LYNNTECH INC. — Department of Defense SBIR Phase I: A214-013

Description

The U.S. Army is interested the development of a Behavioristic Electromagnetic Spectrum Assessment General Learning Engine (BEAGLE) that can monitor RF signals in order to identify, classify, track and predict detected radio frequency (RF)-signals over frequency bands for signal interference detection and signals analysis. There is a need to establish feasible techniques to perform automated RF signal trending, prescribe solutions to RF interference, and predict expected RF environment behavior within an area of interest with RF machine learning (RFML) tools. A set of techniques must be identified and tested for multiple RFML tools with advanced data augmentation to simulate a wide range of environment and receiver conditions (for operational robustness).  In the Phase I feasibility study Lynntech and Virginia Tech’s Hume Center will train and integrate BEAGLE subsystems whose implementation will included deep learning-based RFML tools to meet the needs of RF communication at White Sands Testing Center (WSTC). For each RFML tool we will use data for on a target RF-signal type over a range of conditions. We will simulate multiple possible received versions of the same RF-signal along with advanced data augmentation transformations to train the RFML tools. Importantly, we will evaluate the effectiveness of solutions to RF interference applied to the target RF-signal initially in a digital testbed and by measuring the bit error rate with traditional radio systems.