Non-parametric estimation for the \(M/G/\infty\) queue (Q1300758)

The authors consider the problem of estimating the input rate \(\lambda\) and service-time distribution \(G\) in the \(M/G/ \infty\) queue from data on the queue-length process. Three problems are studied: parametric estimation based on count data (a number of customers in the system), nonparametric estimation based on count data and nonparametric estimation based on indicator data (observations only of the idle and busy periods). In each case, the asymptotic normality of estimators is proved.