A desirability function-based approach for optimising generalised queuing networks

Abstract

The analysis and optimization of queuing networks (QNs) has received increasing attention over the last few years in many manufacturing and service systems such as information technology, transportation, supply chain and traffic control. Robustness is another important concept which refers to the low sensitivity of objective functions to uncontrollable (noise) variables. Most of the existing research works in literature are unable to deal with multi-objective optimization of queuing networks and ignore robustness of objective functions. In this paper an interactive robust multi-objective optimization approach based on desirability function is presented to optimize performance of generalized queuing networks (GQNs). The proposed approach includes computational and decision making phases. The computational phase consists of employing a proper design of experiment and response surface methodology to obtain a primarily design of GQNs. In the second phase, decision maker after evaluating current solution either accepts it or by resetting parameters tries toward optimal design according to his/her preferences. To illustrate the applicability of the suggested approach, a computational analysis is performed on a pizza shop system.