Artificial Heart-Guided Filter Optimization for UWB Radar Cardiac Motion Sensing

Radar-based cardiac motion sensing promises non-contact clinical diagnosis beyond heartbeat rate measurement. However, extracting pure cardiac signals remains challenging due to interference from respiration, body movements, and radar noise. Traditional bandpass filtering (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$0.8-2 ~\text{Hz}$</tex>) often distorts subtle yet clinically significant motion patterns within cardiac cycles. We present an approach that decouples the filtering problem into low-frequency physiological movements and high-frequency motion details with artificial heart as a bridge. By leveraging controlled reference signals from the artificial heart and the characteristic differences between real and artificial cardiac motions, we design optimized filter banks that better preserve important motion details while removing noise. Experiments with UWB radar demonstrate our method reduces signal distortion by 20.8% in Dynamic Time Wrapping(DTW) distance compared to traditional filtering, enabling reliable cardiac motion sensing under realistic conditions.