文献类型：期刊文献

英文题名：Cost-efficient security-aware scheduling for dependent tasks with endpoint contention in edge computing

作者：Li, Zengpeng[1];Yu, Huiqun[1];Fan, Guisheng[1];Tang, Qifeng[2];Zhang, Jiayin[1];Chen, Liqiong[3]

机构：[1]East China Univ Sci & Technol, Sch Informat Sci & Engn, Shanghai 200237, Peoples R China;[2]Shanghai Data Exchange Corp, Shanghai 200436, Peoples R China;[3]Shanghai Inst Technol, Dept Comp Sci & Informat Engn, Shanghai 201418, Peoples R China

年份：2023

卷号：211

起止页码：119

外文期刊名：COMPUTER COMMUNICATIONS

收录：;EI(收录号：20233714724858);WOS:【SCI-EXPANDED(收录号:WOS:001078642300001)】；

基金：This work was partially supported by the National Natural Science Foundation of China (No. 61772200) , Natural Science Foundation of Shanghai, China (No. 21ZR1416300) , Shanghai Municipal Special Fund for Promoting High Quality Development (No. 2021-GYHLW-01007) , Capacity Building Project of Local Universities Science and Technology Commission of Shanghai Municipality, China (No. 22010504100) , and Research Funding from National Engineering Laboratory for Big Data Distribution and Exchange Technologies.

语种：英文

外文关键词：Edge computing; Task scheduling; Endpoint contention; Genetic algorithm; Security

摘要：Edge computing empowers latency-sensitive and computation-intensive applications composed of multiple dependent tasks of mobile devices to execute on nearby edge servers. Despite the advantages, intermediate data across edge servers faces external security threats. Moreover, most existing works on task scheduling are designed on the idealized system model where communications are performed in parallel, they neglect the endpoint contention problem during data transmission, and thereby can hardly meet the user-specified deadline. To tackle these issues, this paper studies the endpoint contention-aware scheduling problem for dependent tasks to minimize the execution cost under the security and deadline constraints in the edge computing environment. We propose a heuristic algorithm named CSSDE that first distributes the deadline to each task and sorts tasks according to probabilistic endpoint contention-aware upward rank, and then allocates bandwidth resources and computation resources to intermediate data and tasks, respectively, finally performs rescheduling to improve the initial scheduling scheme. In addition, based on CSSDE and genetic algorithm, we present a metaheuristic algorithm called C-GA that considers the impact of task ordering and incorporates a multi-population co-evolution mechanism to enhance population diversity and avoid falling into local optimum for a better scheduling scheme. Simulation experiments are conducted on synthetic and real-world applications. The results demonstrate that our proposed algorithms achieve a higher success rate, and C-GA reduces the execution cost by 43.92, 40.52, and 8.56 percent compared with SCAS-E, ELSH, and CSSDE, respectively.