摘要
在大规模分布式网络应用中,对网络节点进行聚类是构建高效网络体系结构的有效办法之一.在利用网络坐标系统Vivaldi得到各个节点的网络坐标的基础上,对网络节点进行K-medoids聚类.然后,针对K-medoids算法对初始中心选值敏感和易陷入局部极值的问题,提出基于免疫克隆算法的K-medoids聚类.实验结果表明,该聚类算法具有良好的可靠性及可扩展性,能对节点进行有效聚类.
In large-scale distributed network applications,nodes clustering is a useful way to construct an effective network infrastructure.The coordinates of network nodes can be get by the network coordinates system Vivaldi,then,network nodes can be clustered by the K-medoids algorithm according to their coordinates.But K-medoids is sensitive to the initial cluster centers and easy to get stuck at the local optimal solutions.In order to improve the performance of the K-medoids algorithm,the K-medoids based on immune clonal algorithm(KICA) is presented in this paper.Experimental results show KICA has good reliability and expansibility,and it is effective for clustering internet nodes.
出处
《微电子学与计算机》
CSCD
北大核心
2011年第8期119-122,共4页
Microelectronics & Computer
基金
中央高校基本科研业务费专项资金(2011YJ)
作者简介
李永男,(1978-),博士研究生,讲师.研究方向为网络体系结构、分布式应用.
余镇危男,(1942-),教授,博士生导师.研究方向为网络体系结构、覆盖网、下一代互联网.