P2P Traffic Classification Using Clustering Technology

Network traffic classification has been always an important technology of network management for a long time. In recent years, the more unknown P2P traffic flows on the Internet new applications produce, the more network bandwidth the traffic flows of P2P network applications will fully occupy. In order to preserve the quality of network service, it is very necessary to classify the P2P traffic flows. In this paper, a P2P traffic flow classification method, based on the clustering technology, is proposed. The system aggregates the labeled traffic flows and similar unknown ones to the same cluster by the distance ratio according to their features. The proposed method does not need to make the decision of the cluster number in advance. In addition, it also solves the classification of unknown traffic flows by checking if the labeled traffic flows exist in the same clusters.