Embedded Low Power Heart Rate Estimation Processor for Flexible Applications

Abstract

In this work, a photoplethysmography (PPG)-based heart rate (HR) estimation processor is designed and implemented for embedded signal processing of flexible heart rate monitoring devices. Compared with the exiting embedded signal processing solutions using Microcontrollers, this customized hardware solution is able to achieve much lower power consumption for long-term wearable health monitoring. Evaluated using the SPC dataset of 12 and 22 PPG recordings, the proposed design achieves a low mean absolute error (MAE) of 1.12 BPM and 2.14 BPM respectively. It consumes only 34.7 μW with a low processing latency of 6.2 ms, which is suitable for long-term wearable health monitoring.