[计算机]流媒体论文：基于前缀缓存和指数分段的代理缓存算法研究.docVIP

流媒体论文：基于前缀缓存和指数分段的代理缓存算法研究 【中文摘要】随着网络和多媒体技术的发展,流媒体应用越来越流行。然而现有的互联网络缺乏对实时通信的良好支持,影响了流媒体的传输性能。同时,流媒体传输对系统I/O和网络带宽的高要求,使得服务器和网络带宽成为系统的瓶颈。流媒体代理缓存技术通过缓存热门节部分或全部数据,为就近的客户端请求提供服务,可以有效的解决上述问题。因此,针对流媒体应用的特点,研究高效的代理缓存策略,以提高缓存命中率、减小客户端启动延迟、降低服务器负载和网络资源消耗,是流媒体代理缓存技术的重要研究内容。如何替换缓存的对象以便用户直接从缓存中访问是缓存替换算法研究的主要内容。现有缓存替换算法主要以媒体访问次数和访问频率来计算缓存效能值,并未考虑对象未来被访问的概率,对效能值评价不够精确,并且效能值的计算也局限于整个媒体对象而没有精确到段。针对这些问题,本文提出了基于前缀缓存的改进替换算法,该算法以指数分段为基础,细化了对于缓存空间的调度粒度,效能值的计算细化到段,考虑了媒体部分有效性、段未来被访问的概率、段的访问频率、段已缓存长度对缓存效能值的影响和段平均访问长度,因而提高了缓存性能。实验结果表明：本文的流媒体代理缓存算法在提高缓存单位字节的命中率、降低启动延迟率等方面有较好的性能,能够提高流媒体的播放质量和传输效率。 【英文摘要】As the development of the Internet and multimedia technology, the application of streaming media is becoming more and more popular. However, the current Internet lacks support for real-time communication, which influencing the performance of media transmission. Furthermore, due to the high demands of streaming media transmission on system I/O and network bandwidth, the server and network and width turn into the bottleneck of the whole system. While such problems can be effectively solved by the proxy caching technology, which caches part or all of the data of hot programs so as to offer services to the users nearby. Therefore, based on the main features of the application of streaming media, studies on efficient proxy caching strategy to increase caching hit rate, decrease the client startup latency and reduce server overloads and network resource consumption are of great significance to the research of proxy caching technology of streaming media.The main research on cache replacement algorithms is how to replace cache object. Existing cache replacement algorithms have the problem of inaccurate evaluation of the caching utility because their standards are access times and frequency of the media, additionally they don抰 consider the probability of access objects in the future, and the evaluation of cache utility is not accurate, the calculation of cache utility is constrained to the whole media ob