High Accuracy Heartbeat Detection from CW-Doppler Radar Using Singular Value Decomposition and Matched Filter

Figure 1. Fundamental mechanism of CW-Doppler radar.

Figure 2. The diagram showing the proposed method for extracting cardiac information.

Figure 3. Simulated chest surface oscillations caused by breathing and heartbeat: (a) shows the signal in the time domain and (b) shows the spectrum in the frequency domain.

Figure 4. Vibration of the chest surface obtained by simulation: (a) shows the signal in the time domain and (b) indicates the spectrum by Fourier transform.

Figure 5. Example of a vibration model of a chest surface to which the proposed method was applied.

Figure 6. Accuracy evaluation of the proposed method for each parameter: (a) illustrates the comparison used the singular vectors, (b) shows comparison for the length of the lagged vector.

Figure 7. An (a) overview and (b) schematic of the developed circuit.

Figure 8. Examples of singular vector generation: (a,b) represent baseband signals from 24 GHz CW-DRS, (c) is the demodulated waveform by Arctangent, (d) shows the filtered PPG signal as a reference of the heartbeat, and (e–h) are the singular vectors of the trajectory matrix created from (c).

Figure 9. Examples of the heart rate enhancement effect by the proposed technique: (a) shows the comparison of the frequency response of the PPG signal, the 3rd singular vector used for heart rate emphasis, and the IIR-type bandpass filter; (b) shows the time-domain waveforms of the PPG signal and the heart rate signal estimated by the proposal method.

Figure 10. Comparison of heart rate interval plots estimated by the proposed method and bandpass filter.

Figure 11. Histograms showing the error of the estimated indices: (a) indicates the error distribution of HR, (b) represents the error distribution of SDHI.