CNS Core: Small: Towards Ubiquitous Sensing With Commodity Wi-Fi

Florida State University, Tallahassee FL

Investigators

Abstract

The prevalence of Wi-Fi devices and ubiquitous coverage of Wi-Fi networks provide the opportunity to extend Wi-Fi's capabilities beyond communication, particularly in human sensing. As Wi-Fi signals travel through space, they modulate with the human body and undergo wave phenomena such as reflection and diffraction, which carry a rich set of information about the physical environment including human activities and gestures. This project aims to reuse commodity Wi-Fi devices to capture such wave phenomena to sense multiple persons simultaneously and to provide quantifiable gesture recognition. Exploiting commodity Wi-Fi devices for human sensing enables ubiquitous sensing for easy and large-scale deployments without requiring users to wear or install any specialized sensors. The outcome of this project can be directly adopted by industry and will have a societal impact by facilitating a variety of applications such as Human-Computer Interaction, smart home, mobile healthcare, and security surveillance. This project will lead to new technologies for human sensing and provide new insights on developing emerging mobile applications. The research activities will be integrated into Florida State University's Computer Science education program and will include curriculum development, female and minority students involvement, and K-12 education and industry outreach. This project focuses on building a commodity Wi-Fi based ubiquitous sensing system that can simultaneously sense multiple persons and provides quantifiable gesture recognition without requiring environment-dependent training. The proposed work enables multiple people tracking and activity recognition by extracting the signal reflections from each individual person based on fine-grained signal reflection paths. The periodical change of the reflection signal is further exploited to derive quantitative measures such as the movement distance and speed for quantifiable gesture recognition. Moreover, the motion-induced Doppler effects that reflect the signal dynamic due to human body movements are leveraged to perform environment-independent sensing. Next, it applies the proposed sensing technologies to develop human-computer-interface applications by considering issues such as environmental interferences. In addition, the project expands the Wi-Fi sensing capabilities from human sensing to sensing bio-information of fruit crops. It leverages the frequency diversity to capture the physiological changes of fruit due to ripening and develops mechanisms to extract multipath-independent features for ripeness detection. Finally, the system will be evaluated using real-scale test-beds. The techniques, algorithms, and software resulting from this research will deepen our understanding of mobile sensing and provide practical value by enabling emerging applications. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.