Steady-state GI/G/\(n\) queue in the Halfin-Whitt regime (Q389068): Difference between revisions
From MaRDI portal
Changed an Item |
Set profile property. |
||
Property / MaRDI profile type | |||
Property / MaRDI profile type: MaRDI publication profile / rank | |||
Normal rank |
Revision as of 00:09, 5 March 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Steady-state GI/G/\(n\) queue in the Halfin-Whitt regime |
scientific article |
Statements
Steady-state GI/G/\(n\) queue in the Halfin-Whitt regime (English)
0 references
17 January 2014
0 references
The paper studies the first-come-first-served (FCFS) GI/GI/\(n\) queueing system in the Halfin-Whitt regime, see [\textit{S. Halfin} and \textit{W. Whitt}, Oper. Res. 29, 567--588 (1981; Zbl 0455.60079)]. The paper proves the tightness of the sequence of steady state queue-length distributions, normalized by the \(\sqrt{n}\), as \(n\to\infty\). The paper derives an upper bound of the large deviation exponent of the limiting steady state queue-length matching. Earlier, this upper bound was conjectured in [\textit{D. Gamarnik} and \textit{P. Momčilović}, Adv. Appl. Probab. 40, No. 2, 548--577 (2008; Zbl 1148.60070)]. Under the assumption that the arrival process is Poisson, a matching lower bound is proved as well. The paper derives new and simple bounds on the basis of new techniques for the FCFS GI/GI/\(n\) queueing system. The bounds are of a structural nature and hold for all \(n\) and all times \(t\geq 0\). The closed form representations for these bounds are intuitively explained as the suprema of certain natural processes weakly converging to Gaussian processes. The aforementioned weak convergence was earlier studied in [\textit{J. Reed}, Ann. Appl. Probab. 19, No. 6, 2211--2269 (2009; Zbl 1181.60137)]. The methodology of the present paper establishes the first non-trivial bound for the aforementioned weak limit process in [Reed, loc. cit.].
0 references
many-server queues
0 references
large deviations
0 references
weak convergence
0 references
Gaussian process
0 references
stochastic comparison
0 references