1、 5G PPP Architecture Working Group View on 5G Architecture Version 3.0, February 2020 Date: 2020-02-14 Version: 3.0 DOI 10.5281/zenodo.3265031 URL http:/doi.org/10.5281/zenodo.3265031 5GPPP Architecture Working Group 5G Architecture White Paper Dissemination level: Public Page 3 / 182 Abstract The 5
2、G Architecture Working Group as part of the 5G PPP Initiative is looking at capturing novel trends and key technological enablers for the realization of the 5G architecture. It also targets at presenting in a harmonized way the architectural concepts developed in various projects and initiatives (no
3、t limited to 5G PPP projects only) so as to provide a consolidated view on the technical directions for the architecture design in the 5G era. The first version of the white paper was released in July 2016, which captured novel trends and key technological enablers for the realization of the 5G arch
4、itecture vision along with harmonized architectural concepts from 5G PPP Phase 1 projects and initiatives. Capitalizing on the architectural vision and framework set by the first version of the white paper, the Version 2.0 of the white paper was released in January 2018 and presented the latest find
5、ings and analyses of 5G PPP Phase I projects along with the concept evaluations. The work has continued with the 5G PPP Phase II and Phase III projects with special focus on understanding the requirements from vertical industries involved in the projects and then driving the required enhancements of
6、 the 5G Architecture able to meet their requirements. The results of the Working Group are now captured in this Version 3.0, which presents the consolidated European view on the architecture design. 5GPPP Architecture Working Group 5G Architecture White Paper Dissemination level: Public Page 4 / 182
7、 Table of Contents 1 Introduction. 25 2 Overall architecture . 26 2.1 Stakeholder roles in the 5G ecosystem . 26 2.2 5G Enhanced Overall System Architecture . 27 2.3 E2E Service Operations Lifecycle Management . 29 2.4 Domain Management Tier 2: edge area with limited computing resources, correspondi
8、ng to, e.g., street cabinets; Tier 3: central area with massive computing resources, corresponding to a datacentre. All tiers provide features for programmability and flexible configuration by generating abstract views on resources of the underlying infrastructure. The solution consists of the utili
9、zation of the SDN paradigm to realize data plane configuration in a way that is agnostic to the underlying hardware infrastructure and fully integrated with management and orchestration plane. The SDN architecture consists in the following layers, cf. Figure 2-7: WAN Resource Manager (SDN Applicatio
10、n) is the functional element that triggers SDN control plane operations. It translates the abstracted view at orchestrator level in a network domain-specific view, ensuring that external link information contained at orchestrator level is translated in a suitable path between NFVI PoPs; Two types of
11、 SDN Controllers, one dedicated to the configuration of the network domain and the second dedicated to the configuration of the RAN domain; each controller is supported by according SDN agents located on the respective network elements; A data-plane consisting of Core NFVI, backhaul network, Edge NF
12、VI, fronthaul network, WLAN Access Points and LTE small cells are the network elements and are considered as part of the infrastructure layer. Figure 2-7: SDN architecture for data plane programmability 2-14 Figure 2-8 depicts a concrete example for realizing the proposed approach for data plane pro
13、grammability in a Cloud Enabled Small Cell (CESC) environment. A two-tier virtualised execution environment in the form of the Edge data centre allows the provision of SDN capabilities. On top, the CESC Manager (CESCM) triggers SDN control plane operations by translating the abstracted view at orche
14、strator level into network domain-specific views. 5GPPP Architecture Working Group 5G Architecture White Paper Dissemination level: Public Page 35 / 182 Figure 2-8: Proposed functional architecture 2-16; DC stands for Data Centre The SDN approach of decoupling control and data plane functions is sui
15、table to make global decisions across several small cells, so called CESC clusters. The Virtualised Infrastructure Manager (VIM) controls the NFV Infrastructure (NFVI), which includes the computing, storage and networking resources of the edge data centres, and creates and controls the CESC clusters
16、. Utilisation of small cells is partitioned into logically isolated slices, offered to different operators or tenants. The CESCM manages and orchestrates the logical cloud environment formed by the so-called “Light Data centre” and the small cell functions. Further, it coordinates and supervises the
17、 use of radio resources and service delivery. It controls the interactions between the infrastructure level and the network operators. For service assurance and fulfilment, CESCM encompasses telemetry and analytics functions for managing the overall network in an efficient and SLA-compliant manner.
