Tutorial: Micro and macro views of discrete-state Markov models and their application to efficient simulation with Phase-type distributions


In this tutorial we discuss methods for efficient system evaluation.

The tutorial is split into two parts. We first describe how the distinction of micro and macro views on CTMC models improves the efficiency of solution methods. The micro view on a CTMC considers the small CTMCs that describe the behaviour of system components, while the macro view concerns the behaviour of the system as a whole. For instance, when modelling a distributed system consisting of a large number of nodes, the micro view gives the CTMCs that describe the behaviour of individual nodes, while the macro view gives the response-time or reliability of the system.

In the second part of the tutorial we give a hands-on introduction to using phase-type distributions in modelling system behaviour. Using examples from practice, we show how phase-type distributions can be used to fit arbitrary measurement data and describe their use in efficient and accurate simulation.

The full description of the tutorial is available as a PDF file here.

Intended Audience

The intended audience of the tutorial are practice-oriented researchers, especially from the networking community, and users of discrete-event simulation. Furthermore, we also invite the attendance of researchers who are adept at using PH distributions in analytical approaches and may be interested in broadening their focus to simulation.

Assumed Background

We assume some working knowledge and practical experience with system evaluation, Markovian modelling and performance analysis.

Length of the Tutorial

Three hours.


Last modified: Tue Mar 27 16:32:01 CEST 2012