在基于分层视频编码技术的P2P点播系统中,由于视频数据的供需不平衡易导致视频服务器带宽开销严重.先从视频数据的带宽供需的角度对视频服务器的带宽开销进行建模,该模型可归结为求解一个线性优化问题.提出了一种基于在线时间相似性的邻居选择及同视频层及跨视频层的带宽资源分配策略,节点可以从位于同视频层的前驱节点及处于其他视频层的“帮助者”获得所需带宽资源。仿真实验表明,邻居选择及带宽分配策略都能有效地降低视频服务器带宽资源开销。
Imbalanced demand and supply of data in a layered video coding based P2P Video-on-Demand system lead to severe bandwidth consumption at streaming server. This paper fast formulates a mathematical model of bandwidth consumption at streaming server through a viewpoint of bandwidth demand and supply of streaming data. This model can be transferred to a linear programming problem. Then a policy for same/cross layer neighbor selection and bandwidth allocation based on online-time similarity is proposed. Peer can retrieve its requested bandwidth resource from the successors located at the same layer and "helper" located at different layers. Simulation result shows that the proposed policy can efficiently decrease the bandwidth consumption at streaming server under different scenario.