LOAD BALANCING IN POLLING SYSTEMS UNDER DIFFERENT POLICIES VIA SIMULATION OPTIMIZATION

Abstract

A polling system is a type of queueing system in which a single server serves many queues. In contrast to classical queueing systems, the server switches between queues and serves the customers under different routing and service policies. The problem addressed in this study is to define the load balancing problem in polling systems, where the balancing behaviour is dependent on server shifts, not the distribution of customers among queues. Although the load balancing problem in multi-server queuing systems is very common, modelling the load balancing problem in polling systems is the novelty of this study. Furthermore, not only the performance of the system has been analysed as in the previous simulation studies but also the balanced queues with optimal routing probabilities for multi-class queues under different routing and service policies are achieved by using Arena-OptQuest.