混合流媒体分发系统中多指标用户群分组算法

A User-group Grouping Algorithm Based on Multiple Indicators in Hybrid Streaming System

针对混合流媒体分发系统中服务节点提供服务时出现额外的跨地域、跨网络开销以及搜索服务节点效率低下的问题,提出一种基于多指标的用户群分组算法,并应用于服务节点选择。依次分别利用节点位置和网络类型对用户群进行分组,将候选服务节点限制在同地域、同ISP范围内,以减少不必要的跨地域、跨网络开销;然后结合节点兴趣对用户群作进一步划分,将搜索服务节点的范围限制在兴趣组内,以减小搜索流量、提高搜索效率。仿真实验表明:提出的算法能有效将用户群分组,提高分发系统服务效率。

To address the problem of additional cross-regional and cross-network cost that appears when the service node provides streaming media services, and the problem of inefficiency in searching service node, a grouping algorithm based on multiple indicators is proposed and used in server selection. As the algorithm goes,the user-group would be grouped by the node location and network type successively,limited the candidate server nodes in the same area and the same ISP, consequently reduced additional cross-regional and cross-network cost. Then the user-group would be divided further based on nodes＇ interest, limited the search scope w/thin an interest group, consequently reduced searching traffic and improved searching efficiency. Simulation demonstrates that the proposed algorithm can effectively divide the user-group, improving service efficiency.