18、The CESCM functions will be built upon the services provided by the VIM for appropriately managing, monitoring and optimising the overall operation of the NFVI resources at the edge data centre. The NFV resources will be ultimately offered via a set of APIs that will allow the execution of network s
19、ervices over the distributed CESCs. 2.5.2 Transport network programmability Data plane programmability has been advocated as the perfect solution to manage the heterogeneity of 5G networks as well as to provide fast and easy network function deployment. In the transport network domain, solutions mus
20、t adapt to the highly variable bandwidth requirements of future RANs, offering at the same time high levels of flexibility as well as resource and energy efficiency. The “DisAggregated RAN” 2-21 is a novel concept adopting the notion of “disaggregation” of HW and SW components across the wireless, o
21、ptical and compute/storage domains. Apart from increased flexibility, disaggregation offers enhanced scalability, upgradability and sustainability potential. These features are particularly relevant when a continuously growing number of devices and services, as well as novel features, such as, the c
22、oncept of flexible functional splits, need to be supported. “Resource disaggregation” decouples hardware and software components creating a common “pool of resources” that can be independently selected and allocated on demand. These components form the basic set of building blocks that can be indepe
23、ndently combined to compose any infrastructure service. To exploit the concept of disaggregation in RAN environments, novel solutions must increase the density and power efficiency of the “pool of resources” and provide high bandwidth connectivity between them 2-22. Such solutions will rely on i) ha
24、rdware programmability: allowing HW repurposing to enable dynamic on demand sharing of resources, and ii) network softwarisation: enabling migration from the traditional closed networking model that focuses on network entities to an open reference platform that instantiates a variety of network func
25、tions. 5GPPP Architecture Working Group 5G Architecture White Paper Dissemination level: Public Page 36 / 182 According architectures take advantage of SDN to exploit the offered reconfigurability of high- performing switching hardware; and NFVs full programmability of network functions via software
26、 on commodity hardware platforms 2-23. They adopt the concepts of transport network slicing and resource and service virtualisation across technology domains in order to develop a unified, programmable control and management framework 2-24 that can be used to coordinate the underlying heterogeneous
27、technology domains and support end-to-end service provisioning across various infrastructure domains. 2.5.3 Network function programmability in RAN The RAN architecture takes the baseline architecture, where the baseline architecture covers 5GPPP Phase 1 consensus and the 3GPP status from the public
28、ation time, i.e., the latest 3GPP Release specification on 5G RAN 2-3 2-4, e.g., addition of Service Data Adaptation Protocol (SDAP) layer and F1 interface with CU-DU split. Here, the Controller Layer is envisioned for RAN 2-25, which provides means to introduce RAN control functions as specific app
29、lication implementations. It is worth noting that such flexibility is already available for the CN thanks to the application functions (AFs) as part of the service-based architecture (SBA) 2-2. A high-level illustration of the RAN architecture is given in Figure 2-9. Therein, the Controller layer is
30、 composed by cross-slice (XSC) and intra-slice controllers (ISC) along with the corresponding applications (APPs) running on the northbound interface (NBI). The control commands and interactions with the gNBs take place via the southbound interface (SoBI). Figure 2-9: High-level RAN architecture wit
31、h the controller layer providing RAN programmability It is envisioned that the Controller layer communicates with the RAN NFs via the RAN Controller Agent (RCA), which is introduced in the CU to interface distributed and centralised NFs to the logically centralised controllers. In general, the RCA a
32、cts a middleware between controller and NFs with a local data-store capable to store most recent monitoring information from the NFs. In this regard, RCA can be considered as one of the common platform functions, cf. Figure 2-2. The amount of the data to be exposed to the Controller layer is thus co
33、ntrolled by the RCA. The SoBI is the unified interface between RCA and the controllers for monitoring and re-configuration of NFs. Each programmable NF in DU and CU supports interaction with RCA for exchanging control information with northbound applications deployed on top of the controllers. The R
