Body Sensor Networks for Activity and Gesture Recognition

Abstract

The last decade has witnessed a rapid surge of interest in new sensing and monitoring devices for health care applications. An important development in this area is that of Body Sensor Networks (BSN) that operate in a pervasive manner for on-body applications. Intelligent processing of the sensor streams from BSN is key to the success of applications that rely on this framework. In this chapter we dwell upon one application of BSN that involved processing of wearable accelerometer data for recognizing ambulatory or simple activities and activity gestures. We elaborate on the different steps such as feature extraction and classification involved in the processing of raw sensor data for detecting activities and gestures. We also discuss various aspects associated with a real-time simple activity recognition system such as computational complexity and factors that emerge considering that the sensors are worn by humans. While some of these factors are common to wireless sensor networks in general, the discussion of the chapter is focused on the BSN system developed by us for recognizing simple activities and activity gestures.