Cell-rate modelling for accelerated simulation of ATM at the burst level

Abstract

The paper presents a new method for the accelerated simulation of ATM networks. The method uses queue and traffic models that manipulate cell-rate information rather than the arrival times of cells. The cell-rate queueing model described in the paper has been validated by comparison with fluid-flow analysis and discrete time queueing analysis with batch geometric arrivals. Results show that the method accurately models the burst-scale component of ATM queueing. Comparisons with cell-by-cell simulation show speed increases of up to five orders of magnitude. This enables the low cell loss probabilities required of ATM networks to be measured within reasonable computing times. The method focuses processing effort on the traffic behaviour which dominates the cell loss, i.e. when the total input rate temporarily exceeds the queue service rate. It produces better speed gains at lower utilisations, and hence at lower cell loss probabilities.