FORTRAN subroutine for an EEG feedback digital filter

A method of constructing digital filters is outlined and illustrated in an application for research uses of EEG feedback. The method is based on least squares estimation with geometric weighting. Old data in the time series are discounted according to age, as new data arrive. Recurrence relations are used, allowing the computations to be performed in real time. A FORTRAN digital filter is formulated that simultaneously estimates (1) the spectral power in a band centered at a target frequency, (2) the total spectral power, and (3) the dominant frequency.