The Study On The Fall Detection System With The Wrist-Worn Sensor

Currently, the aging problem is more and more serious in the world. According to the predication of UN, in 2150 the elderly population will be about 30% of the world’s population. So the health problem of the elder people is the main social concern. With the increasing of the singular elder people, the death due to the falling for the elderly people has increased dramatically. The fall detection might detect the falling of elder person, and let the person obtain the immediate help, which might decrease the death. With the appearance of wearable technique, it provides a new solution for falling detection. But currently, the related technique still has the shortingcoming such as low accuarcy, high energy consumption, and etc. To combat them, this thesis proposes a wrist-worn falling detection system, which uses the 3-axis acclerometer to monitor the inclination, and decide the falling event. It improves the detecting accuracy, energy efficiency, and easily to wear. The algorithm is verified and energy efficiency is analyzed on Shimmer-wearable platform. The main research works follows:(1) Currently the falling detection system based on 3-axis accelerometer has the problems such as low detection accuracy and inconvenient carrying. To combat them, this thesis proposes a wrist-worn system, using the 3-axis acclerometer to detect the falling for elderly. According to the gradient of the orientation data, a two-threshold detection algorithm is designed to decide the falling detection. During the experiments, the sensitivity, specificity and the computation complexity about this algorithm are discussed explicitly. Experimental results show that the sensitivity of this algorithm improves about 25%, while with the similar specificity.(2) The energy efficiency is the key factor for the lifetime of wearable system. In this thesis, the wearable platform Shimmer is used to evaluate the energy efficiency. Based on the tests about the sensing, computing and radio sending components on this platform, a mechanism is proposed with event-driven data transmission and node sleeping. With the online judging of the mean value, the data is transmitted only when the mean value is greater than the threshold, otherwise it is not. Besides, the sleeping of node is controlled based on the inclination of the sensor, which might reduce the energy consumption. Experimental results show that this scheme might prolong the lifetime from 16 hours to 29 hours, about 80%.(3) On the wearable platform-shimmer node, the proposed falling detection algorithm and the energy saving scheme are tested. On the low power consumption,non-real time operating system-Tinyos environment, the precision and transmitted data are lost. The practical sensitivity is less than 20% compared to the theoretical result, and the specificity is nearly similar, while the lifetime prolongs 30 hours.The research of this work paper is promising in the medical and habitation fields..