Application of Deep/Machine Learning to HRRP Radar Data for Target Classification of Fighter Jets and Missiles

This study investigates the classification of aerial targets, crucial for both air defense and civil air traffic control, with a focus on deep learning (DL) methods applied to High-Resolution Range Profile (HRRP) data. Such identification is particularly vital for selecting electronic countermeasure strategies. We present a comprehensive analysis of the classification of five fighter jets and missiles using DL techniques on HRRP data. Our approach includes a hybrid DL-ML method, where features extracted using 2D DL techniques are subsequently classified via Support Vector Machines (SVM). The study utilizes simulated HRRP data, generated from 3D CAD models using an X-band (10 GHz) radar with a 600 MHz bandwidth. Classification performance was assessed using standard metrics: precision, recall, and F1-score. Experimental results indicate that DL classifiers generally exhibit superior average accuracy compared to the hybrid DL-ML method. Specifically, 2D deep learning methods demonstrate reasonable performance in classifying targets with diverse characteristics.