NEWSART: Negative binomial weighted spectral analysis in real time

In this paper, digital filters are described for concurrent adaptive estimation of the spectral powers at several target frequencies. Least-squares estimation is used, with a tapered weight function that discounts time-series data according to age. Frequency selectivity and reliability of spectral estimates are controlled by different parameters. The spectral windows are closely approximated by Student’st distribution, which is useful in the planning of experiments.