大尺度IP网络流量异常特征的多时间序列数据挖掘方法

Multiple time series data mining approach of large-scale IP network abnormal traffic feature analysis

提出了一种大规模通信网络流量异常特征分析的多时间序列数据挖掘方法,把多个网络流量特征参数构成的时间序列作为一个整体进行分析研究,进行多时间序列数据挖掘产生网络流量异常相关的有效关联规则,对整个通信网络的安全威胁进行描述。Abilene网络数据验证了该方法。

This paper proposed a large-scale IP network traffic feature anomaly detection method using time series data mining,analyzed the network traffic feature elements time series as a whole,obtained valid association rules of abnormal network traffic feature using multiple time series data mining,characterized the entire communication network security threats situation accurately.Experiments with Abilene network Netflow data verifies this method.