Searching with Local Information in Complex Networks

Abstract

Searching in complex networks is different from random and regular networks for existing long range connections and hub nodes. So the research on structure and characters of networks will improve the search speed and lower the load of nodes. Though getting the shortest paths is the best choice, in a real network, it is impossible for a node to get global information. For example, there is not a node that has the whole network information in a peer-to-peer network. The shortest paths are available, but the cost, especially in a dynamical network, will be high. The paper first discusses the main characters of complex network and the existing searching strategies with local information, and then defines and analyzes maximum diffuse nodes. After evaluating the stability of maximum diffuse nodes in dynamical network, the paper designs searching strategy based on maximum diffuse principle. To validate the idea, we numerically simulate the most characteristic complex network model of Barabasi and Albert (BA model), and analyze the average path, the network average load and every node’s load in different initial parameter values. The result indicates the new algorithm is effective not only in finding average path but also in load balance.