Network Data-Knowledge Engine: Supporting 6G Endogenous Intelligence with Digital Twin

With the rapid development of 5G cellular systems, both industry and academia are focusing on exploring 6G communication, especially supported by artificial intelligence (AI), which has spurred heated discussions on the convergence of AI technologies and future wireless communications. It is envisioned that 6G will rely heavily on AI to enable data-driven and knowledge-driven solutions in large-scale heterogeneous networks, thereby realizing endogenous intelligence in 6G. To achieve this, digital twin (DT) technology can serve as a powerful enabler for unlocking the full potential of future 6G networks. In this paper, we introduce a six-dimensional framework of the digital twin to support endogenous intelligence in 6G through the integration of AI technology and DT. We propose a self-adaptive hierarchical digital twin network (SAHDTN) based on the above DT framework to achieve self-optimization for the 6G network. Furthermore, we incorporate several key technologies to support endogenous intelligence, including a network data-knowledge engine powered by collective reinforcement learning, transfer learning, and knowledge graphs.