Research on Peer Selection in Peer-to-Peer Networks using Ant Colony Optimization

This paper presents a novel peer selection model named ant peer selection (AntPS) to prevent selecting malicious or selfish peers as service providers in peer-to-peer (P2P) systems. We created two pheromone tables named resource similarity pheromone table and trust similarity pheromone table for every peer in the system. When a peer produces some request messages, a new ant called forward ant (fant) is created at the same time. It is the fant´s task to select which peer is the most suitable server based on the pheromone remained on the path. A new ant called backward ant (bant) will be created when fant finds some useful resources. Bant will update all the two pheromone tables on its way back. We simulated the new peer selection mechanism on the query cycle simulator. After the simulation, we made a comparison among our model and three other known models, i.e., flooding, top-k and eigentrust. From three aspects of the comparison, we made a conclusion that our model seems better than the others, and the performance is excellent.