Likelihood ratio sensitivity analysis for class of queuing systems in transient analysis

Simulation is an essential tool for performance evaluation of many practical systems where planners typically want to know how the system will perform under various parameter settings. Since large-scale simulation may require great amounts of computer time and storage, appropriate statistical analysis can become quite costly. In this paper, we develop an interpolating technique as an effective tool for estimating system responses to parametric perturbations in simulation. We also analyze the usefulness of the continuous--time Markov chain framework to find the likelihood ratio (Radon- Nikodym derivative) for Markovian classes of queuing models. We provide numerical experiments that demonstrate how the interpolating technique significantly outperforms the likelihood ratio performance extrapolation technique in the context of the Markovian queuing models in transient analysis