Nonlinear Fiber Compensation Based on Neural Network in Reflective Coherent Detection System

Abstract

We study artificial neural network used for fiber nonlinear mitigation in single-fiber bidirectional reflective coherent detection system. In this system, the transmitter of this system is located at the receiving end, and the transmission distance of carrier is double the length of actual optical fiber channel, the required optical emission power increases which leads to intensifying of nonlinear effect. Also, using neural network to compensate fiber nonlinear can be processed without knowing the system parameters preferentially. The simulation results show that fiber nonlinear mitigation based on neural network can effectively reduce the bit error rate and improve the transmission performance of the system.