The node-cache structure in a P2P-based streaming Video-on Demand (VOD) system is different from the ones traditional client/server systems. The real-time synchronization of cache management at each node has been a challenging issue. Usually, the capability of an embedded computing system within the node structure is low and its memory is usually small. In order to reduce the time used in information controlling and data transmission time, a realistic data exchange mechanism and a scheduling algorithm are needed. In this paper, we propose an algorithm method that uses a segmented rolling window and a pre-obtained data to predict information-cache replacement. The method maximizes the cache-hit rate, thus to ensure the quality of VOD service. We apply the algorithm to several scenarios on a local area network (LAW). We demonstrate the experimental results indicating that the node processing overhead is small after deploying our algorithm. The method provides an insight into developing an innovative methodology for higher quality of service of VOD systems.
展开▼