基于混合传送机制的P2P节点缓存策略的研究

发布时间：2018-01-12 00:06

本文关键词：基于混合传送机制的P2P节点缓存策略的研究　出处：《北京邮电大学》2012年硕士论文　论文类型：学位论文

【摘要】：随着通信与网络技术的快速发展,传统的C／S模式加单播方式来推送媒体流的流媒体服务系统由于服务器性能和网络带宽的限制。往往造成系统瓶颈。在有限的带宽实现高质量的音频、视频文件的传输。最好的解决方案就是采用P2P的传输方式。 P2P是"Peer-to-Peer"的缩写,意义是对等计算或对等网络。其核心思想就是把P2P技术应用于实时流媒体数据的分发中去,把网络层的组播功能转移到应用层实现,充分有效地利用各个节点的资源。P2P流媒体的应用也是在P2P文件交换的基础上产生的,用户在互联网上观看视频流媒体,按照以前的技术是用户从服务器端下载了一段,然后用户的电脑上再播放一段,用户只能从服务器端下载资源,而P2P流媒体的应用,使用户在观看视频流媒体时,可以与同时在观看该视频的用户之间进行资源利用,用户可以从这些用户中同时下载该视频,这比过去只能从服务器端下载,明显在效率与速度上要快许多。大多数现有的P2P直播系统采用了基于Chunk的设计。一个视频被分割成许多Chunk。网络中的节点,即Peer利用自己的带宽,在有相同需求的Peer之间上传和下载Chunk。本文将首先介绍P2P流媒体直播系统的发展,优点和关键技术,在建立P2P流媒体直播的概念的基础上。分析P2P系统的工作流程和原理,使读者进一步熟悉P2P系统。然后将从缓存策略的角度对P2P进行分析。在P2P系统中每一个Peer都有一个缓存,缓存中储存着将要播放的Chunk。这个缓存的大小关系着播放延迟和播放流畅度。大的缓存意味着比较好的播放流畅度而播放延迟也相应的会大。当播放流畅度的要求是一定的时候,有必要使用较小的缓存。现在已经有许多关于缓存策略的研究。比如说稀少优先级策略和贪婪优先级策略,两者各有优缺点。而混合策略是将两者的优点结合,综合表现比两者要好。已经有人对这种策略进行建模分析,但是模型比较简单而且没有充分考虑到真实的网络条件,真实的网络带宽很有可能是有一定限制的,因此不一定能满足所有Peer要求下载Chunk的请求。我们将针对原有模型的不足,经过分析在带宽限制条件的网络环境下,建立一个新的模型,重新计算在新的模型中各个Chunk选择策略和播放流畅度的关系。根据这种关系,经过数学推导得出了在播放流畅度为一定的情况下,计算每个Peer最小缓存的方法,并利用这种方法进行了数值仿真。仿真结果进一步验证了我们的结论。根据仿真的图,我们可以更加直观的观察出这些关键参数之间的关系。根据数学表达式和仿真结果我们得到了在播放流畅度一定的情况下,如何降低最小缓存大小的方法的几种途径。
[Abstract]:With the rapid development of communication and network technology. The traditional C / S mode plus unicast to push media stream streaming service system, due to the server performance and network bandwidth constraints, often lead to system bottlenecks, in limited bandwidth to achieve high-quality audio. Video file transmission. The best solution is to use P2P transmission. P2P is the abbreviation of "Peer-to-Peer", which means peer-to-peer computing or peer-to-peer network. Its core idea is to apply P2P technology to the distribution of real-time streaming media data. The multicast function of the network layer is transferred to the application layer, and the application of P2P streaming media is also generated on the basis of P2P file exchange to make full and effective use of the resources of each node. Users watch video streaming on the Internet. According to the previous technology, users download a segment from the server, and then play it on the user's computer. The user can only download resources from the server. The application of P2P streaming media enables users to use resources between watching video streaming and users watching the video at the same time, users can download the video from these users at the same time. This can only be downloaded from the server side. Most of the existing P2P live broadcast systems are designed based on Chunk. A video is divided into many Chunk. nodes in the network. Peer uses its bandwidth to upload and download Chunk. among Peer with the same needs. Firstly, this paper introduces the development, advantages and key technologies of P2P streaming streaming system. Based on the concept of P2P streaming streaming, this paper analyzes the workflow and principle of P2P streaming system. Make readers more familiar with P2P system. Then we will analyze P2P from the point of view of cache policy. In P2P systems, each Peer has a cache. The size of the cache is related to the playback delay and the playback fluency. A large cache means better playback fluency and the playback delay is also larger. The requirement for a degree of ease is a certain amount of time. It is necessary to use smaller caches. There is a lot of research on caching policy. For example, sparse priority strategy and greedy priority strategy have their own advantages and disadvantages. The hybrid strategy is a combination of the advantages and disadvantages. Comprehensive performance is better than both. This strategy has been modeled and analyzed, but the model is relatively simple and does not fully take into account the real network conditions, the real network bandwidth is likely to have some limitations. Therefore, we will not be able to meet all the Peer requests to download Chunk. We will analyze the shortcomings of the original model and build a new model under the bandwidth constraints of the network environment. Recalculate the relationship between each Chunk selection strategy and the playback fluency in the new model. The method of calculating the minimum buffer of each Peer is used to carry out the numerical simulation. The simulation results further verify our conclusion. We can more intuitively observe the relationship between these key parameters. According to the mathematical expression and simulation results, we can get a certain degree of fluency. Several ways to reduce the minimum cache size.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP393.02;TP333

【相似文献】