Recursive approximation of the energy spectral density

In a previous paper, R.Z. Kham et al. (ibid., vol.35, p.102-5, Jan. 1988) described a fast technique for estimating the energy spectral density of a noisy waveform via Taylor series. A fast sliding-time window approach for recursively updating the series is introduced in this work. A method for automatically determining the required series degree using the estimated truncation error is given. The application of the method is demonstrated in an example. The technique is shown to be more efficient than interpolation of the DFT modulus