基于离散粒子群的SDN动态流调度算法  

SDN Dynamic Flow Scheduling Algorithm Based on Discrete Particle Swarm Optimization

作　　者：刘威 高新成 王启龙 张宣 王莉利[1] LIU Wei;GAO Xincheng;WANG Qilong;ZHANG Xuan;WANG Lili(School of Computer and Information Technology,Northeast Petroleum University,Daqing 163318,Heilongjiang Province,China;Center of Modern Education Technique,Northeast Petroleum University,Daqing 163318,Heilongjiang Province,China)

机构地区：[1]东北石油大学计算机与信息技术学院,黑龙江大庆163318 [2]东北石油大学现代教育技术中心,黑龙江大庆163318

出　　处：《吉林大学学报（理学版）》2023年第5期1139-1146,共8页Journal of Jilin University:Science Edition

基　　金：国家自然科学基金(批准号:61702093);中国高校产学研创新基金(批准号:2021ITA02011);黑龙江省教育科学规划重点项目(批准号:GJB1423357).

摘　　要：针对数据中心网络中流量路径分配不合理、易导致大流冲突的问题,提出一种基于离散粒子群的软件定义网络(SDN)流量调度算法.该算法重定义粒子群内部的寻解过程,以最优化网络性能为目标,动态地为数据中心的流量分配最优路径,以减少大流量冲突;并引入Metropolis设计多元化最优调度方案,确保数据中心流量的合理调度.与其他流量调度算法进行对比验证的实验结果表明,该算法提升了网络质量,降低了大流时延,实现了更好的负载均衡.Aiming at the problem of unreasonable traffic-path allocation and elephant flow collision in data center networks,we proposed a software defined network(SDN)flow scheduling algorithm based on discrete partical swarm optimization.The algorithm redefined the search process within the partical swarm with the goal of optimizing network performance,dynamically allocated optimal paths for data center traffic to reduce elephant flow collision,and introduced Metropolis to design diversified optimal scheduling scheme to ensure reasonable scheduling of data center traffic.The experimental results of comparing and verifying with other flow scheduling algorithms show that the algorithm improves network quality,reduces elephant flow delay,and achieves a better load balancing of network.

关 键 词：数据中心 软件定义网络(SDN) 离散粒子群 流调度 负载均衡 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]