A novel fast simulation method for ATM networks

A high-level abstracted hybrid analytical simulation model, based on intensive investigation of packet queuing behaviour, is presented in this paper. The approach we present here substantially speeds up the basic simulation of ATM networks. Comparing the results from the model against a testbed leads to some difference between the results of the model and those of the testbed. A neural network is applied to 'fine-tune' the model accordingly. The results indicate that the proposed simulation method achieves fast simulation and also remains accurate. This approach is particularly promising for use in the area of large-scale network designing and planning, where concern is more about the overall performance of the network than the detailed structure of a network node.