Optimizing Resource Allocation and Task Offloading in Multi-UAV MEC Networks

Abstract

Uncrewed aerial vehicle (UAV)-assisted multi-access edge computing (MEC) is increasingly being adopted to meet the rising demand for low-latency and efficient data processing, particularly in environments with limited ground infrastructure. However, optimizing task offloading and resource allocation in such networks remains a significant challenge as the number of UAVs and connected devices grows. To address this, we propose a Trajectory-Based Task Offloading in UAV-Assisted MEC (TB-TUAV) scheme, which leverages UAV mobility to enhance resource utilization and reduce latency. Unlike existing methods, TB-TUAV integrates a deep reinforcement learning (DRL) framework based on a Markov Decision Process (MDP) to dynamically optimize task offloading and resource allocation in multi-UAV networks. Our approach effectively balances exploration and exploitation, improving learning stability and ensuring fast convergence. By incorporating trajectory optimization through random path exploration, the proposed scheme efficiently distributes computational tasks while mitigating processing delays. Simulation results demonstrate that TB-TUAV significantly improves resource efficiency and reduces latency compared to state-of-the-art baseline methods. This research presents a scalable and adaptive solution for real-time MEC applications in dynamic multi-UAV environments, ensuring improved performance even under resource constraints.