An experimental evaluation of the matrix-geometric method for theGI/PH/1 queue
- 1 January 1989
- journal article
- research article
- Published by Taylor & Francis in Communications in Statistics. Stochastic Models
- Vol. 5 (4) , 629-667
- https://doi.org/10.1080/15326348908807128
Abstract
A GI/PH/1 queue is a single server queueing model with general interarrival time distribution and phase type service time distribution. It provides a unified framework to include a large number of special cases of single server models that are commonly used. A class of algorithms for computing the performance measures for such queues goes under the name “ matrix-geometric solution ” and has been implemented as a FORTRAN package by V. Ramaswami. Reported here are the results of a numerical experiment with that package. Using a carefully selected set of examples, we address many issues related to the computational complexity and numerical accuracy of the matrix-geometric method. Among other things, some insights into modeling service times using two moment approximations are also obtained. The examples reflect a wide variety of characteristics in the interarrival and service time distributions and may therefore be of independent interest as possible test problems to evaluate the performance of algorithms for single server queuesKeywords
This publication has 5 references indexed in Scilit:
- Moments Of the stationary waiting time in the Gi/Ph/1 queueJournal of Applied Probability, 1988
- The caudal characteristic curve of queuesAdvances in Applied Probability, 1986
- Stationary waiting time distribution in queues with phase type service and in quasi-birth-and-death processesCommunications in Statistics. Stochastic Models, 1985
- OpsearchCommunications in Statistics. Stochastic Models, 1985
- Markov chains with applications in queueing theory, which have a matrix-geometric invariant probability vectorAdvances in Applied Probability, 1978