Wireless systems that extend our senses

Abstract

Wireless signals, such as Wi-Fi, are traditionally used for communications. In this thesis, we show that these signals can also be used as sensing tools that enable us to learn about our environment without physically reaching out to the various objects in it. Specifically, as these signals travel in the medium, they traverse occlusions like walls and bounce off different objects and humans before arriving at a receiver; hence, they carry information about the environment. This thesis presents algorithms and software-hardware systems that extract this information to deliver a variety of new sensing capabilities. We deliver four fundamental contributions: We present the first design that uses Wi-Fi signals to see through walls, enabling us to detect people behind walls by relying purely on the reflections of Wi-Fi signals off their bodies. Next, we demonstrate how we can use radio frequency (RF) reflections to track people's 3D locations and gestures in indoor environments without requiring them to wear or carry any devices. Beyond localizing people, we introduce the first system that can recover human silhouettes through walls; the captured silhouettes enable us to track the 3D positions of human limbs and body parts and to distinguish between different people behind a wall. Finally, we show how smart environments can monitor their inhabitants breathing and heart rates by relying purely on how the human body modulates reflected RF signals. To deliver these contributions, we exploit physical properties of RF signals, work across software-hardware boundaries, and introduce new systems and new algorithms that require redesigning the entire computing stack, from the hardware to the applications. We implement and evaluate these systems demonstrating how they can enable many new real-world applications including baby monitoring, elderly fall detection, non-invasive vital sign tracking, gesture control, and human identification through walls..