Model Predictive Control for the acquisition queue and related queueing networks

Model Predictive Control (MPC) is a well established method in control theory and engineering practice. It is often the method of choice for systems that need to be controlled in view of constraints. The main idea of MPC is to solve an optimization problem over a given time horizon at each control epoch, and to use the obtained solution for controlling the system until the next control epoch. In many cases this optimization problem can be formulated as a tractable quadratic programming problem. In this paper we apply MPC to discrete-time queueing networks that incorporate multiple product routes, delays, self-generated arrivals and multi-job processing. We compare our MPC-based controller to simple threshold policies and show improved performance. We also mention some open theoretical problems related to this control method.