The \(M/G/1+G\) queue revisited (Q543544): Difference between revisions

The authors consider the \(M/G/1\) queue with generally distributed impatience times, where patience refers to the waiting time, and distinguish between customers balking and reneging. In the balking case, an arriving customer knows in advance the evolution of the system, so that he never enters the system if he is to run out of patience; while in the reneging case, a customer always enters the system and may leave it either because he has ran out of patience or because he has completed his service, whichever occurs first. First, the authors briefly review the steady state workload distribution, which equals the steady state waiting time distribution. Next, they obtain the number of customers in the system by using a limiting argument that permits them to adequately adapt the methodology of the ordinary \(M/G/1\) (FCFS) queue of exploiting the relation between the sojourn time and the number of customers in the system. They continue their analysis by proving an elegant closed form expression for the tail distribution of the steady state maximum workload during a busy period. They conclude their analysis by presenting an iterative scheme for the calculation of the busy period LST demonstrating the similarities between this result and the functional equation for the busy period LST of the ordinary \(M/G/1\) (FCFS) queue. Finally, it should be pointed out that the authors also treat the analogous \(M/G/1\) model in which customers leave the system when their virtual sojourn time, i.e., the amount of work seen upon arrival plus the amount of work they contribute to the system, is larger than a certain random patience time.