LAUSR

The fast detection of heart rate (HR) is challenging when using the noncontact continuous-wave (CW) Doppler radar. Applying the Fourier transform (FT) to the baseband signal analysis, the accuracy is degraded due to the insufficient frequency resolution if using less than 5-s time window to realize fast detection. Moreover, respiratory harmonic peak might be incorrectly picked as the heartbeat signal if its magnitude is larger than the heartbeat peak in frequency spectrum. In this paper, a wavelet-transform-based data-length-variation technique is proposed to realize the fast detection of HR. With this technique, HR can be extracted with 3–5-s data length, and the respiratory harmonics can be distinguished from heartbeat signals, because the frequency of wavelet harmonic is not as tolerant of the change of the data length as heartbeat in the wavelet frequency spectrum. The algorithm is verified by simulation using numerical computing tool and demonstrated by human tests utilizing a 5.8-GHz CW Doppler radar platform. Compared to the traditional frequency domain method using FT, the proposed technique reduces the average error of HR from 26.7% to 3.5% using 3–5-s length of data varied in the range of ±0.5 s.