Comments on 'A weighted least-squares method for the design of stable 1-D and 2-D IIR digital filters' [and reply]

摘要

For original paper by Lu et al. see IEEE Trans. Signal Processing, vol.46, p.1-10 (1998 January). The iterative procedure by Lu et al. for designing IIR filters is recognized as the Sanathanan-Koerner or Steiglitz-McBride iteration, which was proposed originally in 1963 and 1965, respectively, for system identification purposes. We re-examine some claims and issues related to Lu et al.'s paper in view of various properties that have been deduced for the Steiglitz-McBride iteration in the identification literature. In particular, the Steiglitz-McBride procedure is known not to minimize a least-squares distance measure, contrary to Lu et al.'s claim. It can, however, provide good models and remain stable during successive iterations under specific conditions which are reviewed. A reply is given by Lu et al.