Digital Compensation of Signal Distortions in Fiber Communication Lines Based on Perturbation Theory and Machine Learning

Information transmission throughout long distances requires to use fiber-optical communication lines (FOCLs) as long as their benefits include higher transmission rate, lower absorption [1] and prospects of capacity increase due to implementation of special signal modulations [2] and multiplexing. However, FOCLs are still associated with a number of undesireable effects which include nonlinearity in fiber and built-in elements [3] resulting in distortion of optical signal and corresponding growth of Bit Error Rate (BER). To minimize the influence of nonlinearity onto the accuracy of data transmission [4], a procedure of compensation [5] should be introduced. One of possible ways is to perform compensation digitally after the signal was received at the exit of the FOCL. In this case, the precise solution of the inverse task of light field propagation can be obtained by integrating the Nonlinear Schroedinger Equation (NSE) over a grid of coordinates by Split-Step Fourier Method (SSFM) [6].