Survey, Analysis and Comparison of Radar Technologies for Embedded Vital Sign Monitoring

Marco Giordano,Gamze Islamoglu,Viviane Potocnik,Christian Vogt,Michele Magno

DOI: https://doi.org/10.1109/EMBC48229.2022.9871847

Abstract:Contactless vital sign monitoring systems are becoming increasing in demand for a wide range of biomedical applications. Millimetre-wave radars and embedded signal processing are the most promising technologies to enable non-contact vital signs monitoring. In this work, the challenging task of heart rate estimation from radar data has been addressed. Three different radar systems from Infineon, Texas Instruments and Acconeer, and four algorithms, FFT, Median-FFT, STFT and Median-STFT, have been analysed and compared against a reference sensor. Accuracy, as well as power figures, have been reported for all the radar systems. A dataset of 16 volunteers has been acquired, yielding a total of 400 minutes of radar-recorded vital sign data. The accuracy of the four investigated algorithms has been reported on average and per subject for every radar. The algorithm exploiting the Short Time Fourier Transform (STFT) is able to achieve an error as low as 0.02% on a single person and of 6.4% in heart rate estimation on average across the whole dataset.