Asymptotic sampling properties of effective bandwidth estimation for admission control

In measurement based admission control, measured traffic parameters are used determine the maximum number of connections which can be admitted to a resource within a given quality constraint. It has been pointed out that the certainty equivalent formulation, in which the measured parameters are assumed to be the true ones, can compromise admission control. This is because the measured parameters are themselves random quantities, and so contribute additional variability to the attained quality. This paper analyzes the asymptotic sampling properties of admission controllers that use effective bandwidth estimation in a large deviation setting. This analysis applies to both bufferless and buffered resources; in the many sources asymptotic and in the large buffer asymptotic