Signal Recovery and Detection of Certain Wideband Signals Using Multiple Low-Rate ADCs

Abstract

The focus of this paper is to develop an innovative approach to receiver design for wideband signals with resonant frequency components. It is known that there are certain wideband signals that occupy quite large bandwidths but may have dominant amplitudes in certain frequency bands. Many studies have applied signal processing techniques to high bandwidth signals to lower the sampling rate. Compressive sensing (CS) is one such method and has been shown to succeed but only in a special case. The CS approach requires the signal to be sparse in some domain. In this work, we focus on signal classes where previously developed reconstruction and detection techniques may fail. In other words, we focus on signals that have large bandwidths with dominant frequency components, but are not necessarily sparse in time. We show that these signals can be effectively sampled by a lower sampling rate compared to what is required by the Nyquist-Shannon sampling theorem. In our design resonant frequency components are channelized into separate receiver paths (subchains) and sampled at a lower sampling rate than is required for the entire wideband signal. Both the signal and noise energy are reduced, although not equally, by the receiver. Noise energy outside of the selected bands is greatly attenuated while these dominant bands are filtered into subchains for processing. It is shown that the signal's sum squared error (SSE) improves and the probability of detection performance experiences little or no degradation as the effective noise power is attenuated by the proposed method.