The Research Of Fall Detection Based On Tri-axial Accelerometer

The definition of unintentional fall is that some part of one's body, except feet, hit the ground or any other lower planes accidentally, and the one could not react in time. Fall is one of the great threats against people's health, even lives. The detection of fall is becoming more and more urgent as the aged population growing bigger and bigger. The aim of the fall detection system (FDS) is to discriminate falls and activities of daily life (ADL), to detect fall and give an alarm if necessary, thereby to shorten the time before help coming, to minimize the harm caused by fall, especially the 'long-lie', to reduce the false alarm and to upgrade the quality of life finally.This paper presents a real-time fall detection system based on tri-axial acceleration that is made up of MCU MSP430F149, tri-axial accelerometer MMA7260 and communication module.This research was divided into two parts. First, to define the fall and ADL patterns, and to analyze and compare the different patterns' tri-axial acceleration data which are recorded in the laboratory-based experiments. Second, based on the analysis result, to extract the threshold-based algorithms, that is based on SVM or SMA and postural orientation to distinguish fall and ADL, with the result indicating an overall accuracy over 90% through the laboratory environment trials.Besides, to present the assumes of gait feature extraction and fall forecast that is based on tri-axial acceleration, and to analyze and demonstrate briefly.