Heart Rate Extraction Technique With Mitigation of Respiration Harmonic for Bio-Radar Sensors
Joo-Eun Park, Geon-Haeng Lee, Inseong Lee, Jong‐Ryul Yang
Abstract
A methodology is proposed to effectively extract heart rate from bio-radar sensors based on radar technology. Heart and respiration rates are extracted separately by calculating intrinsic mode functions (IMFs) obtained from the empirical mode decomposition (EMD) method applied to the received radar signals. Due to the frequency difference between respiration and heartbeat signals, it is possible to selectively generate respiration harmonics by exploiting the characteristics present in different IMFs. These harmonics are used to eliminate the noise caused by respiratory activity around the heartbeat signal, thereby improving the signal-to-noise ratio (SNR) of heartbeat detection. In the experiment utilizing the in-house 5.8-GHz continuous-wave (CW) radar sensor, the proposed technique exhibited an average heart rate extraction accuracy of 97.5% for the stationary subjects at a distance of 1 m, in comparison to the reference electrocardiogram (ECG) sensor. Compared to the conventional method using only fast Fourier transform (FFT) and the EMD-based method without respiration harmonic rejection, the proposed technique achieved an average noise suppression of 22.3 and 20.3 dB around the heart rate signal, respectively. In addition, the proposed technique demonstrated the potential to mitigate the effects of random body motion noise, achieving an average heart rate accuracy of 92.8% when applied to the walking subject at a speed of 2 km/h.