Tongxin xuebao (Jun 2022)
6G-ADM: knowledge based 6G network management and control architecture
Abstract
Objectives: With the consensus reached on the 6G vision of three-dimensional coverage, extreme performance, virtual real integration and ubiquitous intelligence, problems such as personalized service customization, proliferation of network element types and changeable scene superposition will bring more severe challenges to the network management and control system. For 6G networks, network elements, protocols, applications and architectures will be highly heterogeneous and complex. Native Intelligence and lite networks provide feasible ideas for 6G network architecture and function design. The 6G network with native intelligence, whose network elements have different levels of intelligence, can independently generate strategies to complete the network functions realized by the traditional manual configuration strategies, providing the basic conditions for more efficient network management and control. Starting from the network design idea of "Da Dao Zhi Jian", starting with efficient network management and control, the network architecture can be simplified, complex protocols can be simplified, manual operation and maintenance can be reduced, and the full scene network on-demand services can berealized through the rapid and efficient organization and allocation of multi-level resources in the whole network. At present, 5G network management, control and operation and maintenance systems are closed independently, focusing on solving specific scenario problems. Their security, intelligence and collaboration lack global planning and unified design, and it is difficult to meet the future immersive and personalized full scenario services and performance requirements. It is urgent to build a network management and control system and make breakthroughs in key technologies for 6G on-demand services. Methods: Facing the design requirements of intellicise network, this paper introduces intelligent knowledge space into the 6G control system. The knowledge space is responsible for collecting and extracting the network control experience and knowledge generated by super intelligent network nodes through intelligent computing to form a network control knowledge space. The perception of network needs, the sharing of network resources and the generation of global network control strategies play a role of a super brain. Finally, there is only one layer of knowledge space above the next generation network implementation infrastructure. The 6G management and control architecture as the management and control decision-making layer has native intelligence and minimalist optimization. Based on this, this paper proposes a 6G network management and control system based on knowledge space, which is called 6G-ADM(6G admin) for short. Results: In order to realize the personalized customization of 6G services and improve the performance of 6G services, the integration trend of network management, control and operation was explained. The development of network management and control system will conform to the trend of "intellicise network" and form an "intellicise integration of management and control system". The management and control system is native intelligent. The concept and strategy generation process of intelligent management and control tends to be simplified, but resources will be allocated in a more granular manner. 6G network will support personalized and immersive services in the whole scene and provide users with extreme performance experience such as extremely low delay and high reliability. 6G resources are abundant, but still limited. There is a contradiction between demand growth and resource consumption, which poses an important challenge to the high adaptation of fine-grained resources. This paper proposes 6G-ADM to improve network management and control knowledge, aiming to form a closed loop to support on-demand services, and effectively deal with the contradiction between demand growth and resource consumption in 6G networks. This paper considers that the sustainable on-demand service can be realized through native intelligence of network knowledge, and establishes a knowledge space to coordinate artificial intelligence and traditional artificial definition. Conclusion: As two key technologies to realize the closed-loop control function of 6G-ADM, this paper proposes a new network slicing method and an anomaly detection method based on knowledge space. 6G-ADM converts the closed-loop service policy including resource allocation and anomaly detection into the execution behavior of the global network elements. Simulation results show the effectiveness of the proposed methods.
