Vital Sign Detection in Any Orientation Using a Distributed Radar Network via Modified Independent Component Analysis

Abstract

Recent advances in radar systems have demonstrated the possibility of noncontact human vital signs detection. Most prior research has focused on vital sign detection with specific subjects’ orientations. So far, very few reported studies have been carried out on vital sign detection in any orientation, which is necessary to be applicable in various scenarios, e.g., assisting in the daily life of elderly persons living alone. To tackle this problem, this article presents a scheme that uses a distributed radar network, which transmits stepped frequency continuous-wave (SFCW) signals, then collecting the vital sign data returned from different directions of the subject and subsequently obtaining the respiratory and heartbeat rates from these collected data. A new respiratory and heartbeat model is proposed here for the theoretical analysis and simulation of vital sign detection for any orientation utilizing a distributed radar network. A modified independent component analysis (ICA) method noted as derivative independent component analysis (DICA) is proposed here to estimate the respiratory and heartbeat rates from multiple channels of the distributed radar network. Use of the ICA method was theoretically validated first to prove that the problem formulation satisfies the three conditions required for its implementation. Vital signs of a stationary subject were detected in simulation and measurement using multiple receiving antennas that are fan-shaped distributed around the subject from different directions. In this scenario, both the heartbeat and respiratory rates can be accurately estimated using the proposed DICA algorithm. Moreover, we also experimentally investigated the performance of the DICA method with different combinations of channels as input signals to the DICA algorithm. Both the simulated and experimental results have demonstrated that, by using a distributed radar network and the DICA method, a successful and accurate respiratory and heartbeat rate detection of a subject regardless of subject’s orientation can be achieved due to the independence of their non-Gaussian sources.