Valery A. Granovsky, Tatiana N. Siraya
NOISE CHARACTERISTICS IN ELECTRICAL MEASUREMENTS: METROLOGICAL APPROACH
Metrological approach to the analysis of noise is presented. It is based on the time-domain approach for random processes, which is valid both for stationary and non-stationary cases. The asic functional model is proposed as reproducing kernel (RK) Hilbert space H(R), produced by the correlation function R(s, t) of random process x (t). This space with the specific inner product provides an isomorphic representation of the random process x (t). Several typical models of noise are suggested, including White noise, Brownian motion, Markovian, and 1/f – type processes. The corresponding RK-Hilbert spaces are studed. These RK-norms are recommended as the main characteristics for the noises, which are subordinated to the corresponding typical processes. In particular, the set of the norms and estimates includes both well-known characteristics, such as sample variance and Allan variance, and some new characteristics, such as “Markovian” norms. The case is also considered when several norms (characteristics) from this set are valid. Thus one can employ several norms simultaneously.