Fir filter design for area efficient implementation /

Abstract

In this dissertation, a variable precision algorithm based on sensitivity analysis is proposed for reducing the wordlength of the coefficients and/or the number of nonzero bits of the coefficients to reduce the complexity required in the implementation. Further space savings is possible if the proposed algorithm is associated with our optimal structures and derived scaling algorithm. We also propose a structure to synthesize FIR filters using the improved prefilter equalizer structure with arbitrary bandwidth, and our proposed filter structure reduces the area required. Our improved design is targeted at improving the prefilters based on interpolated FIR filter and frequency masking design and aims to provide a sharp transition-band as well as increasing the stopband attenuation. We use an equalizer designed to compensate the prefilter performance. In this dissertation, we propose a systematic procedure for designing FIR filters implementations. Our method yields a good design with low coefficient sensitivity and small order while satisfying design specifications. The resulting hardware implementation is suitable for use in custom hardware such as VLSI and Field Programmable Gate Arrays (FPGAs).

FIR filters are preferred for many Digital Signal Processing applications as they have several advantages over IIR filters such as the possibility of exact linear phase, shorter required wordlength and guaranteed stability. However, FIR filter applications impose several challenges on the implementations of the systems, especially in demanding considerably more arithmetic operations and hardware components. This dissertation focuses on the design and implementation of FIR filters in hardware to reduce the space required without loss of performance.