Related Experiment Video
Updated: Dec 25, 2025
07:12
Real-Time Monitoring of Neurocritical Patients with Diffuse Optical Spectroscopies
Published on: November 19, 2020
2.4K
Deep neural network based OSNR and availability predictions for multicast light-trees in optical WDM networks
Optics Express
|April 1, 2020
Summary
Deep neural networks accurately predict optical signal-to-noise ratio (OSNR) and availability for multicast light-trees, enabling faster network design and guaranteed quality of service (QoS).
Area of Science:
- Optical networking
- Telecommunications
- Machine learning applications
Background:
- Light-tree quality of transmission (QoT) is affected by physical impairments and varying destination distances.
- Real-time optical network states are dynamic, making accurate QoT and availability prediction challenging.
- Advance QoT/availability assessment is crucial for quality of service (QoS) and network optimization.
Purpose of the Study:
- To develop and implement deep neural network (DNN) models for predicting optical signal-to-noise ratio (OSNR) and availability in multicast light-trees.
- To address the challenge of rapid and accurate QoT and availability determination in dynamic optical networks.
Main Methods:
- Leveraging deep neural networks (DNNs) for predictive modeling.
- Developing and implementing DNN-based methods for OSNR and availability prediction in optical WDM networks.
Main Results:
- The DNN-based OSNR prediction method achieved approximately 95% accuracy.
- The DNN-based availability prediction method demonstrated high accuracy exceeding 98%.
Conclusions:
- DNN models offer a fast and accurate approach for predicting OSNR and availability of multicast light-trees.
- These methods facilitate efficient light-tree construction and enhance network management.
