Multi-Radar Classification of Unmanned Aerial Vehicles

Unmanned Aerial Vehicles (UAVs) have risen to the forefront of modern combat and pose a serious threat to NATO forces globally. While several emergent technologies offer solutions for defeating UAV threats, the reaction time for operators is slowed by tasks such as object classification. The Forward Area Air Defense Command and Control (FAAD C2) system is designed to correlate air track data from multiple sensors to produce an integrated air picture that enables the operator to make fast, well-informed decisions while defeating UAVs. While the FAAD C2 works with a range of radar types, they are not equal in their ability to classify objects as UAV (versus human-operated), and they do not report a common measure of confidence in the classification. This research proposes a method for fusing the input of multiple radars to produce a common report of classification and confidence. Exploratory findings from synthetic data indicate that Long Short-Term Memory (LSTM) networks are well suited for the task. In application, this may improve the classification of UAV reported by the FAAD C2 and empower the operator with real-time, decision-relevant information to expedite counter-UAV engagements.