Empirical performance evaluation

Description

This lecture discusses how the performance of a system can be evaluated. We will concentrate on empirical means, embracing both simulation and experiments. We shall look at dynamic system with discrete spaces, suitable to model communication systems as well as many other relevant examples. 

The lecture has two main parts. The first part is fairly practical, discussing simulation techniques and implementation concepts for so-called discrete event systems. Examples will use C/C++. The second part is much more mathematical, treating the statistical foundations and justifications for empirical work. Background on stochastic will be provided as needed, although basic familiarity with stochastics is required. (E.g., Mathe III). 

At the end of the lecture, a participant should be able both to design a concrete simulation and to use it to derive statistically justified conclusions about the performance of the so-modelled system. 

Teaching language is English.

Prerequisites

You should have a reasonable working knowledge of basic stochastics (e.g. random variables, density and distribution functions, expected value, variance and standard deviation). A little bit of C or C++ can't hurt for the examples, but we will not do anything particularly fancy in this context.