Field and lab experimental demonstration of nonlinear impairment compensation using neural networks

Abstract

Fiber nonlinearity is one of the major limitations to the achievable capacity in long distance fiber optic transmission systems. Nonlinear impairments are determined by the signal pattern and the transmission system parameters. Deterministic algorithms based on approximating the nonlinear Schrodinger equation through digital back propagation, or a single step approach based on perturbation methods have been demonstrated, however, their implementation demands excessive signal processing resources, and accurate knowledge of the transmission system. A completely different approach uses machine learning algorithms to learn from the received data itself to figure out the nonlinear impairment. In this work, a single-step, system agnostic nonlinearity compensation algorithm based on a neural network is proposed to pre-distort symbols at transmitter side to demonstrate ~0.6 dB Q improvement after 2800 km standard single-mode fiber transmission using 32 Gbaud signal. Without prior knowledge of the transmission system, the neural network tensor weights are constructed from training data thanks to the intra-channel cross-phase modulation and intra-channel four-wave mixing triplets used as input features.

Introducing Load Aware Neural Networks for Accurate Predictions of Raman Amplifiers

“…Mostly known for its use at network operation level, such as in software-defined networking (SDN) [1], ML has also been successfully applied in different fields of optical communications. It has been proposed for nonlinear compensation [2], optical performance monitoring (OPM) and modeling [3], fault detection/prevention [4] and control of Erbium Doped Fiber Amplifiers (ED-FAs) [5].…”

Machine Learning for Fiber Nonlinearity Mitigation in Long-Haul Coherent Optical Transmission Systems : Invited Paper

“…However, the selection of the model appears to be case by case. Leaky RELU as an experimentally demonstrated example, gives the best performance under the NN configuration of [11], for which case the generated model can also be employed at either Tx or Rx with around 1dB Q-factor gain at Tx side compared to it employed at Rx side.…”

Complex-Valued Neural Network Design for Mitigation of Signal Distortions in Optical Links

“…M ACHINE learning (ML) techniques have been recently proposed as a promising tool to address various challenges in optical communications. In particular, neural network (NN) based algorithms have demonstrated their potential to mitigate nonlinear transmission impairments in optical communication links [1][2][3]. The knowledge of physical effects, and underlying mathematical models, can be used to build the respective NN's architecture for specific transmission systems to improve their performance.…”