Chun-Fang, Su.Su.Chun-FangLI-CHEN FUYi-Wei, JienJienYi-WeiTing-Ying, LiLiTing-Ying2021-09-022021-09-022018https://www.scopus.com/record/display.uri?eid=2-s2.0-85073167917&origin=resultslisthttps://scholars.lib.ntu.edu.tw/handle/123456789/581394In order to track the trajectory of the elderly with dementia in indoor environment, it is important to constantly to monitor the position of the elderly with dementia. Therefore, we proposed an indoor positioning method based on hand free device -smart bracelet. The challenge of the indoor positioning basing on smart bracelet is that the Bluetooth signal is not only subject to indoor environments, but also affected by the motion of hand. Indoor items, body shading and other factors will cause the signal diffraction, reflection or even loss. For this challenge, we use median filter to reprocess the original signal and integrated the RSSI features with the features of the wrist's gestures. After that we use random forest to establish the position model and adopt the finite state machine(FSM) to calibrate the result of the classify model. It's proved that the proposed system has good robustness and can accurately locate the elderly in the indoor environment. ? 2018 IEEE.Automation; Computer networks; Decision trees; Finite automata; Information systems; Information use; Intelligent buildings; Median filters; Neurodegenerative diseases; Wearable technology; Indoor environment; Indoor positioning; Original signal; Position model; Random forests; Smart homes; Wearable devices; Indoor positioning systemsconference paper10.1109/ICNISC.2018.000202-s2.0-85073167917