Design of a method for encrypted P2P traffic identification using K-means algorithm

作者：Du Ye^*; Zhang Ruhui

来源：Telecommunication Systems, 2013, 53(1): 163-168.

DOI：10.1007/s11235-013-9690-5

• K-means
• P2P
• Traffic Identification
• Flow-information extraction

Peer-to-Peer (P2P) plays an important role in the domain of Internet application, such as file sharing and instance communication. However, for the characteristic of bandwidth-intensive, P2P traffic identification is especially important in network management, which has been a hot topic in recent years. Based on the review of shortages of current existed technologies, a k-means algorithm based method was brought forward to identify the P2P traffic. For the reason that K-Means is one of the quickest and simplest clustering methods, it was selected as the core algorithm. In addition, the process of flow-information extraction and the descriptions of selected traffic attribute were also introduced in detail. Since the BT file-sharing traffic mainly runs over TCP, we only extract the flow-information of TCP connections. Finally, the experiments on 3 common used P2P applications, which are BitTorrent, BitSpirit and eMule, demonstrated the validity and efficiency of the method.

• 出版日期2013-5
• 单位北京交通大学