SWS: Small World Based Search in Structured Peer-to-Peer Systems

The maintenance of overlay topologies in structured P2P (DHTs) will lead to great message overhead because of the sharp dynamic of the peers. And the DHTs are less usable because the strategy of DHT discards the useful application-specific information. Here we present a usable and easy to maintain search algorithm for structured P2P systems, Small World based Search (SWS). In SWS, the structure of semantic tree is introduced to organize the peers. The semantic distance of two peers is formed according to their relative position in semantic tree. The different distance between two peers will result in different connection probability, which lead to the sense of routing orientation. The mechanisms of "lazily-maintained" and "passively-modified" are used to decrease the overhead of overlay maintenance. The final simulation experiment shows, SWS can improve the usability if given the ontology definition of the semantic tree, and remarkably decrease the cost of the topology maintenance