Non-Wearable UWB Sensor for Human Identification in Smart Home

摘要

Human identification is one of the main challenges for health monitoring systems in smart home. The majority of existing systems employ either wearable tags or video cameras to identify multiple residents in home environment. However, due to the inflexibility and inconvenience of wearable devices, and potential intrusiveness of video cameras, the adoption rate of these approaches remains low. In this paper, we propose a novel non-wearable identification system to recognize multiple residents in a home environment, through ambient non-intrusive ultra-wide band (UWB) sensors. For each room, over the entrance, we propose to install a UWB sensor to monitor and identify individuals who enter or exit the room. As each individual has different body figure and walking gait, they are “seen” differently by UWB sensor. Through the proposed Region of Interest (ROI) extraction approach, our system generates a unique UWB signature for each individual which will be used for their identification. We evaluated our system in a controlled laboratory environment with eight individuals including six males and two females. Our UWB-based approach achieved 88% accuracy in identifying individuals among eight subjects, while a similar ultrasound-based technique [1] achieved only a 54% accuracy rate.