Posture recognition algorithm for the elderly based on BP neural networks

Wireless-sensor-network-based home monitoring for elderly activity behavior involves functional assessment of daily activities. In this paper, a mechanism for estimation of elderly postures through various sensors is presented. The data from the triaxial acceleration sensor placed on the waist and the pressure sensors placed under the insoles are consolidated by the coordinator to distinguish the posture of the elderly, such as standing, sitting, walking, and falling down. In the data processing, an algorithm based on BP neural network is presented. The developed system for monitoring and discrimination of the elderly postures is tested through the obtained data in the laboratory and the results are encouraging in distinguishing postures of the elderly.