Vital Sign Detection via Angular and Range Measurements With mmWave MIMO Radars: Algorithms and Trials

Vital signs such as respiration rate and pulse (heart rhythm) rates are very useful in determining health problems and managing well-being. Heart rate measurement techniques have evolved from counting pulses to contact-based approaches such as electrocardiography (ECG) and photoplethysmography (PPG). The current availability and proliferation of low-cost and easily integrable millimeter wave (mmWave) radar sensors make contact-free simultaneous detection of weak vital sign signals highly promising. However, radar echoes reflected from the mechanical behavior of the heart (i.e. heartbeat) are complex and mixed with other (usually stronger) mechanical signals (respiration, body movements, etc.). Such echoes, as captured by radar sensors, comprise a range of Doppler frequencies with varying densities. These frequencies can be directly related to the respiratory and heart rate patterns. Conventional spectral-based approaches adopted directly from radar signal processing do not always yield an accurate estimations of these frequencies. For example, the harmonics of the respiration signal, intermodulation, and cross-product terms complicate the robust detection of heart rate signals. In this study, we show that spatially scanning a human participant using digital beamforming methods provides a more reliable estimation technique for heart rate detection. Prior studies using radar sensors that can filter participants in the range use a range selection strategy based on either the maximum return signal or the highest variance. We developed a strategy to incorporate multiple range and angular bins to reduce the probability of erroneous heart rate estimation. The algorithm, tested on multiple participants, provides an accurate heart rate estimation compared to existing methods.