A pilot study on low power pulse rate detection based on compressive sampling

Abstract

Low power consumption is one of the key design challenges for various pervasive healthcare systems. Compressive Sampling (CS) is an emerging technique for reconstructing signals from data sampled under the Nyquist rate. CS has great potentials for low power pulse rate detection based on photoplethysmograph (PPG) signals, since by reducing the PPG data sampling rate the LEDs could be turned off for a prolonged period of time. Obviously the higher CS rate, the lower power consumption and lower pulse rate measurement accuracies. In this paper, a feasibility study of using CS for low power pulse rate detection was conducted. A miniature PPG measurement device based on our body sensor networks platform was employed for signal acquisition. Experiments for evaluation the pulse rate estimation and the power consumption were completed. Results suggested that the Gradient Projection for Sparse Reconstruction (GPSR) algorithm is a highly efficient for retrieving pulse rate from PPG signals. It was suggested that the CS rate should be approximate 3 for low power pulse rate detections with averaging estimation mean-square error being less than 5.