Multi-task neural network for joint estimation of optical impairments in coherent communication systems
Optics Continuum · 15.11.2026 · 5 (11) · 3743
DOI: 10.1364/optcon.616845 ↗Kurz gefasst
Fachartikel in der Zeitschrift Optics Continuum, online erschienen am 12.11.2026. Verfasst von Estabrak Salah Mahdi und 2 weiteren Autor:innen. Die Arbeit nennt 14 Literaturangaben.
Abstract (Originalsprache)
This paper presents a hard-parameter-sharing multi-task artificial neural network (MT-ANN) for simultaneous estimation of optical signal-to-noise ratio (OSNR), chromatic dispersion (CD), polarization mode dispersion (PMD), and launch power in coherent optical systems. The same architecture, 52 engineered signal features, and training protocol are independently evaluated for QPSK, 16-QAM, and 64-QAM. The results demonstrate that estimation performance depends on the target parameter and the modulation format, with PMD and CD generally estimated most accurately, followed by OSNR and launch power. NMAE ranges from 0.56% for PMD with QPSK to 18.78% for launch power with 64-QAM. The proposed model uses 34,468 trainable parameters and provides a compact baseline for multi-impairment OSQM.
Zitieren
APA
Mahdi, E. S., Jamel, T. M., & Nourildean, S. W. (2026). Multi-task neural network for joint estimation of optical impairments in coherent communication systems. Optics Continuum, 5(11), 3743. https://doi.org/10.1364/optcon.616845
Vancouver
Mahdi ES, Jamel TM, Nourildean SW. Multi-task neural network for joint estimation of optical impairments in coherent communication systems. Optics Continuum. 2026;5(11):3743. doi:10.1364/optcon.616845
BibTeX
@article{Mahdi2026,
author = {Mahdi, Estabrak Salah and Jamel, Thamer M. and Nourildean, Shayma Wail},
title = {Multi-task neural network for joint estimation of optical impairments in coherent communication systems},
journal = {Optics Continuum},
year = {2026},
volume = {5},
number = {11},
pages = {3743},
doi = {10.1364/optcon.616845}
}
EUVIMED