基于突变理论的网络异常行为分析方法

给出了把异常数据流从网络数据流中按区间分离出来的方法,基于分离出来的每个时间序列异常数据流集，给出了网络异常行为突变模型设计、参数估计和检验等方法。为了推断未来网络异常行为出现情况。给出基于RBF网络的模型外推方法。根据外推得到的突变模型,可以方便地计算有关网络异常行为的各种参数和评估指标。

An Approach to Analyzing Network Anomalous Behaviors Based on Catastrophe Theory

A method to pickup anomalous data set according to time-interval from network data flow is represented, a catastrophic model is designed based on the distilled time-serial data set, the methods of parameters estimating and checking on the catastrophic model are given. In order to forecast occurrence of future network anomalous behaviors, a catastrophic model extrapolation approach is put forward using RBF neural network. Using the extrapolated catastrophic model the parameters about future network anomalous behaviors can be calculated conveniently.