AI-Native Network Digital Twin for Intelligent Network Management in 6G (2410.01584v2)

Abstract: As a pivotal virtualization technology, network digital twin is expected to accurately reflect real-time status and abstract features in the on-going sixth generation (6G) networks. In this article, we propose an AI-native network digital twin framework for 6G networks to enable the synergy of AI and network digital twin, thereby facilitating intelligent network management. In the proposed framework, AI models are utilized to establish network digital twin models to facilitate network status prediction, network pattern abstraction, and network management decision-making. Furthermore, potential solutions are proposed for enhance the performance of network digital twin. Finally, a case study is presented, followed by a discussion of open research issues that are essential for AI-native network digital twin in 6G networks.