多边缘节点场景下的计算任务卸载算法

Computational Task offloading Algorithms for Multi-edge Node Scenarios

针对多边缘卸载环境下存在异地边缘服务器设备空闲从而导致资源浪费以及效率低和能耗大的问题,提出了一种基于改进蛙跳算法的多边缘多设备卸载模型,随机生成终端设备位置并判断其可卸载服务器的列表,将时延与能耗的加权和作为判断卸载决策优劣的目标函数,且为满足计算任务卸载模型的需求,对标准蛙跳算法进行改进,加入自适应权重同时基于异或操作进行青蛙个体位置的更新,并引入遗传算法的变异思想。最后,将提出的卸载算法与4种其他主流卸载算法进行对比,仿真实验结果表明提出的卸载方案得到的卸载决策更优,目标函数值即时延和能耗的优化明显优于其他算法。

In this study,we propose a multi-edge and multi-equipment unloading model based on an improved frog leaping algorithm in order to solve the issues of resource waste,low efficiency,and high energy consumption caused by idle remote edge server equipment in a multi-edge unloading environment.For this,the location of terminal equipment is randomly generated,and the list of its uninstallable servers is examined.The quality of the unloading decision is examined using the objective function,which includes the weighted sum of delay and energy consumption.In addition,the standard frog leaping algorithm is improved to meet the needs of the calculation task unloading model.After the addition of the adaptive weight,the position of the frog individual is updated based on XOR(Exclusive OR)operation.Moreover,the concept of mutation in the genetic algorithm is also introduced in this study.Finally,the performance of the proposed unloading algorithm is compared with that of four other mainstream unloading algorithms.Our simulation results show that the proposed unloading scheme obtains better unloading decisions,better objective function value instant delay and better energy consumption optimization than the other algorithms.