Author Login Editor-in-Chief Peer Review Editor Work Office Work

Computer Engineering ›› 2006, Vol. 32 ›› Issue (16): 103-104,. doi: 10.3969/j.issn.1000-3428.2006.16.038

• Networks and Communications • Previous Articles     Next Articles

Net Flow Clustering Analysis Based on SOM Artificial Neural Network

YANG Zhe   

  1. Department of Computer Science and Engineering, Tongji University, Shanghai 200092
  • Received:1900-01-01 Revised:1900-01-01 Online:2006-08-20 Published:2006-08-20

基于SOM人工神经网络的网络流量聚类分析

杨 哲   

  1. 同济大学计算机科学与工程系,上海 200092

Abstract: Network performance and use pattern are important for the applications. Through net flow analysis, use pattern can be gotten, there are lots of redundant information in net flow datum. Through data mining, some potential used patterns can be distinguished. The SOM clustering algorithm is used to analyze the net flow. And most users in same local network have the normal use pattern, and there still exist some exceptional use patterns, which are called outlier.

Key words: Data mining, SOM, Net flow, Clustering analysis

摘要: 网络性能及使用模式是影响网络应用的关键因素。通过对网络流量的分析,能够反映出网络的使用模式,但网络流量数据中包含大量的冗余信息。通过数据挖掘的方法,能提取出潜在的网络使用模式。使用SOM人工神经网络对局域网流量进行聚类分析,发现同一局域网内的用户,其对网络的使用模式基本相同。同时发现个别不同的网络使用模式,存在少量的使用模式上的“奇点”。

关键词: 数据挖掘, SOM, 网络流量, 聚类分析