FIR filter design over discrete coefficients and least square error

The difference routing digital filter (DRDF) consists of an FIR filter foll owed by a first-order integrator. This structure with power-of-two coeffici ents has been studied as a means of achieving low complexity, high sampling rate filters which can be implemented efficiently in hardware. The optimis ation of the coefficients has previously been based on a time-domain least- squares error criterion. A new design method is proposed that includes a fr equency-domain least-squares criterion with arbitrary frequency weighting a nd an improved method for handling quantisation of the filter coefficients. Simulation studies show that the new approach yields an improvement of up to 7 dB over existing methods and that oversampling can be used to improve performance.