Peer-to-peer video-on-demand with scalable video coding

Peer-to-peer has evolved into a promising communication paradigm for large-scale content sharing. It has recently been suggested for on-demand video streaming as well. Such peer-to-peer Video-on-Demand (VoD) makes effective use of local disk space and upload bandwidth distributed across peers to relieve the server load, which has long been a bottleneck of conventional VoD systems that demand enormous storage and network resources. It is well-known that the startup delay of state-of-the-art peer-to-peer VoD remains much longer than powerful client/server-based systems. In this paper, we present a novel peer-to-peer VoD system that utilizes Scalable Video Coding (SVC) for delay minimization, and to deal with heterogeneous user capabilities as well as dynamic end-to-end resources availability. It brings two tangible benefits: first, starting from the SVC base layer only, the startup delay for a peer to join the system and successfully initialize video playback can be reduced; second, by dynamically adding or dropping SVC enhancement layers, the occurrences of frame freezing due to the temporal network congestion or the insufficient peer bandwidths can be minimized, and quick recovery from such freezing can be expected too. We mathematically formulate the transmission scheduling problem for peer-to-peer VoD with SVC. We strike a balance between startup delay and playback quality, trying to maximize the overall playback quality that all peers experience. We develop a practical scheduling strategy that allows each peer operate locally and efficiently. It implements a zigzag like importance allocation mechanism to determine the transmission order, taking advantage of the supplying peers with more layers and larger bandwidth. We have extensively evaluated our proposed system under diverse network and peer configurations. Our simulation results demonstrate that it makes effective utilization of network resources, and outperforms conventional P2P VoD systems in terms of startup delay and playback quality. (C) 2010 Elsevier By. All rights reserved.