Employing QoS Driven Neighbor Selection for Heterogeneous Peer-to-Peer Streaming

Peer-to-peer (P2P) streaming is attracting much attention. It requires neighbor selection to construct an overlay and delivers streaming data on it. Therefore, neighbor selection determines the performance of P2P streaming system. There is obvious heterogeneity of bandwidth and latency in current Internet. However, there is still a lack of neighbor selection approach suited for heterogeneous P2P streaming. Most current neighbor selection methods do not consider the heterogeneity or do not consider bandwidth and latency simultaneously. In this paper, we propose a new QoS driven neighbor selection method derived from the Metropolis-Hastings algorithm. The main contribution of this paper is: i) we put forward a new metric for the first time, called bandwidth latency ratio (BLR), which is a tradeoff between bandwidth and latency. We testify the effectiveness of BLR on evaluating QoS sensitive P2P streaming system. Our neighbor selection could optimize the distribution of peers´ BLR; ii) in heterogeneous environment, our approach could effectively improve the QoS of peers´ neighbors, which will enhance the transmission performance and user experience of P2P streaming system. We provide theoretical analysis of our method´s improvement in terms of BLR, compared with SCAMP. Simulation results also demonstrate that the throughput, packet delay and playback quality of the P2P streaming system with our neighbor selection can be significantly improved.