RH-AVSWM: A Multimodule Framework for Robust RR and HR Estimation Using FMCW Radar

The detection of respiration rate (RR) and heart rate (HR) using frequency-modulated continuous wave (FMCW) millimeter-wave (MMW) radar has numerous applications across different domains. Most existing studies focus on one or two aspects of the processing pipeline. This article presents RH-AVSWM, an algorithm consisting of four modules that provide more accurate estimates of RR and HR. The adaptive random body movements (RBMs) inhibition algorithm is used to eliminate RBM through range-bin selection and phase compensation. The variational mode decomposition optimized via the sparrow search algorithm (SSA) enhances vital sign signal extraction. Weight reconstruction is employed to mitigate clutter and respiration harmonics in HR signals, while multiple signal classification improves spectral resolution for distinguishing RR and HR. Comprehensive experiments were conducted using a public dataset and data were obtained with a realistic FMCW MMW radar, considering variables such as monitoring distance, incident angle, and body motion. The results indicate that the mean absolute error (MAE) is 0.49 bpm for RR and 1.24 bpm for HR, reducing to 0.23 and 0.70 bpm, respectively, under optimal conditions (0.5 m distance, no body motion, and direct incidence). Through ablation experiments, the contribution of each module to the overall performance (regarding accuracy and robustness) of RH-AVSWM has been assessed. Given that RH-AVSWM achieves comparable or superior performance relative to state-of-the-art results, it holds promise for a wide range of applications in healthcare.