Quantitative stability estimates of the Erlang system under a perturbation of the input flow of customers (Q1088309)

In the present paper we give an algorithm for the computation of the stationary probabilities of the states of a system with losses, for which the input flow of customers is ''close'' in a certain sense to a Poisson flow. Namely, we consider the queueing system GI/G/n/0, i.e. on n servicing channels one has a recurrent flow of customers, defined by the distribution function F(x). In at the moment of the entrance of a current customer into the system one has already n customers, then the newly arrived customer leaves the system. The service times of the customers are random variables, independent of the flow and independent among themselves, with distribution function G(x) and finite mathematical expectation \(\tau\).