Research On Topology And Information Transmission Problems Based On Complex Networks

The studies on network dynamics can be divided into two groups: the evolution mechanisms of networks and the behavior on networks. According to the latest studies, many real networks such as Internet,WWW etc are not complete regular networks or random networks, on the contrary, they are complex networks that both have certainty and uncertainty. Therefore, those conclusions that based on random networks or regular networks may not apply to complex networks. In this paper, the studies mainly focus on the evolution mechanisms of networks and information transmission problems based on the latest research on complex networks1. A new evolving network model named"the information transmission model based on agent"is proposed. New evolution mechanisms are designed by summarizing the common topological properties of information networks and by using the methord of network dynamics theory for reference. Compared to other models, this model could not only incorporate small-world property, scale-free property and community property into the same framework but also control the evolution process of network by adjusting the parameters.2. SA algorithm is applied to find community structure of networks and a new algorithm that based on the roles of nodes (PANT) in complex networks is proposed for the same purpose. The basic idear of PANT algorithm is as follows: Firstly, stochastic algorithm is adopted for initial partition. Then, nodes are classified into four different types and the connections between different communities are classified into three types. According to different types of connctions, further partitions of the network are made until the algorithm reaches optimization. Experimental results show that this algorithm could find the community structure efficiently.3. How topological properties of complex networks such as small-world property, scale-free property and community property affect the transmission capacity of a network are studied. Experimental results show that (1) as the average degree of the a network increases, the transmisison capacity of a network increases either (2) when the average degree of a network keeps unchanged, as the number of long-range connections increases, the capacity of a network increases at first and then decreases when the quantity of long-range connections is much more than local conncetions.(3) when exponentialγincreases, the capacity decreases.(4) the stronger the modularity is , the lower efficiency the transmission capacity has.4. It is found thatω=-1 responds to the optimal local routing strategy when the capacity of each node is a constant andω=1 responds to the best local routing strategy when the capacity of each node is not a constant. In order to further maximize the capacity of the network, a glocal routing strategy based on"effective betweenness"is proposed. Experimental results show thatβ=1 responds to the best routing strategy and routing strategies based on effevtive betweenness run better than those routing algorithms based on the degree of nodes.