P2P Traffic Identification Based on the Signatures of Key Packets

Peer-to-Peer (P2P) protocol is widely used in many network applications and P2P traffic is becoming dominating in the current Internet and may cause serious congestion. Recently, P2P applications tend to intentionally disguise their traffic flows by using arbitrary ports, which leads the accurate identification of P2P traffic to become a very difficult job and hot research topic in network traffic control. In this paper, we employ the longest common subsequence to identify the key packets from the traffic flows, and present a new method to identify the P2P flows based on the signatures of key packets with only direction and payload length in the first few packets of a flow. We test the new method with four popular P2P applications in the actual network environment and the results show that the new method has high accuracy and efficiency since only a few packets in each flow needs to be analyzed. The new method is suitable for real time traffic management and monitoring.