On the fast design of high-order FIR digital filters

The computational cost for the design of FIR digital filters with short or medium length impulse responses is insignificant. However, the computational cost of conventional FIR filter design programs is o(L^{2}) , where L denotes the length of the impulse response. Several hours of CPU time may be required to compute the coefficients in a very long filter, as for example in the case of L = 1024 . Here we discuss an improved method which can significantly reduce the computational requirements for the design of very long FIR digital filters.