Deep Learning-Based Target Localization and Deghosting in Multistatic Radar Multistatik Radarlarda Derin Ö?grenme Temelli Hedef Konumlandirma ve Hayalet Giderme
34th Signal Processing and Communications Applications Conference, SIU 2026, İstanbul, Türkiye, 7 - 10 Temmuz 2026, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.1109/siu71813.2026.11637065
- Basıldığı Şehir: İstanbul
- Basıldığı Ülke: Türkiye
- Anahtar Kelimeler: Deep learning, Ghost target reduction, Multistatic radar, Target detection and localization
- Orta Doğu Teknik Üniversitesi Adresli: Evet
Özet
This study presents a deep learning method for target detection, localization and deghosting using range observations obtained from a multistatic radar system. To mitigate the ghost target problem, which emerges as a byproduct of the employed target detection algorithm, motion information is incorporated into the system and compared against the baseline case without motion. Cartesian grids generated from three consecutive coherent processing intervals (CPI) are fed into the proposed network as input channels. Within the scope of the performance analysis, U-Net++ and ConvLSTM U-Net architectures are compared in terms of parameter count, inference time, and ghost target reduction. The proposed spatio-temporal approach successfully distinguishes true targets from structurally similar ghost intersections.