TY  - JOUR
T1  - Accurate Localization of Elderly People Based on Neural and
Wireless Sensor Networks
AU - Kamel Gharghan, Sadik AU - Ali Hashim, Huda AU - Latif Mohammed, Salim 
JO  - Journal of Engineering and Applied Sciences
VL  - 14
IS  - 11
SP  - 3777
EP  - 3789
PY  - 2019
DA  - 2001/08/19
SN  - 1816-949x
DO  - jeasci.2019.3777.3789
UR  - https://makhillpublications.co/view-article.php?doi=jeasci.2019.3777.3789
KW  - Accuracy
KW  -localization
KW  -MAE
KW  -neural network
KW  -wireless sensor network
KW  -Zigbee
AB  - This study aimed to localize the elderly while moving in a health care center or at home. Elderly
localization was achieved by using a combination of the Received Signal Strength Indicator (RSSI) of Zigbee
Anchor Nodes (ANs) and an artificial neural network. A Feed-Forward Neural Network (FFNN) was selected
on the basis of the Levenberg-Marquardt (LM) training algorithm to train, test and validate data with the
MATLAB Software. Two experiments were conducted in an indoor environment. The first and second
experiments used three and four ANs, respectively. The effect of the numbers of ANs and neurons in each
hidden layer of the FFNN on localization error was examined in terms of statistical analyses. Results show the
better elderly localization accuracy achieved with four ANs compared with that obtained using three ANs. The
four ANs achieved a localization error of 0.232 m (for testing) and improved by 65% compared with the three
ANs. The results also reveal that the increase in the numbers of ANs and neurons can improve elderly
localization accuracy. The second experiment (four ANs) provided a lower minimum localization error than the
first experiment. Comparison of the results showed that our proposed method outperformed the other
procedures in related literature in terms of localization error.
ER  -