Machine Learning for Mitigation of Nonlinear Intersymbol Interference in Digital Coherent Optical Communication Systems

This paper explores the interaction between dispersion-induced intersymbol interference and self-phase modulation in coherent optical communication systems using raised cosine pulse shaping. In order to address this interaction, we trained a multilayer perceptron with adjacent symbols, demonstrating that utilizing information from multiple symbols enhances the system's ability to compensate for nonlinear distortion.