34、CA is interfacing the so-called RAN data analytics function (RAN-DAF), which is responsible for collecting monitoring information related to both UEs and RAN, such as Channel Quality Indicator (CQI), power level, path loss, radio link quality, radio resource usage, Modulation and Coding Scheme (MCS)
35、, Radio Link Control (RLC) buffer state information, etc. The RCA can 5GPPP Architecture Working Group 5G Architecture White Paper Dissemination level: Public Page 37 / 182 forward the information obtained from RAN-DAF to the controllers and further to northbound applications, such as, slow inter-sl
36、ice RRM, slice-aware RAT selection, elastic resource control, etc. RCA also routes re-configuration information from controller to the respective NFs in the CU and DU. 2.6 Vertical-specific architecture extensions The following subsections elaborate on service-specific extensions of the overall arch
37、itecture shown in Figure 2-2. They demonstrate that the 5G system can be flexibly extended and customized to serve the requirements of vertical industries. This is illustrated using the examples energy utilities, vehicular communications, as well as media content production and delivery. 2.6.1 Exten
38、sions for energy utilities The aim of the proposed extensions to the overall architecture as shown in Section 2.2 is to enable energy utilities in their transition towards more decentralized systems focusing on renewable energy and accelerate their digitalization. Relevant extensions include several
39、 VNFs offering SaaS and IaaS, Self-X functions as well as smart energy (application) VNFs. Since energy grids constitute a core part of critical infrastructures, guaranteed quality of service is crucial and self- optimization processes considering energy grid KPIs must be provided. Referring to Figu
40、re 2-2, the extensions focus on the Radio Access Network, the Management of Domain Resources and Functions and the E2E Service Creation. Extensions in the Radio Access Network realise new methods for IoT device identification and optimization of data routing for small and very small devices. On the
41、Resources and Function Level, this comprises application specific VNFs, deployed at the edge. These VNFs focus on i) the extensive monitoring of the energy grid and networking infrastructure, ii) the digitization of the existing control of the energy grids, iii) the decoupling of the smart grid asse
42、ts from the physical devices by means of employing so-called digital twin technologies, iv) the introduction of blockchain technologies towards storing critical data in an unambiguous, traceable manner, v) the acceleration of infrastructures maintenance- and security-oriented media, and vi) the enab
43、ling of high-accuracy mobility management services allowing for better management of next-generation devices such as drone swarms for automated inspection. Extensions in the Management of Domain Resources and Functions enable the service-aware configuration and orchestration of specific resources an
44、d functions. On the Network Level, such adaptations can be used to create isolated end-to-end network slices on the same infrastructure for simultaneous use by heterogeneous services. Indeed, depending on the operational environment, the energy utility vertical uses all three 5G flavours: eMBB (e.g.
45、 drones for remote infrastructure inspection), mMTC (e.g. connecting 5G-ready advanced smart metering infrastructure deployments), and URLLC (e.g. connecting scalable installations of phasor measurement units). Additionally, to better coordinate different resource categories, analytics-based optimiz
46、ation mechanisms, which are controlled by a utility-based policy, govern the behaviour of network services and also consider application-level metrics, e.g., energy grid-related KPIs. To this end, two new interfaces are introduced linking the Analytics component with the Service Operations component
47、 (Operations-Analytics, Os-An interface) and with the Domain Management component (Analytics-Management, An-Ma interface), respectively. Finally, in the E2E Service Creation, various multi-tenant applications and specific Smart Energy as a Service applications are deployed. Indeed, 5G-enabled energy
48、 grids would enable killer applications such as advanced metering infrastructure as a service, predictive maintenance as a service, as well as dispatchable demand response as a service, which have the potential of revolutionising the operations workflow of energy utilities. 5GPPP Architecture Working Group 5G Architecture White Paper Dissemination level: Public Page 38 / 182 2.6.2 E
sgpjbg002
工作日 8:30 - 17:30
会员动态：
报告分类
新质生产力
ChatGPT
新能源汽车
大模型
Sora
机器人
微短剧
芯片
薪酬
白皮书
低空经济
数据要素
创新药
人工智能
AI产业
信息科技
互联网
消费经济
汽车交通
电商零售
传媒娱乐
医药健康
投资金融
能源环境
地产开发
传统产业
全球化
其它
扫码登录
手机快捷登录
账号登录
