APPLYING 1/F NOISE FOR DRIFT ESTIMATION

Abstract

The work introduces a new 1/f noise theory that focuses on limited signals. Usually, 1/f noise represents drift, because 1/f noise is the spectral power density of the drift. The subjects of the new theory are signals that have limited value and duration. Therefore the basis of the new theory of 1/f noise corresponds to the real properties of any device and signal comprising 1/f noise. On this basis, the standard deviation of 1/f noise was derived, which is the most important parameter of 1/f noise. This standard deviation has good consistency with (a) the widely used Hurst approximation, (b) the square root of time dependence of Brownian motion displacement, and (c) theoretically derived Brownian displacement. Comparison of an existing 1/f noise theory and the new one shows that both theories are incompatible because subjects of the existing 1/f noise theory are limitless signals, which do not exist.