网络虚拟化环境下的业务聚类方法

Methods for Traffic Clustering in Network Virtualization Environment

为了满足未来丰富多样化的网络业务需求,研究者们提出了许多解决方案,这些方案的本质都是基于网络虚拟化技术的。本文在网络虚拟化环境下展开业务聚类研究,并对业务聚类的需求和聚类算法进行了分析,提出了基于近邻传播聚类算法(AP)的网络业务聚类方法,针对算法存在的不足,提出了优化的AP聚类算法(OAP)。实验结果表明采用该方法在业务聚类过程中获得了较好的性能。

In order to meet the demand of abundant network traffic in the future, many solutions are proposed by researchers, most of which were based on network virtualization technology. We studied the traffic clustering under network virtualization environment, made analysis to the demand of traffic clustering and some clustering algorithms, and proposed a traffic clustering method based Affinity Propagation（ AP） algorithm. Considering the shortcomings of the AP algorithm in traffic clustering,an Optimized Affinity Propagation（ OAP） algorithm was designed. Experimental results prove that the OAP algorithm has fine performance in traffic clustering.