Research On P2P Search Algorithm & Improvement Of Chord

Since Napster appeared in 1998, during several years, Peer-to-Peer (P2P) technique has developed from simple file sharing to the domain of distributed compute, information searching and real time communication.Due to the inherent expansibility and reliability, P2P technique gradually changes existing network service which used C/S model as kernel. Specially the complete distributed structured P2P systems which are based on Distributed Hash Table (DHT) overlay can provide high searching success rate, and can limit the number of routing hops within effective bounds. Therefore, DHT networks, such as Chord, CAN, Pastry and Tapestry, have become researching hotspot.In current DHT network model, Chord has been applied widely for simple building, algorithm provable correctness and effective routing capability, but traditional Chord's routing capability cannot completely satisfy real-time demand, and doesn't consider the matching between upper topology and bottom physical network, thereby it cannot attain good physical delay.For improving Chord's routing capability, the paper present an algorithm, called Neighbors'Neighbors Chord (NN-Chord), base upon analyzing traditional algorithm, to improve Chord's logic routing capability. And, its core idea is using nodes'Finger Table,and get the information from the neighbors'Finger Table which are included in the local Finger Talbe, to make up a routing table possessed more neighbors in existed Chord networks. During resource searching, NN-Chord can attain the next hop much more approaching target resource, thus shorten route-lengths. Moreover, it only needs to increase little overhead for constructing and maintaining routing table.And, in order to resolve the problem of the matching between application layer's network topology and bottom physical network, the paper presented an Advanced NN-Chord (ANN-Chord) which can attain better physical delay by considering both logic distance between nodes and target resource, and actual distance among nodes during routing process.Simulation and analytic proved that the average searching routing complexity by NN-Chord presented in this paper is O (log 4N ), and that by Chord is O (log N ). Therefore, NN-Chord improved obviously on routing capability in contrast with Chord. And in ANN-Chord, it imported matching relationship between physical network and logical topology. Simulation indicated that ANN-Chord had better performance on routing delaying.