Air–Ground Coordinated MEC: Joint Task, Time Allocation and Trajectory Design

The incorporation of uncrewed aerial vehicles (UAVs) presents a promising shift of paradigm for traditional mobile edge computing (MEC) with ground infrastructure. An air-ground coordinated MEC framework provides extensive signaling coverage and additional computing power for user equipments (UEs), especially in remote areas or emerging scenarios. This paper proposes a novel air-ground coordinated MEC system, comprising a laser-powered UAV and a ground access point (AP). In this setup, the UAV serves both as an MEC server and a relay, while the AP supplies energy to the UAV through laser charging. Given the uncertainties in wireless energy harvesting, our goal is to minimize the UAV's long-term average energy consumption by collaboratively optimizing its trajectory, task allocation between the UAV and AP, and time for energy harvesting. To tackle this non-convex optimization problem, we introduce a two-step alternating optimization algorithm that decouples the optimization variables into two subproblems. Subproblem 1 focuses on computation task and energy harvesting time allocation, formulated as a linear programming (LP) problem. Subproblem 2 addresses real-time trajectory scheduling for the UAV, tackled using the deep deterministic policy gradient (DDPG) algorithm. Simulation results demonstrate the algorithm's convergence and its effectiveness in reducing UAV energy consumption compared to benchmark schemes across various scenarios. Additionally, the optimal policies pertaining to UAV trajectory and task allocation ratios are identified and discussed, highlighting key features and implications.