Step Rate Estimator from Wearable Photopletysmography Signal

Smart wearable watch (smartwatch) currently becomes popular device as personal activity companion. Nowadays, most of smartwatch has been equipped for gait monitoring. However, the performance of the system for step measurement is relying on conventional sensor such as accelerometer. In this paper we describe our new novel system about step rate estimation by using Photopletysmography (PPG) signal. One of our main goal is to reduce the number of sensor used. Our backbone method are based on an adaptive intrinsic mode function (IMF) selection method for Complete Ensemble Empirical Mode Decomposition (CEEMD) in step measurement. We made our own public dataset recorded from 5 subjects with 3 different activity state (stand still, walk, and run). The experimental results show our system achieve overall over 90% accuracy for all activity.