Computationally Efficient Mismatched Filters Based on Sidelobe Inversion

Abstract

A computationally efficient mismatched filter comprised of a matched filter in cascade with a multi-stage filter based on v is proposed. For this approach to work, the autocorrelation of the given code has to satisfy certain conditions that are derived in this work. If in addition, the sidelobes are sparse and of small integer values, such as in Barker codes and Huffman sequences, then the proposed filters are shown to be computationally very efficient. When implemented in VLSI, they require significantly smaller chip area and less power compared to the length-optimal filters achieving comparable sidelobe suppression.