Track Based UAV Classification Using Surveillance Radars

Sarikaya T. B., Yumus D., EFE M., SOYSAL G., Kirubarajan T.

22nd International Conference on Information Fusion (FUSION), Ottawa, Kanada, 2 - 05 Temmuz 2019 identifier identifier

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Basıldığı Şehir: Ottawa
  • Basıldığı Ülke: Kanada
  • Anahtar Kelimeler: UAV, classification, Radar clutter, deep learning, feature extraction, air defense, NEURAL-NETWORKS
  • Ankara Üniversitesi Adresli: Evet

Özet

With their rapid technological developments, Unmanned Aerial Vehicles (UAVs), and micro UAVs have become a significant threat for homeland security and air defense systems. In air defense applications, classification of UAVs and discrimination of UAV tracks from non-UAV ones have become necessary. The classification of UAVs in the presence of heavy spatio-temporally varying radar clutter and interference across rural and urban areas, or under different weather conditions such as rain or snow is particularly challenging. In this paper, a classification method for UAV vs. non-UAV target discrimination, and micro-UAV classification using kinematic and Radio Frequency (RF) characteristics is proposed. For three different classes, namely, micro UAV-1, micro UAV-2 and non-UAV, various classifier architectures are implemented and the results based on real radar data are presented.