An improved class of multiplierless decimation filters: Analysis and design

Abstract

In this paper, we present a class of low-complexity decimation filters for oversampled discrete-time signals. The proposed class of filters improves the frequency response of classical comb filters in two respects. First, it introduces extra-attenuation around the so-called folding bands, i.e., frequency intervals whose spurious signals are folded down to baseband during the decimation process. Second, this class reduces the passband distortion via an effective droop-compensator block, thus increasing the passband of the decimation filters. Like comb filters, the proposed class can be realized through multiplierless architectures, which are also discussed thoroughly in the paper. Unlike comb filters, the proposed filters have superior spurious signal rejection and a greatly reduced droop in the signal passband. These features make the proposed filters suitable for multistage decimation applications, such as reconfigurable software radio receivers, as well as for decimating oversampled digital signals produced by @[email protected] A/D converters. The paper discusses several useful techniques for designing the proposed filters in a variety of architectures with emphasis on non-recursive architectures. Design examples are discussed to highlight the key frequency features along with implementation issues aimed at reducing the computational complexity of the filters.