A Distributed Offloading Scheme With Flexible MEC Resource Scheduling
Yanfei Lu, Zhiyuan Zhao, Qinghe Gao · 2021
Mobile edge computing (MEC) has been a promising technique to reduce the total system latency by allowing computations at distributed edge servers instead of centralized cloud servers. With the increasing computation capability of mobile devices, a device is also able to help other devices complete computing tasks. In this paper, we consider a dispersive task offloading scenario, where a device can arbitrarily divide one task into multiple subtasks to the MEC server or other devices. To facilitate the collaboration between multiple devices and the edge server for minimizing the total system latency, we formulate a mixed integer nonlinear programming (MINLP) problem. To solve this problem, we propose a comprehensive task scheduling (CTS) algorithm, where the limited resource of the edge server is allocated on demand of actual sizes of a task rather than on blocks. The CTS algorithm can effectively solve the above subtask offloading problem based on the matching theory. Extensive evaluation results demonstrate that CTS can achieve the lowest latency as compared to the other baseline schemes under different network loads.