Deep Autoencoder-Based Denoising for Telecommunication Signal Processing
Journal of Robotics and Control (JRC) · 05.11.2025 · 6 (6) · 2645-2652
DOI: 10.18196/jrc.v6i6.28736 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Robotics and Control (JRC), online erschienen am 05.11.2025. Verfasst von Thi-Nguyet Vu und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-SA 4.0).
Abstract (Originalsprache)
Reliable recovery of digitally modulated waveforms remains challenging under additive white Gaussian noise, where classical smoothing filters often blur symbol transitions and degrade detector performance. The research contribution is a compact, reproducible benchmark that quantifies deep autoencoder denoising against moving-average and median baselines on a canonical binary phase-shift keying link. The framework, implemented in MATLAB R2023b, generates 50,000 random symbols, applies binary phase-shift keying, and sweeps the energy-per-bit to noise-density ratio from 0 to 12 dB using identical noise realizations across methods. A lightweight fully connected autoencoder (rectified linear units, mean-squared error loss) is trained once at a fixed signal-to-noise ratio and evaluated alongside the baselines. Performance is reported via two complementary metrics, bit error rate and signal-to-noise ratio improvement, relative to the theoretical binary phase-shift keying bound. The autoencoder consistently lowers bit error rates across the sweep and closely tracks the theoretical curve at moderate to high signal-to-noise ratios, whereas moving-average and median filtering exhibit persistent error floors and negative signal-to-noise ratio improvement. Headline gains include near-theoretical error performance and large positive signal-to-noise ratio improvement at higher operating points, while classical filters further degrade signal quality. The benchmark isolates denoising effects under controlled conditions and is intentionally simple to replicate, providing a foundation for extending analysis to fading channels, higher-order constellations, and alternative neural architectures.
Zitieren
APA
Vu, T., Nguyen, X., & Nguyen, T. (2025). Deep Autoencoder-Based Denoising for Telecommunication Signal Processing. Journal of Robotics and Control (JRC), 6(6), 2645-2652. https://doi.org/10.18196/jrc.v6i6.28736
Vancouver
Vu T, Nguyen X, Nguyen T. Deep Autoencoder-Based Denoising for Telecommunication Signal Processing. Journal of Robotics and Control (JRC). 2025;6(6):2645-2652. doi:10.18196/jrc.v6i6.28736
BibTeX
@article{Vu2025,
author = {Vu, Thi-Nguyet and Nguyen, Xuan-Kien and Nguyen, Thanh-Tung},
title = {Deep Autoencoder-Based Denoising for Telecommunication Signal Processing},
journal = {Journal of Robotics and Control (JRC)},
year = {2025},
volume = {6},
number = {6},
pages = {2645--2652},
doi = {10.18196/jrc.v6i6.28736}
}
EUVIMED