Pre-impact & impact detection of falls using wireless Body Sensor Network

Falls in elderlies are a major health and economic problem. This paper aimed at finding the best body position to place inertial sensors and the best feature for pre-impact and impact detection of fall using wireless Body Sensor Network. Waist acceleration maybe the optimal formula for fall detection, under the conditions of existing inertial sensor´s best precision. We set two thresholds for acceleration, 5 m/s² could get 500ms lead-time and 35 m/s² ensure the specificity up to 100%. We also analyzed the critical phase in fall events, and subdivided it into three periods, make a fall event more intuitive to people.