ORT image and IMT-MIMO-ResNet-based multi-mode multi-parameter joint optical performance monitoring in MDM-EON systems

Area of Science:

• Optical communication systems
• Signal processing
• Machine learning in telecommunications

Background:

• Mode division multiplexing (MDM) enables higher capacity in elastic optical networks (EONs).
• Accurate optical performance monitoring (OPM) is crucial for reliable MDM-EON operation.
• Existing OPM schemes often monitor only a single spatial mode, limiting their application in MDM systems.

Purpose of the Study:

• To propose a novel joint OPM scheme for multiple spatial modes in MDM-EONs.
• To enable simultaneous monitoring of key performance parameters across different spatial modes.
• To enhance the efficiency and accuracy of OPM in complex optical network environments.

Main Methods:

• Integration of Optimal Radon Transform (ORT) images with an Improved Multi-Task Multi-Input Multi-Output Residual Neural Network (IMT-MIMO-ResNet).
• ORT images are used to enhance parameter differentiation across spatial modes.
• The IMT-MIMO-ResNet model is designed based on system output characteristics and spatial mode correlations.

Main Results:

• Simultaneous monitoring of bit rates (BRs), modulation formats (MFs), mode coupling (MC) intensities, and optical signal-to-noise ratio (OSNR) for five spatial modes.
• Achieved 100% accuracy in baud rate identification (BRI) and modulation format identification (MFI) for each spatial mode.
• Demonstrated robustness against chromatic dispersion, differential group delay, and fiber nonlinearities.
• Average MAEs for OSNR and MC intensities were below 0.4 dB and 0.010, respectively.
• Computational complexity is approximately 1/7 of typical OPM schemes.

Conclusions:

• The proposed joint OPM scheme effectively monitors multiple parameters across spatial modes in MDM-EONs.
• The integrated ORT and IMT-MIMO-ResNet approach offers high accuracy and efficiency.
• This method provides a robust and computationally efficient solution for advanced optical network monitoring.