基于演化合成神经网络的光通信MIMO检测算法

MIMO DETECTION ALGORITHM OF OPTICAL COMMUNICATION BASED ON EVOLUTIONARY SYNTHESIS NEURAL NETWORK

针对光通信系统中多模式传输所引起的非线性耦合问题,提出一种基于演化合成神经网络的光通信多输入多输出(Multiple Input Multiple Output, MIMO)检测算法。给出光通信系统MIMO的深度神经网络模型,利用深度神经网络强大的非线性学习能力解决多模光通信的非线性耦合问题。采用演化合成技术训练全连接神经网络,以光通信MIMO的特点作为演化合成的环境因子,逐渐缩小神经网络的规模,提高神经网络的结构效率。基于真实光通信系统的仿真实验结果显示,该算法提高了神经网络的效率,并且实现了较好的光通信MIMO检测效果。

Aiming at the problem of the nonlinear coupling caused by the multiple modes transmission in optical communication system,we propose a multiple input multiple output(MIMO)detection algorithm of optical communication based on evolutionary synthesis neural network.The deep neural network model of optical communication system MIMO was giveh,and we took advantage of strong nonlinear fitting ability of deep neural network to solve the nonlinear coupling problem of multiple mode optical communication.We adopted evolutionary synthesis technique to train the full connected neural network,and set the attributes of optical communication MIMO as the environment factors of evolutionary synthesis.It helped to reduce the size of neural network and improve the structure efficiency of neural network.The simulation experimental results based on real optical communication system show that the proposed algorithm improves the efficiency of neural network,and it also realizes ideal effect for optical communication MIMO detection.