Overview

In formulating any statement about a "system", the engineer needs to acquire as precise as possible knowledge about it and about the environment in which it is to work. Such knowledge cannot result from pure reasoning only, and has to rely on observation and measurement. In the domain of performance evaluation, service times, as well as traffic offered, are examples of such input data. For reliability studies, equipment lifetime, system availability, etc. are measured. In a real network or during lab experiments, response delays or load levels can be observed. Similar measurements can be made in the preliminary design step, during simulation studies.

It results from this that a certain amount of data is to be collected on the system. The methods of descriptive statistics help in choosing the parameters of interest and in presenting in a synthetic way the set of results - in short, how to visualise them.

Now, exhaustive measurements are clearly impossible to carry out. The methods of mathematical statistics aim at providing tools to analyze data in order to extract all the possible information from them. For instance, estimation theory helps in estimating the confidence level to be associated with the prediction of a parameter of interest. At last, hypothesis testing helps in making decisions about the population under study, such as comparing two different samples or deciding the conformance of the measurements with a given theoretical distribution function.

Statistics are concerned with a set of elements, called the