Studies On Coherent Optical Communication System Based On Neural Network Compensating Fiber Nonlinear Effect

Optical fiber communication is developing towards the goal of high speed,large capacity and long-distance transmission.Due to optical fiber loss,optical amplification is needed.In order to increase the transmission distance,more and more optical amplifiers must be used correspondingly.Optical amplifiers also amplify noise,and in order to maintain a high signal to noise ratio,the signal optical power needs to be increased.However,when the power of the optical signal exceeds a certain value,the nonlinearity in the optical fiber will increase,resulting in nonlinear interference noise.In the optical communication system with long distance and high laser transmitting power,the accumulated nonlinear effect in the optical fiber becomes the main factor affecting the communication performance of the system,and the signal is distorted due to the nonlinearity,which reduces the communication quality.In common traditional nonlinear compensation algorithms,such as digital back propagation(DBP)algorithm and phase conjugate two-wave propagation algorithm,there are some shortcomings such as high complexity of calculation and reduction of transmission capacity.Compared with the above nonlinear compensation algorithm,the machine learning has the advantages of relatively small computation,fast calculation speed and strong nonlinear fitting ability.In this work,the artificial neural network algorithm in machine learning is used to compensate the accumulated nonlinear effects in optical fibers.In this work,the source and principle of nonlinear effect in optical fiber will be discussed at first,then the structure and working principle of neural network model summarized,and then proposes an artificial neural network algorithm to compensate the accumulated nonlinear effect in optical fiber is investigated after that.Finally,the nonlinear equalization ability of neural network is verified by software simulation optical communication system.The simulation results show that the error rate of the system is reduced and the system constellation is good.Two kinds of optical communication systems are involved.One is the conventional high-order high-speed coherent optical communication system,and the complex network structure is used to balance the complex signal transmitted in the system.The other one is a single fiber bidirectional reflective modulated coherent optical communication system studied in the laboratory in the past two years,and the neural network model is applied to this reflective system structure for the first time.It shows the effectiveness of compensating the nonlinear effect of optical fiber based on artificial neural network.