Future of autonomous agents like drones and robots need accurate, robust and compact sensing solutions to map and understand their surrounding environments. Today’s solutions are insufficient. Vision- and acoustic-based solutions are sensitive to lighting conditions and background noise, while more robust laser- and radar-based solutions require expensive and bulky hardware. Alternative solutions are necessary.
Networking radios provide a great opportunity. The same radio waves that we use for networking can also be used to perform mobile sensing. As these radio waves bounce off different objects, they carry information about the objects, which can be extracted by a capable receiver. Since these radios are already installed on mobile devices to enable networking, the cost of radio-frequency (RF) sensing is minimal.
The development of mobile RF sensing systems must also consider the impact of adversarial sensing attacks. As radio waves bounce around us, they also carry information of our physical status. The radio waves, when intercepted by attackers, can potentially reveal our private information. We need to develop robust defenses against such attacks.
In this dissertation, we explore the above two aspects of mobile RF sensing. Different from existing works that rely on specialized hardware, we focus on commodity networking hardware that is installed on mobile devices for networking. Along this line, we make two key contributions. First, we design, implement and evaluate an environmental mobile imaging system using 60GHz networking radios. Leveraging highly directional 60GHz signals, our system can reliably detect the location, size, shape and material of the nearby objects, while navigating itself in unknown environment. Second, we propose and evaluate a silent reconnaissance attack. This leverages the presence of nearby commodity WiFi devices to track users inside private homes and offices, without compromising any WiFi network, data packets, or devices. We then evaluate potential defenses, and propose a practical and effective defense based on AP signal obfuscation.
In summary, this dissertation explores two key aspects of RF sensing for future mobile agents. It addresses a key challenge faced by today's mobile environmental sensing systems. It then identifies a new reconnaissance attack that invades our physical security and privacy, and proposes practical defenses against the attack. We hope our work sheds light on the development of future mobile sensing systems.
