Toward Energy Efficiency and Fairness in UAV-Based Task Offloading

The rising demand for compute-intensive mobile applications challenges the limited energy and processing power of user equipment (UE). While Mobile Edge Computing (MEC) enables task offloading to nearby servers, deploying fixed MEC infrastructure is often impractical in settings like disaster zones or temporary high-density events. Furthermore, challenges such as high task delays, limited UE battery life, and unfair load distribution persist. To address these issues, we propose a system where Unmanned Aerial Vehicles (UAVs) serve as mobile relays between UEs and MEC servers. This results in a joint optimization framework combining (i) a Multi-Agent Deep Deterministic Policy Gradient (MADDPG) algorithm for UAV trajectory control to enhance service coverage and energy efficiency, with (ii) a low-complexity task offloading algorithm for UEs. The framework is explicitly designed to minimize UE energy consumption while promoting fairness in task allocation and data rates. Simulations demonstrate that our approach significantly outperforms state-of-the-art benchmarks, reducing UE energy consumption by 25–30% and improving fairness indices by up to 90%. The proposed system proves scalable and robust, making it suitable for real-time deployment in resource-constrained environments with dynamic workloads.