Perturbation Analysis: Basics for Deriving Higher Order Propagation Rules

Perturbation analysis of discrete event dynamic systems (DEDS) is a time domain based approach for the analysis and optimization (especially, the sensitivity analysis) of DEDS. It views a queueing network as a stochastic dynamical system evolving in time, and observes the sample realization of its trajectory. To this extent this is a similar viewpoint (and hence enjoys the same advantages) as the simulation approach to queueing systems. However, by observing a sample realization of the network trajectory, one uses analytic formulas, i.e., the so-called perturbation generation and propagation rules to derive answers to the question “What will happen if we repeat the sample trajectory exactly except for a (parameter) perturbation at some time t” The efficiency of this approach lies in the fact that one can answer a multitude of such what-if questions simultaneously while the sample trajectory is being observed. Thus compared with the brute force simulation study, perturbation analysis of DEDS has a computational advantage near by M: 1 where M is the number of what-if questions asked.