Dht-based P2p Search Engine Research

In recent years, P2P network develops extremely fast. It uses the resources of the nodes at the edge of Internet to provide distributed computing power and storage capability. The development of P2P network changes the way of sharing files on Internet, which results in a vast application in computation distribution, work collaboration, research, file exchange and so on. However, how to find the nodes resources in a p2p network which has no central node is a problem. At present, there are three solutions: 1) Using hybrid-p2p networks which has a central node; 2) Using flooding method in pure-p2p; 3)Using DHT in structured P2P. Napster is representative software of hybrid-p2p, but the central node is a bottle-neck. Flooding is applied in Gnutella. It broadcasts data in the network. The overwhelming data resulted by the increasing network size will use up the bandwidth of the network. To avoid redundant messages produced by flooding, researchers put forward structured P2P network with a search algorithm based on DHT. DHT uses a distributed hash table to find resources, which is fast and has good reliability. Many DHT algorithms are introduced including Tapstry, Pastry, CAN, Kademlia, and Chord. Especially, Chord introduced by MIT shows good performance in networks in which nodes join and exit with high frequency.After the deep research on the p2p resources search algorithms, especially DHT based Chord algorithm and its efficiency, this paper proposes a new three-level Chord method of looking up and caching resources. The three-level Chord makes every node in the Chord network withhold more routing information, and the cache keeps recently accessed content. In this way, looking up messages can be sent to the target node faster. It improves the efficiency of the Chord algorithm.At last, this paper uses p2psim to simulate the new method. The simulation result proves the new method reduces the messages transmission in the network without sacrificing the Chord performance because the reduction of messages transmission leads to the reduction of looking up delay, which improves the resources searching efficiency. In conclusion, the new method has better performance and reliability.