Pre~2VOD: A VCR Supported P2P VOD System

With the development of network technology and bandwidth of the network, video application has been paid more and more attention in recent years and a large number of streaming media applications have appeared. Traditional streaming media application is based on C/S mode which need a central service node to provide video streaming and take charge of node manager. To overcome the problems exposed in C/S mode such as heavy loading of server and limited scalability,P2P technology was used in streaming media applications.At the beginning, P2P streaming applications mainly used for living video services, with people's further hope to watch their favorite programs on the Intemet,P2P technology for video on demand appears.Many problems exiting in traditional C/S mode, such as sever overload and scalability, have been solved by P2P VOD system, however, the characteristics of data distribution and uncertainty of frequent VCR (video cassette recorder) operations make a VOD system hard to locate new supplying peers with low latency when VCR operations occurred, causing long resuming time and a waste of data buffer. To achieve a good interaction in P2P systems, we need to do optimization in content locating, content delivery and content storage, in which content locating is the most important. In this paper, we propose a purposeful pre-fetch scheme with the assistant of a tree structured vector quantization based prediction model. We link all the active nodes, some of which are in the R-B tree indexing overlay, into a viewing list, sorted according to their playback position, in order to accelerate the new supplying peers locating.Experimental results show that:The prediction approach proposed in our paper can predict the user behaviors more accurately and cache utilization is effectively improved according to the pre-fetch. In addition, the red-black tree index structure also played a good role in the data location.