Investigation of a cellular automaton model for computer network

We have studied the collective behavior of computer network nodes by using of a cellular automaton model. It is shown that when the load of network is constant, the probability distributions for throughputs and buffer content s of nodes are power-law in both space and time. And the power spectrum of the number of nodes bearing a constant part of the system load has the feat ures of a 1/f noise. It can be seen as yet another example of self-organize d criticality. Power-law decay in the distribution of buffer contents impli es that large congestion in network occurs with less probability. The tempo ral power-law distribution for throughput may become another explanation fo r the observed self-similarity in computer network traffic.