Radio Frequency-Activity Modeling and Pattern Recognition (RF-AMPR)

Current solutions for automated RF data analysis and information discovery in in RF sensor data struggle to cope with large volumes of sensor data and unknown emissions and behavior underlying the emissions. Our proposed solution for Radio Frequency Modeling and Pattern Recognition (RF-AMPR) leverages state of the art machine learning based RF emission and event detection, activity modeling and anomaly detection, and automated unknown emission clustering, sorting, curation, categorization, data management, and training using generative adversarial deep learning models for both unlabeled and partially labeled wideband RF datasets. This solution builds upon proven OmniSIG Sensor capabilities by providing a rich set of algorithms for activity modeling, anomaly detection, and unknown emission type auto-learning.