Chapter 6: Modeling noise
Daniela Calvetti, Erkki Somersalo · Society for Industrial and Applied Mathematics eBooks · 2012
Almost every field of science where quantitative observations are used for validating models must deal with the consequences of noise; thus the concept of noise, and the art of modeling it properly, must be an important component of the model itself. Before becoming fully immersed in the details of noise models, let's pause and think briefly about what is really meant by noise. A standard answer that can be found, e.g., on Wikipedia suggests that noise is unwanted addition to a wanted signal. While this definition corresponds well to the generic and everyday meaning of the word, and covers some commonly used quantitative models of noise, it is nonetheless imprecise and in some cases misleading. As we will see in the examples discussed in this section, sometimes the signal itself has a random characteristic, in which case the concept of “wanted” remains vague and obscure.