Research And Improvement On Mechanisms Of Searching Resources In Peer To Peer Network

Peer-to-Peer Network is one of the most popular researching fields of computer network technology. And P2P application has become the major application among all kinds of computer network applications in terms of traffic. Resources' sharing, among all kinds of P2P applications, is the most prevalent application. It is clear that to find efficient, reliable mechanism of searching resources with consuming less bandwidth and computing ability is a crucial issue for P2P networks. This thesis researches the popular models of P2P network and their mechanisms of searching resources in depth, with emphasis on Chord network and its mechanism and algorithms of searching resources.Chord network is a typical model of structured P2P network. It distributes index information of resources into peers based on the basic idea of distributed hash table. By using an amazing routing table for routing query messages, it has the characteristics such as reliable resources searching, scalability, less times for forwarding query messages and so on. However, there are some problems for the mechanism of searching resources in a Chord network, such as too much redundant information in the routing tables of peers, the deviation among needed hops for searching resources whose key values locate in different areas of identifier space is too large, some available routing information being not used and the difficulty for supporting complex querying, e.t.After analyzing the mechanism of searching resources in a Chord network, this thesis proposes a scheme to improve the mechanism. The scheme includes a mathematic model for adding routing information to the routing tables of peers. Meanwhile, it holds down the scale of routing tables by merging the items that have redundant information. In most cases, it makes the peers in a Chord network have actually even less items in the routing tables than they used to. Compared to the standard mechanism of searching resources in Chord model, it also has the advantage that the deviation of hops which are needed for routing query messages to find resources whose key values locate in different identifier spaces becomes smaller. Thus, it makes the performance of searching resources more stable, the average hops of routing query messages deminished. It also makes the routing tables of peers consumes less memory and the time for peers to route query messages get shorter.For the purpose of delving into the organization mechanism and certifying the performance of the improved measures, a Chord simulator is also developed by the author of this thesis. It is proven that the simulator can simulate the standard Chord model and the model proposed in this thesis correctly by testing.