Staffing to Stabilize Blocking in Loss Models with Non-markovian Arrivals

Previous work has shown that (i) it is not possible to find a time-varying staffing policy (number of servers) to stabilize blocking probabilities in a multi-server loss model with flexible staffing and a nonhomogeneous Poisson arrival process to the same extent as in a corresponding delay model, because the blocking probabilities necessarily change dramatically after each staffing change, but nevertheless (ii) a variant of the established modified-offered-load staffing algorithm used for delay models again performs well for loss models if we either randomize the times of staffing changes or average the blocking probabilities over a suitably small time interval. In this paper we use simulation to show that (i) these conclusions still hold for the more general multi-server loss model having a non-Poisson arrival process with a time-varying arrival rate and (ii) the required staffing can be quite different with non-Poisson arrivals.