A mobility-aware task scheduling by hybrid PSO and GA for mobile edge computing

Yongxuan Sang,Jiangpo Wei,Zhifeng Zhang,Bo Wang,Wei, Jiangpo

DOI: https://doi.org/10.1007/s10586-024-04341-6

2024-03-26

Cluster Computing

Abstract:Mobile edge computing (MEC) is considered one of the key technologies for large-scale network services. Task scheduling helps to improve the task completion rate of MEC, by properly mapping tasks generated by devices onto MEC resources. However, the mobility of devices introduces complexities, potentially resulting in failed task offloading or unavailable task results. To tackle this issue, we propose a mobile-aware task scheduling scheme. We first model the trajectory of mobile devices and introduce a strategy for the fastest task offloading, coupled with an efficient result return method. Subsequently, to improve the task completion rate, we present a task scheduling model based on task migration and formulate the relevant problem as a Mixed Integer Non-linear Programming (MINLP) problem. To achieve a solution within a reasonable time complexity, we propose a Particle Swarm Optimization and Genetic Algorithm with a Rescheduling operator (PSOGAR). In PSOGAR, particles update their positions using a mating operator, while maintaining diversity by a mutation operator. In addition, a rescheduling operator is used to further improve the task completion rate. Finally, through simulation experiments, compare PSOGAR with state-of-the-art and classic algorithms. The experimental results show that PSOGAR can improve the task completion rate by 18–31% and can be applied to scenarios with tight task deadlines.

computer science, information systems, theory & methods