Abstract: Network traffic is composed of many connections between hosts, which send all kinds of traffic data. Most studies focus on the statistic of integrated traffic, ignoring the microscopic relationship of different network flows. The paper focuses on analyzing the integrated traffic in a microscopic view, trying to find the relationship of network flows which are based on source and destination IP addresses and studying the difference of normal network flows and attack network flows in correlation.

Key words: Network flow, Principal components analysis, Eigenvalue, Eigenvector

摘要： 网络流量是由主机间的大量连接组成，并且包含各种各样的数据。大部分研究集中在整体流量的统计性质上，忽视了整体流量中不同网络流之间的微观关系。该文从微观的角度将整体流量进行分析，找出了基于源和目的IP地址的网络流之间的关系，研究了正常网络流和攻击网络流在相关性上的不同。

关键词: 网络流, 主成分分析, 特征值, 特征向量

CLC Number: 

• TP393.01