AP-HR: Amplitude and Phase Joint Optimization-Based Heartbeat and Respiration Separation Algorithm using IR-UWB Radar

Recently, non-contact vital signs monitoring based on bio-radar has made great progress. However, because the wireless signal is greatly interfered by noise, subjects need to face the radar sensor in the forward direction for measurements. In this paper, an Amplitude and Phase joint optimization-based Heartbeat and Respiration separation algorithm using IR-UWB radar (AP-HR) is proposed. Different from other research, we explore the possibility of different information fusions of IR-UWB radar. The complementarity of amplitude and phase is introduced for vital signs monitoring. The correlation between the two signals determines the position index of vital signs. After the feature fusion, we calculate breathing rate using FFT and employ a neural network based on Temporal Convolutional Network (TCN) for heart rate prediction. A radar signal dataset containing 13 persons is set up to evaluate the performance of AP-HR, encompassing three positions: sitting forward (SF), sitting sideways (SS), and sitting with their back to the radar (SB). The experimental results demonstrate that the average Mean Absolute Error (MAE) for heart rate prediction is 1.18 bpm, and for breathing rate prediction is 2.15 bpm, significantly lower than traditional methods. These results indicate that the method can achieve contactless and accurate heart rate and breathing rate monitoring in different scenarios. The source code and radar signal dataset are public.