一种新的P2P直播系统共享机制的测量、分析与建模

不同的P2P 直播系统具有不同的设计理念和特点，以往对直播系统的研究大多基于PPLive，而PPStream由于其协议机制不对外公开，研究者知之甚少。本文的主要工作是：（1）通过网络测量破解了PPStream直播系统的缓存结构和共享机制，发现：与PPLive 系统滑动的共享窗口不同，PPStream具有跳跃的共享窗口---系统将节目源全局地划分成多个相位，下载同一相位的数据的用户之间才能进行数据共享；（2）利用“窗口跳跃模型”分析了PPStream 直播系统的缓存结构和共享机制对节点间的互惠关系（数据共享）的影响，发现：该种跳跃的共享窗口机制将造成节点相位波动；（3）根据该相位波动特征，我们建立了预估模型来描述系统中处于不同相位的人数。数值评估与系统仿真实验都表明，这个预估模型能很好地反映带宽资源争抢造成节点相位波动的特征。这有利于进一步完善PPStream 直播系统的带宽分配策略以保证节点的播放进度。

Measurement,Analysis and Modeling of a New Sharing Mechanism in P2P Living Streaming System

Different P2P live streaming systems represent different design conceptions and present obviously different characteristics. Currently, most studies on P2P living streaming systems are based on PPLive, while PPStream is hardly known to researchers because its protocols are not open to public. The contributions of this paper are: (1) peers’ buffer structure and sharing mechanism in PPStream live system are figured out in this paper. It is found that peers have jumping sharing window in PPStream which is different from PPLive with sliding sharing window. In PPStream, video chunks are classified into different windows (or phases). Peers downloading chunks in the same phase can share data with each other; (2) authors use a window jumping model to analyze the impact of buffer structure and sharing mechanism on the reciprocal relation between peers. It is found that the sharing mechanism in PPStream can result in peers’ phase fluctuation; (3) based on the features of phase fluctuation, authors develop a predicting model to describe the user population of different phases. Numerical evaluation and system simulation both show that the model can well reflect the fact that the contention on bandwidth resources contributes to the peers’ phase fluctuation. It can bring insights to the work of further improving the bandwidth assignment strategy and peers’ playback quality.