Joint Offloading Decision, User Association, and Resource Allocation in Hierarchical Aerial Computing: Collaboration of UAVs and HAP

Abstract

In recent years, applications are becoming increasingly computation-intensive and delay-sensitive owing to the rapid growth of Internet of Things (IoT) devices among ground users (GUs). Mobile edge computing (MEC) presents crucial computational support, but conventional MEC services often fail in remote areas and in disaster scenarios. This study presents a hierarchical aerial computing platform leveraging uncrewed aerial vehicles (UAVs) and high-altitude platform (HAP) to meet the computation demands and latency requirements of various IoT applications for GUs. We propose a joint offloading decision, user association, and resource allocation (JOUR) scheme, utilizing binary offloading from GUs to UAVs and partial offloading from UAVs to HAP. The proposed scheme minimizes the energy consumption and latency while maximizing the load balancing. A matching game-based algorithm addresses the GUs offloading decision and GUs-UAVs association, followed by an enhanced soft actor-critic (ESAC) algorithm for UAV partial offloading decision, UAV computation resource allocation, and HAP computation resource allocation. Our simulation results demonstrate the effectiveness of the JOUR scheme in reducing the energy consumption and latency, while improving the load balancing and task completion rates. This demonstrates its potential for optimizing the hierarchical aerial computing platforms in dynamic IoT environments.