Peak-constrained design of nonrecursive digital filters with low passband/stopband energy ratio

A new approach for designing nonrecursive digital filters is proposed. The method is able to compromise the minimax efficiency of maximizing the stopband attenuation and the ability of the weighted least-squares (WLS) method to minimize the total stopband energy. The approach uses a time-varying weight function which is made constant, at each iteration, inside a given frequency interval. For that matter, a partially WLS-like and partially equiripple stopband is achieved along with an equiripple passband. Efficient implementations of the new method present computational complexity comparable to minimax-based approaches. Examples are included demonstrating the good results achieved with the proposed scheme