Approximate models for the study of nonstationary queues and their applications to communication networks

The authors present numerical-methods-based techniques to model multiclass Markovian traffic, with and without priority, in ATM (asynchronous transfer mode) networks. The use of these methods is validated, under stationary and nonstationary traffic patterns, with simulation. The models developed are quite accurate, and provide substantial saving in computational time over the simulations done to validate the model. An illustrative application of the models to an ATM network is presented.<<ETX>>