Radar can watch heartbeats and breathing without touching the body, but walls and moving objects can hide signals. New methods make tracking more accurate.
Tracking heart rate and respiration without attaching sensors to the body could improve healthcare, smart homes, and vehicle monitoring. But real-world environments make this challenging. Reflections from walls, furniture, and moving objects often overwhelm radar signals, making it difficult to identify the correct person. Even when the target is detected, breathing produces strong periodic signals that interfere with the weaker heartbeat signal. Conventional clutter-suppression methods can remove the slow chest movements that carry vital-sign information, reducing accuracy and reliability.
To address these issues, researchers at Xiamen University in Xiamen, China have developed two complementary algorithms. The first, the Matrix Coefficient Selection Method (MCSM), works in the range-Doppler domain to locate the target. It suppresses static clutter and low-frequency noise, then applies autocorrelation to identify the distance bin most likely to contain periodic vital-sign motion. This approach improves accuracy and stability when multiple people or objects are present. The second algorithm, Recursive Least Squares Respiratory Harmonic Suppression (RLSRHS), works in the frequency domain. It adaptively filters out respiratory harmonics without manual tuning, allowing the weaker heartbeat signal to be measured reliably even as breathing patterns change.
The methods were tested using a 77 GHz FMCW radar system. Simulation studies showed that MCSM reduced distance errors by about 40 percent compared with conventional techniques, while RLSRHS improved heart-rate detection accuracy to 83 percent. In indoor environments with strong reflections and moving objects, radar measurements closely matched readings from ECG monitors, wearable devices, and respiratory sensors, with an average error of around 4 percent. Simple post-processing, like median filtering, further stabilized heart-rate readings over longer periods.
By addressing both target localization and frequency-domain interference, these algorithms make radar-based vital-sign monitoring more accurate in environments that previously challenged such systems. While tested indoors, the methods could also be applied in smart homes, vehicles, and other settings where contact-free monitoring is useful.
