NeTS: Medium: Collaborative Research: Exploiting Fine-grained WiFi Signals for Wellbeing Monitoring

点击翻译按钮获取中文摘要

While proliferating WiFi networks are usually used for wireless Internet connections, they have great potential to capture environment changes and identify human motions of various scales. Examples of such motions range from performing daily activities to breathing and heartbeat during sleep. These various scales of motions can be captured by fine-grained WiFi signals to perform continuous wellbeing monitoring. Wellbeing monitoring leveraging existing WiFi infrastructure is particularly attractive as it requires neither wearing body instrumentation nor active monitorng by the user. Such an approach would facilitate a broad range of healthcare related applications at home environments without frequent hospital visits, such as real-time prediction and prevention of certain health problems (e.g., cardiovascular diseases and sleep apnea). Using existing WiFi infrastructure for wellbeing monitoring not only advances and extends the applications that could be supported by WiFi networks but also enables easy and large-scale deployment in non-clinical settings due to the proliferation of WiFi networks. Additionally, the educational efforts include curriculum development, outreaching to high school students, and engaging both undergraduate and graduate students in research.This project focuses on building a WiFi enabled continuous wellbeing monitoring framework for fine-grained sleep monitoring and vital signs tracking at home environments. Users do not need to wear any sensors or actively participate in the monitoring process. The proposed framework targets to advance techniques in device-free fine-grained sleep events identification and vital signs tracking during sleep by utilizing existing WiFi signals. The proposed framework develops device-free localization strategies, vital signs tracking methods and statistical learning techniques to depict a comprehensive picture of users' wellbeing. Such wellbeing information is further utilized to assist in real-time disease prediction by leveraging today's ever-growing mobile environments. A hierarchical multivariate logistic regression model is developed to effectively mine through health conditions and identify risk factors of certain diseases. Chances of developing certain health problems, such as cardiovascular diseases, is promptly predicted. The project also provides user-centric access control of archived wellbeing monitoring information to ensure data privacy and coping with distrusted servers.