Based On Cellular Automata Self-regulating Network Model

With the deeper research and application of Internet, there appear some new complex phenomena, such as phase transition, power phenomenon, self-similarity phenomenon. For the convenience sake, the simulation of network has become an important way to test and research the traits of network. But some cost much. So people establish many network models to study the network, which becomes a new research.The theory of Cellular Automata was proposed for the discrete numerical value computations at the first. Now, the researches and applications of CA are deeper and larger. It shows that with its own characters and rules, CA can establish network models. This paper establishes a new one based on CA taking the NaSch model and Yuanjian's model for reference. And we name it Self-Adjusting Network Model based-on Cellular Automata, SAN-CA for short. The features of the SAN-CA are as follows: 1) based on CA. It takes the packets and cache as a whole to update according to the steps.2) self-adjusting. It can control the rate of packet transition with the mechanism of feedback to realize the self-adjusting.3) with the random parameters. It can simulate the disorder as a result of the incidental expenses of the real network. So, based on the SAN-CA, we give the steps of simulation and an algorithm of it. Then simulate it with MATLAB. The results show that the bandwidth and load influent the phases of network flow into free one and congestion. And we analyze the phase transition and the criterion of it. The results also prove that SAN-CA can simulate the network simply and availably. And it provides a good model for the further research of other traits as power phenomenon, self-similarity phenomenon and has a promising foreground.