Localization and Collision Prediction via RF Transceiver and Radar Methods

Abstract: This dissertation explores the implementation of sensor-equipped products in various environments to enhance safety, particularly in workplaces, homes, and transportation systems. We explore multiple facets of building automated systems and data-driven algorithms, in attempts to minimize human errors, leading to safer interactions. The focus is on safety-critical applications such as collision avoidance systems (CAS) for vehicles and smart wearables, where both centralized and distributed algorithms are employed for decision-making. Additionally, the importance of prediction over detection in collision avoidance systems is emphasized, highlighting the need for advanced RF-sensors and algorithms to foresee potential hazards before they occur. Egocentric sensing, which involves sensing the entire environment for potential threats, is discussed, along with the challenges and solutions associated with different types of sensors like LiDARs, ultrasonic sensors, and radars. Privacy preservation is also identified as a crucial aspect, especially in RF sensing, with various techniques proposed to ensure that user privacy is maintained while still providing quality service. Techniques such as encryption, differential privacy, and fingerprinting methods are explored to address privacy concerns in RF-based services. Overall, the dissertation aims to pave the way for a safer and more networked world through the deployment of sensor-based technologies while addressing privacy considerations.