Resource discovery in locality-aware group-based semantic overlay of peer-to-peer networks

The resource discovery is a critical component of the P2P file-sharing network. However, because of the huge overhead of locating operation or management, neither the traditional methods, such as Flooding, EPS and Random Walks, provide high performance for this process, nor do the recent ones such as Chord, CAN, and so on. To conquer this problem, locality of the underlying network should be taken into account when constructing the P2P networks. It can shorten the length of routes in network layer and reduce the bandwidth consumed when locating the resource. Meanwhile, semantic overlay is another powerful way to organize the P2P nodes. In the semantic overlay, the nodes with semantically similar content are "clustered" together, which can facilitate the resource discovery. Based on these two characters, we propose a new architecture of resource discovery which incorporates the underlying locality into the semantic overlays using decentralized group concept. The query is processed in the group one by one and the major management operations are in the group. In this way, the globe status maintenance can be avoided. The mathematical analysis and simulation results also show that the performance of new mechanism has been enhanced largely, including average diameter, average management overhead, average searching overhead, and so on.