The proposed research is concerned with both optimal design and optimal control of networks of queues. A major theme is the development of models for systems that do not meet the criteria for product-form probabilities, in particular, for systems in which buffer capacities are finite and/or service-time distributions are neither exponentially distributed nor independent from one queue to another. Research on optimal design and control of networks of queues is in its infancy. The analytical and numerical difficulties are formidable. Building on the Principal Investigators previous research, a primary focus of this work will be on efficient numerical computations, approximate solutions, and the evaluation of sub-optimal design and control schemes of a simple form. Queueing networks exist in manufacturing, computer/communications, traffic flow, and multi-level logistics systems. This research relates directly to the design and control of such systems.
