QOS - LIQUIDSTREAM: SCALABLE MONITORING AND BANDWIDTH CONTROL IN PEER TO PEER (P2P) LIVE STREAMING

Abstract

The vast majority of research in P2P live streaming systems focuses on system architectures that offer to participating peers: high upload bandwidth utilization, low delays during the video stream diffusion, robustness and stability under dynamic network conditions and peers behavior. On the other hand in order to guarantee the complete and on time video distribution to every participating peer, the average upload bandwidth of the participating peers should be always greater than the playback rate of the video stream. Most of the approaches do not take into consideration this requirement. Thus, in this paper we propose a very scalable monitoring mechanism of the total upload bandwidth of the participating peers, which is dynamic, accurate and with low overhead. Moreover, by exploiting this monitoring mechanism we present and evaluate an algorithm that allows the accurate and on time estimation of the minimal required additional bandwidth that an external set of resources (e.g. auxiliary peers) have to contribute. In this way we guarantee the uninterrupted the stream delivery and provide high Quality of Service (QoS) in live streaming.