Priority-Aware Task Offloading and UAV Trajectory Optimization for Aerial Access Network - Assisted Mobile Edge Computing

With the surge in computational data, Mobile Edge Computing (MEC) is set to become a crucial technology for reducing communication latency and congestion. However, the widespread adoption of MEC faces several challenges. Aerial Access Networks (AANs), comprising hierarchical High Altitude Platforms (HAPs) and low-altitude Unmanned Aerial Vehicles (UAVs), offer a groundbreaking framework for MEC task offloading, particularly enhancing the service experience of Internet of Things (IoT) devices in disaster zones, battlefield healthcare, or remote areas. In this paper, we propose an MEC task offloading framework supported by AANs to serve IoT devices distributed on the ground. We define system gain based on the energy consumption and latency of tasks with varying priorities. Our objective is to maximize workload fairness among UAVs and overall system gain by jointly optimizing UAVs' flight trajectories, IoT devices' computational task offloading decisions, and service fairness. We propose a multi-agent proximal policy optimization (MAPPO)-based algorithm to solve this joint optimization problem. Experimental results validate the effectiveness of the proposed approach, and numerical analysis evaluates system performance.