改进遗传算法的边缘计算卸载策略  

Offloading strategy for edge computing based on improved genetic algorithm

作　　者：葛海波 李文浩 冯安琪 王妍 GE Haibo;LI Wenhao;FENG Anqi;WANG Yan(School of Electronic Engineering,Xi'an University of Posts and Telecommunications,Xi'an 710121,China)

机构地区：[1]西安邮电大学电子工程学院,陕西西安710121

出　　处：《西安邮电大学学报》2020年第3期7-13,共7页Journal of Xi’an University of Posts and Telecommunications

基　　金：陕西省自然科学基金项目(2011JM8038);陕西省重点产业创新链(群)项目(S2019-YF-ZDCXL-ZDLGY-0098);西安邮电大学研究生创新基金项目(CXJJLZ2019029)。

摘　　要：针对经典遗传算法处理移动边缘计算卸载存在系统开销较大的问题,提出一种基于改进遗传算法的边缘计算卸载策略。建立了包含时延、能耗的多目标优化模型。将每一个卸载策略作为一条染色体,每条染色体上的基因对应一个本地、边缘和云端计算任务。将系统整体开销的倒数作为适应度函数,选择适应度函数值较小的染色体进入种群。利用随机锦标赛方法进行选择操作以提高种群的质量。使用正态分布交叉(normal distribution crossover,NDX)算子确定搜索步长进行算法的交叉操作。随机选择任务序列中的任务作为突变点,以突变概率决定是否需要突变。根据适应度函数值的变化趋势确定迭代次数,避免算法的过早收敛。仿真结果表明,与All-local、Full-edge、Full-cloud和经典遗传算法卸载策略相比,提出的卸载策略总开销较小。To deal with the problem of large system overhead in the classical genetic algorithm for mobile edge computing offloading,an edge computing offloading strategy based on improved genetic algorithm is proposed,and a multi-objective optimization model including time delay and energy consumption is established.In this strategy,each offloading strategy is regarded as a chromosome,and the gene on each chromosome corresponds to a task such as local computing task,edge computing task and cloud computing task.The reciprocal of the overall cost of the system is used as the fitness function,and the chromosome with a smaller fitness function value is selected to enter the population.The random tournament method is then used for selection operations to improve the quality of the population.The normal distribution crossover(NDX)operator is used to select the search step to perform the crossover operation of the algorithm,randomly select tasks in the task sequence as mutation points,and decide whether to mutation or not according to the mutation probability.According to the change trend of fitness function value,the reasonable iteration times are determined to avoid premature convergence of the algorithm.Simulation results show that,compared with All-local,Full-edge,Full-cloud and classical genetic algorithm offloading strategies,the proposed offloading strategy has the smallest total cost.

关 键 词：边缘计算 计算卸载 遗传算法 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]