Learning for Perturbation-Based Fiber Nonlinearity Compensation

TL;DR

This paper critically evaluates machine learning methods for perturbation-based fiber nonlinearity compensation, finding that learned linear processing outperforms neural network solutions in this context.

Contribution

It demonstrates that simple learned linear processing can be more effective than neural networks for PBNLC, challenging previous assumptions.

Findings

Learned linear processing is preferable over neural networks for PBNLC.

Numerical results favor linear methods for perturbation triplet processing.

Critiques the claimed benefits of ML methods in fiber nonlinearity compensation.

Abstract

Several machine learning inspired methods for perturbation-based fiber nonlinearity (PBNLC) compensation have been presented in recent literature. We critically revisit acclaimed benefits of those over non-learned methods. Numerical results suggest that learned linear processing of perturbation triplets of PB-NLC is preferable over feedforward neural-network solutions.