40 research outputs found
ML-Assisted Gaussian Noise Modeling of NLI Accumulation in Dispersion Managed Optical Links
No full textCoherent accumulation of non-linearity on QAM modulated signals can be se-
vere in dispersion-managed links. We propose a machine learning assisted, spatially disag-
gregated model able to provide a fast and accurate non-linearity estimation in this scenari
Spatially Disaggregated Approach to Cross-Channel Interference in Dispersion-Managed Optical Links
No full textWe propose a spatially disaggregated approach to observe and model the XCI coherent accumulation when routing coherent channels through dispersion-managed optical multiplex section enabling real-time digital twin and large capacity gains on legacy deployed infrastructure
Failure Prediction in Optical Transport Networks Through the Integration of Digital Twins and Deep Learning
No full textThis work introduces an advanced comprehensive
framework for predictive maintenance by integrating Digital
Twin (DT) with multiple Deep Learning (DL) models with
the aim of predicting amplifier failures in optical networks.
Using GNPy (Gaussian Noise model in Python), an open source
framework, a DT is created to emulate network behavior under
both normal and failure conditions, enabling the generation
of synthetic datasets representative of amplifier degradation
fault scenarios. These datasets are used to train DL models
based on Convolutional Neural Networks (CNN), Long Short-
Term Memory (LSTM), and Long- and Short-Term Time-Series
Networks (LSTNet). A comparative analysis shows that all models
exhibit strong performance, with LSTM achieving an accuracy
of 99% and LSTNet, CNN models closely following at 98% and
96% respectively, demonstrating the ability of these DL models
to identify complex temporal and statistical patterns in network
telemetry data, facilitating accurate prediction of early failures.
This integrated solution provides a scalable and data-driven
approach for proactive fault management, improving operational
capabilities in optical transport networks
Leveraging Digital Twins to Build Comprehensive Network Failure Databases for Predictive Machine Learning in Optical Transport Networks
No full textSOP-Based Anomaly Detection Leveraging Machine Learning for Proactive Optical Restoration
No full textThis paper presents an experimental proof-of-
concept for detecting malicious mechanical vibrations in optical
networks using Machine Learning (ML) techniques. The study
leverages the State of Polarization (SOP) as a real-time sensing
mechanism for the identification of anomalous disturbances, like
those caused by drilling, which can lead to fiber cuts and sig-
nificant network disruptions. The proposed ML-based approach
is able to continuously monitor SOP fluctuations, enabling the
early detection of vibrations and proactive mitigation of poten-
tial network failures. By leveraging advanced ML algorithms,
the model effectively identifies between normal environmental
vibrations and critical, harmful disturbances. This capability
ensures timely intervention to protect the network infrastructure.
The ML model achieved a vibration detection accuracy of 95%,
which demonstrates it’s high reliability in distinguishing benign
anomalies from disruptive anomalies. This level of precision
significantly enhances the stability, resilience, and operational
efficiency of the optical network. This leads to a reduction in
the likelihood of service outages and physical infrastructure
damage. The results show the potential of combining real-time
SOP monitoring with ML-based analytics to advance network
management strategies
Machine Learning Agents Leveraging Digital Twins for Failure Prediction in Optical Networks
No full textThis study proposes an advanced framework for predicting optical amplifier failures in optical networks by integrating Digital Twins (DT) and Machine Learning (ML). Utilizing the GNPy open-source framework, DTs replicate amplifier behavior under various conditions, resembling faults and captures the network conditions and performance metrics of the optical networks. The telemetry data generated from these simulations represents both short-term dynamics and long-term trends in amplifier performance, enabling the training of a Long Short-Term Memory (LSTM) model. The ML model demonstrates
an amplifier fault level classification accuracy of 98%, effectively identifying soft failures and assessing fault severity. By leveraging the ability to model complex fault scenarios in a controlled environment, the framework provides a comprehensive solution for generating datasets that are otherwise difficult to obtain from live networks. This approach enables early detection and intervention, minimizes service disruptions, and enhances network reliability. The integration of DTs and LSTM-based ML offers a scalable and data-driven solution for improving the resilience, efficiency, and operational continuity of modern optical communication systems
Machine Learning for Predictive Multi-Event Detection in Fiber Optic Systems
No full textWe apply Machine Learning (ML) to detect and
classify anomalies in optical fiber systems. Fiber optic systems
are highly sensitive to external factors such as mechanical stress,
vibrations, and malicious manipulations, leading to potential
service disruptions and data loss. Using State of Polarization
(SOP) data and its angular speed (SOPAS), we developed a
ML framework capable of detecting multi-event anomalies,
including small hits, up and down fiber movement, and shaking.
