Application of Window Functions for Estimation of Spectral Correlation Functions

This paper deals with the problem of enhancing spectral correlation function (SCF) estimation, which is the frequency domain characteristic describing the second-order statistical properties of cyclostationary processes. The algorithm based on two-dimensional FFT procedure is chosen as a base SCF estimation algorithm. It is shown that this algorithm can be transformed so as to involve the use of a window function in order to reduce the variance of the SCF estimates. The results of the application of some well-known window functions in addition to the rectangle are shown altogether with the analysis of the estimation accuracy dependence on the window functions shape and width. It is shown that according to the type of cyclostationary process under study, the optimal width and shape of the window function can be chosen in terms of the minimum root mean square error criterion. As an example of cyclostationary process, it is considered base-band BPSK signal observed in an additive Gaussian noise. Monte Carlo simulations are performed and SCF estimation accuracy is analyzed