Efficient neighbor selection through connection switching for P2P live streaming

Abstract

Due to the advantages of high scalability and low cost, P2P techniques have been known as a promising solution to a large-scale live streaming system. In the conventional P2P mesh-pull structure, however, a newly joined peer is more likely to connect to peers that have joined most recently and thus have relatively long playback lags. To reduce the average playback lag in a P2P live streaming system, we therefore propose an efficient scheme to select neighbor peers when a new peer joins. In our proposed scheme, even peers whose numbers of connectable neighbor peers have already reached the maximum can be selected as neighbor peers of a new peer through connection switching. Since this makes the playback range of the system denser, data duplication among peers increases considerably. In addition, to prevent the degradation of playback quality of the two peers whose connection is switched to a new peer from each other, connection switching is performed only when all of their neighbor peers have already buffered sufficient data. Through extensive simulations, we show that our proposed scheme reduces playback lag and startup latency significantly compared with the conventional policy.