Safety-Critical Task Offloading Heuristics for Workflow Applications in Mobile Edge Computing

作     者：Wang, Yushen Sun, Tianwen Yang, Guang Yang, Kai Song, Xuefei Zheng, Changling 

作者机构：China Mobile Grp Jiangsu Co Ltd 59 Huju Rd Nanjing 210029 Jiangsu Peoples R China China Mobile Res Inst 32 Xuanwumen West St Beijing 100053 Peoples R China 

出 版 物：《JOURNAL OF CIRCUITS SYSTEMS AND COMPUTERS》 (电路、系统与计算机杂志)

年 卷 期：2023年第32卷第11期

核心收录：

学科分类：0808[工学-电气工程] 08[工学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

主　　题：Mobile edge computing task offloading safety-criticality workflow applications 

摘      要：As the fundamental mechanism in mobile edge computing (MEC), task offloading strategy is of great significance to the quality of computing services provided by MEC systems. When coping with workflow applications, the precedence relations among tasks increase the difficulty in developing task offloading strategies. This paper studies the problem of safety-critical task offloading for workflow applications in a MEC environment. Considering the precedence constraints on workflow tasks and the overhead of security services, we formulate the safety-critical workflow offloading model with the objective of jointly optimizing the total completion time and energy consumption. By using a task sequence to represent a feasible solution to the optimization model, we introduce a family of heuristics to solve the safety-critical workflow offloading problem under precedence constraints upon workflow tasks. Depending on whether the offloading solution satisfies the precedence relations among workflow tasks, task sequences can be classified into two categories, i.e., precedence-aware and precedence-unaware offloading solutions. With the satisfaction of precedence constraints, a family of heuristics by using a precedence-aware strategy and a precedence-unaware strategy is designed to offload safety-critical workflow tasks. Given an offloading sequence and the operating conditions of MEC servers, the heuristic algorithms select the currently best MEC server to offload workflow tasks. Experimental results justify the performance of the proposed algorithms in solving the safety-critical workflow offloading problem under precedence constraints.