Delay-aware Distributionally Robust Trajectory UAV-assisted MEC with Uncertain Task Size
Mengdi Zhan, Shubin Xu · 2024
This paper focuses on enhancing the robustness of task offloading within UAV-assisted Mobile Edge Computing (MEC) systems. We consider a UAV traversing predefined flight paths to provide computational support for IoT devices. The most significant challenges in this context arise from the jitter during UAV motion and the uncertainty in task size. Jitter in UAV motion is a practical concern, and to mitigate its impact, we propose Distributionally Robust Trajectory Constraints (DRTC), ensuring stable UAV trajectories while considering uncertain factors. Furthermore, we acknowledge the inherent variability in task sizes that arises from dynamic environmental conditions and specific task requirements. Traditional deterministic methods may increase system overhead and task retransmissions, compromising robustness. To address these issues, we utilize a data-driven approach that effectively captures the uncertainty related to task sizes in UAV. This approach forms the basis of our Distributionally Robust Offloading and Trajectory Optimization (DROTO) algorithm, which is crucial in achieving near-optimal solutions. Thus, it ensures both the efficiency and robustness of the system. Our findings, supported by extensive simulations, provide valuable insights by comparing them with alternative strategies, thus highlighting the algorithm's effectiveness in UAV -assisted MEC systems.