Monitoring of heart pulse rate of non-stationary subjects using mmWave radar

• Main Author: Fu, Xuemei
• Other Authors: Andy Khong W H
• Format: Thesis-Master by Coursework
• Language: English
• Published: Nanyang Technological University 2025
• Subjects: Engineering; Heart pulse rate; Millimeter-wave radar
• Online Access: https://hdl.handle.net/10356/182036
• Institution: Nanyang Technological University

The millimeter-wave (mmWave) radar-based contactless heart rate estimation method effectively solves the problems associated with traditional heart rate monitoring. Its applications are not limited to medical use, showing potential in areas such as home health monitoring. However, environmental complexity and human motion states increase the difficulty of extracting and estimating human heart rate signals. This study conducts research on heart rate monitoring algorithms for non-stationary humans using an AWR2243 frequency-modulated continuous-wave (FMCW) radar operating at 77GHz. The main contents are as follows: 1. Based on FMCW theory, analyze the phase form in radar system echo signals to facilitate subsequent processing of actual signals. Optimized the signal preprocessing workflow, including target range localization and phase signal extraction, using beamforming, phasor mean cancellation, and direct current (DC) offset compensation to enhance signal-to-noise ratio (SNR) and improve heartbeat detection accuracy. 2. The signals from multiple targets at the same range but different angles are difficult to separate. To address this issue, a target separation method based on beamforming in virtual channel and fast-time dimensions was adopted. To address the issue of phase signal outliers introduced by random target jitter, an energy-based adaptive tunable Q-factor wavelet trans form (EA-TQWT) was proposed. 3. Collected real data from scenarios involving single and multiple persons at different ranges and angles, as well as at the same range but different angles, to verify the effectiveness and stability of the proposed algorithm. Experimental results show that the proposed method can effectively extract heart rate signals, with an average root mean square error (RMSE) of 1.70 bpm, maintaining stability across different scenarios.