High-speed recursive filtering using the frequency-response masking approach

A novel approach for high-speed recursive filtering is introduced. In this approach, the filter structure is composed of a periodic model filter, its complementary periodic filter, and two masking filters. Earlier, this structure has been used for FIR filters in order to reduce the arithmetic complexity. Here, two allpass filters in parallel are used for the model filter, whereas linear-phase FIR filters are used for the masking filters. The allpass filters are functions of z/sup M/ resulting in filter algorithms with high maximal sample frequencies. By realizing the allpass filters as a cascade of low-order sections, it is possible to obtain highly modular and parallel filter algorithms, which is advantageous from an implementation point of view. Further, by using wave digital filters for the allpass filters, it is possible to maintain stability under finite-arithmetic conditions. Using the new approach, it is possible to increase the maximal sample frequency to an arbitrary level, for arbitrary bandwidths.