Research On Unstructured P2P Searching Technology Based On Ant Colony Algorithm

With the rapid development of computer network technology in the recent years,Internet plays an increasingly essential role in people’s life, and becomes an importantsource of access to information resources. Nevertheless, with Web usersproliferating, content of network information growing explosively, the traditionalserver as the core of the growing disadvantages of C/S model, P2P network is widelyused for its unique advantages, and becomes a research hotspot.The P2P network is also called peer-to-peer network, each node is equal, both aserver and client, data transmission between nodes no longer rely on a central server,therefore the interactive network is more direct and easily. However, join or leave ofthe dynamic node in unstructured P2P network changes resources in the network.How to find and locate information resources in large-scale and distributed P2Pnetworks lacking of centralism is a difficult issue for sharing P2P resources. Existingsearch mechanisms are mainly by means of flooding and K-Random walks, theblindness of flooding explorer resulted by the query information when transmittingthe node, the redundant messages in the network become more and more, K-RandomWalks randomly transmits the query message to K neighbor nodes, reducing theamount of message forwarding but the search results are inferior to flooding explorer,therefore, in unstructured P2P network, resource search mechanism needs furtherstudy.Ant colony algorithm is a new optimizing algorithm which successfully solvesthe TSP problem, by means of the basic characteristics of ant colony algorithm, thisthesis puts forward an improved ant colony search algorithm based on the existingresearch, the main work is as follows:1. Applying user search resource similarity and ant colony foraging behavior inunstructured P2P network, the thesis presents a search algorithm based on animproved ant colony algorithm in unstructured P2P cyber source, algorithm is dividedinto local resources search algorithm and network routing algorithm. The local searchalgorithm based on semantic similarity between searching keywords and documentresources realize the semantic query function. In the network routing algorithm, thethesis proposes a new concept of "interest pheromone", the users can select the search path according to the interest in pheromone and choice of transition probability tocalculate the preference information,.2. Interest pheromone is constituted by resource matching rate and response rate.Resource matching rate reflects the ability of contribution of nodes on the keyword, ifthe value is higher, the contribution to the pheromone concentration is greater.Response rates reflect the path length and the search time node; if the search path isshorter, response time will be shorter, the response rate of node, thus the pheromoneconcentration is higher. Both of them make pheromone can effectively guide thequery forwarding direction message forwarding, providing basis for the next searchants.3.The preference is constituted by node correlation and the number ofcommunications. Node correlation reveals the keyword and node similarity, thegreater the similarity is, the higher the probability of choosing nodes is. Through thevariable settings, the random initial node selection can effectively be avoided. Thenumber of communications can reflect the stability of the node, if the times of thenode’s communication is higher, the node will be more stable, therefore the variablesettings can effectively avoid the access node failure, improve the efficiency ofsearch.4. Based on the PeerSim simulation platform, through the experimental method,the thesis compares the unstructured P2P cyber source in this algorithm and classicalsearch algorithms--flooding algorithm (Flooding) and K-random walks (K-RandomWalks) algorithm, the results show that the algorithm can reduce the number ofredundant messages in a certain extent, and speed up the searching response time,improve the efficiency of search.