THE DESIGN OF COMPUTATIONALLY EFFICIENT FINITE IMPULSE RESPONSE DIGITAL FILTERS

Abstract

The main purposes of this research work are to develop computationally efficient methods for the design of FIR filters. First, the frequency response masking approach is generalized for the design of optimum one- and two-dimensional FIR filters. Second, four methods are presented to reduce the complexity of the masking filters. One of the methods utilizes a half-band filter to serve as one of the masking filters for achieving further saving in the number of multipliers. The other methods are based on the fact that the frequency responses of the masking filters are similar. Hence, each pair of the masking filters can be realized as a cascade of a common subfilter and a pair of equalizers. Third, a novel structure for implementing narrow as well as moderately wide transition-width computationally efficient FIR filters is presented. A strategy for determining the values of the design parameters for the proposed structure is also formulated. Finally, a new prefilter structure is presented. The frequency response of the prefilter is derived from the combination of two cosine functions. The new prefilter structure can also be used in linear-phase digital audio tone control system.