Topology Dynamics in a P2PTV Network
Abstract. In recent years, a number of commercial peer-to-peer TV (P2PTV) applications have been launched. Yet, their mechanisms and characteristics are unknown. In this paper, we study SopCast, a typical proprietary P2PTV system. Treating SopCast as a black box, we perform a set of experiments that are suitable to analyze SopCast in depth. We attempt to disclose the SopCast protocol. The dynamic nature of the SopCast overlay, in terms of node degree, is also addressed in this paper. Our approaches in analyzing the SopCast mechanism and characterizing its topological properties reveal important design insights in SopCast, and may help to better understand similar P2PTV systems.
Introduction
The success of peer-to-peer (P2P) file-sharing systems has spurred the deployment of P2P technologies in many other bandwidth-intensive large-scale applications. Peer-to-Peer Television (P2PTV) has become a popular means of streaming audio and video content over the Internet. Example applications are CoolStreaming [8], TVAnts1, TVU2, SopCast3, etc. It is important to evaluate the traffic impact of such applications, while modeling their behavior. However, P2PTV streaming systems, such as SopCast, are developed for commercial purposes: thus, very little is known about their architectures. Some papers claim that SopCast is based on similar principles as those underlying CoolStreaming, e.g. [2], some refer to it as a BitTorrent-based P2PTV system, e.g. [1], but all without substantiating their claims. Furthermore, SopCast traffic is encoded, which makes understanding the protocol even more challenging. In this paper, we will investigate the SopCast P2PTV system by answering the following two questions. What is the operational mechanism in SopCast? How are topology dynamics reflected in the SopCast overlay?
The rest of the paper is organized as follows. In Section 2, related work is discussed. Section 3 describes the measurement settings in PlanetLab4 network. Based on the experiments conducted in Section 3, we present our understanding of the SopCast protocol in Section 4. Our methodology of modeling the timevariant SopCast overlay and the derived results are provided in Section 5. In Section 6, we conclude the paper.
Download file here
