Wearable UWB Technology for Daily Physical Activity Tracking, Detection, and Classification

Abstract

A methodology for daily physical activity tracking, detection, and classification is proposed using wearable ultrawideband (UWB) technology. The compact UWB antennas act as sensors to provide useful information and metrics regarding the activity, derived from channel information (path loss and rms delay spread) and received signal features. The activities consist of daily activities, such as walking, sitting, running, opening/closing door, and ascending/descending stairs, which are considered as main activities in everyday life. Experimental validation is carried out showing effective tracking capabilities based on the interdistance variation, giving an estimate of the range of movement during various phases of the activity cycle. The computation of patterns of different daily activities cycle provides unique recognition/classification characteristics, which have been validated through template-based and machine learning techniques. The proposed method provides high accuracy, is cost-effective, and has low computation complexity, making it suitable for various healthcare applications, such as elderly/remote monitoring, rehabilitation, patient activity tracking, and general fitness.