Research On The Caching & Scheduling Strategy For P2P Streaming System

In recent years, the research on eliminating the bottleneck of the streaming media service of P2P system has become a hotspot in the application field. Moreover, how to guarantee the quality of service of streaming media and promote the system performance on the premise of taking full advantage of the resources of many nodes in the P2P network is one important aspect in the field.Based on analyzing and summarizing system architecture and the application characteristics of the existing P2P streaming media system, this paper studies the optimization of the performance of the P2P on-demand streaming media system. The details are as following:First of all, the paper analyzes and summarizes the main factors that affecting the node startup-latency, playback continuity, and the server load performances. The paper also discourses the important effects of node caching management and data scheduling strategy on the performances of P2P streaming system.Secondly, the paper studies the existing caching management penetratingly. Meanwhile, the paper analyzes the advantages of the probability caching mechanism from the views of the streaming media demanding of buffer management and the usage of buffer memory space. Moreover, in order to overcome the drawback of cache replacement strategy of the existing probability caching mechanism, a novel strategy based on the services supply-demand difference and urgency, which is called SDDU strategy, is proposed to further improve the system performances.Subsequently, to overcome the drawback of the existing data scheduling strategy, the paper proposed a novel data scheduling strategy, which is based on the cache offset of playback and called SSCOP. After that, the paper describes how to use the new strategy in the probability caching mechanism.Finally, the new probability caching management mechanism using SSCOP and SDDU strategies is simulated in a simulator. In the simulation, the existing probability caching management mechanism is also simulated with the same parameters. Through comparing the new with the old in startup-latency, playback continuity, and the server load, the paper confirms that the improved probability caching management mechanism can effectively promote these main performances.