Mitigation of UWB Radar Self-Motion for mm-Scale Vibration Detection

To reliably detect a target's mm-scale vibration with a handheld ultra-wideband (UWB) radar system, it is essential to mitigate radar self-motion (RSM) caused by unavoidable hand shaking. In this paper, to further mitigate the effect of stationary objects surrounding the target in real deployment environment, a more practical RSM signal model is proposed. Affected by adjacent objects and RSM, traditional variance-based target positioning method becomes unreliable and the quality of target signal degrades severely as the target's position is often estimated inaccurately. To tackle this problem, an extended bin selection strategy is adopted by considering multiple range bins around the target and independent component analysis (ICA) is leveraged to separate target motion and RSM. Experiments are conducted to verify the proposed technique, and a mm-scale mechanical vibration is reliably detected from 0.5-2m away with the handheld UWB radar system, achieving a median frequency estimation error rate lower than 3.7% under different adjacent environments.