Use of complex fully connected neural networks to compensate for nonlinear effects in fibre-optic communication lines

A scheme is proposed for processing optical signals in a receiver of a communication system, based on complex fully connected neural networks. The influence of the main characteristics of the neural network on the efficiency of nonlinear distortion compensation is studied. A significant advantage of the proposed scheme over real-valued neural networks is demonstrated.

Automated summary

A scheme utilizing complex fully connected neural networks for processing optical signals in a communication system receiver is proposed. The study focuses on the influence of neural network characteristics on the efficiency of nonlinear distortion compensation, demonstrating a significant advantage over real-valued neural networks.