Abstract. The problem of reduction of computing expenses for digital estimation of power spectral density of a random process with correlogram method using correlation windows is considered. A solution to the problem is obtained with the use of a signanalog stochastic quantization as a primary transformation of the random process under investigation. Estimates of the correlation function calculated from the sign-function signals and discrete-time representation of these signals make is possible to carry out an analytical calculation of the integral cosine transformation of correlation window function when spectral estimation algorithm is designed. The well-known window functions by Bartlett, Hann, Hamming, Blackman, and Nuttall are considered as examples. The developed algorithm for spectral power density estimating does not require direct calculation of the correlation function estimates to be carried out preliminary. The algorithm uses logical operations and simple arithmetic operations of addition and subtraction, and therefire reduces the complexity of digital estimation of power spectral density. 

Keywords: power spectral density, random process, stochastic quantization, correlation window, sign-function signal, time readout