Real-time activity monitoring using wireless wearable sensors

With the emerging of wireless sensor network, continuous monitoring of daily activities for the elderly people to provide timely and personalized health care service is becoming feasible. The algorithms to monitor the physical activities using wireless wearable sensors are proposed in this paper. A tree classifier is proposed to classify the daily physical activities of standing, sitting, lying, and walking in real-time. The detection of emergent status such as falling and abnormal walking pattern is investigated. New features are proposed to describe the walking pattern. This is the first time that the abnormal walking pattern is studied to prevent falls according to available literature. Real-time implementation of the activity monitoring algorithms is performed. The classification result is satisfactory, based on our experiment. The alert can be generated for the emergent status to provide timely healthcare service. The system also provides an ideal platform to build very large database for further healthcare researches.