Fuzzy Based Approach for P2P File Sharing Detection

Abstract

P2P file sharing is a very popular network application. Many studies have focused on the identification of P2P file sharing. In this paper, a fuzzy based approach was proposed to identify the P2P file sharing application. In this fuzzy based approach, four features are selected and normalized as the input parameters; these features are default P2P port, UDP packet count, traffic volume, and distinct IP connections. The fuzzification, rule evaluation, and defuzzification are three steps to obtain the final crisp value. Host running P2P file sharing application can be identified by thresholding on the crisp value. Experimental results show that this proposed fuzzy based approach can identify the P2P file sharing application more accuracy than other intelligent methods such as flexible neural tree and decision tree algorithms.