Research Of Complex Networks Evolving Modeling And Topology Optimization

The study of complex networks is crucial for understanding structure and behavior of complex systems. Recently, the complex network research has become a new field. In this thesis, we studied the evolving law of complex networks and topology architecture optimization according to real networks properties. In addition, one of properties of complex networks, the vulnerability of scale-free has also received attention of researcher. We proposed that the hubs forming are controlled and topology structures are altered, which can increase the robustness of networks.The first part of this thesis investigated the evolving model of complex networks; this part analyzed the shorting of scale-free model. The preference attachment of nodes is only a possible case, which doesn't have general representation. The two evolving model are proposed in this part. One is based on self-similarity, the other is based on that the different nodes would have different attractive factor, which will affect the networks evolved. In self-similarity model, the nodes that have commonness would be connected, which the local network is similar to local network, and local network is similar to the whole network. We used volume dimension to calculate the dimension of network. Although volume dimension can calculate the dimension of network, about how many nodes in boxes, it can not calculate precisely. To calculate dimension objectly, we used information dimension to forecast the self-similarity of networks.In addition, many complex networks evolving model ignore some factors, for example, the attractive of nodes. In fact, the number of links of nodes not always changes with time. Some old nodes can acquire great links, because they enter network earlier. But some nodes enter network later, they also can acquire great number of links. For instance, some novel webs and can have great links in short time. Hence, we proposed a new evolving model for the recognition of attractive factors.The second part researches about topology structure optimization of networks. In this part, two strategies are proposed to avoid hubs formed, one is hierarchy structure; and the other is super-cube structure. In hierarchy, if some nodes reached a certain valve, the virtual nodes would be setup to form hierarchy structures. In super-cube, if some nodes reached certain valve values, the virtual nodes also would be setup to form three dimension cubes. The hierarchy of nodes has better agility and fault tolerance and the super-cube structure has fault reconfiguration compare with hubs. Both have higher robustness and steadiness shown by simulation.