Abstract:

An $M/M/1$ queue whose server rate depends on the state of some Markov process (jump process or a diffusion) is studied by means of perturbation analysis. A probabilistic approach is proposed to analyze the duration of busy periods as well as the mean number of customers.