Research On Resource Discovery Mechanism In Peer-to-Peer Network

Peer-to-peer network is a new sharing mode different from the traditional centralized C/S,whose whole network nodes’interactivity and equality have winned more and more researchers’eyes. According to the structure types, peer-to-peer network can be divided into structured modeand unstructured mode, structured resource discovery locates the target resource according to themapping of the DHT table, but the factors such as the algorithm defects of setting prior naturalseat and the differences of computer capacity are bound to its large-scale application, relatively,unstructured model describing the peer’s joining and leaving in natural state is more real, but theresource discovery is often completed through request message’relay between the neighbornodes, sometimes without the aid of any historical information, its discovery is almost blind andrandom, which is easy to aggravate the burden of network traffic, so how to improve the resourcediscovery mechanism in distributed and charming peer-to-peer network ensuring rapidpositioning likable resources and effective saving network bandwidth has became particularlyimportant.Based on the overall considerations of drawbacks of too many redundant messages and lowdiscovery efficiency lying in unstructured flooding, this paper has drawn on the the smart antcolony technology and the traditional optimization ideas in this field and has made thecorresponding improvement in unstructured network resource discovery mechanism to betterbalance the efficiency of resource discovery and congestion problems. Specifically, the paper hasmainly done the following work:1. Theoretical background knowledge of peer-to-peer network has been comprehensivelyexpounded.The paper firstly has described the background cause of choosing this topic, as well as theresearch progress on the subject at home and abroad, has made a depth introduction in the orderof the origin, development, and application of peer-to-peer network, and has classified accordingto topology and made an comparative summary of the current P2P system of high popularitydegree and its idea on resource discovery, and then has mainly studied the classic unstructuredpeer resource discovery algorithm, and has made comparative analysis of the pros and cons ofeach algorithm.2. An efficient and low redundant resource discovery mechanism in unstructuredpeer-to-peer network has been put forward to solve the inadequate of too many redundant messages and low discovery efficiency.On basis of the detailed study of the previous unilateral improvement ideas and modernintelligent ant colony technology and inspired by the both basic theory, the nonstructured sourcediscovery mechanism has been improved in three aspects:Firstly,the selection mechanism of thenext station has been set to exclude the neighbors not meeting the conditions to avoid loopdiscovery according to the two order neighbor table and busy marking, and then calculate thenext station selection probability from the remaining neighbor set according to pheromonepositive feedback, which has made the discovery message’s forwarding more guidable, reducedthe unnecessary computational overhead, and improved the hit rate; Secondly, the congestionavoidance mechanism has been set, every node sets the threshold value according to its serviceability, when the request message’s number exceeds the threshold then informs his neighbor itsbusy mode, coordinates the local congestion;the combination of TTL and TS manages the overallregulation, TTL controls the hops depth of discovery message, and TS is responsible for theoverall time control to avoid the waste of network bandwidth; Finally, To ensure quick discovery,we also set up a rapid response table recording route node’s hit resource information, in order tofastly response to replace the target node when facing similar queries. The process of resourcediscovery is composed of two modules of local resource discovery and distributed resourcediscovery, when local semantics matching failed, then screens the neighbor set to compute nextstation selection probability, starts the distributed resource discovery, when the response messageis returned, downloads the proper resources in reference for each resource’s download asrecommendation.3. The experimental demonstration about the efficiency of the improved resource discoverymechanism has been conducted on PeerSim.To make better comparative analysis of the performance of the improved resource discoverymechanism, the according experiment has been simulated on the platform of PeerSim to verifyits high discovery efficiency and low redundancy rate, the experimental data has showed that theimproved resource discovery mechanism has behaved better than the traditional floodingdiscovery and K-random walker in the properties of average hit rate, the number of consumedmessages and response time and has proved to be an intelligent and efficient resource locationmechanism.