A Neighbor Selection Algorithm Based on Relative Capacity for P2P Media Streaming System*

摘要

Data-driven P2P media streaming system is the most popular system at present. However, without considering the peers' bandwidth, delay and other factors, the traditional neighbor peer selection can cause the long system delay. Taking fully account of the available bandwidth, delay and playback quality, this paper presents a Neighbor Selection algorithm based on Relative Capacity (NSRC). First, from the view of the peer itself, a relative capacity model is proposed for evaluating the neighbor relationship between peers. Second, a directional gossip method is applied in order to distribute the service capacity information by which peer can find neighbors efficiently. Moreover, the neighbor updated mechanism is elaborated and the message overhead is analyzed. Simulation results demonstrate that the algorithm can significantly reduce the system delay and improve the system playback finality with low message overhead.