BAYESIAN INFERENCE AND PREDICTION FOR THE GI/M/1 QUEUEING SYSTEM

This article undertake Bayesian inference and prediction for GI/M/1 queueing systems. A semiparametric model based on mixtures of Erlang distributions is considered to model the general interarrival time distribution. Given arrival and service data, a Bayesian procedure based on birth-death Markov Chain Monte Carlo methods is proposed. An estimation of the system parameters and predictive distributions of measures such as the stationary system size and waiting time is given