Identification of errors-in-variables models

The problem of estimating mathematical models from noisy data when all the measured variables are corrupted by additive noise is considered. This class of models, called errors-in-variables (EIV), plays an important role in many engineering applications like, for instance, fault diagnosis, array signal processing, blind channel equalization, image processing, etc. The well-known least squares method cannot be applied to EIV models as the obtained estimates are not consistent. In this course, the identification of both static and dynamic EIV models will be discussed and some of the main approaches will be described.