Abstract:
In this paper, we consider retrial queueing system. Two classes of customers come to the system according Poisson arrival processes. The first flow is a flow of priority customers, the second flow is a flow of non-priority customers. Service times have exponential distributions.
If a priority customer finds the server occupying by the customer of the same class, it goes to an orbit (orbit for priority customers) and makes a repeated attempt after a random delay. Interretrial times have exponential distributions. If an arrival priority customer finds the non-priority
customer on the server, it can interrupt its service and starts servicing itself. The preempted
customers moves into the orbit for non-priority customers. If a non-priority customer finds the
server occupying, it goes to an orbit (orbit for non-priority customers). The customers from the
orbit behave the same way. Customers are submitted to the system after successful completion
of service. We propose an asymptotic-diffusion analysis of the system. Probability distribution
of the number of customers in a non-priority orbit and in a priority orbit are obtained.