P2P流媒体系统中并发请求的数据分发算法

Data Distribution Scheme for Concurrent Query in P2P Streaming Media Systems

大规模并发请求是流媒体直播系统面临的一个挑战,也是视频点播系统中亟待解决的一个问题。本文针对相同数据的并发请求问题,提出了一种高效,低带宽消耗、低延迟的数据分发算法。该算法在所有节点的输出带宽只满足一个节点的下载需求带宽的情况下,能够以Td+log2N*Tc的延迟让N个节点收到数据时长为Td、最小数据片段为Tc的数据。

Concurrent query is an urgent problem to be resolved in living streaming systems and in Video-on-Demand systems to alleviate the stress of the server, especially when nodes＇ output bandwidth is not enough to support more than one peer. A new algorithm with high efficiency, low bandwidth requirement and low delay is presented in this paper to settle the problem above. With a well designed node selection algorithm, even though each node＇s output bandwidth can only satisfy one node＇s downloading bandwidth requirement, all nodes can receive the required data block with length Td within the delay of Td＋log2N＊Tc seconds, where N is the number of peers, and Tc is the least data piece＇s time.