Morphology Transformation and Content Selection of Near-Field RF Sensing by Complex Vector Injection

Abstract

Near-field radio-frequency(RF) sensing can realize wearable touchless vital-sign sensors with touchless and flexible deployment. By placing a transmitter (Tx) antenna in the near-field region of dielectric boundary motion, both the magnitude and phase of the Tx signal will be modulated and be recorded as a complex signal at the receiver (Rx). Direct-path interference (DPI) from Tx to Rx without motion modulation often dominates and can vary due to different environmental factors and initial transceiver conditions, which can lead to inconsistent waveform morphology of the magnitude and phase at Rx, sometimes even yielding low signal quality or phase reversal. Here we proposed a complex vector injection (CVI) method, which adds a constant complex phasor to the Rx quadrature signal to regularize the waveform morphology selectively. To demonstrate the effectiveness of the CVI algorithm, we designed a torso phantom for sensor benchmarking. By using different objective functions to select the optimal injection vector, we can enhance different features in the real-number waveform. Furthermore, we applied the algorithm in the study of 21 human subjects for heartbeat monitoring, and showed that we could effectively enhance the signal quality and regularize the waveform morphology which is critical for time-domain feature recognition in physiological and pathological studies.