Dictionary Learning for Radar Classification of Multiple Micro-Drones

• Published in: 2019 International Radar Conference (RADAR) pp. 1 - 4
• Main Authors: Zhang, Wenyu, Li, Gang, Baker, Chris
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.09.2019
• Subjects: Dictionary learning; Micro-Doppler; Micro-drone recognition
• ISSN: 2640-7736
• DOI: 10.1109/RADAR41533.2019.171299

Micro-Doppler signatures (MDS) for classifying micro-drones has been the subject of much research in recent years. Most existing work on radar classification of micro-drones assumes that the received signal is reflected from a single micro-drone. However, this assumption may be violated in a number of practical cases. When multiple micro-drones exist in the observed scene, the superimposition of their micro-Doppler signatures greatly increases the difficulty of their classification. In particular, it is challenging to determine if a specific kind of micro-drone exists when there is more than one type of drone observed. In this paper, a method for detection and classification of multiple micro-drones based on their micro-Doppler signatures via dictionary learning is proposed. First, the dictionary is learned for each kind of micro-drone based on cadence-velocity diagrams (CVD) of training samples. Then, a sparse representation of CVD for multiple micro-drones is obtained via the orthogonal matching pursuit (OMP) algorithm. Finally, a threshold detector is applied to the sparse solution in order to recognize the components of multiple micro-drones. The results of experiments, using a 25GHz radar and three types of micro-drone, show that the proposed method achieves a successful classification performance of 92%.