Skip to Main Content
For original paper see ibid., vol. 47, no. 10, p. 1227-1237 (2000 ). In Theorem 1 of the aforementioned paper, it is claimed that the correlation matrix of the lattice backward predictor error vector is always better conditioned than that of the wide sense stationary input signal. The proof of the result, unfortunately, is based on an incorrect reasoning. In this comment, we shall give a rigorous proof of the result for the optimal lattice filters.