On the distribution of the successful and blocked events in the \(M/M/c\) retrial queue: a computational approach (Q2383907): Difference between revisions

A queuing system is considered in which customers are allowed to conduct retrials. The paper deals with the exponentially distributed inter-retrial times. The pool of retrial customers is considered as a hidden queue. New performance descriptors are introduced for such systems: the number of successful retrials during a busy period, the number of blocked retrials, the number of blocked primary arrivals, the number of successful primary arrivals. The system is described by a bidimensional Markov process \((C_t,N_t)\) where \(C_t\) is the number of busy servers (channels), \(N_t\) is the number of customers in the hidden queue. The model with infinite hidden queue is approximated by a model with finite retrial group. Generating functions and direct computation algorithms are proposed for the computation of moments and the probability mass function of the performance descriptors. Numerical examples are presented.