边缘计算系统的任务调度策略

Task scheduling strategy of edge computing system

如何有效的解决云边混合计算中的计算卸载问题已逐渐成为互联网和物联网领域的研究重点。计算卸载问题属于NP-hard优化问题,标准蚁群优化算法作为传统的启发式算法可以用来解决类似计算卸载问题,但是传统蚁群优化算法存在着诸多不足,针对这些不足对传统蚁群优化算法进行了改进,旨在得到时延优化的计算卸载策略。针对单边缘计算节点和云服务器之间的协同计算进行了研究,首先根据整体任务卸载模型进行了数学建模;然后根据各个数学模型利用改进后的蚁群优化算法进行了实验仿真。仿真实验结果表明改进后的蚁群优化算法比传统蚁群优化算法收敛更快,其中改进蚁群优化算法在17次迭代后开始收敛,传统蚁群优化算法在23次迭代后开始收敛;并且使用改进后的蚁群优化算法调度任务与传统蚁群优化算法调度任务、任务只在边缘计算节点执行和任务全部卸载到云服务器上执行进行了对比,结果表明使用改进后的蚁群优化算法进行任务调度大大缩短了任务的完成时间,其中当任务数量为50个时,改进蚁群算法执行任务比上述3种方法分别大约短6 s,15 s,50 s。

How to effectively solve the problem of computing offloading in cloud-side hybrid computing has gradually become the research focus of the Internet and the Internet of Things.The problem of computing offloading is an NP-hard optimization problem and the standard ant colony optimization algorithm as a traditional heuristic algorithm can be used to solve similar calculation offloading problems.But the traditional ant colony optimization algorithm has many shortcomings.In view of these shortcomings,this paper improves the traditional ant colony optimization algorithm to obtain the delay optimization calculation offloading strategy.This paper studies the collaborative computing between single-edge computing nodes and cloud servers.Firstly,the mathematical modeling was performed according to the overall task offloading model;then the experimental simulation was performed using the improved ant colony optimization algorithm according to each mathematical model.Simulation results show that the improved ant colony optimization algorithm converges faster than the traditional ant colony optimization algorithm.The improved ant colony optimization algorithm started to converge after 17 iterations,and the traditional ant colony optimization algorithm started to converge after 23 iterations;and this article uses the improved ant colony optimization algorithm to schedule tasks and compares it with traditional ant colony optimization algorithm to schedule tasks,tasks are executed only at edge computing nodes and tasks are all offloaded to the cloud server.The results show that task scheduling using the improved ant colony optimization algorithm greatly shortens the task completion time;When the number of tasks is 50,the improved ant colony algorithm performs tasks about 6 s,15 s,and 50 s shorter than the above three methods,respectively.