Physiological Signal Monitoring Based on Frequency Modulated Continuous Millimeter-Wave Radar

The demand for real-time health monitoring devices is growing due to an aging population and unhealthy lifestyles. Traditional wearable or contact-based sensors have limitations for certain groups and may pose risks. Non-contact millimeter-wave radar-based devices have emerged as a promising alternative. This paper reviews traditional signal processing methods for extracting respiratory and heartbeat signals from radar data, highlighting their limitations. We propose an innovative approach by extending heart rate analysis to higher harmonics, introducing an optimal fundamental-harmonic group method to accurately extract heartbeats despite respiratory harmonic interference. Additionally, we integrate convolutional neural networks (CNN) and long short-term memory (LSTM) deep learning techniques with spectral analysis, significantly improving accuracy and opening new avenues for the advancement of health monitoring technologies.