A generalized Remez method for the design of FIR digital filters

A generalized Remez method for the design of finite impulse response (FIR) filters is proposed. The method is based on a new problem formulation which largely eliminates certain difficulties brought about by an undetermined approximating polynomial. The new method can be used to design maximal-ripple (MR), extra-ripple (ER), and weighted-Chebyshev filters satisfying prescribed specifications, and, with the addition of some simple techniques, filters can be designed that are free from transition region anomalies. The method incorporates a new initialization strategy and a selective search technique to reduce the amount of computation needed to carry out a design. Extensive experimental results show that the new method is robust and at least as efficient as existing methods for the design of weighted-Chebyshev filters. For MR as well as ER filters, the new method is both robust and very efficient.>