Energy-Efficient Dependency-Aware Task Offloading in Mobile Edge Computing: A Digital Twin Empowered Approach
Huan Zhou, Lingxiao Chen, Kai Yong Jiang, Yuan Wu · 2022
In the 6G era, integration of Digital Twin (DT) and mobile edge computing (MEC) has been expected to significantly improve the service quality of many mobile applications. However, existing studies rarely consider service caching and task dependency together, resulting in a degraded system performance. In addition, the collaboration between Edge Servers (ESs) should also be taken into account because of their limited computing resources as well as caching capacities. In this paper, we investigate a DT-empowered MEC architecture, which intelligently offloads tasks of Mobile Users (MUs) to collaborative ESs with the assistance of DT, while accounting for the service caching and task dependency. Accordingly, we formulate the problem as a Mixed Integer Non-linear Programming (MINLP) problem, aiming to minimize the system-wise energy consumption. In order to solve this problem, the Asynchronous Advantage Actor-Critic (A3C)-based algorithm is proposed. Extensive simulation results demonstrate that our proposed algorithm can reduce the long-term energy consumption of the system greatly, and outperforms the other benchmark algorithms under different scenarios.