Customer-oriented finite perturbation analysis for queueing networks (Q1581035): Difference between revisions
From MaRDI portal
Changed an Item |
Set OpenAlex properties. |
||
(One intermediate revision by one other user not shown) | |||
Property / MaRDI profile type | |||
Property / MaRDI profile type: MaRDI publication profile / rank | |||
Normal rank | |||
Property / full work available at URL | |||
Property / full work available at URL: https://doi.org/10.1023/a:1008345726373 / rank | |||
Normal rank | |||
Property / OpenAlex ID | |||
Property / OpenAlex ID: W2090128718 / rank | |||
Normal rank |
Latest revision as of 09:53, 30 July 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Customer-oriented finite perturbation analysis for queueing networks |
scientific article |
Statements
Customer-oriented finite perturbation analysis for queueing networks (English)
0 references
6 December 2000
0 references
This concerns a FIFO multiserver queueing network with blocking. An appropriate point of view allows a sample path to be completely defined by a sequence of departure times of individual customers from the various stations (the outside world is modelled as a station). A recursion, called the sample path formula, is derived for the successive departure times defining a path. Different choices of service time distributions and buffer sizes are of iterest. These choices are modelled such as to correspond to the different elements of a parameter space. Suppose a sample path, i.e. a sequence of departure times, is generated by simulation under a given parameter. Then the sample path formula allows computing simultaneously an associated sequence of departure times corresponding to a perturbation of the given parameter. This is shown to be valid for certain types of perturbations and under additional assumptions for the queueing network, some of which are rather restrictive. Thus one simulation only is needed for producing a set of sample paths for different parameters. This saves work when estimating performance measures or looking for optimal parameter values.
0 references
queueing networks
0 references
perturbation analysis
0 references