6G Network Architecture: QoS Paradigms and Data Lifecycle Management for Next-Generation Networks

In contrast to the limited artificial intelligence (AI)-related research in fifth-generation networks, the usage scenarios of AI and communication in sixth-generation (6G) networks encourage the discovery of greater value from AI, both in the direction of AI for networks (AI4NET) and networks for AI(NET4AI). How can the architecture of the 6G system better enable native AI capabilities, which are increasingly critical challenges? In this article, we present and validate the advantages of representative use cases of AI4NET and NET4AI through prototypes. To meet the requirements of 6G AI service quality assurance, internal data collection, and AI model transfer needed by native AI, we propose a 6G network architecture from a standardization perspective. This architecture provides a new quality of service (QoS) model, enabling new paradigms for multidimensional services. The test results for the 6G data plane protocol stack demonstrate the high efficiency of the proposed 6G network architecture for large amounts of internal data.