Evaluation of the performance of a multiphase queuing system with retrial orbit using neural network model

Multiphase queuing systems are suitable model for evaluating the performance of wireless networks with linear topology. Analytical studies of such queuing systems had been provided in previous researches. However, for a variant of multiphase queuing system with single common retrial orbit queue, exact analytical results are not available and simulation method is slow in the case of complex model. We investigate the possibility of using a neural network model to predict the performance of a multiphase queuing system with single common retrial orbit queue. Experiment results have shown that by choosing suitable neural network architecture for the corresponding queuing system model, the accuracy of neural network model was significantly improved.