Research Of Complex Network Evolving Model And Self-similar Network Traffic

In recent years, with further study of complex networks, it was found that more and more real world systems,such as research cooperation networks of social network, the power grid of technology networks, Internet and World Wide Web, can be described by complex network. However, how to properly summarize and abstract these complex networks to reconstruct the evolution of the real networks has gradually become the key of researching complex network. Meanwhile, in the application of complex network research, it has been a trend that the theoretical knowledge of complex networks combines with the knowledge in other fields. Motivated by this, we construct the network traffic model to predict the actual traffic data, by the self-similarity of network traffic.The main work of this thesis is as follows:1. Based on the BBV scale-free model, the Directed Model for Weighted Networks Based on Limited Node Strength is introduced. Our experiment shows that the value of out-strength and in-strength distribution and in-degree and out-degree of nodes obey distribution of the exponent of power-law.2. Through analyzing the traffic of a campus network, we found the exponent of power-law can be adjusted between 2 to 3. Therefore, the presented model can be applied in the campus network traffic.3. First,the average of daily outflow and inflow of the core node in campus network is handled with the discrete wavelet transform. Then, the parameters of Auto-Regressive and Moving Average Model(ARMA) is predicted with the damping least squares. With these work, we present a campus network traffic model. Finally, we use this model to research the security event location of the self-similar network traffic with the concept of degree in complex network theoretical knowledge.