Real-Time Detection of Encrypted Thunder Traffic Based on Trustworthy Behavior Association

Abstract

Thunder, as the most popular P2P download software in China, has token up a large amount of bandwidth. And it is almost impossible to identify the encrypted thunder traffic. This paper proposes a method to detect encrypted Thunder traffic, featuring high precision and small computational cost. At the same time, this method doesn't depend on content inspection, nor does it violate users' privacy, which can be used flexibly in high-speed network environment, and deal with changes of statistical traffic properties. We implement a prototype system based on this algorithm, which can detect multiple versions of encrypted Thunder traffics in real time, achieving a precision rate above 95% and a recall rate above 95%. © Springer-Verlag Berlin Heidelberg 2013.