PHASE EQUALIZATION OF ONE-DIMENSIONAL AND 2-DIMENSIONAL RECURSIVE FILTERS

An effective optimization algorithm is proposed to design multidimensional recursive phase equalizers. The optimization algorithm uses the modified version of the Rosenbrock pattern search method to minimize the maximum group delay ripple. This numerical method eliminates the increased complexity that derivative methods show when multiple variables are to be found. The criteria used to guarantee the stability of the equalizer are based on the DeCarlo-Strintzis theorem. By applying this theorem, the stability test procedure in m dimensions is reduced to m one-dimensional stability test routines. Using this stabilization technique and the Rosenbrock pattern search method for optimization, the intensity of computations is considerably reduced in higher dimensions.