DRL-Driven Digital Twin Function Virtualization for Adaptive Service Response in 6G Networks

Yihang Tao, Jun Wu, Xi Lin, Wu Yang

Abstract: Digital twin networks (DTN) simulate and predict 6G network behaviors to support innovative 6G services. However, emerging 6G service requests are rapidly growing with dynamic digital twin resource demands, which brings challenges for digital twin resources management with quality of service (QoS) optimization. We propose a novel software-defined DTN architecture with digital twin function virtualization (DTFV) for adaptive 6G service response. Besides, we propose a proximal policy optimization deep reinforcement learning (PPO-DRL) based DTFV resource orchestration algorithm on realizing massive service response quality optimization. Experimental results show that the proposed solution outperforms heuristic digital twin resource management methods.