\(\mathrm{G}/\mathrm{GI}/N (+ \mathrm{GI})\) queues with service interruptions in the Halfin-Whitt regime (Q261545)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | \(\mathrm{G}/\mathrm{GI}/N (+ \mathrm{GI})\) queues with service interruptions in the Halfin-Whitt regime |
scientific article |
Statements
\(\mathrm{G}/\mathrm{GI}/N (+ \mathrm{GI})\) queues with service interruptions in the Halfin-Whitt regime (English)
0 references
24 March 2016
0 references
A standard \(\mathrm{G}/\mathrm{GI}/N\) multiserver queueing system under non-idling FCFS is subject to interruptions due to an external alternating renewal environment which generates up-down cycles for the complete system, i.\,e., servers break down concurrently and are repaired after the same time interval. The interruption of the service follows a preemptive-resume regime and customers are held in their service channel when interrupted. Customers waiting in line (not being assigned to a server) may abandon the queue due to expiring of their patience time (which is generally distributed), even during down times of the system. Whenever a customer enters a service channel he will no longer abandon even during down times. During down times, newly arriving customers are admitted to the system, and either send to a free server, if any, to wait there for service (without impatience behaviour) when the next up-period starts, or attend the queue possibly departing before their service starts due to impatience. Several variants of the model are studied in the Halfin-Whitt regime, where the arrival rate and the number of servers grow appropriately, while service time distributions remain unchanged. The up-times and their distributions are of the same order as the service times while the down times become asymptotically negligible. The results for the queueing processes are: functional weak law of large numbers and functional central limit theorems, and for the virtual waiting time process: functional central limit theorem. For technical reasons and from the structure of the model, the proofs of the convergence utilize the Skorokhod \(M_1\)-topology.
0 references
multiserver queues
0 references
service interruptions
0 references
impatient customers
0 references
Halfin-Whitt regime
0 references
functional central limit theorem
0 references
functional weak law of large numbers
0 references
Skorokhod \(M_1\)-topology
0 references
0 references
0 references
0 references
0 references