Dependency-Aware Flexible Computation Offloading and Task Scheduling for Multi-access Edge Computing Networks

Abstract

With continuous emergence of the new mobile applications, multi-access edge computing (MEC) is generally regarded as a promising technology to enable the computing-intensive and delay-sensitive services at the mobile devices by pushing more computing resources to the network edge. However, computation offloading, which has been a hot topic for MEC networks, is still facing the challenges due to the diversified task characteristics of the new mobile applications and the multidimensional resource conditions of the MEC networks. In this paper, we take the time-dependent logic characteristics of the tasks into consideration and propose a more flexible computation offloading and task scheduling strategy based on the multi-connectivity technology to further minimize the MEC network cost. We model our problem as a multi-objective optimization problem and propose a genetic algorithm-based flexible computation offloading and task scheduling algorithm (GA-FCOTS) to search for the optimal solution iteratively. Simulation results verify the convergence of the proposed algorithm, and show that the proposed algorithm can balance multiple performances and reduce the network cost effectively compared with the other baseline schemes.