Sparse identification for nonlinear optical communication systems: SINO method.

Area of Science:

• Optical Communications
• Machine Learning
• Signal Processing

Background:

• Nonlinear effects in optical systems significantly degrade transmitted data quality.
• Existing methods for nonlinearity mitigation can be computationally complex and may lose crucial signal features.
• There is a need for efficient and accurate methods to manage signal impairments in optical networks.

Purpose of the Study:

• To introduce a low-complexity machine learning method for nonlinearity mitigation in optical systems.
• To identify interactions between symbols and select minimal perturbation terms for adaptive mitigation.
• To enable efficient recovery of transmitted data by simplifying analysis without losing key features.

Main Methods:

• Utilized a machine learning approach based on lasso regression, promoting sparsity.
• Developed the Sparse Identification method for Optical systems (SINO) to identify symbol interactions.
• Applied the method to select the minimum number of relevant perturbation terms for nonlinearity mitigation.

Main Results:

• Successfully demonstrated the application of the SINO method in standard fiber communication links.
• Achieved over 3 dB gain in standard fiber communication systems.
• Showcased successful application in few-mode spatial-division-multiplexing systems.

Conclusions:

• The SINO method provides an optimal, minimal set of degrees of freedom for adaptive nonlinearity mitigation.
• This approach simplifies analysis while preserving essential features for data recovery.
• SINO is effective for both standard and advanced optical communication systems.