Gossip-Based Dynamical Adaptive Search Selection in Hybrid Peer-to-Peer Networks

In hybrid Peer-to-Peer (P2P) networks, the decision of whether to use flooding or DHT depends mainly on the popularity of desired data. Previous work either used only local information, or do not consider the dynamic factors of P2P systems. In this paper, an improved algorithm called dynamic adaptive hybrid based on Gossip (DAHG) is presented. In DAHG, a P2P ultrapeer tosses a coin when an end node joins or leaves the P2P networks, and uses a gossip-style algorithm to collect global statistics about document popularity. Therefore the dynamics of the resources is taken into consideration by DAHG, which it can be used to get the exact popularity of resources in a dynamic P2P network. Simulation shows that DAHG outperforms existing approaches and also scales well.