Research On Resource Searching Based On Ant Colony Algorithm In Unstructured P2P Networks

Peer to peer (P2P) technology effectively integrates resources of the Internet, provides a more effective way for people to obtain target resource on a large-scale information. The unstructured P2P network has the advantages of simple network topology, easy maintaining, and fuzzy queries supporting, but because of its simple topology and loose structure, its resources search and location efficiency have been research concerns.The traditional unstructured P2P network resource-searching algorithm uses flooding to locate resources. But with the expansion of the network, this flooding method will produce a lot of redundant information, resulting in increased network load and low search efficiency. In order to solve the efficiency problems of unstructured P2P network resource-searching, a feasible method is to precisely select neighbor nodes to query and forward, thereby reducing the generation of redundant information and the searching time, and ultimately improving the efficiency of the searching algorithm. Ant colony algorithm derived from observation of the foraging behavior of ants, is a simulated evolutionary algorithm. It has a positive feedback mechanism and a good parallelism. This paper, which is on the basis of the traditional ant colony algorithm to solve the problem of network resource-searching of unstructured P2P network, gives an improved ant colony algorithm for unstructured P2P resource-searching algorithm.The traditional ant colony algorithm is only the guidable query and forwarding of one pheromone without considering the association between nodes, and in the latter part of the searching, the algorithm will only search a part of network resources due to the accumulation of pheromone, making the search prematurely stalled. To solve these problems, the paper follows two improvements about traditional ACO:(1) Setting node interest pheromone. On the basis of keyword pheromone, introducing the node interest pheromone. The former reflects the search behavior for a keyword of nodes; the latter reflects the communication frequency and resource similarity between nodes. Using these pheromones to guide query and forwarding considers the links between nodes, and through this positive feedback mechanism, it can accurately select neighbor nodes to query and forwarding, and reduce the generation of redundant information.(2) Adding the pseudo-random proportion rule, balancing the relationship between the ants using existing historical information and exploring the new path, avoiding the algorithm just searching in a part of the internet and plunging into prematurely stalled.The paper uses PeerSim network simulator to test the algorithm, and makes a comparison between algorithm with the traditional unstructured P2P resources-search algorithm, that is flooding and random walking algorithm. The experimental results show that the algorithm improves the success rate of resources, and in the case of the same searching results, reduces the spreading of redundant packet in the network.