中文摘要：

P2P流媒体点播系统中,由于用户的VCR操作存在不确定性,在查找合作节点时存在一定困难。针对该问题,提出了一种基于预测的邻居节点选择算法。描述了基于虚拟直播频道的点播系统覆盖网构建方式,将系统中的节点通过DHT协议的方式组织起来。提出了一种基于用户行为的学习预测策略模型,根据该模型预测结果提前查找所需邻居节点。仿真实验表明,该算法能改善用户跳转响应延时,促进节点协作。

英文摘要：

In P2P VoD（Video on Demand） system,it is difficult to find neighbors because the users’VCR operations are uncertainty.Aiming at this problem,a neighbor search algorithm based on prediction is proposed in this paper.A method based on virtual live channel is described to build the overlay,in which peers are organized by DHT（Distributed Hash Table） protocol.A learning prediction model based on users’behavior is proposed.According to the predicted results,the peer searches the neighbors in advance.The simulation results demonstrate that the algorithm can improve the VCR response latency and promote the peers’cooperation.