THE PROBLEM OF GENERATING PSEUDORANDOM SEQUENCES OF COMPOSITE DISTRIBUTIONS FOR QS SIMULATION

Abstract

This article presents the results obtained on developing a pseudo-random sequence generator for simulation modeling of a QS with a hyper-Erlangian input distribution of the second order. In foreign and Russian-language scientific literature, including web resources, data on this domain were not found by the authors. It is known that this distribution law is the most general distribution of a continuous nonnegative random variable within a wide range of variation coefficients. The latter plays an important role in estimating a delay of requests in a queue in queuing systems, since an average delay of requests in a queue is related to the coefficients of variation of arrival and service intervals via the quadratic dependence. Using this higher order distribution is difficult to derive a solution for an average waiting time due to the increasing computational complexity. For the hyper-Erlangian distribution law of the second order, the authors previously obtained some numeric and analytic results based on the method of spectral solution to the Lindley's integral equation. The article presents the obtained algorithm and program on GPSS WORLD for simulating the functioning of a QS with a hyper-Erlangian input distribution. The adequacy of the obtained results was confirmed by comparing the simulation results with the results of numerical simulation in the Mathcad environment. The authors hope that the presented results will be in demand by specialists in the field of simulation modeling in the GPSS WORLD environment.