Periodic Collaboration and Real-Time Dispatch Using an Actor–Critic Framework for UAV Movement in Mobile Edge Computing

The increasing need for communication capabilities in mobile devices has led to the recognition of mobile edge computing (MEC) as a critical solution for addressing computationally intensive and latency-sensitive tasks due to its widespread distribution of resources close to devices. However, in scenarios such as disaster response and emergency rescue, the rapid deployment of edge servers to handle tasks may be challenging. Therefore, unmanned aerial vehicle (UAV)-assisted MEC systems have garnered significant interest due to their ease of deployment and high mobility. Nonetheless, the limited computational resources and sensing radius of UAVs give rise to the challenge of optimizing target area coverage and mission data processing timeliness within a restricted time period. In response to this challenge, we present PCRDAC, a novel reinforcement learning-based mobility management framework for UAVs. This framework periodically instructs UAVs to collaboratively update their decision networks, thus determining their movement patterns. This framework can also control UAVs to support worst-case scenarios. Comprehensive simulation experiments validate the efficacy of our framework. Our framework promotes efficient collaboration among UAVs and significantly reduces data staleness in the system. As a result, edge devices can collect ambient data that is fresh enough.