A novel caching mechanism for peer-to-peer based media-on-demand streaming

In recent years, peer-to-peer networks and application-level overlays without dedicated infrastructure have been widely proposed to provide on-demand media services on the Internet. However, the scalability issue, which is caused by the asynchronism and the sparsity of the online peers, is a major problem for deploying P2P-based MoD systems, especially when the media server's capacity is limited. In this paper, we propose a novel probabilistic caching mechanism for P2P-based MoD systems. Theoretical analysis is presented to show that by engaging our proposed mechanism with a flexible system parameter, better scalability could be achieved by a MoD system with less workload imposed oil the server, and the service capacity of the MoD system could be tradeoff with the peers' gossip cost. We verify these properties with simulation experiments. Moreover, we show by simulation results that our proposed caching mechanism could improve the quality of the streaming service conceived by peers when the capacity of the server is limited, but will not cause notable performance degradation under highly lossy network environments, compared with the conventional continuous caching mechanism. (C) 2007 Elsevier B.V. All rights reserved.