Machine Learning for Radio Frequency (RF) Signatures Detection and Classification System

Vadum will develop a prototype Context-Aware Machine Learning Signal classifier (CAMLS) to recognize emissions and predict interference to RF systems for the Army's Next Generation Combat Vehicle (NGCV) manned platforms. The CAMLS system will employ offline trained classifiers, and online signal and emitter characterization techniques that to detect signals and emitters of interest. Signal detection will use an intelligence or data-derived Electronic Order of Battle (EOB) with expected signal parametrics to aid signal detection prior to detecting unanticipated signals from enemy or civilian sources. CAMLS signal classifiers and characterization functions will create a Dynamic EOB (D-EOB) that stores spectrum usage history and drives interference analysis for NGCV RF receivers. The Phase I interference analysis will focus on the NGCV Active Protection System (APS) radar by investigation of the radar impact on NGCV communications receivers, and interference from all inband signals in the APS operating frequency range. The CAMLS output would inform a spectrum situational awareness decision aid for Army operators. This would improve vehicle and crew survivability and effectiveness through better decisions regarding vehicle RF systems state and the presence of enemy forces since Army vehicles do not have this capability.