Deep Learning-Based Automatic Modulation Recognition of Harmonic-Modulated Signals Harmonik Modülasyonlu Sinyallerin Derin Ö?grenme Tabanli Otomatik Modülasyon Taninmasi
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/siu71813.2026.11636998
- Basıldığı Şehir: İstanbul
- Basıldığı Ülke: Türkiye
- Anahtar Kelimeler: Automatic modulation recognition, convolutional neural network, deep learning, harmonic modulation, low probability of intercept (LPI)
- Ankara Yıldırım Beyazıt Üniversitesi Adresli: Evet
Özet
In this study, the problem of deep learning-based automatic modulation recognition (AMR) of harmonic-modulated signals is investigated. Within the scope of the study, a dataset consisting of M-PSK signals under both standard and harmonic modulation effects is generated and analyzed using a convolutional neural network (CNN) based classifier. In the proposed approach, raw I and Q components of the signals are used as input data; the training and testing processes are conducted under various SNR values. The obtained results show that while the deep learning-based approach achieves high accuracy for low-order modulations, the classification performance drops to the 30%-40% range as the modulation order increases due to the micro-phase variations induced by harmonic components. These findings demonstrate that harmonic modulation provides a high low probability of intercept (LPI) capability against artificial intelligence-based surveillance systems.