Activity recognition in healthcare monitoring systems using wireless sensor networks

Abstract

Advanced researches using sensor networks for monitoring home care environments have become a significant concern over the past years. Tracking posture presents many challenges due to the large number of degrees of freedom of the human body. Using the system developed by Brusey [1], in this paper was added an additional hardware component and implemented a new system for posture analysis. An algorithm with multiple threshold methods for detecting and classifying the postures of the subjects in free living live was used. Our new design, which targets a Verdex XM4-bt Gumstix at 400 Hz with a netmicroSD-vx board, is able to collect data and perform posture analysis. Four healthy subjects carry out a set of postures/movements and seven basic positions have been identified by this system: standing, kneeling, sitting, crawling, walking, lying up, and lying on one side. © 2011 Springer-Verlag.