Efficient calculation of signal and noise variances

The performance of most algorithms in communications and signal processing is dependent upon the input signal-to-noise ratio (SNR). It is therefore useful to estimate input signal and noise variances. In this study, a method is developed for estimating these variances, given the input and output energies for a zero-phase band-pass filter and knowledge of the shapes, but not the amplitudes, of the signal and noise power spectral densities (PSD´s). Expressions for the means and variances of the variance estimates are developed.