2010•International Journal of Communication SystemsRequires access

Performance modeling of DiffServ meter/markers

Rafał Stankiewicz, Andrzej Jajszczyk

Open publisher page 4 citations

Abstract

Abstract The paper presents analytical models of the following meter/markers used as building blocks of DiffServ architecture supporting Assured Forwarding: Single Rate Three Color Marker (srTCM), Two Rate Three Color Marker (trTCM), and Time Sliding Window Three Color Marker (TSW3CM). The model of the latter involves finding the distribution of the TSW‐averaged traffic rate approximated by the rate estimator. A respective probability density function that, if integrated in appropriate bounds, allows finding packet marking probabilities is proposed. Models for srTCM and trTCM, which use the token bucket mechanism, are based on the classical queuing theory. Namely, models based on the M/M/1/K and D/M/1/K queuing systems are considered. The models enable quick finding of meter/marker characteristics under a particular configuration and traffic volume. It can be also quickly checked how parameter manipulations or changes in the traffic rate affect the characteristics. The models are validated by simulations with two types of traffic: TCP/ftp and web‐like traffic. Copyright © 2010 John Wiley & Sons, Ltd.

About this research paper

What this paper is about

Abstract The paper presents analytical models of the following meter/markers used as building blocks of DiffServ architecture supporting Assured Forwarding: Single Rate Three Color Marker (srTCM), Two Rate Three Color Marker (trTCM), and Time Sliding Window Three Color Marker (TSW3CM). The model of the latter involves finding the distribution of the TSW‐averaged traffic rate approximated by the rate estimator. A respective probability density function that, if integrated in appropriate bounds, allows finding packet marking probabilities is proposed. Models for srTCM and trTCM, which use the token bucket mechanism, are based on the classical queuing theory. Namely, models based on the M/M/1/K and D/M/1/K queuing systems are considered. The models enable quick finding of meter/marker characteristics under a particular configuration and traffic volume. It can be also quickly checked how parameter manipulations or changes in the traffic rate affect the characteristics. The models are validated by simulations with two types of traffic: TCP/ftp and web‐like traffic. Copyright © 2010 John Wiley & Sons, Ltd.

Why it matters

OpenAlex reports 4 citations for this work. Citation counts describe recorded attention and do not establish research quality.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

Abstract The paper presents analytical models of the following meter/markers used as building blocks of DiffServ architecture supporting Assured Forwarding: Single Rate Three Color Marker (srTCM), Two Rate Three Color Marker (trTCM), and Time Sliding Window Three Color Marker (TSW3CM). The model of the latter involves finding the distribution of the TSW‐averaged traffic rate approximated by the rate estimator. A respective probability density function that, if integrated in appropriate bounds, allows finding packet marking probabilities is proposed. Models for srTCM and trTCM, which use the token bucket mechanism, are based on the classical queuing theory. Namely, models based on the M/M/1/K and D/M/1/K queuing systems are considered. The models enable quick finding of meter/marker characteristics under a particular configuration and traffic volume. It can be also quickly checked how parameter manipulations or changes in the traffic rate affect the characteristics. The models are validated by simulations with two types of traffic: TCP/ftp and web‐like traffic. Copyright © 2010 John Wiley & Sons, Ltd.

Key concepts: Token bucket, Computer science, Queueing theory, Network packet, Estimator, Sliding window protocol, Throughput, Real-time computing

Related papers

Back to paper searchBrowse research topicsOriginal source
Performance modeling of DiffServ meter/markers — Research Paper | ScholarLens