A Low-Diameter Unstructured P2P Overlay and Corresponding Search Algorithm

Abstract

Unstructured peer-to-peer networks are frequently used as the overlay in various middleware toolkits for emerging applications, from content distribution to query result caching to distributed collaboration. There are two main aspects in unstructured P2P system researches: (1) How to build a scalable and robust topology and (2) how to present a low cost and more efficient query algorithm. Though normally a search algorithm is independent of the topology, a good topology can greatly improve the performance of the search algorithm. However, most unstructured P2P systems can not combine the robustness with the efficiency and searching accuracy. The main contribution of the paper is to combine the researches on both topology and search algorithm. At first, a low-diameter unstructured P2P overlay called MPO with small world and power-law characteristics is presented. Detailed experiments show that current popular search algorithms can obtain better performances in MPO. Then, a new search algorithm based on the overlay is presented. We compared the new algorithm with popular search algorithms in the unstructured P2P systems. The results show that the algorithm is better than others in MPO according to several parameters. Thus, we draw a conclusion that the combination of the new search algorithm and the MPO has better performances than that of other search algorithm and other overlay.