Entropy Based P2P Flow Detection

Abstract

As an efficient distribution mechanism, peer-to-peer technology has providing users with cheap and powerful communicating facilities. With the growth of P2P applications, Statistics suggest that P2P flow accounts for a significant fraction of the Internet traffic. Existing approaches to identify P2P traffic have well known drawbacks. In this paper, a novel approach to identify P2P traffic based on information entropy theory has been proposed. We investigate the problem of estimating the entropy in a P2P streaming computation model and find that using entropy can aid P2P stream monitoring. Experimental results show that the characteristic entropy changes can be used to construct a threshold based detector for P2P stream applications.