IIR filters based on frequency-response masking approach

In this paper, IIR digital filter design based on the frequency-response masking approach is presented. The overall filter structure is based on the periodic model filter, complementary periodic model filter, and masking filters. The periodic model filter is obtained by replacing each delay in the prototype low-pass IIR filter by M delays. Masking filters are designed to eliminate unwanted bands from the periodic model filter or complementary periodic model filter. The transition bandwidth of the overall filter is M times smaller to that of the prototype filter. This technique can be used to design sharp low-pass, high-pass, band-pass, and band-stop filters. IIR model filters are implemented as low-sensitivity wave lattice digital filters (WDF-s) consisting of two all-pass branches in parallel. In all-pass sections, each delay is replaced with M delays to obtain periodic model filter. This way maximal sample frequency of the filter is M times increased. We show that the transfer function of the model filter is an elliptic minimal Q factors (EMQF) filter suitable for the efficient WDF implementation.