Aerial Reconfigurable Intelligent Surface-Aided MEC Networks with Ultra-Reliable Low-Latency Communications: A Digital Twin Approach

The integration of digital twin (DT) and mobile edge computing (MEC) has emerged as a promising paradigm for enabling the rapid development of edge intelligence in 6G, which has been recognized as the key enabler for ultra-reliable low-latency communications (URLLC). However, the limited resources of MEC nodes and dynamic network environments pose significant challenges to conventional resource allocation schemes. This paper investigates a DT-driven aerial reconfigurable intelligent surface (RIS)-aided MEC network with ultra-reliable low-latency communications, where an unmanned aerial vehicle (UAV) is deployed to carry RIS for assisting computation offloading from user equipments (UEs) to edge servers. To minimize the system latency and energy consumption, we design a deep reinforcement learning framework based on multi-agent deep deterministic policy gradient (MADDPG) that enables edge nodes and UEs to automatically learn optimal strategies by exploiting DT-provided information, which effectively optimizes offloading decisions, computation resource, UEs’ transmission power, RIS phase shifts, and the UAV’s trajectory. Simulation results demonstrate that our approach significantly reduces system latency and energy consumption compared to benchmark methods.