Smart environment monitoring through data analytics processing of WiFi to detect activities of daily living

点击翻译按钮获取中文摘要

The number of older adults in Canada and around the world is expected to double over the next 20 years**exerting pressure on our healthcare systems and taxpayers. As a result, there is growing interest in designing**smart environments to support aging-in-place, while including the monitoring required to ensure the well-being**of older adults. Monitoring the Activities of Daily Living (ADL) routines of home occupants reveals important**information about their health and well-being. Older adults tend to have routine ADL patterns. Any deviations**from these ADL routines may indicate important changes in their health status necessitating an intervention.**Although wearable sensors are an alternative, older adults can often forget to carry them on their person (e.g.**while going to the washroom during the night, or while bathing). Ambient sensors are thus a preferred option**for older adults. With respect to smart environments, the integration of ambient sensing capabilities is gaining**ground as a viable mechanism to facilitate unobtrusive monitoring with non-wearable devices. ADLs have**been monitored using ambient sensors (such as motion sensors, cameras, and occupancy sensors) that are**deployed in the environment. Cameras are not preferred because of privacy concerns. Other ambient sensors**require sensor installation and incur operational and maintenance costs. WiFi is a promising alternative to**ambient sensors and is extremely sensitive to occupant activity. It can detect high-resolution movements and**obviates cost of additional sensors. Yet, while WiFi-based activity recognition can build on both WiFi's**pervasiveness and low cost, it encounters challenges for monitoring multiple home occupants and identifying**multiple activities of daily living. This project aims at tackling these challenges by proposing data analytics**solutions for the development of subject-specific ADL models that capitalize on changes to the baseline CSI**measurements from both intrinsic (e.g., configuration of the network interface card) and extrinsic factors (e.g.,**configuration of the environment). Making use of these novel models, the system will be capable of identifying**abnormal activity and provide appropriate alerts.