Likelihood-Free Parameter Estimation for Dynamic Queueing Networks
Many complex real-world systems such as airport terminals, manufacturing processes and hospitals are modelled with dynamic queueing networks (DQNs). To estimate parameters, restrictive assumptions are usually placed on these models. For instance arrival and service distributions are assumed to be time-invariant, which allows for likelihood-based parameter estimation, but realistic DQNs often violate this assumption. We consider the problem of using data to estimate the parameters of a DQN. We combine computationally efficient simulation of DQNs with approximate Bayesian computation and an estimator for maximum mean discrepancy. Forecasts are made which account for parameter uncertainty. We motivate and demonstrate this work with an example of an international airport passenger terminal.
READ FULL TEXT