A Note on 'A Secondary Path Modeling Technique for Active Noise Control Systems'

A seeming paradox in a paper by Kuo and Vijayan is explained in terms of the prediction error filter notch bandwidth in comparison with the resolution of the model filter. In [1], a pseudo-random probe signal is used in conjunction with an active noise control (ANC) system for identification of the secondary path transfer function S(z). The estimated secondary path S(z) is required by the FXLMS algorithm, which is commonly used in such applications to provide the main ANC function. However, when used on-line, the identification method is usually confounded by the presence of much stronger primary path noise. Accordingly, [1] proposes a method for eliminating this contamination that utilizes an adaptive prediction error filter.