Self-learning Routing in Unstructured P2P Network

Since the appearance of Napster in 1999, peer-to-peer networks which don't distinguish clients with servers, have become one of the fastest growing and most popular Internet applications. Content location is a key function, but it presents a very difficult and challenging problem for large-scale peer-to-peer systems. A number of different methods are currently in use. However, new mechanism which gains high success rate with low search scope and retains the simplicity and robustness of Gnutella-like systems is still under research. In this paper, we introduce a new self-learning algorithmSLPS. SLPS learns interest similarity of nodes from history search results, and computes friend relations which can be used to locate content effectively. Simulation results show that, compared to the flood-based algorithm, SLPS improves query efficiency by up to ten times without a significant increases in load.