Traffic analysis in concurrent multi-channel viewing on P2PTV

Abstract

In recent years, peer-to-peer (P2P) video streaming services (P2PTV) have attracted much attention because of the ability to decrease the load on the content servers by distributing data delivery function to peers. On the other hand, P2P overlay networks are oblivious to the physical network topology and thus may cause undesirable traffic straddling on some Internet service providers (ISPs). To optimize P2PTV traffic, several traffic measurements have been studied and revealed the characteristics of P2PTV traffic. However, these studies did not focus on users’ behavior and traffic flow of each user. In this paper, we focus on PPTV that is one of the most famous P2PTV services and collect traffic data when multiple channels are viewed at the same time. Through this measurement, we observed the changes of the number of peers, traffic flows, and packet arrival time. As a result, we found the new characteristics of PPTV such as the transmission state of PPTV by monitoring the variation in the number of new arrival peers. Moreover, we could detect video servers by simultaneously analyzing multi-channel PPTV traffic.