Heterogeneous Search in Unstructured Peer-to-Peer Networks

Resource search or discovery is a fundamental issue in peer-to-peer (P2P) and grid studies.1 Search objects, or resources, can be cycles, storage spaces, files, services, addresses, and so on. In general, systems are employing three categories of P2P-network architectures to improve search performance: centralized (such as Napster, http://www.napster.com/), decentralized but structured (such as Chord, http://www.pdos.lcs.mit.edu/chord),2 and decentralized and unstructured (such as Gnutella, http://gnutella.wego.com/). Most popular P2P applications operate in the latter category, and grids are essentially P2P systems.3 Resource search technology in these complex, fully unstructured and decentralized P2P networks presents a great challenge to the further usability of P2P networks and grids.