Performance prediction and ANFIS computing for unreliable retrial queue with delayed repair under modified vacation policy

Abstract

This paper deals with the analysis of MX/G/1 retrial queue with impatient customers, modified vacation policy and Bernoulli feedback. When the incoming customer finds the server busy, on vacation or in the state of breakdown, he joins the virtual queue called retrial orbit, otherwise the service is provided to the customer who is at the head of the queue. The service is provided in l(0 ≤ i ≤ l) phases where first is compulsory service and remaining services are optional. When the system becomes empty, server leaves for the vacation of arbitrary length and can take at most J number of vacations. When server comes back from the vacation and finds at least one customer in the queue, he starts providing service to the customer. Supplementary variable technique (SVT) and probability generating function (PGF) method is used to derive the system size distribution and other performance indices. We have also approximated the analytical results using adaptive neuro-fuzzy interface system (ANFIS) soft computing technique, which can identify parameters using supervised learning methods.