EURO Working Group on Stochastic Modelling

FIRST MEETING: 19-21 APRIL 2006, AMSTERDAM

meeting 1
     program
     practical information
     register
     participants
meeting 2
meeting 3

Abstracts

Haralambie Leahu, Estimating the Effect of Uncertainties in Stochastic Models

When building a stochastic model there is often uncertainty about the exact underlying distributions. For example, while the type of distribution of the service times of a certain server in a queuing network may be known, the distributional parameters, such as the mean, may only be approximately known.
This paper establishes bounds on the effect of a perturbation of a distributional parameter, on performance measures stochastic models that depend on finitely many stochastic variables, such as the transient waiting time in a queuing network. The key observation is that, under quite general conditions, such bounds can be obtained in a simple way from the Lipschitz constant of the weak derivative of the probability distribution of the random variables.
Our results apply to the parameter-free setting as well. We derive upper bounds for the performance effect of replacing in a stochastic model a given probability measure by another one.