LANCStream: A Locality-Aware and Network Coding-Based P2P Live Streaming System

The locality-aware P2P live streaming system is a promising solution for reducing the cross-ISP traffic, due to its topology formation considering the underlay network information. However, they usually do not consider the scheduling algorithm simultaneously and thus suffer from bad performance. In this paper, we design a locality-aware and network coding-based P2P live streaming system (LAN Stream) to simultaneously reduce the cross-ISP traffic and improve the performance. We solve the problem through considering the two aspects: the topology formation and the scheduling algorithm. We introduce a new hybrid topology formation combined tree and mesh structure and we respectively design a seamless push scheduling algorithms in the tree and mesh structure. We model the push scheduling strategy based on network coding as an optimization problem and put up a greedy algorithm to solve it. The extensive simulation shows that LAN Stream can outperform two classical P2P live streaming systems, in terms of the transmission performance and the reduction of cross-ISP traffic.