Black River Systems Company is pleased to submit this Phase I SBIR proposal to research, evaluate, and prototype an innovative framework to autonomously detect and bring to an operators attention novel and unusual activities in a RF environment. This framework, Autonomous Detection of Anomalous RF-Patterns for Threat-Alerting (ADAPT), can interrogate multiple sensor data streams in near real-time to identify anomalies within the spectrum. Our proposed approach utilizes self-supervised learning to autonomously learn emitter behaviors and RF Patterns of Life (PoL) from the data feeds of existing sensors. These learned RF behaviors and PoL are used to automatically detect anomalous spectrum use in congested environments. The RF spectrum observed by a sensor is often congested and the observed RF PoL are unique to the individual sensors location. Our proposed approach is trained in a self-supervised manner (without need for human labeled data) and utilizes online learning to autonomously learn sensor specific RF PoL while continually adjusting to temporal changes in the normalcy RF PoL (daily, weekly, seasonal, etc.). The proposed approach will facilitate near real-time operation without the need for user interaction and alleviate the laborious requirements to label, model, and train RF PoL manually for each sensor emplacement.
Benefit: Black River Systems anticipates that the proposed Phase I work will demonstrate the feasibility of our approach to autonomously detect and alert an operator to novel and anomalous RF spectrum use. By alerting the operator to anomalous signals our framework will reduce the amount of information that an operator must process at any given time. Thus, the operator will be able to focus on the most important aspects of the RF environment more efficiently, and not waste resources repeatedly evaluating the same RF signals. Deploying and sustaining new sensing hardware is very costly. Existing sensors are already deployed for force protection and ISR missions and monitor the RF spectrum. For this reason, our approach is sensor agnostic and leverages the data feeds from existing sensors to develop PoL and detect anomalies. Detected anomalies will be prioritized based on risk of threat and potential threat impact and broadcast through existing alert systems whether tactical, regional, or national. This scalable framework can be deployed in ISR, tactical or base (port) defense systems making it suitable for a wide array of applications from early missile launch warning, predicting insurgent attacks, identifying out of band communications, hostility change indication, etc.
Keywords: Novel Signals, Novel Signals, Autonomy, Spectrum Monitoring, Operator Attention, Congested RF Environments, Anomalous Spectrum Use, Cognitive Radio, Patterns of Life
