Dynamic Resource Scheduling for Air-Ground Collaborative Communication: A Hierarchical Multi-Agent Method for Metaverse

With the rapid advancement of 5 G communication and the Internet of Things (IoT), unmanned aerial vehicles (UAVs) has shown its potential to play a pivotal role as aerial base stations within the evolving landscape of future wireless networks and edge computing systems. However, in such UAV-assisted communication networks, the complexity of multi-agent scenarios creates challenges due to the problem of learning and decision-making in complex tasks. To address this issue, we propose a novel algorithm that optimizes UAV trajectory planning and autonomous clustering of user equipment (UE) for communication. The proposed algorithm leverages hierarchical reinforcement learning and multi-agent cognitive consistency to improve UE communication decisions in interference-prone environments. Simulation results confirm the convergence of the proposed algorithm and reveal special strength in coordinating spectrum utilization in complex air-ground environments.