A Novel Position-Aware Attention–Enhanced Bi-GRU Detector for Faster-than-Nyquist Signaling
IEEE COMMUNICATIONS LETTERS, cilt.30, ss.2317-2321, 2026 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Cilt numarası: 30
- Basım Tarihi: 2026
- Doi Numarası: 10.1109/lcomm.2026.3704639
- Dergi Adı: IEEE COMMUNICATIONS LETTERS
- Derginin Tarandığı İndeksler: Scopus, Materials Science & Engineering Collection (ProQuest), Technology Collection (ProQuest), Aerospace Database, Science Citation Index Expanded (SCI-EXPANDED), Compendex, INSPEC
- Sayfa Sayıları: ss.2317-2321
- Ankara Yıldırım Beyazıt Üniversitesi Adresli: Evet
Özet
This study proposes a deep learning-based detector for faster-than-Nyquist (FTN) signaling, integrating Bidirectional Gated Recurrent Units (Bi-GRU) with a novel Position-Aware Attention (PAA) mechanism. Unlike prior approaches that implicitly model inter-symbol interference (ISI), the proposed PAA introduces a structured and domain-informed bias. This is achieved by assigning attention weights according to symbol positions relative to the detection target. Conventional attention mechanisms typically compute sample-specific, content-based relevance scores. In contrast, the PAA module leverages the inherent properties of FTN signaling by learning a globally shared, position-indexed attention distribution. This design exploits the physical insight that ISI strength decays with symbol distance from the target, effectively guiding the Bi-GRU to focus on the most influential interference components. To the best of our knowledge, this is the first application of a structured attention mechanism tailored explicitly for FTN detection. Simulation results demonstrate near-optimal bit error rate (BER) performance for τ ≥ 0.5, where τ denotes the time acceleration factor, closely matching the benchmark BCJR algorithm. Furthermore, hardware analysis confirms the scalability and efficiency of the proposed method, achieving up to 72.3% reduction in Look-Up Table (LUT) usage.