Non-contact physiologically related motion sensing with ultra-wideband impulse radar

Impulse Radio Ultra-wideband (IR-UWB) radar is a promising tool for non-contact, non-invasive sensing of physiological parameters, which could be valuable in identifying or preventing clinical disorders. This research investigates the use of IR-UWB radar in monitoring of physiological parameters, e.g. respiration rate, heart rate, and cardiopulmonary chest wall displacement. The dissertation has two major parts. The first part consists of novel theoretical analysis of the radar received signal due to arbitrary periodic displacement of the heart and lungs. The second part consists of algorithm development, where estimation methods are developed to estimate a range of physiological parameters. In particular, for heart rate detection, estimation algorithms ranging from heuristic regime to sequential Bayesian filtering have been developed, leveraging HR harmonics, tracking, and motion artifact removal. The performance of these algorithms are evaluated with real-world experimental data from human subjects lying on a bed with an under-the-mattress UWB radar. This research is also concerned with the reconstruction of chest wall displacement from the IR-UWB radar signal. There are several novel aspects of the proposed method. First, only a tiny fraction of the bandwidth of the received spectrum is needed to reconstruct the entire displacement waveform. Since only a tiny fraction of the UWB bandwidth needs to be retained, the required sampling rate is substantially reduced. Second, the maximum likelihood estimator of the displacement is obtained, and the bias and Cramer-Rao lower bound are derived. Finally, in order to further improve estimation accuracy, a denoising method such as empirical mode decomposition is applied on the maximum likelihood estimates. This dissertation also proposes a method that quantifies body-macro movement, such as limb movement, and detects a posture change. The method is set up as a hypothesis test, where the decision threshold between the null and alternative hypotheses are calculated from statistics obtained from experimental data.Ph.D