A Sub-Nyquist Rate Compressive Sensing Data Acquisition Front-End

Abstract

Abstract-This paper presents a sub-Nyquist rate data acquisition front-end based on compressive sensing theory. The front-end randomizes a sparse input signal by mixing it with pseudo-random number sequences, followed by analog-to-digital converter sampling at sub-Nyquist rate. The signal is then reconstructed using an L1-based optimization algorithm that exploits the signal sparsity to reconstruct the signal with high fidelity. The reconstruction is based on a priori signal model information, such as a multi-tone frequency-sparse model which matches the input signal frequency support. Wideband multi-tone test signals with 4% sparsity in 5~500 MHz band were used to experimentally verify the front-end performance. Single-tone and multi-tone tests show maximum signal to noise and distortion ratios of 40 dB and 30 dB, respectively, with an equivalent sampling rate of 1 GS/s. The analog front-end was fabricated in a 90 nm complementary metal-oxide-semiconductor process and consumes 55 mW. The front-end core occupies 0.93 mm .

Hardware-Algorithms Co-Design and Implementation of an Analog-to-Information Converter for Biosignals Based on Compressed Sensing

“…In other words, one can amply relax the system requirements and still be able to correctly process the RF input signals. For instance, the circuits in [18] and [19], both implemented in 90 nm CMOS technology, process, respectively, multitone BPSK signals up to 500 kHz and radar pulse signals up to 2 GHz, with a power consumption of only 55 mW and 506 mW (without considering the final Nyquist ADC, that was not embedded in either circuits).…”

Towards a Standard Mixed-Signal Parallel Processing Architecture for Miniature and Microrobotics

“…For example, sparsity is often inherent such as in the frequency domain in wireless communications [ 33 ] and the wavelet-domain for images [ 34 ]. A compressive sensing analog front-end utilizes basis functions Φ i ( t ) that are pseudo-random and emulate white noise [ 35 , 36 ]. By mixing the input signal with a randomized basis function, the signal is randomized and the information in each channel spreads over the entire bandwidth.…”

Context-Adaptive Sub-Nyquist Sampling for Low-Power Wearable Sensing Systems

“…Compressive Sensing (CS) [2] aims to reduce power and memory footprint for sensing applications by taking few random samples at the sensor, and by forwarding this sparse signal to a receiving unit that uncompresses the signal again for further processing [3]. While some variants of CS focus on data compression for storage and transmission [4], energy consumption in a sensor device could be reduced by controlling the rate at which the analog-to-digital converter operates [5]. The basic concept is to sample a signal at its information rate, rather than its maximum bandwidth, which is also called sub-Nyquist sampling (SNS) [6], [7].…”