Sunday, January 24, 2021

News

Home » News » GTI Release the 5G Intelligent Network White Paper with Huawei and Industrial Partners
  • E-mail
  • Print

GTI Release the 5G Intelligent Network White Paper with Huawei and Industrial Partners

BEIJING - GTI officially released the 5G Intelligent Network White Paper V1.0 at the 29th GTI workshop held on November 23, 2020. At the beginning of 2020, China Mobile initiated the network intelligence project at GTI in collaboration with a number of mainstream industry partners, including Huawei, Ericsson, Nokia, Datang Telecom, and ZTE. The white paper emphasizes the indispensable role of network intelligence in efficient and quality construction, deployment, and operation of 5G networks. It describes how to achieve the ultimate goal of highly intelligent networks — level by level — in terms of standards, use cases, intelligent network level evaluation, network architecture, and function requirements for network management. It also covers typical cases of 5G coverage optimization, 5G traffic optimization, energy saving, alarm root cause analysis, and customer complaint handling. Mainstream intelligent network grading methods are applied to these cases so that the latest standardized achievements can be used to inform live network evaluation. As one of the members of GTI, Huawei actively participates in the network intelligence project and contributes in multiple chapters in the white paper, such as the network intelligent classification and application cases, in the white paper, according to the official website of Huawei. 

To advance the development and evolution of highly intelligent mobile communications networks, GTI calls for a consensus on the concept of intelligent networks and the methods used to evaluate them right across the industry. This requires all parties to specify the development phases and objectives of intelligent networks based on the requirements of various intelligence capability levels, and to jointly incubate intelligent network applications for faster commercial use.

Methods for Classification of Intelligent Network Levels and Evaluation of Use Cases

Several notable industry standardization organizations, such as 3GPP, ITU, CCSA, GSMA, and TM Forum, have already initiated projects related to intelligent autonomous networks. The prevailing belief is that the implementation of fully intelligent networks is a long-term process due to the complex communication networks involved. Based on the organizations' grading standards for intelligent autonomous networks, the GTI white paper proposes a hierarchical evolution path from L0 (manual operating networks) to L5 (fully intelligent networks). In the white paper, typical carrier use cases on the live network are evaluated by closed-loop intent management, perception, analysis, decision-making, and execution, and capability enhancement requirements for further evolution are analyzed. The results confirm that certain use cases within the industry have already achieved L2 or L3.

Hierarchical Evolution of Intelligent Network Architecture for Multi-level Closed-loop Control

To gradually realize fully intelligent networks, architecture-based hierarchical evolution is required for multi-level closed-loop control, without resulting in more complicated networks. The white paper demonstrates the layered architecture of intelligent networks in many typical use cases, including the cross-domain coordination layer, single-domain autonomous layer, and NE layer. The higher cross-domain coordination layer has high requirements on computing capabilities for the collection of large amounts of data, as well as the centralized analysis and processing of such data. As such, this layer is suitable for cross-domain global policy training and reasoning which does not have high requirements for real-time performance. The single-domain autonomous layer is responsible for domain-specific management and control on the network, implementing intelligent closed-loop control of a single domain, while the NE layer is suitable for scenarios involving low requirements for computing capabilities but high requirements for real-time performance, enabling the processing of relatively small amounts of data and support for partial reasoning. Likewise, cross-domain closed-loop and single-domain closed-loop control need to support scenario-specific open interfaces (such as open APIs and SDKs) for full workflows, in order to facilitate information exchange and interconnection between different applications and solutions.

Initial Standards and Industry Consensus Drive Implementation of Intelligent Network Use Cases

Today's global carriers, equipment vendors, and third-party vendors have begun to explore intelligent network applications, encompassing 5G coverage optimization, fault root cause analysis, and intelligent energy saving. Of these, 12 typical application cases are included in the white paper, which are classified by planning, construction, maintenance, and optimization of the entire life cycle of carrier networks, and by the network elements (NEs) and operations support systems (OSSs) functional entities. Currently, the white paper primarily focuses on network maintenance and optimization, and provides the background and requirements of each application case, feasible solutions, and live-network application effect and benefit evaluation, thereby serving as a useful reference for the industry.

Network intelligence is fast becoming an important driving force for 5G business success, as the industry has gradually reached a consensus on standards, architecture, and evolution path, and many innovative applications and collaborations have emerged. The implementation of technical standards and solutions on carriers' live networks requires continuous cooperation, and the joint efforts of all parties within the industry are necessary to achieve the shared goal of ideal intelligent networks. Only through such steadfast collaboration can quality and guaranteed communication services be delivered to network users, further accelerating the informatization of society.