Minimizing Probing CostWith mRMR Feature Selection in Network Monitoring

Periodically monitoring the state of internal links is important for network diagnosis. In this letter, we consider the problem of minimizing probing cost in tomography-based network monitoring. We propose a new algorithm to select the informative probing paths by means of maximum relevance minimum redundancy (mRMR) method. mRMR has been widely used in the field of feature selection. Through the proposed scheme, the paths are regarded as features, and the ones that strongly relevant with link state are picked. We conduct simulation experiments of congested detection, demonstrating that the most useful paths for locating the congested links are chosen through the proposed scheme.