Spectral Parameter Approximation Based Tunable Digital Filters on Zynq SoC

Spectral parameter approximation based tunable digital filters (SPA-TDFs) are in demand for applications such as software defined radio, cognitive radio in wireless networks as well as radar, sonar and speech signal processing. SPA-TDFs offer variable frequency as well as group delay responses and have advantages such as linear phase, lower group delay, and fewer tunable parameters over other TDFs. This paper deals with the design of an efficient pipelined architecture of SPA-TDF using Farrow structure and modified coefficient decimation method (MCDM) for an implementation on Zynq System on Chip (ZSoC) testbed. The proposed work considers fully parallel, serial-parallel and fully serial configurations of SPA-TDF as opposed to only parallel configuration considered in the literature. To the best of our knowledge, the proposed work is the first attempt to analyze the serialization effect on the throughput and power consumption of the SPA-TDFs. Such analysis is useful in the selection of appropriate serialization level to meet the desired throughput for a given power constraints and vice-versa.