The Study Of Segment Caching Mechanism In VoD System

Video-on-Demand Streaming is one of the most prevailing applications in the Internet. As the increase amount of Internet users and the exploding of video content in the Internet, VoD significantly changed our ways to get information and communication. Interiorly, VoD get a variety of opportunity by network convergence, and meanwhile, it focuses the attention of industry and academia.The introduction of P2P partly solves the problems that the traditional C/S Causes video distribution way causes, such as the heavy workload of the sever, and the poor play quality. In a P2P on-demand system, the overall performance is mainly depend on the distribution of data segments in the network, a peer not only cache a certain mount of data, but also adjust the data judging by the global supply and demand. Therefore, it is essential to find a effective data caching strategy to change the data distribution, to improve the share degree, to decrease the workload of sever and to improve the quality of the service.By studying the average consistence problem in asynchronous network environments, we proposes a flow based asynchronous average consistence protocol(FBAA). This protocol is meaningful because each node that follow this protocol will independently get real-time information about some kinds of resources without global control. Experiments shows that FBAA protocol can converge to the average value relatively fast, and the converge time is unrelated to the network scale.We propose a distributed caching strategy that based on data segments popularity. In this strategy, we integratedly consider the data segment popularity in a P2P on-demand system, the imbalance of system resource supply and the disturbance user VCR operators brings. Simulation experiment shows that the algorithm can balance the supply and demand of the data segments flexibly, improve the service quality effectively, reduce the node workload meanwhile it is independent of a global coordinator.