This paper presents an innovative multi-resident activity detection sensor network that uses the Bluetooth Low Energy (BLE) signal emitted by tags worn by residents and passive infrared (PIR) motion sensors deployed in the house to locate residents and monitor their activities. This measurement system solves the problem of monitoring older people and measuring their activities in multi-resident scenarios. Metrics are defined to analyze and interpret the collected data to understand daily habits and measure the activity level (AL) of older people. The accuracy of the system in detecting movements and discriminating residents is measured. As the sensor-to-person distance increases, the system decreases its ability to detect small movements, while still being able to detect large ones. The accuracy in discriminating the identity of residents can be improved by up to 96% using the Decision Tree (DT) classifier. The effectiveness of the measurement system is demonstrated in a real multi-resident scenario where two older people are monitored during their daily life. The collected data are processed, obtaining the AL and habits of the older people to assess their behavior.
The Problem of Monitoring Activities of Older People in Multi-Resident Scenarios: An Innovative and Non-Invasive Measurement System Based on Wearables and PIR Sensors
Naccarelli R.;Casaccia S.;Revel G. M.
2022-01-01
Abstract
This paper presents an innovative multi-resident activity detection sensor network that uses the Bluetooth Low Energy (BLE) signal emitted by tags worn by residents and passive infrared (PIR) motion sensors deployed in the house to locate residents and monitor their activities. This measurement system solves the problem of monitoring older people and measuring their activities in multi-resident scenarios. Metrics are defined to analyze and interpret the collected data to understand daily habits and measure the activity level (AL) of older people. The accuracy of the system in detecting movements and discriminating residents is measured. As the sensor-to-person distance increases, the system decreases its ability to detect small movements, while still being able to detect large ones. The accuracy in discriminating the identity of residents can be improved by up to 96% using the Decision Tree (DT) classifier. The effectiveness of the measurement system is demonstrated in a real multi-resident scenario where two older people are monitored during their daily life. The collected data are processed, obtaining the AL and habits of the older people to assess their behavior.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
