Optimising the service capacity by using a speed-up simulation

In this work we provide an effective method based on a speed-up simulation+optimisation tool that allows us to optimise the capacity of a queue model system subject to the completion of a Quality of Service (QoS) criterion. This criterion is expressed in terms of a very small probability of losing a customer due to system overload. To evaluate the performance of the system, multiple simulations with different capacity levels have to be made. Because of the characteristics of these kinds of systems, any speed-up tool will be necessary in order to keep the computational time reasonably bounded. Our tool uses a speed-up technique, called RESTART, once and it does not need another entire execution of the method when a new capacity level is considered. Our tool only requires a stage (partial execution) of the RESTART method in order to update the probabilities when the capacity of the system is changed, so this involves a considerable saving of computational time. [Submitted 6 November 2007; Revised 25 January 2008; Accepted 25 January 2008]