P2P流媒体直播系统数据调度算法的研究

发布时间：2018-04-02 00:16

本文选题：P2P　切入点：数据调度　出处：《吉林大学》2015年硕士论文

【摘要】：随着互联网的高速发展，超大规模流媒体应用已成为互联网和多媒体技术发展的趋势和前沿性方向，对等网络（peer to peer）的出现为大规模流媒体应用提供了技术支持，P2P流媒体系统以其优异的可扩展性和易部署等优点，迅速占领了市场，但是P2P系统也存在着诸多问题，需要继续进行研究。 P2P流媒体系统中，参与数据传输的结点数量多，传输数据量大，而且流媒体数据非常注重时效性，因此必须设计合理的数据调度策略，本文对P2P流媒体系统数据调度的问题进行了讨论，主要研究内容如下：为了适应互联网终端带宽的异构性，引入分层视频编码的思想，它将原始的视频流编码为一个基础流和多个增强流，基础流保证了最基本的视频质量，，因此需要保证能够被单独解码，而增强流是为了增加视频的质量，使用户得到更好观看享受，用户根据自身带宽情况请求相应的视频层次。建立分层数据调度模型，定义分层数据调度中数据块的优先级，并建立多级BM（Buffer Map）模型，阐述分层数据调度的基本流程，并通过实验验证分层数据调度在性能上的优越性。为了将结点缓冲区中的数据块进行更细致的划分，提出了缓冲区分段的思想，将缓冲区分为抛弃区、紧急区、策略选择区、数据调度区。根据RTT设定阈值规定播放点之后一段区间内的缺失的数据块不再请求以节省带宽资源，给出紧急区边界的计算方法，并推导出不同层被选中概率的计算方法，提出针对不同的区段设计不同的调度策略。分析了P2P系统中数据调度的问题，将数据调度问题形式化，分析了最优数据调度的形式，通过理论分析将最优调度问题转化为等价的可在二项式时间内计算的最小费用流问题，通过推理证明验证了最优数据调度问题与最小费用流问题求解的一致性，验证了观点的正确性和可行性。使用P2P仿真软件对上述观点进行了实验验证，通过调整实验中的参数设定，得出不同条件层次条件下分层视频调度的性能差异，并通过实验验证分段调度在性能上优于传统的调度算法，通过使用Goldberg的“CS2”库模拟实验验证基于最大流的数据调度算法在性能上也更优于传统的数据调度算法。
[Abstract]:With the rapid development of the Internet, the application of super-large scale streaming media has become the trend and the leading direction of the development of Internet and multimedia technology. Peer-to-Peer (Peer-to-Peer) provides technical support for large-scale streaming media applications. P2P streaming media system has quickly occupied the market because of its excellent scalability and easy deployment, but there are also many problems in P2P systems. Further research is needed. In P2P streaming media system, the number of nodes involved in data transmission is large, the amount of data transmitted is large, and the streaming media data pay great attention to timeliness, so it is necessary to design a reasonable data scheduling strategy. This paper discusses the problem of data scheduling in P2P streaming media system. The main research contents are as follows:. In order to adapt to the heterogeneity of Internet terminal bandwidth, the idea of layered video coding is introduced. The original video stream is encoded into one basic stream and multiple enhanced streams, and the basic stream guarantees the most basic video quality. Therefore, it is necessary to ensure that the video can be decoded separately, while the enhanced stream is designed to increase the quality of the video and make the user enjoy the video better. According to the bandwidth of the user, the user requests the corresponding video hierarchy and establishes a hierarchical data scheduling model. The priority of data blocks in hierarchical data scheduling is defined, and a multilevel BM(Buffer Map model is established. The basic flow of hierarchical data scheduling is expounded, and the superiority of hierarchical data scheduling in performance is verified by experiments. In order to divide the data blocks in the node buffer more carefully, the idea of buffer segmentation is put forward. The buffer is divided into jettisoning area, emergency area and policy selection area. Data scheduling area. According to the threshold set by RTT, the missing data blocks in the interval after playing point are not requested to save bandwidth resources, the method of calculating the boundary of emergency area is given, and the method of calculating the selection probability of different layers is deduced. Different scheduling strategies are proposed for different segments. This paper analyzes the problem of data scheduling in P2P system, formalizes the problem of data scheduling, and analyzes the form of optimal data scheduling. The optimal scheduling problem is transformed into an equivalent minimum cost flow problem which can be calculated in binomial time by theoretical analysis. The consistency between the optimal data scheduling problem and the minimum cost flow problem is proved by reasoning. The correctness and feasibility of the view are verified. P2P simulation software is used to verify the above viewpoint. By adjusting the parameters in the experiment, the performance differences of layered video scheduling under different conditions and hierarchies are obtained. The performance of piecewise scheduling is better than that of traditional algorithms, and the performance of the data scheduling algorithm based on maximum flow is better than the traditional data scheduling algorithm by using the "CS2" library of Goldberg to simulate the performance of the algorithm.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP393.02

【参考文献】