Wideband RF Sensing Algorithms for Detection of Priority Ground RF-Enabled Threats

OUSD (R&E) CRITICAL TECHNOLOGY AREA(S): Advanced Computing and Software, Trusted AI and Autonomy OBJECTIVE: Leveraging emergent combinations of commercially available high-rate Analog-to-Digital Converters (ADCs) and advanced Field Programmable Gate Arrays (FPGAs) to detect a broad spectrum of priority Radio Frequency (RF)-enabled threats. DESCRIPTION: This effort explores novel applications of new, state of the art Commercial Off the Shelf (COTS) ADCs tightly integrated with advanced FPGAs providing revolutionary increases in wideband direct digital Radio Frequency (RF) sampling. Rapidly changing threat environment with a multitude of signals, both threat and non-threat, in close proximity and covering an ever-increasing swath of spectrum, is an ever-present challenge. This presents the difficult task of assessing and identifying a wide variety of signal types accurately and quickly across an extremely wide range of frequencies. Additionally, as systems are forced to address an increased number of threats concurrently, false detections can compromise protection, requiring identification algorithms to be more precise. With the increase in spectrum data that RF systems now need to ingest as a result of advances in state-of-the-art ADC coupled RF FPGAs, systems are further in danger of wasting these gains through inefficient detection and identification techniques. These emerging COTS ADCs will enable new algorithms for wideband threat detection that has the potential to increase the efficiency and efficacy of future next gen systems. This is extremely important for ground platforms face a complex cluttered environment and are increasingly hindered by platform space and power constraints. Systems in the future will also need to identify and characterize unknown threats, these enhanced algorithms will provide rapid detection and better effectiveness at the tactical edge. Decreasing false detections and misclassifications reduce unnecessary RF emissions and reduce output power and increase systems interoperability. All these factors require an innovative set of detection and identification algorithms capable of leveraging advanced RF components to provide accurate and efficient threat characterization across an extremely wideband of RF frequencies. PHASE I: Identify novel algorithms and techniques for threat/signal identification enabled by wideband COTS hardware to detect representative Radio Frequency (RF) enabled threats targeting ground platforms. Use representative class of threats to define an extendable proof of concept, algorithm (or suite of algorithms) for wideband threat identification on identified COTS hardware. Define requirements for algorithm and validate method/framework for identifying and characterizing threats. Measure performance of algorithm against such metrics as speed, accuracy, and rate of false characterization, as well as proficiency in successfully characterizing out of library threats. PHASE II: Implement prototype algorithm developed in Phase I and an extended to include additional classes of threats. Demonstrate generalized identification algorithms for all types of RF-threats to ground platforms within a digital M&S environment provide report documenting findings. Additional class(es) of threats will be assessed based on how well it demonstrates an extension of the base algorithms, as well as how different the class(es) of threat(s) is(are) from the original class used in Phase I. Algorithms will be assessed based on effectiveness criteria mentioned above, as well as ease and speed of algorithm retraining. Identify exemplar threat and demonstrate and test algorithms in a Hardware in the Loop (HITL) environment. PHASE III DUAL USE APPLICATIONS: Implement and test algorithms from Phase II on wideband representative hardware and demonstrate in relevant open-air environment. Demonstrate performance gains that superior detection and identification algorithms can provide on hardware. REFERENCES: 1. https://www.turing.ac.uk/research/interest-groups/machine-learning-radio-frequency-applications 2. https://ieeexplore.ieee.org/abstract/document/8879545 3. https://ieeexplore.ieee.org/abstract/document/8913640 4. https://ieeexplore.ieee.org/abstract/document/8988509 KEYWORDS: Machine Learning, Ground Platforms, Threat Identification, Survivability