Power and Modulation Format Transfer Learning for Neural Network Equalizers in Coherent Optical Transmission Systems

TL;DR

This paper introduces transfer learning for neural network equalizers in coherent optical systems, significantly reducing training time and data requirements across different transmission conditions.

Contribution

It demonstrates a novel transfer learning approach that adapts neural network equalizers to various launch powers and modulation formats, improving efficiency.

Findings

Up to 92% reduction in training epochs

Up to 90% reduction in training dataset size

Effective adaptation across different transmission scenarios

Abstract

Transfer learning is proposed to adapt an NN-based nonlinear equalizer across different launch powers and modulation formats using a 450km TWC-fiber transmission. The result shows up to 92% reduction in epochs or 90% in the training dataset.