An interval-based approach to model input uncertainty in M/M/1 simulation

Abstract

This paper presents an interval-based discrete-event simulation (IBS) mechanism to help improve the robustness of simulations that incorporate input uncertainty. This proposed simulation mechanism is based on imprecise probability and models the parameter and model-form uncertainty without the need of sensitivity analysis in traditional simulation practice. The imprecise probabilistic measure given in an interval form is used as the input of simulation in our proposed IBS mechanism, where the statistical distribution parameters have interval values. A parameterization technique for interval probability distributions is proposed in this paper. An interval random variate generation method is used to run the simulation. Uncertainty propagation is achieved by applying Kaucher interval arithmetic. The proposed mechanism is illustrated with an M/M/1 queueing system.