Overlay Optimization for Mesh-based Peer-to-peer Streaming Applications

Mesh-based peer-to-peer (P2P) streaming has proven to be a robust and scalable way for delivering live video streams over the Internet. In this paper, we investigate the overlay optimization problem for mesh-based peer-to-peer streaming. By modeling the peer-to-peer overlay as a directed graph, we propose a new overlay optimization method that imposes a strict partial order on the P2P overlay. The partial order can be defined according to different capability metrics of the heterogeneous peers. We have also designed overlay construction and optimization protocols that can promote large capability nodes to higher levels while keeping the overlay robust and efficient. Experimental results show that our optimization method can efficiently promote the high performance peers and improve efficiency in the P2P overlays. Compared with the non-optimized case, the P2P streaming applications that use our method have reduced average latency by 25-40% and increased acceptance rate up to 125%.