WMAN中的边缘服务器放置研究  

Research on edge server placement in WMAN

作　　者：赵兴兵 赵一帆 李波[1] 陈春 丁洪伟[1] ZHAO Xingbing;ZHAO Yifan;LI Bo;CHEN Chun;DING Hongwei(School of Information Science and Engineering,Yunnan University,Kunming 650000,China;School of Electrical and Information Technology,Yunnan Minzu University,Kunming 650000,China;College of Applied Technology,Yunnan Minzu University,Kunming 650000,China)

机构地区：[1]云南大学信息学院,云南昆明650000 [2]云南民族大学电气信息工程学院,云南昆明650000 [3]云南民族大学应用技术学院,云南昆明650000

出　　处：《现代电子技术》2022年第13期22-27,共6页Modern Electronics Technique

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

摘　　要：以有效减少用户请求的传输时延、提高用户体验为目的,针对移动边缘计算中WMAN环境下的边缘服务器(ES)放置问题进行研究,提出基于灰狼优化(GWO)算法优化K⁃Means的算法予以解决,该算法借助GWO的全局寻优能力,开始时对聚类中心进行优化,改善了K⁃Means对初始聚类中心敏感的问题;针对放置问题,对传统聚类标准欧氏距离进行改进,并以所有类内距离之和作为适应度函数;通过迭代更新α狼优化聚类中心,避免了陷入局部最优的问题。仿真结果表明,与其他算法相比,该算法的时延平均降低了40%左右,负载均衡平均降低了约34.2%。In order to effectively reduce the transmission delay of user requests and improve user experience,the placement of edge server(ES)in WMAN environment in mobile edge computing is studied.An algorithm of gray wolf optimization(GWO)algorithm optimizing K⁃Means is proposed.With the help of the global optimization ability of GWO,the clustering center is optimized at the beginning,which improves the sensitivity of K⁃Means to the initial clustering center.The traditional clustering standard Euclidean distance is improved to solve the placement problem,and the sum of all intra⁃class distances is used as the fitness function.By iteratively updating theαwolf optimization clustering center,the problem of falling into local optimization is avoided.The simulation results show that,in comparison with the other algorithms,the delay of the proposed algorithm is reduced by about 40%on average,and the load balancing is reduced by about 34.2%on average.

关 键 词：移动边缘计算 边缘服务器 放置 平均时延 负载均衡 K⁃Means GWO 

分 类 号：TN911.1-34[电子电信—通信与信息系统] TP391[电子电信—信息与通信工程]