An Algorithmic Approach for Multiserver Retrial Queues with Two Customers Classes and Non-preemptive Priority

Abstract

Retrial queueing models with multiple servers and two classes of customers arise in various practical computer and telecommunication systems. The consideration of retrials (or repeated attempts) introduces analytical difficulties and most of works consider either models with preemptive priority or non-preemptive priority in the single server case. This paper aims to propose a recursive algorithmic approach for the performance analysis of a multiserver retrial queue with non-preemptive priority and two customers classes: ordinary customers whose access to the service depends on the number of available servers and who join the orbit when blocked; and impatient priority customers who have access to all servers and are lost when no server is available. In addition, we develop the formula of the main stationary performance measures. Through numerical examples, we study the effect of the system parameters on the blocking probability for ordinary customers and the loss probability for priority customers.