An Ultra-low Power, Robust Photoplethysmographic Readout Exploiting Compressive Sampling, Artifact Reduction, and Sensor Fusion

Abstract

This paper describes an ultra-low power yet robust photoplethysmographic (PPG) readout exploiting various mixed-signal processing techniques. Firstly, compressive sampling (CS) enables to reduce the LED driver power consumption by up to 30x. Feature extraction is performed in the compressed domain, using a Lomb-Scargle periodogram (LSP) to extract the average heart rate and variability, without requiring complex signal reconstruction techniques. Secondly, we demonstrate, in simulations, increased robustness through digital motion artifact reduction for PPG signals, using a spectral subtraction technique. Finally, simulations show further signal enhancement through sensor fusion, enabling electrocardiogram (ECG)-assisted PPG acquisition for cuffless blood pressure (BP) monitoring. The power consumption gains of compressive sampling and feature extraction directly from the compressed domain are demonstrated through a 172 μW compressive sampling PPG acquisition ASIC fabricated in a 0.18 μm CMOS process. The ASIC achieves up to 30x reduction in LED driver power consumption while extracting heart rate with an accuracy conforming to ANSI-AAMI standards.