FREQUENTIST INTERVAL INFERENCE FOR TRAFFIC INTENSITY IN SINGLE-SERVER MARKOVIAN QUEUES

Queueing theory analyzes systems where the service capacity is insufficient to meet demand immediately. One of the queueing models of interest is the single-server Markovian queue (M/M/1). The traffic intensity is a parameter that completely defines the stochastic behavior of M/M/1 queues. This paper focuses on the interval estimation of the traffic intensity for M/M/1 queues by analytical and numerical methods. In particular, asymptotic and bootstrap confidence intervals were developed and compared. It was observed that the performance of the bootstrap confidence intervals is superior to the asymptotic confidence interval, but there is no performance gain when using parametric algorithms, compared with the non-parametric algorithms analyzed. Nevertheless, the asymptotic confidence interval may be favorable for larger samples, due to its computational efficiency.