系统聚类树算法在网络拓扑判定中的研究

System clustering tree algorithm in network topology identification

网络拓扑结构的判定是网络层析成像技术在大规模网络中应用的关键问题.主要讨论通过对网络上的主机进行单播的测量来获得网络的逻辑拓扑,提出运用系统聚类分析法进行拓扑判定的新方案.首先介绍了基于延时的"三明治"网络测量方案及相关网络拓扑判定方案,在此基础上提出更为高效准确的运用统计聚类模型进行拓扑判定的系统聚类树算法,最后通过实验将该算法与合并似然树算法进行了比较与分析.

Network topology identification is a key issue in network tomography. This paper considers the problem of discovering network topology solely from host-based, unicast measurements, without internal net- work cooperation. First, the authors introduce a delay-based measurement scheme that does not require clock synchronization and a maximum penalized likelihood criterion for topology identification. Secondly, the authors propose a system clustering tree algorithm for topology identification. Finally, the performance of new identification algorithm is explored through experiments.