Real-Time Monitoring of Physiological Signals Using Millimeter-Wave Radar and Convolutional Neural Network

This paper presents a study on real-time monitoring technology for respiratory and heartbeat processes based on millimeter-wave radar. The core focus of this research is on achieving high real-time signal monitoring and recognition, capable of identifying the peak and smooth intervals of the respiratory process, as well as the moments of P, R, and T waves in the heartbeat process, all with minimal latency. This approach provides a more convenient and user-friendly solution for medical gating in CT, MRI, and other smart home applications. Through an in-depth investigation of millimeter-wave radar signal processing methods, an efficient algorithm for respiratory and heartbeat detection is proposed, which includes noise suppression, real-time respiratory-heartbeat separation, heartbeat signal reconstruction based on deep learning, signal peak finding, and signal gradual period identification. The algorithm is verified in real-world environments, and the fusion of traditional signal processing techniques with deep learning methods enables real-time monitoring and recognition of respiratory and heartbeat signals using cost-effective 24 GHz millimeter-wave radar.