Abstract:
The success of P2P file sharing system highly depends on the scalability and versatility of its search mechanism Existing structured P2P networks (such as CAN) supporting distributed Hash table (DHT) functionality are scalable but they can’t support partial match queries effectively On the opposite, unstructured P2P networks (such as Gnutella) rely on flooding for search, thus supporting partial match queries, but such flooding does not make the systems scalable A new architecture for P2P networks is proposed Based on content addressable networks (CAN), a hybrid scheme called pGroup for building semantic peer to peer networks (SPNs) is presented where nodes with the same classes of contents are connected to each other According to the different querying, corresponding search algorithms are also proposed As is shown by experiments, scalability and search efficiency are improved greatly compared with Gnutella