UC San Diego

This thesis addresses the problem of securing environments from RF threats by providing a framework to identify and differentiate expected signals from anomalies. A signal is classified as a threat if it is anomalous and cannot be reclassified as benign. Developing such a system involves several key requirements: intercepting RF signals and converting them into IQ sample streams, processing these streams in real time, detecting signals within noise, resolving unique signals across time and frequency, localizing signals for further characterization, and applying both local and global operations to signals. This dissertation introduces three contributions: the channogram, which focuses on resolving signals in time and frequency; Searchlight, which enables signal detection and localization in the time-frequency plane; and Lighthouse, which facilitates real-time processing and application of both local and global signal operations. The challenge of signal interception is acknowledged but lies beyond the scope of this work.