Efficient task offloading and resource allocation in an intelligent UAV-MEC system

Abstract

Nowadays, the functional integration of digital twin (DT) technology and artificial intelligence (AI) methodologies has enabled reliable predictions of many random processes, supporting efficient control and optimization procedures. In line with this trend, this paper explores the joint use of these technologies in an AI-empowered DT framework for an unmanned aerial vehicle-aided multi-access edge computing (UAV-MEC) system. Specifically, this approach defines an intelligent UAV-MEC system capable of significantly improving service quality and deployment flexibility. The focus is on a UAV-MEC network consisting of multiple elementary service areas, where DTs efficiently orchestrate and reduce congestion levels by utilizing UAVs with onboard processing capabilities. A potential architecture for the DTs is outlined, conceptualizing each DT as a collection of basic cyber entities. Additionally, a suitable framework utilizing a matching game approach is proposed to effectively manage task offloading, channel allocation, and the dynamic assignment of UAV support to congested service zones within the same area. Finally, comprehensive simulation results validate the efficacy of the proposed intelligent UAV-MEC system, as indicated by metrics such as task completion delay and accuracy in congestion prediction.