关键词: 流量测量与分析 网络监控与管理 主成分分析

Abstract:

Reducing the objects to be measured is one of the optimization targets when monitoring the traffic flows on large-scale networks. Traditional methods  often reduce the objects to be measured according to the flow＇s dependence in the space. The paper presents an algorithm of finding critical paths base   ed on principal component analysis named PCAR, which finds the critical paths in the network by analysing the time-and-space dependence of the network t raffic. We evaluate the algorithm using a large collection of real traffic flows collected in the Abliene network and our results demonstrate that the a lgorithm are effective and features a low error ratio compared with other algorithms.

Key words: traffic flow measurement and analysis, network monitor and management