Balanced dynamic buffering for scalable video-on-demand streaming on peer-to-peer networks

In peer-to-peer video-on-demand streaming networks, the alleviation of server load depends on reciprocal stream sharing among peers. In general, on-demand video services enable clients to watch videos from beginning to the end in a pipeline fashion. As long as a client still caches the initial part of the video, it can provide on-demand service to the subsequent clients without intervention of a server. Therefore, the key challenge is how to maximize the availability of a video for stream relay by keeping the initial part of a video in a peer’s buffer as long as possible. In addition, to address the issues of delivering data on lossy network and providing scalable video quality of services for clients, the adoption of multiple description coding has been proven as a feasible resolution by much research work. In this paper, we propose a novel caching scheme for P2P on-demand streaming called Balanced Dynamic Buffering with multiple description coding. The proposed balanced dynamic buffering gradually reduces the number of buffered descriptions once the buffer of a peer gets full. In addition, peers are grouped based on the number of descriptions cached in their buffer. The intra-group balancing and inter-group balancing are applied for well distributing descriptions and balancing their forwarding bandwidth on peer-to-peer on-demand streaming networks. In the paper, we also propose the failure recovery for the balanced dynamic buffering scheme. Simulation results showed that the balanced dynamic buffering significantly outperformed other dynamic buffering schemes and the CoopNet in server bandwidth utilization, and a longer buffer in a peer did favor the failure recovery on peer-to-peer on-demand streaming networks.