Simulation methodology - an introduction for queueing theorists (Q1116316)

The author gives many useful hints on the use of simulation in queueing theory. First, the problems of constructing simulations, the generation of arrival streams and service times, the fact that the uniform distributed variates obtained by an algorithm may not be sufficiently independent, are discussed. An analysis of output simulation sequences is made to answer questions about the equilibrium behavior of the system. For determinating the variability of output values a lot of techniques based on dependence of output is proposed; substituting correlations, time series methods, batching, standardized time series or a regenerative simulation. A special section is dedicated for finding some simulation tricks for variance reduction of output results. So, for queueing systems the author studies the following methods: antithetic variates, control variates, conditioning importance sampling and indirect estimation. Different techniques, as the principle of blocking, the factorial experimentation, the response surfaces method or considering the dependence between experiments, are also briefly presented for designing simulation experiments. A lot of actual references is given in this very well- informed paper.