论文部分内容阅读
基于异步延迟采样和人工神经网络统计学习提出了一种光通信性能监测方法。通过对高速光信号进行异步延迟采样,获得信号二维幅度直方图,然后提取其中特征参数并对人工神经网络进行训练,最后以人工神经网络的预测输出实现对光信号损伤的监测。构建10Gb/s非归零码开关键控,40Gb/s光学双二进制码和归零码差分移相键控光通信仿真系统,并对光信噪比、色散和偏振模色散损伤进行监测。仿真结果表明,所提方法对被监测光信号的速率、码型调制格式透明,可同时准确监测多种并存的传输损伤,损伤参数监测误差小于5%。该方法具有电域处理带宽要求低、采样机制简单的特点,适用于分布式在线光性能监测。
Based on asynchronous delayed sampling and artificial neural network statistical learning, a method of optical communication performance monitoring is proposed. Through the asynchronous delayed sampling of high speed optical signal, the two-dimensional amplitude histogram of the signal is obtained, and then the characteristic parameters are extracted and the artificial neural network is trained. Finally, the optical signal damage monitoring is realized by the prediction output of artificial neural network. The simulation system of 10Gb / s non-return-to-zero on-off keying, 40Gb / s optical binary code and zero-return differential phase-shift keying optical communication was constructed and the optical signal to noise ratio, dispersion and polarization mode dispersion damage were monitored. The simulation results show that the proposed method is transparent to the rate and pattern modulation of the optical signals being monitored, and can simultaneously and accurately monitor a variety of transmission impairments. The monitoring error of the damage parameters is less than 5%. This method has the characteristics of low processing bandwidth in the electric domain and simple sampling mechanism, which is suitable for the distributed on-line optical performance monitoring.