Resources NeuroSearch in Peer-to-Peer networks

Unstructured peer-to-peer (P2P) networks as Gnutella are dynamic, distributed systems without any centralizing point favoring failure tolerance and strength. However, resource search in these systems is an important problem. Gnutella´s breadth-first search algorithm is flooding-based and generates a large amount of traffic thus making scalability difficult. This paper proposes a new search algorithm in which nodes, assisted by their local neuronal networks, selectively send the query to the most appropriate subsets of neighbors only. Hence, Gnutella algorithm is significantly improved and provides a greater percentage of findings with less amount of traffic generated on P2P network.