A 4.4μW Cuffless Blood Pressure Measurement Processor Based on Event-Driven and Module-Level Asynchronous Scheme

This paper proposes a cuffless blood pressure measurement processor based on photoplethysmogram (PPG) signals and deep neural networks (DNN). To reduce system power consumption, a module-level asynchronous scheme is designed. Moving average filter, feature extraction, and DNN computation units are all driven sequentially using event-driven wake-up. The moving average filtering unit is driven by a signal detector, which consists of a comparator and a counter. Each unit operates at different frequencies, effectively reducing the overall power consumption. The processor achieves mean error of $4.9 \pm 6.2$ mmHg and $3.4 \pm 4.4 \mathrm{mmHg}$ for systolic blood pressure (SBP) and diastolic blood pressure (DBP), respectively. Designed with a standard 55 nm CMOS technology, the digital core occupies an area of $0.64 \mathrm{~mm}^{2}$, operates at a voltage of 1.2 V, with an estimated power consumption of $4.4 \mu \mathrm{~W}$.