Two-fold normalized square-root Schur RLS adaptive filter

SVD-based method for band limitation. In doing so, we redefined finite-extent band limitation. Our method was chosen to have the correct dimension specified by Slepian's result [ 11, to specialize correctly to the DTFT as the sample size tends to infinity, and to have maximum energy concentration. We presented an efficient method, based on the Lanczos algorithm, for performing the proposed band limitation of sequences with up to 4096 or 8192 samples on present work stations. The SVD-based method is less dependent on input frequency than the DFT, but requires storage for the low-frequency singular vectors.