Sensitivity analysis of regenerative queuing models (Q1896700): Difference between revisions

Discrete event dynamical systems may be considered as a different view of using supplemented variables in standard queueing models. For applications these discrete event dynamical systems are models for discrete event simulation programs, e.g., of reliability of queueing systems. Instead of elaborating system properties by focussing on an underlying Markov structure of the processes describing their time evolution the authors evaluate the systems as regenerative processes driven by asynchronous sequences of events. This makes it possible to derive for sensitivity evaluation of performance measures in simulations mathematically rigorous statistical procedures. The aim is to derive sensitivity coefficients defined as (partial) derivatives (of higher order) of desired stationary characteristics of the systems (mean waiting times, loss probabilities, \(\dots\)) with respect to parameters of the distributions of the driving sequences. The approach via regenerative processes allows to weaken assumptions (i.i.d. sequences as driving processes) put on the system before.