Asynchronous Transfer Mode(ATM) switching networks should support a large variety of services with different traffic characteristics and quality of services. However, it is very difficult to control traffic requirements by conventional mechanisms in this complex traffic environment. As an alternative approach, a multi-layer perceptron neural network model is proposed in this study as an intelligent control mechanism like a traffic control policeman" in order to perform ATM connection admission control. The burstiness, the number of connected calls, and the utilization are input variables to neural nodes and the cell loss rate is assigned to output variables. A learning mechanism based on "quantized" data table is also proposed. The proposed neural control model is analyzed by simulations in homogeneous and heterogeneous traffic environments and this simulation result shows the effectiveness of the control mechanism, compared with the results of analytical method and Hiramatsu's neural control model. This study can be further extended to applying the neural controls to other ATM traffic control mechanisms."
