Activity recognition using smartphones and wearable wireless body sensor networks

Abstract

This paper explores automated activity recognition using a WIreless Sensor nEtwork (WISE) connected via Bluetooth to a smartphone. Automated activity recognition enables patients, such as diabetics, to keep more accurate logs of their activities (intensity and duration of the activity) and so to prevent short-terms complication, such as hypoglycaemias. We developed a platform records motion using two wearable sensory devices equipped with 3-axis accelerometers, worn on the waist and the shank, and a wireless heart rate monitor. Data are transmitted via Bluetooth to a smartphone, annotated and analyzed to recognize user activity. WISE platform architecture is described along with recognition accuracy performed by multiple classifiers. © 2012 ICST Institute for Computer Science, Social Informatics and Telecommunications Engineering.