Remote Multi-Person Heart Rate Monitoring with Smart Speakers: Overcoming Separation Constraint

Figures

• 

  Figure 1. Practical scenarios of multi-person heart rate monitoring: (a) two people sitting in line and (b) Two people sharing a bed.

• 

  Figure 2. Transmitted and reflected FMCW signals.

• 

  Figure 3. Heart rate–distance heatmap showing heart rates and interference from the reflected signals. The x-axis represents the distance D from the users to the device. In the figure, the two people with heart rates of 72 and 67 bpm located in front of the device cannot be distinguished visually from the heatmap.

• 

  Figure 4. Overview of system for detecting the heart rates of k users.

• 

  Figure 5. Amplitude changes and frequency domain of distances: (a) FFT amplitude changes by distance and (b) breathing and heart rate obtained by applying FFT at 1.08 m.

• 

  Figure 6. Interference removal: (a) original heatmap S; (b) heatmap S after step 1; (c) heatmap S after step 2; (d) smoothed heatmap S.

• 

  Figure 7. Three heart rates of 69 bpm, 83 bpm, and 98 bpm: (a) heatmap with three people and (b) the three brightest blobs.

• 

  Figure 8. Source and circular microphone array with L = 6.

• 

  Figure 9. Device and example showing the experimental setup: (a) the device and (b) one of the experimental setups.

• 

  Figure 10. Extracted heartbeats of an individual located at 1 m and ground truth from ECG. The signal extracted from our system then undergoes FFT to obtain the heart rate in bpm.

• 

  Figure 11. Overall evaluation of the system: (a) detected and ground truth heart rates in bpm and (b) cumulative distribution function of the error.

• 

  Figure 12. Impact of distance.