Unveiling Aerial Threats: Enhancing UAV Classification Through Radar Digital Twins

Unmanned Aerial Vehicles (UAVs) present significant security concerns due to their widespread utilization in various nefarious activities, including acts of terrorism and military operations involving explosive payloads. Conventional radar-based methods for detection and classification primarily rely on range-Doppler signatures, which may result in misclassification, particularly in distinguishing UAVs equipped with explosives. To mitigate this challenge, this study proposes an algorithm based on Inverse Synthetic Aperture Radar (ISAR) for classification. The proposed algorithm is developed and validated using radar digital twins to generate extensive datasets. Initially, a Machine Learning (ML) algorithm is trained on a dataset containing range-Doppler information to differentiate between a standard commercial quadcopter and the same quadcopter modified to carry explosives. However, the ML model exhibits limited accuracy in classifying instances where the quadcopter is laden with explosives based solely on range-Doppler data. Subsequently, the ML model model is retrained using a dataset incorporating ISAR images for both scenarios. Upon application to a distinct dataset featuring ISAR images of a quadcopter carrying explosives, the model demonstrates enhanced classification accuracy. This study offers valuable insights for the future development of robust countermeasures to mitigate the evolving security challenges posed by UAVs in sensitive environments.