Abstract
Numerical evaluation of waiting time distributions for M/G/1 systems is somewhat difficult. This paper examines a simple variation of the heavy traffic formula which may be useful at modest levels of traffic intensity. One can justify the heavy traffic approximation by expressing the Laplace transform of the service time distribution as a Maclaurin series and then truncating to three terms. The spectrum factorization and inversion leads in a straightforward fashion to the heavy traffic approximation. If one carries two additional terms from the Maclaurin series, the characteristic equation is a cubic with exactly one real negative root. This root provides an easy way to extend the heavy traffic formula to cases where the traffic is not so heavy. This paper studies the quality of this approximation and includes some numerical evaluation based on data actually encountered.
Similar content being viewed by others
References
V.E. Benes, On queueing with Poisson arrivals, Ann. Math. Statist. 28(1956)670.
A.A. Fredericks, A class of approximations for the waiting time distribution in a GI/G/1 queueing system, The Bell System Technical Journal 61, 3(1982)295.
J.F.C. Kingman, On queues in heavy traffic, J. Roy. Statist. Soc. B24(1962)383.
W.G. Marchal, An approximate formula for waiting time in single server queues, AIIE Trans. 8, 4(1976)473.
W.G. Marchal and Klungle, The delay distribution in queues with Poisson arrivals, ORSA/TIMS San Diego Meeting (1982).
W.L. Smith, On the distribution of queueing times, Proc. Cambridge Philosophical Society 49(1953)449.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Marchal, W.G. An empirical extension of the M/G/1 heavy traffic approximation. Ann Oper Res 8, 93–101 (1987). https://doi.org/10.1007/BF02187084
Issue Date:
DOI: https://doi.org/10.1007/BF02187084