异构环境下的P2P流媒体节点选择算法

Peer selection algorithm for P2P streaming media in heterogeneous environment

针对异构环境的复杂性和不稳定性,提出一种异构环境下的点对点(P2P)流媒体节点选择算法。利用模糊认知图理论研究异构环境下影响节点性能的多方面因素之间的关系,计算节点的综合服务能力,并选择服务能力强的节点作为邻居节点;为保证邻居节点具有较强的实时服务能力,利用马尔科夫蒙特卡洛方法进行随机行走,周期性地更新邻居节点列表,采用Metropolis-Hastings算法计算转移矩阵以满足随机行走的期望静止概率分布。研究结果表明:该算法能在选择优质邻居节点,提高视频服务质量的同时,保证节点的负载均衡,降低系统消耗,显著提高了系统性能。

In view of the complexity and the instability of heterogeneous environment, a peer selection algorithm for P2 P streaming media system was proposed. The relationship between the factors which affect the performance of joints under heterogeneous environment was studied, the comprehensive service ability of peers was calculated through the fuzzy cognitive maps theory, and the peers with high service ability were selected as the neighbors. In order to guarantee that the neighbors have a high real time ability, the random walk process was utilized to update the list of neighbors periodically by using Monte Carlo methods. In addition, transition probability matrix was calculated by the MetropolisHastings methods to satisfy the expected stationary distribution of random walk. The results show that the proposed algorithm can select excellent peers and ensure the load balance of peers, as well as reduce the consumption of the system and improve the quality of video service and significantly improve system performance.