MMT-HEAR: Multiple Moving Targets Heartbeats Estimation and Recovery Using IR-UWB Radars

• Published in: 2020 42nd Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC) Vol. 2020; pp. 5733 - 5736
• Main Authors: Yang, Xiuzhu, Zhang, Xinyue, Qian, Hongyu, Ding, Yi, Zhang, Lin
• Format: Conference Proceeding Journal Article
• Language: English
• Published: United States IEEE 01-07-2020
• Subjects: Heart Rate; Radar; Signal Processing, Computer-Assisted; Vital Signs
• ISSN: 1558-4615; 2694-0604
• DOI: 10.1109/EMBC44109.2020.9175318

Populations around the world are rapidly ageing. Age-friendly environments address the significance of continuous inhome vital sign monitoring. Impulse Radio Ultra-WideBand (IR-UWB) radar serves as a household healthcare assistance, providing non-contact vital sign monitoring without privacy issues and illumination limitation. However, the body movements bring difficulty in extracting heartbeat from radar signals, let alone obtaining complete information with body occlusions among multiple targets. This paper proposes a Multiple Moving Targets Heartbeat Estimation And Recovery (MMT-HEAR) approach to extract vital signs using IR-UWB radars. CLEAN and Joint Probability Data Association (JPDA) algorithms are firstly performed on each radar to estimate target-to-antenna distances of multiple targets. Considering signal obstruction and attenuation for targets occluded by others, the location-based distance optimization is proposed to refine these distances by combining information from all radars. Then the mapping from signal amplitudes to refined distances is introduced and combined with the Variational Nonlinear Chirp Mode Decomposition (VNCMD) to extract vital signs with body movements. To the best of our knowledge, this is the first attempt to monitor vital signs of multiple moving targets with radars. The averaging accuracy for two moving targets heartbeat monitoring during a 20-minutes observation is 85.93% with MMT-HEAR. Compared to two other conventional methods, the MMT-HEAR approach yields improvements of 16.11% and 10.16%, revealing the efficiency and robustness of this proposed approach.