Experimental data from polarization signature monitoring were
classified using supervised ML models, namely Random Forest,
Support Vector Machine, Logistic Regression, and Extreme
Gradient Boosting (XGBoost). A Weighted Performance Metric
(WPM) was employed to evaluate model performance by bal-
ancing accuracy and training time. The results demonstrate the
superior effectiveness of Random Forest and XGBoost in achiev-
ing reliable anomaly detection while maintaining computational
efficiency. This study underscores the transformative role of ML
in predictive anomaly detection, enabling proactive maintenance
and enhancing the reliability of optical communication system
Intelligent Detection of Overlapping Fiber Anomalies in Optical Networks Using Machine Learning
No full textWe propose a machine learning approach leveraging
state-of-polarization dynamics to detect overlapping fiber anoma-
lies. Simulated disturbances and XGBoost classification achieve
near-perfect accuracy under noise, enabling precise identification
of concurrent events and enhancing both fault detection and
physical layer security in optical communication network
Digital Twin-Integrated Binary Classifier ML Model for EDFA Failure Prediction
No full textThis study presents a digital twin-enabled framework integrated with a binary classification Machine Learning (ML) model for forecasting failures in Erbium-Doped Fiber Amplifiers (EDFAs). The framework utilizes GNPy, an opensource optical network planning tool, to construct a digital twin that serves as a virtual replica of the physical EDFA system. This digital twin facilitates the estimation of Quality of Transmission (QoT), along with the collection and analysis of key operational parameters. A binary classification model, based on Long Short-Term Memory (LSTM) networks, is trained on the data generated by the digital twin to predict potential EDFA failures, achieving a high prediction accuracy of 98 %. This predictive capability enables early fault detection and proactive maintenance, thereby minimizing unplanned downtime and service disruptions. By incorporating real-time analytics and predictive insights, the proposed approach significantly enhances the reliability, availability, and intelligence of optical network management
Resilient Anomaly Detection in Fiber-Optic Networks: A Machine Learning Framework for Multi-Threat Identification Using State-of-Polarization Monitoring
No full textWe present a thorough machine-learning framework based on real-time state-of-polarization (SOP) monitoring for robust anomaly identification in optical fiber networks. We exploit SOP data under three different threat scenarios: (i) malicious or critical vibration events, (ii) overlapping mechanical disturbances, and (iii) malicious fiber tapping (eavesdropping). We used various supervised machine learning techniques like k-Nearest Neighbor (k-NN), random forest, extreme gradient boosting (XGBoost), and decision trees to classify different vibration events. We also assessed the framework’s resilience to background interference by superimposing sinusoidal noise at different frequencies and examining its effects on the polarization signatures. This analysis provides insight into how subsurface installations, subject to ambient vibrations, affect detection fidelity. This highlights the sensitivity to which external interference affects polarization fingerprints. Crucially, it demonstrates the system’s capacity to discern and alert on malicious vibration events even in the presence of environmental noise. However, we focus on the necessity of noise-mitigation techniques in real-world implementations while providing a potent, real-time mechanism for multi-threat recognition in the fiber networks
