【作者】 付锦红；

【导师】 刘卫忠；

【作者基本信息】 华中科技大学 ， 微电子学与固体电子学， 2009， 硕士

【摘要】 随着Internet网络的快速发展,人们已不再满足于普通的网页、邮件等应用,流媒体应用正在逐渐成为关注的热点,而在流媒体的各项应用中,网络电视备受瞩目。但是流媒体资源数据量大,对服务器的带宽要求高,因此在流媒体点播系统中采用P2P技术势在必行。本文首先分析了基于P2P网络流媒体点播系统的基本原理,并给出了系统实现的整体架构、工作模式以及工作流程。在此基础上,重点研究了缓存淘汰与替换以及数据调度算法。在本文的最后,结合提出的缓存淘汰与替换以及数据调度算法,实现了基于P2P网络的流媒体点播系统。在缓存淘汰与替换方面,传统的LRU(最早使用)和LFU(最少使用)算法只考虑了节点自身的缓存情况,并没有考虑整个系统资源的缓存情况。本文提出了基于冗余度的缓存淘汰与替换算法,该算法优先将系统中冗余度高的数据淘汰和替换掉,使得系统中含量较少的数据资源始终保留在有限的缓存空间上,以充分发挥缓存空间的作用,更好地为其他节点提供服务。仿真结果表明基于冗余度的缓存淘汰与替换算法可以大量减轻流媒体服务器的负担。在数据调度方面,分析了传统的顺序调度算法和最少优先调度算法,本文中提出将顺序调度与最少优先调度结合起来的调度算法,在预选窗口的前面部分采用顺序调度算法以保证播放质量,而在后面部分采用最少优先调度算法,这样可以增加系统中含量少的片断数量,从而更好的为其他节点服务,减轻流媒体服务器的负担。仿真结果显示顺序调度与最少优先调度相结合的调度算法继承了顺序调度与最少优先调度两者的优点,可以在保证一定播放质量的同时减轻流媒体服务器的负担。更多还原

【Abstract】 With the rapid development of Internet networks, it is no longer satisfied with the ordinary web pages, e-mail applications. Now streaming media applications are increasingly becoming the focus of attention, and in the applications of streaming media, network television concerns a great deal of attention. However, streaming media has large amount of data, and high bandwidth requirements on the server. So in the streaming media system, it is imperative to use P2P technology.This paper firstly analyzed the basic principle of the streaming media on demand system based on P2P network and gave the architecture, working mode and the working process of the system. On this basis, this paper primary did research on the cache replacement and transmission algorithms. In the end of this paper, combining with the cache replacement and transmission algorithms, the streaming media on demand system based on the P2P network has been implemented.There are two traditional cache replacement algorithms, one is LRU (Least Recently Used) and the other is LFU (Least Frequently Used). For one single peer, the two algorithms maybe are the best, but they just considered the cache condition of their own. So this paper proposed a cache replacement algorithm which is related to the redundant of the chunks in the system. When the disk cache is full, the chunks having high redundancy rate will first be removed. And the chunks which are less in the system will be kept in the disk cache, so the peer can provide these chunks for other peers. Simulation results show that this cache replacement algorithm can reduce the overhead of the streaming media server.There are two traditional transmission algorithms, one is sequential and the other is rarest first. The sequential algorithm transmits the chunk one by one according to their playing sequence, this algorithm can keep the video playing fluently. And the rarest first algorithm transmits the chunk first that is the rarest in the system, this algorithm helps speeding up the spread of chunks, hence improves the quality of the whole system. This paper presented a transmission algorithm that combining the sequential and rarest first algorithms. For the emergency chunks, the sequential algorithm will be chose to keep the video play fluently. And for the none emergency chunks, the rarest first algorithm will be chose to make the chunks uniform distribution in the system. Simulation results show that this transmission algorithm can guarantee the video playing fluently and reduce the overhead of the streaming media server.更多还原