BERT: Remote Sensing of Vital Signs for Bioradar With an Efficient Recursive Technique

Recursive techniques have demonstrated excellent computational efficiency in classic remote sensing tasks but are rarely applied in emerging remote sensing tasks in the IoT field, such as bioradar-based remote sensing of vital signs (VS). As a fundamental but important problem, VS sensing lacks an effective state model, making it difficult to implement using recursive techniques for a bioradar system. However, this paper presents an efficient recursive technique (BERT) that benefits remote VS sensing and accelerates computational speed with bioradar. Specifically, the recursive technique is derived from an efficient Markov state model based on the features of bioradar VS and is significantly helpful for algorithm implementation. This technique fills the recursive application gap in the VS sensing task and motivates further exploration of its rationale. Thanks to the strong capability of the recursive technique in computation, the proposed BERT requires less time and demonstrates greater accuracy compared to other methods in simulation results. We further conduct extensive experiments on two real datasets — one comprising 50 children and the other involving 30 adults across various scenarios, and the results show that the BERT algorithm significantly accelerates computation speed, reducing processing time by nearly 41% compared to the state-of-the-art algorithm, while maintaining superior estimation accuracy. Furthermore, this work also offers a fresh perspective on interpreting remote VS sensing for bioradar.