Noise covariance matrices in state-space models: A survey and comparison of estimation methods-Part I

What is it about?

The paper surveys the techniques and methods for identification of the noise properties affecting a real-world system on the basis of measured data. Methods, developed over last six decades, are introduced, compared, and discussed in a unified way. The sample implementations of the key methods are provided.

Featured Image

Why is it important?

Correct knowledge of the properties of the noises affecting the system is essential for design of any optimal algorithm for automatic control, signal processing, and fault detection and exclusion. However, description of the noise using a first principle on the basis of various mathematical, physical, chemical and other laws is extremely difficult or even impossible, and thus the description has to identified. The paper and implementations should help the reader to select and apply a suitable method for identification of the noise properties for the particular considered system and task.

Perspectives