9500471 Srinivasan This research is focused on the development of decomposition methods and queuing analysis of polling (i.e., cyclic-queuing) models. In a polling queuing system, a single serve attends to a system of multiple queues. The server travels from queue to queue in some prescribed manner to attend to the customers in the queues. In decomposition, the system model, which is usually large and complex, is broken down into simpler models. The simpler models are analyzed and subsequently linked together to obtain a solution to the larger model. The objectives of this research are to produce practical algorithms for the performance analysis of polling models, and to generate theoretical knowledge and insight about polling models, vacation models, and the phenomenon of decomposition. The methods to be used are the mathematical methods of queuing theory, and the tools and insights from vacation models and polling models with and without switchover times. The research will investigate whether decompositions, of various types, hold for various kinds of polling models. There are many real world systems in which a single server attends to arriving entities in a given queue in a noncontinuous basis. An example of such a system can be found in manufacturing where an operator attends to multiple machines. Material handling systems that employ mobile transporters are another example. The results obtained from this research will be of significant value to researchers and engineers concerned with the modeling and analysis of manufacturing systems, telecommunication systems, and other systems that provide services to a random demand.
