卷积神经网络参数优化的光纤通信信道均衡研究

Research on optical fiber communication channel equalization based on convolutional neural network parameter optimization

为了提高光纤通信信道均衡的收敛速度,研究了卷积神经网络参数优化的光纤通信非线性信道均衡方法。分析光信号传输模型和引起信道失衡的主要因素,色散和扰动影响易导致光纤信号失真。考虑信道非线性失真因素,引入量子粒子群优化算法优化卷积神经网络。通过该算法搜索训练卷积神经网络最优超参数,以提高收敛性。优化后获取全局最优解,实现了光纤通信非线性信道均衡。实验结果表明,所提方法具有较好的均衡性能和较快收敛速度,在100次迭代时均方误差已趋于平稳,均方误差值为10-1。在信道波动后能够较快恢复到平稳状态,具有较强跟踪能力。

In order to improve the convergence speed of optical fiber communication channel equalization,a non-linear channel equalization method based on convolutional neural network parameter optimization is studied.The trans-mission model of optical signal and the main factors causing channel imbalance are analyzed.Dispersion and disturb-ance are easy to cause fiber signal distortion.Considering the nonlinear distortion of the channel,the convolution neu-ral network is optimized by introducing the quantum particle swarm optimization algorithm.The algorithm is used to search and train the optimal hyperparameters of convolutional neural networks to improve the convergence.After opti-mization,the global optimal solution is obtained,and the nonlinear channel equalization of optical fiber communication is realized.The experimental results show that the proposed method has better equilibrium performance and faster con-vergence speed.After 100 iterations,the mean square error has become stable,and the mean square error is 10-1.It can quickly recover to a stable state after channel fluctuation and has strong tracking ability.