Right now, around the world, 5G networks are being deployed. Compared with 4G networks, 5G networks can support more diverse service scenarios and applications. However, 5G networks also bring the challenges of increasing CAPEX and OPEX for mobile operators. In 5G Era, O&M mode innovation required by increased complexity, and accelerated service innovation raising higher requirements for network intelligence. 5G networks have a large amount of available data, including transport layer data (channel, spectrum, and customer link), network layer data (signalling and management data), and various types of application layer data. Based on such data, operators can leverage AI technologies to cope with the challenges of 5G networks.
5G network planning and design should quickly and concisely reflect operators’ intention. Service provisioning requires fewer manual configuration errors and quick service rollout. In terms of network O&M, breakthroughs need to be made to remove the limitations of the traditional expert-experience-based O&M mode. The automatic network operation capability will become the indispensable 4th dimension of the 5G era together with eMBB, mMTC, and URLLC, and become one of the most important driving forces for 5G service innovation and development.
Many operators, infrastructure vendors, and third-party vendors have started to explore intelligent autonomous networks. Cases are including network traffic prediction, automatic base station deployment, automatic fault locating, and on-demand experience optimisation emerges one after another. Typical examples of AI technologies applied in network could be categorised in network planning and construction, maintenance and monitoring, configuration optimisation, service quality assurance, energy-saving, security protection, and operational service. This Beta Labs report collects the above use cases in China, together with results and suggestions.
The path towards a fully intelligent autonomous network will be a long-term process of gradual evolution. A unified understanding of the intelligent autonomous network and its development path across the entire industry will achieve the goal. Based on the concept of “layered autonomy and vertical collaboration with openness “, the industry needs to continuously clarify the connotation and extension of the intelligent autonomous network through various cases. Also, all parties in the industry need to pay attention to the critical problems found in the practice of these cases, such as incomplete standardised interfaces, non-optimized algorithms, and poor model portability. All parties should work together to solve these problems and promote the development of intelligent autonomous networks.