化转型的目标在于为其企业客户、消费者提供ROADS (Real-time, On-Demand, All online, DIY, Social) 体验，这需要通过端到端协同整体架构才能够实现，需要在各个环节都实现敏捷，自动化和智能化。

运营商的网络、运营系统、业务的全面云化是必要条件和实现手段。

ᵒ在同一套物理基础设上基于不同的业务需求生成逻辑隔离的独立运行的网络切片，通过基于数据中心的云化架构支撑多种应用场景。

ᵒ利用CloudRAN对无线接入网络进行重构，满足5G 时代多技术连接以及RAN功能按需部署的需求。

ᵒ通过控制面和用户面（CP/UP）分离，功能模块化以及统一的数据库管理技术简化核心网络架构，实现网络功能的按需配置。

ᵒ基于应用驱动来自动的生成，维护，终止网络切片服务，利用敏捷的网络运维降低运营商的运营成本。

“全面云化”将带来硬件资源池化、软件架构分布化、部署自动化的系统优势。

在全面云化的战略下，运营商网络将转型为“以数据中心为中心”的架构，所有的网络功能和业务应用都运行在云数据中心上，即原生云化（Cloud-Native）的架构。

5G时代将以一张物理的基础网络支撑多种不同的商业需求，云化的端到端网络架构通过以下几个方面实现上述需求：5G 将强化电信生态系统5G 时代新的通信需求对现有网络提出了包括技术上的，商业模式上的种种挑战，需要下一代移动网络来满足。