RF emissions’ detection, classification, and spectro-temporal localization are essential not only for understanding, managing, and protecting the radio frequency resources, but also for countering today's security threats such as jammers. Achieving this goal for wideband, real-time operation remains challenging. In this article, we present WRIST, a Wideband, Real-time, Spectro-Temporal RF Identification system. WRIST can detect, classify, and precisely locate RF emissions in time and frequency using RF samples of 100 MHz spectrum in real-time. The system leverages anone-stage object detectionDeep Learning framework, and transfer learning to a multi-channel visual-based spectral representation. Towards developing WRIST, we devised an iterative training approach which leverages synthesized and augmented RF data to efficiently build a large dataset with high-quality labels. WRIST achieves over$99 \%$class detection accuracy,$94 \%$emission precision and recall, with less than 0.08 bandwidth and time offset ratios in a large anechoic chamber over-the-air environment. In the extremely congested in-the-wild environment, WRIST still achieves over$80 \%$precision and recall. WRIST currently supports five 2.4 GHz technologies (Bluetooth, Lightbridge, Wi-Fi, XPD, and ZigBee) and is easily extendable to others. We are making our curated dataset available to the whole community. It comprises over 10 million labelled RF emissions from off-the-shelf wireless radios spanning the five classes of technologies.

Drones are increasingly associated with incidents disturbing air traffic at airports, invading privacy, and even terrorism. Wireless Direction of Arrival (DoA) techniques, such as the MUSIC algorithm, can localize drones, but deploying a system that systematically localizes RF emissions can lead to intentional or unintentional (e.g., if compromised) abuse. Multi-Party Computation (MPC) provides a solution for controlled computation of the elevation of RF emissions, only revealing estimates when some conditions are met, such as when the elevation exceeds a specified threshold. However, we show that a straightforward implementation of MUSIC, which relies on costly computation of complex matrix operations such as eigendecomposition, in state of the art MPC frameworks is extremely inefficient requiring over 20 seconds to achieve the weakest security guarantees. In this work, we develop a set of MPC optimizations and extensions of MUSIC. We extensively evaluate our techniques in several MPC protocols achieving a speedup of 300-500 times depending on the security model and specific technique used. For instance a Malicious Shamir execution providing security against malicious adversaries enables 536 DoA estimations per second, making it practical for use in real-world setups.