TY - GEN
T1 - SmartFDS
T2 - 9th IEEE International Conference on Internet of Things, 12th IEEE International Conference on Green Computing and Communications, 9th IEEE International Conference on Cyber, Physical, and Social Computing and 2016 IEEE International Conference on Smart Data, iThings-GreenCom-CPSCom-Smart Data 2016
AU - Li, Xiao
AU - Wang, Yufeng
AU - Zhou, Qicai
AU - Jin, Qun
N1 - Publisher Copyright:
© 2016 IEEE.
PY - 2017/5/1
Y1 - 2017/5/1
N2 - Falling accidents cause severe damage to human health, especially to the elderly. Nowadays, smartphones are ubiquitous and widely used around the world, in which rich sensors are embedded. Thus, a smartphone based falling detection system, SmartFDS is proposed in this paper, which can recognize the associated individual's falling behavior and send out an emergent message containing the location for immediate help. Generally, SmartFDS includes offline training phase and online activity recognition phase. Specifically, in training phase, utilizing the embedded accelerometer sensor, falling data for various situations are collected to extract the desired features that can appropriately characterize the falling behavior. Considering that the raw data are intrinsically prone to noise and error, those data are preprocessed by weighted smoothing. Then, 6 time-domain features are elaborated to determine the discriminative features for characterizing falling, and two features, maximum and vertical velocity are selected. Then, 4 different classification algorithms are investigated, and SVM is selected. The prototype of SmartFDS is implemented on Android phones, and can detect falling in real time accurately.
AB - Falling accidents cause severe damage to human health, especially to the elderly. Nowadays, smartphones are ubiquitous and widely used around the world, in which rich sensors are embedded. Thus, a smartphone based falling detection system, SmartFDS is proposed in this paper, which can recognize the associated individual's falling behavior and send out an emergent message containing the location for immediate help. Generally, SmartFDS includes offline training phase and online activity recognition phase. Specifically, in training phase, utilizing the embedded accelerometer sensor, falling data for various situations are collected to extract the desired features that can appropriately characterize the falling behavior. Considering that the raw data are intrinsically prone to noise and error, those data are preprocessed by weighted smoothing. Then, 6 time-domain features are elaborated to determine the discriminative features for characterizing falling, and two features, maximum and vertical velocity are selected. Then, 4 different classification algorithms are investigated, and SVM is selected. The prototype of SmartFDS is implemented on Android phones, and can detect falling in real time accurately.
KW - Classification
KW - Falling detection
KW - Feature selection
KW - Smartphone
UR - http://www.scopus.com/inward/record.url?scp=85020195911&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85020195911&partnerID=8YFLogxK
U2 - 10.1109/iThings-GreenCom-CPSCom-SmartData.2016.95
DO - 10.1109/iThings-GreenCom-CPSCom-SmartData.2016.95
M3 - Conference contribution
AN - SCOPUS:85020195911
T3 - Proceedings - 2016 IEEE International Conference on Internet of Things; IEEE Green Computing and Communications; IEEE Cyber, Physical, and Social Computing; IEEE Smart Data, iThings-GreenCom-CPSCom-Smart Data 2016
SP - 402
EP - 405
BT - Proceedings - 2016 IEEE International Conference on Internet of Things; IEEE Green Computing and Communications; IEEE Cyber, Physical, and Social Computing; IEEE Smart Data, iThings-GreenCom-CPSCom-Smart Data 2016
A2 - Liu, Xingang
A2 - Qiu, Tie
A2 - Li, Yayong
A2 - Guo, Bin
A2 - Ning, Zhaolong
A2 - Lu, Kaixuan
A2 - Dong, Mianxiong
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 16 December 2016 through 19 December 2016
ER -
