A simple analysis on P2P streaming with peer playback lags

Abstract

Nowadays, mesh-based live streaming on P2P network is becoming more and more popular because of its scalability, robustness and no additional infrastructure requirement (e.g. CoolStreaming, PPlive, PPstream, Gridmedia). How to design a high fluency and small delay system is a challenge task. In this paper we analyze the impact of playback lags between peers on the playback fluency via a simple stochastic model. For every position in arbitrary peer’s buffer, we calculate its probability of containing the corresponding chunk, which can indicate the fluency of a peer’s playback on the video. Through our analysis and simulation, we get insights into playback lags and some useful tips for designing a real P2P live streaming system more effective.