Robustness properties of perturbation analysis estimators for discrete event systems with unknown distributions

Sample-path-based stochastic gradient estimators for performance measures of discrete event systems rely on the assumption that a probability distribution of the random vector of interest (e.g. a service or interarrival time sequence) is given. The authors address the issue of dealing with unknown probability distributions and investigate the 'robustness' of such estimators with respect to possibly erroneous distribution choices. It is shown that infinitesimal perturbation analysis can be robust in this sense, and, in some cases, provides distribution-independent estimates. Comparisons with other gradient estimators are provided.<>