A Gossip-based Adaptive and Reliable Multicast Model on Hybrid P2P Overlay

Online games are receiving more and more attention due to the potential market. But how to implement information dissemination reliably and efficiently in gaming platform along with large or larger size of game users is urgent to be solved. Peer-to-peer (P2P) technology becomes more and more popular because it can massively reduce loads on backbone of Internet, pushing intelligence to the edge of Internet. Nowadays, P2P is turning towards hybrid model which integrates advantages of central model with fully distributed model. Hybrid model is also called a semi-structured model which considers the difference among peers. Peers are often divided into two branches, superpeer group and leafpeer group. Superpeers are often to designate more important tasks. It’s widely applied in streaming, storage and IM etc. But how to disseminate messages on hybrid P2P reliably and efficiently is an open issue. Previous works often take tree-based multicast or flat probability model which cannot make a trade-off in reliability and efficiency. Doualcast, a probability information dissemination model on hybrid P2P system, is presented in this paper which considers ability among peers and can select suitable peer to forward messages, improving reliability of multicast. Simulations demonstrate DouLaCast functions better than flat gossip method.