Heartbeat Measurement with Millimeter Wave Radar in the Driving Environment

Millimeter wave radar has proven to be an effective method to measure human vital signs in a non-contact manner. However, the driving context is complex and noisy, and this fact must be considered in the algorithm design as well as in testing and verification methodologies. We implemented an estimation algorithm based on arctangent demodulation and added refinements to optimize performance. Three different algorithm choices, peak spectrum, MUSIC, and particle filter were tested while driving in an on-road vehicle as well as in a high-fidelity driving simulator. Accuracy was explored over a parameter space by varying window size and error tolerance. We present results that help to select the method and parameter. Additionally, we conclude that a high-fidelity simulator is an appropriate environment to conduct radar tests for in-cabin monitoring.