Latency Optimization-based Joint Task Offloading and Scheduling for Multi-user MEC System

• Published in: 2020 29th Wireless and Optical Communications Conference (WOCC) pp. 1 - 6
• Main Authors: Yang, Tiantian, Chai, Rong, Zhang, Liping
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.05.2020
• Subjects: completion latency; computing mode; Minimization; Mobile edge computing; Processor scheduling; Search problems; Servers; Simulation; task offloading; task scheduling; Transforms; Wireless communication
• ISSN: 2379-1276
• DOI: 10.1109/WOCC48579.2020.9114942

Mobile edge computing (MEC) has been recognized as a promising technique which provides mobile devices (MDs) with enhanced computation capability. In this paper, we consider a multi-user, multi-server MEC system which consists of a number of MDs and multiple base stations (BSs) deployed with MEC servers. We assume that computation tasks can be executed locally at the MDs or be offloaded to the MEC servers. Further assume that each MEC server may execute computation tasks for multiple MDs, however, the tasks sharing one MEC server should be scheduled sequentially. We jointly study computation task offloading and scheduling scheme for the MDs and formulate the problem of joint task offloading and scheduling as a task execution latency minimization problem. Since the optimization problem is a mixed integer nonlinear problem which cannot be solved using conventional methods, we transform it into two subproblems, i.e., task partition subproblem and task scheduling subproblem. Under the assumption that task scheduling strategy is given, task partition subproblem is a set of single variable optimization problems, which can be solved easily. To tackle the task scheduling subproblem, we propose a heuristic algorithm, which first determines complete local computing mode for the MDs, then calculates local optimal strategy for the MDs. In the case that multiple MDs may share one MEC server, various priorities are then assigned to the MDs and corresponding computing mode and task scheduling strategy are determined for the MDs with different priorities. Numerical results demonstrate the effectiveness of the proposed scheme.