边缘计算中的边缘服务器放置策略  

Edge server placement strategy in edge computing

作　　者：赵兴兵 李波[1] 杨志军[2] 保利勇[1] 丁洪伟[1] ZHAO Xing-bing;LI Bo;YANG Zhi-jun;BAO Li-yong;DING Hong-wei(School of Information,Yunnan University,Kunming 650000,China;Teaching Instrument and Equipment Center,Yunnan Provincial Department of Education,Kunming 650000,China)

机构地区：[1]云南大学信息学院,云南昆明650000 [2]云南省教育厅教学仪器装备中心,云南昆明650000

出　　处：《计算机工程与设计》2022年第11期3008-3014,共7页Computer Engineering and Design

基　　金：国家自然科学基金项目(61461053);云南大学研究生科研创新基金项目(2020306)。

摘　　要：针对移动边缘计算中边缘服务器(edge server,ES)放置性能差和放置效率低的问题,建立ES放置的多目标优化数学模型,提出一种基于改进遗传算法(modified genetic algorithm,MGA)的放置方法予以解决。采用多轮随机不重复解策略产生初始种群;使用优秀父代和子代共同竞争的策略产生新的子代,保证寻优过程中的种群多样性,提高算法的收敛性。基于上海市电信局的真实网络数据集进行仿真,仿真结果表明,MGA的负载均衡和平均时延优于其它对比算法,能够应对各种复杂WMAN中的ES放置问题。Aiming at the problems of poor placement performance and low placement efficiency of edge server(ES)in mobile edge computing,a multi-objective optimization mathematical model of ES placement was established,and a placement method based on improved genetic algorithm(modified genetic algorithm,MGA)was proposed to solve the problems.The initial population was generated using multi-round random non-repetitive solution strategy,and the excellent parent and offspring competition strategy was used to generate new offspring,which ensured the population diversity in the optimization process and improved the convergence of the algorithm.The simulation was carried out based on the real network data set of Shanghai Telecommunication Bureau.The simulation results show that the load balancing and average delay of MGA are better than that of other comparison algorithms,and it can deal with the problem of ES placement in all kinds of complex WMAN.

关 键 词：移动边缘计算 边缘服务器 放置 遗传算法 负载均衡 平均时延 

分 类 号：TN915.45[电子电信—通信与信息系统]