Research On P2p Network Based On Cayley Graph And Related Technology

In recent years, Peer-to-Peer file-sharing technology has been applied to many areas. Resource locating and grouping become especially important as the shared files and the need of cluster increase dramatically. Therefore, the research work of this dissertation is focused on how to make P2P network support efficient resource locating and grouping scheme. The research includes P2P overlay topology, grouping and grouped media streaming application.This paper presents a extend butterfly method to design and analyze a novel structured P2P network named BuNet based Cayley Graph. BuNet has many excellent properties such as small routing table, short query path and high robustness. Moreover, it is suitable for providing file browsing service, while the recent structured P2P network can hardly support this service because of omitting the connection of the peers that has the same interests. BuNet can achieve the theoretical lower bound of log(n) on node routing table size, log(n)/loglog(n) on network diameter. Compared to existing DHT-based schemes with similar network size, BuNet maintains the better level of robustness. The performance is evaluated using both analysis and simulations.BuNet network support grouping. Based on BuNet topology, a design approach of grouping is presented and a protocol on Gossip within groups is designed. In Gossip Protocol, nodes are connected by random method. But the nodes with local knowledge in ex-SCAMP gossip-based protocol can build global random graph. The ex-SCAMP protocol has the features of simple implement, good scalability and high fault tolerance. In addition, the problem of grouped live media streaming is tried to solve in this paper. Based on ex-SCAMP protocol, Coolstream media streaming system is built. Simulation results show that streaming continuity index of the system is good, and the proformance of ex-SCAMP protocol is better in network with big network, large traffic and high real time capability.