Design of efficient parallel hybrid FIR filters using dynamic programming and subset selection methods

The methods of Boudreaux and Parks (1980) for hybrid FIR (finite impulse response) filter design are extended to include the IIR (infinite impulse response) structure pole location optimization problem within the dynamic programming (DP) formulation and allow multiple parallel cascade structures. In addition, the authors propose a second design method in which they apply complex basis function subset selection methods in the frequency domain to design the optimal weighted L₂ hybrid approximation for a given degree of complexity. If desired, one can perform a sequence of these weighted L₂ approximations to converge to the L_P approximation. Both DP and subset selection approaches are applicable over a broad class filters, including both linear- and nonlinear-phase filters