Detail publikace

Enhancing fiber security using a simple state of polarization analyzer and machine learning

TOMAŠOV, A. DEJDAR, P. MÜNSTER, P. HORVÁTH, T. BARCÍK, P. DA ROS, F.

Originální název

Enhancing fiber security using a simple state of polarization analyzer and machine learning

Typ

článek v časopise ve Web of Science, Jimp

Jazyk

angličtina

Originální abstrakt

The paper focuses on the security of fiber-optic cable infrastructures by detecting vibrations using an optical state of polarization analyzer. The developed system can detect various security breaches. The system only detects abnormal events without any event classification. The proposed system relies on an analyzer evaluating optical polarization differences caused by mechanical or acoustic vibrations analyzed by a machine-learning model for real-time anomaly detection. The main goal of experiments is to find the best combination of the normalization method and anomaly detector. The proposed system achieves an F1-score over 95.65%, which proves the solution’s suitability for protecting fiber-optic infrastructures.

Klíčová slova

Communication system security;Machine learning;Neural networks;Optical sensor;Optical polarization

Autoři

TOMAŠOV, A.; DEJDAR, P.; MÜNSTER, P.; HORVÁTH, T.; BARCÍK, P.; DA ROS, F.

Vydáno

10. 6. 2023

Nakladatel

Elsevier Ltd.

ISSN

1879-2545

Periodikum

OPTICS AND LASER TECHNOLOGY

Ročník

167

Číslo

2023

Stát

Spojené království Velké Británie a Severního Irska

Strany počet

9

URL

https://www.sciencedirect.com/science/article/pii/S0030399223005613

BibTex

@article{BUT183730,
  author="Adrián {Tomašov} and Petr {Dejdar} and Petr {Münster} and Tomáš {Horváth} and Peter {Barcík} and Francesco {Da Ros}",
  title="Enhancing fiber security using a simple state of polarization analyzer and machine learning",
  journal="OPTICS AND LASER TECHNOLOGY",
  year="2023",
  volume="167",
  number="2023",
  pages="9",
  doi="10.1016/j.optlastec.2023.109668",
  issn="1879-2545",
  url="https://www.sciencedirect.com/science/article/pii/S0030399223005613"
}