Service-Oriented Segmented Trajectory Design for Low-altitude UAV-Assisted MEC Networks

This paper investigates the integration of Unmanned Aerial Vehicles (UAV) with Internet of Things (IoT) infrastructure to enhance Mobile Edge Computing capabilities in urban environments. While UAVs offer promising solutions for mobile edge computing, their deployment in high-rise urban areas presents significant challenges, particularly in computational resource balancing, energy-efficient trajectory planning, and dynamic IoT service provisioning. We propose a comprehensive low-altitude UAV-assisted mobile edge computing framework that jointly optimizes UAV trajectory planning, the assignment of offloaded tasks to specific UAVs, and the strategic deployment and energy management of the UAV fleet to maximize system utility. We first formulate this as a multi-objective optimization problem and prove its NP-hardness due to its non-convex and integer linear programming nature. To tackle this challenge, we develop a decomposition-based approach that systematically addresses the coupled variables. We then propose a novel Variable Strategy Reinforcement Learning-based Lin-Kernighan-Helsgaun algorithm that synergistically combines Q-learning, Sarsa, and Monte Carlo methods with the LKH algorithm. The proposed solution is further enhanced by incorporating two refined trajectory optimization mechanisms, the Trajectory Refining Algorithm and the Service-Oriented Segmented Trajectory Refining Algorithm, specifically designed to improve the robustness and reliability in solving the Computation Offloading Trajectory Optimization Problem. Extensive simulation results demonstrate that our proposed algorithms consistently outperform state-of-the-art approaches, achieving faster convergence, higher energy efficiency for UAVs, and lower computational latency for IoT devices.