摘要
针对软件定义网络中的多控制器部署问题,首先通过K-means++算法对网络节点聚类,得到网络中初始控制域和控制器位置,然后使用粒子群算法以最小化时延和负载均衡为优化目标,多个粒子并行搜索最优解,进一步优化控制域和控制器位置。在小、中、大型网络拓扑上与随机算法、K-means++算法、粒子群算法的多控制器部署方法比较,仿真结果表明,在中小型网络中,比其他3种算法在平均传播时延和负载均衡上更加稳定且时延更低,在大型网络中,平均传播时延,最坏传播时延和控制器的负载均衡上均优于其他3种算法。
To solve the problem of multi-controller deployment in software defined networking,a multi-controller deployment method based on K-means++and particle swarm optimization is proposed.Firstly,the K-means++algorithm is used to cluster the network nodes to obtain the initial control domain and controller position in the network,and then the particle swarm optimization algorithm is used to minimize delay and load balancing as the optimization goal,and multiple particles search for the optimal solution in parallel to further optimize the control domain and controller position.Compared with the multi-controller deployment methods of random algorithm,K-means++algorithm and particle swarm optimization algorithm in small,medium and large scale network topology,simulation results show that it is more stable and less delay than the other three algorithms in average propagation delay and load balancing in small and medium scale network,and average propagation delay,the worst propagation delay and the load balancing of the controller are better than the other three algorithms in large scale network.
作者
徐慧
吴美连
XU Hui;WU Meilian(School of Computer Science,Hubei Univ.of Tech.,Wuhan 430068,China)
出处
《湖北工业大学学报》
2025年第1期43-48,共6页
Journal of Hubei University of Technology
基金
国家自然科学基金(61602162)。
关键词
软件定义网络
多控制器部署
K-means++
粒子群算法
时延
负载均衡
Software Defined Networking
multi-controller deployment
K-means++
particle swarm optimization algorithm
delay
load balancing
