Transfer Learning for Neural Networks-based Equalizers in Coherent Optical Systems

TL;DR

This paper demonstrates that transfer learning enables rapid and efficient adaptation of neural network equalizers in coherent optical systems, reducing training data and time requirements while maintaining performance across various system changes.

Contribution

It introduces the application of transfer learning to neural network equalizers in optical communications, showing significant reduction in training effort for different system conditions.

Findings

Transfer learning reduces training data to as low as 1%.

Neural networks can adapt to different fiber types and system parameters.

Transfer learning maintains performance across various transmission scenarios.

Abstract

In this work, we address the question of the adaptability of artificial neural networks (NNs) used for impairments mitigation in optical transmission systems. We demonstrate that by using well-developed techniques based on the concept of transfer learning, we can efficaciously retrain NN-based equalizers to adapt to the changes in the transmission system, using just a fraction (down to 1%) of the initial training data or epochs. We evaluate the capability of transfer learning to adapt the NN to changes in the launch power, modulation format, symbol rate, or even fiber plants (different fiber types and lengths). The numerical examples utilize the recently introduced NN equalizer consisting of a convolutional layer coupled with bi-directional long-short term memory (biLSTM) recurrent NN element. Our analysis focuses on long-haul coherent optical transmission systems for two types of fibers: the standard single-mode fiber (SSMF) and the TrueWave Classic (TWC) fiber. We underline the specific peculiarities that occur when transferring the learning in coherent optical communication systems and draw the limits for the transfer learning efficiency. Our results demonstrate the effectiveness of transfer learning for the fast adaptation of NN architectures to different transmission regimes and scenarios, paving the way for engineering flexible and universal solutions for nonlinearity mitigation.