Field Demonstration of a Dual-Band Photonic Radar for Real-Time ISAR Drone Classification

This paper presents a dual-band (X/Ku-band) photonic radar system implemented on a single hardware platform to detect and classify low-Radar Cross Section (RCS) aerial targets. The system uses a reconfigurable photonic module to generate coherent Linear Frequency Modulation (LFM) signals for both frequency bands from a common laser source. Each band operates with a 500 MHz bandwidth to acquire high-resolution Inverse Synthetic Aperture Radar (ISAR) images.We use a Sagnac loop interferometer for the high-sensitivity reception of weak echo signals. A ResNet-18 model deployed on a Neural Processor Unit (NPU) processes the achieved ISAR images for real-time, automatic target classification. We conducted field experiments to validate the system’s ability to detect and track a drone at a range of 2.3 km, achieving a classification accuracy of 94.51% for rotary-wing drones. This field validation of a single-platform system integrating a photonic dual-band radar with an NPU-based classifier demonstrates a practical approach for identifying low-size, low-speed aerial threats.