Multitarget Vital Signs Estimation Based on Millimeter-Wave Radar in Complex Scenes

Regarding the issue that existing methods for detecting vital signs using millimeter-wave radar primarily focus on extracting information from single stationary targets in ideal static scenarios, a novel sparse representation-based approach is developed to detect multiple moving target vital signs in dynamic clutter scenes. The echo data is initially preprocessed to acquire point clouds of both targets and backgrounds. Subsequently, a dynamic clutter suppression model is established to eliminate the dynamic interference. Furthermore, the echo data are assigned to their corresponding targets, enabling multitarget tracking using an extended Kalman filter to extract phase information from the chests of multiple moving targets. In addition, a data-driven adaptive dictionary construction approach is proposed to effectively separate respiratory and heartbeat signals by leveraging their sparsity in the frequency domain. Finally, high-precision estimation of multitarget vital signs can be achieved using a sparse reconstruction method. Numerous experimental results conducted in real-life scenarios demonstrate that compared to state-of-the-art methods, the proposed approach can effectively perceive the vital signs of multiple targets in complex dynamic clutter scenarios.