Recuperação de fase de sinais ópticos baseada em machine learning.

Abstract

Coherent optical communications systems have become the state of the art for creating new optical communications system architectures. However, new algorithms with greater economic and computational feasibility have been researched to be implemented, mainly, in short and medium distance systems. The advancement and popularization of new programming languages, such as the Python language, and also of specialized frameworks for the creation of machine learning algorithms and artificial neural networks present an excellent development environment. Considering the need for signal reception and detection techniques, this work has as general objective the study and implementation of a digital signal processing technique using machine learning algorithms, based on artificial neural networks, for optical signal phase detection by direct detection using supervised learning. The implemented neural networks showed good results against the Kramers-Kronig algorithm present in the literature.