Research on a method of P2P traffic identification based on multi-dimension characteristics

With the increasingly development of P2P in recent years, the operational mechanism of P2P application has varied, it´s very difficult to distinguish P2P traffic from normal traffic by using single traffic characteristics. The paper put forward a method of P2P traffic identification based on multi-dimension characteristics. Extracting multi-dimension characteristics of P2P application such as ratio of UDP/TCP protocol packet, packet changing ratio, average data transmitting size and ratio of up/down traffic, to indentify P2P traffic under off line status by using decision tree. Test result indicated that P2P traffic in network could be effectively identified by using this method, with an accuracy over 97 percent.