Embedded system for optical spectral optimization based on a genetic algorithm

In the past few years, R&D of optical communications advanced to provide ever-higher optical signal rates for ultra-long-haul and unrepeatered links. However, higher signal rates and longer transmission distances come along with penalties to the optical signal, such as nonlinear distortions. Signal spectral shaping is a well-known alternative to mitigate nonlinearities and optimize signal reception. Most spectral shaping methods to date use fixed-filter profiles, or slow converging variable-filter profiles. In this work, we develop a fast optical spectral shaping optimization method for nonlinearity mitigation in unrepeatered links. The profile is optimized based on a genetic algorithm that mitigates nonlinear effects for unrepeatered optical transmissions. The designed embedded system and C++ implemented algorithms converge to the best filter profile within minutes, contrasting with previous implementations whose convergence time was in the order of several hours. As an embedded system, it can be easily deployed in legacy links to improve signal reception. Finally, we investigated the performance of the developed system in an experimental unrepeatered link. The method allowed increasing the optical launch power by 1.3 dB at a BER of 10^-3, enabling longer transmission reaches.