Research Of Resources Searching In Unstructured P2P Networks Based On Ant Colony Optimization Algorithm

With the rapid development of Internet, various information resources is going through a rapid expansion too. How to help the users get resources which they are instrested in efficiently in the large-scale network, becomes a serious challenge which the Internet is facing now. Peer network (Peer-to-Peer, P2P) technology could provide fast, accurate inquiry service, which could be a good solution to the problem of large-scale information resource sharing. Due to the flexibility and the adaptability of the dynamic environment, unstructured P2P networks has been widely applied in the Internet. However, unstructured P2P networks adopt the flooding as the main search method, so there is a certain degree of blindness in course of query messages forwarding between nodes. Because of the redundant information, the search efficiency is relatively low., and network resources search of unstructured P2P still needs to be studied further.Ant colony algorithm is a novel optimization method basing on swarm intelligence, and it has successfully solved a series of combinatorial optimization problems. This paper bases on the study of search algorithm in unstructured P2P network, and puts forward a unstructured P2P network search algorithm basing on ant colony optimization by the introduction of the idea of it, to solve the problem of redundant messages, low searching efficiency and insufficient consideration of malicious nodes behavior in Unstructured P2P network resource search now. This algorithm simulats the foraging behavior of ants in order to searches the documents that users need. It adds the credibility of the node when setting pheromone, so that the node's pheromone could reflect the node's credibility either. It uses pheromone left by ants to guide forwarding of query messages, and forwards user's query request to the nodes which will most likely respond. By modifying the pheromone, It guides the ants behind to select forwarding nodes, and speed up convergence time by the use of positive feedback mechanism.In this paper, the author uses the OPNET network simulation tool for the modeling and simulation of the algorithm, and details the modeling process in three aspects:the network domain, domain and process node. By comparing the simulation results with Fooding and K-Random Walker, the author analyzes the performance of the algorithm in three aspects:the average hit rate, average response time and average bandwidth utilization. The analysis and comparison shows that the algorithm proposed in this paper does a good job in reducing the number of redundant messages in the network and speeding up the search response times, and makes the search more efficient.