Joint Optimization of Sensing, Communication, and Computing for Collaborative Multi-UAV Edge Computing System

Unmanned aerial vehicle (UAV) and high-altitude platform (HAP)-enabled aerial edge computing (AEC) networks facilitate diverse Internet of Things (IoT) applications. In this paper, we investigate the average task completion time and energy consumption by jointly optimizing sensing, communication and computing in cooperative AEC networks facilitated by multiple UAVs and HAP. The sensing times, multi-UAV trajectories, transmission power control, offloading strategy and communication resource allocation are jointly optimized. We transform the original optimization problem into minimizing the average task completion time while ensuring energy consumption stability by introducing Lyapunov optimization theory. The problem is then decomposed into multiple subproblems, which are solved through numerical analysis, successive convex approximation, and the Dinkelbach algorithm, respectively. These algorithms are embedded into the proximal policy optimization (PPO)-based multi-agent deep reinforcement learning (MADRL) framework to speed up the convergence performance of the MADRL model. Simulation results demonstrate that the proposed algorithm achieves superior performance in terms of average task completion time and energy consumption.