A Note on the Benefits of Buffering

DOI10.1081/STM-120028390MaRDI QIDQ4451555zbMATH OpenFDO

Authors M. R. H. Mandjes

Publication date 26 February 2004

Published in Stochastic Models (Search for Journal in Brave)

Full work available at URL https://handle.uba.uva.nl/personal/pure/en/publications/a-note-on-the-benefits-of-buffering(02c22799-a43c-44a2-80f1-43b15e7eb146).html

zbMATH Keywords

decay rate functionGaussian sourcesinput correlation structurelarge deviation asymptoticsoverflow probability

Mathematics Subject Classification ID

Queueing theory (aspects of probability theory) (60K25)

Review text

The note deals with a queueing system fed by a large number \(n\) of Gaussian sources and emptied at constant service rate \(nc\). The large deviation asymptotic of the (tail) probability of the overflow of a buffer level \(b\) is considered. A duality relation between the shape of the rate function \(I\) and correlation structure of the input is obtained. A key notion is the most likely duration \(t(b)\) of a busy period preceding overflow over buffer level \(nb\) (time-scale overflow). The main result is that the decay rate function \(I(b)\) at some buffer size \(b\) is convex (concave) if and only if the sources exibit negative (positive) correlation on the time scale \(t(b)\). A few models (e.g. fBM, periodic arrivals, Ornstein-Uhlenbeck input) are discussed and numerical examples are also included.

Recommendations

• Large Deviations for Complex Buffer Architectures: The Short-Range Dependent Case
• Queueing Networks with Gaussian Inputs
• Overflow behavior in queues with many long-tailed inputs
• Buffer overflow asymptotics for a buffer handling many traffic sources
• Patterns of buffer overflow in a class of queues with long memory in the input stream

Cites work

• A new technique for analyzing large traffic systems
• Large deviations for Small buffers: An insensitivity result
• Large deviations, the shape of the loss curve, and economies of scale in large multiplexers
• Sample path large deviations for queues with many inputs

Cited in

(2)

• Long range correlations in network traffic: Implications for the utilisation of network resources
• Sample-path large deviations for tandem and priority queues with Gaussian inputs

This page was built for publication: A Note on the Benefits of Buffering

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4451555)