Heart Rate Monitoring Using FMCW Radar with Dilated Convolutions and Residual Blocks

This article introduces an innovative deep learning method for heart rate (HR) detection using a Frequency Modulated Continuous Wave (FMCW) radar system. The proposed approach seamlessly integrates feature extraction and HR monitoring into a single computational model, thereby eliminating the need for the complex handcrafted feature extraction modules required by previous methods. The system utilizes dilated convolutions and residual blocks to construct the Dilated Residual Convolutional Network (DRCN) model, which directly processes chest micro-motion phase signals obtained from radar data. Experimental results demonstrate that the DRCN model significantly outperforms state-of-the-art methods in HR detection accuracy.