Low-Complexity Adaptive Chromatic Dispersion Estimation Scheme Using Machine Learning for Coherent Long-Reach Passive Optical Networks

Jin Li,Danshi Wang,Min Zhang

doi:10.1109/jphot.2019.2936426

Abstract

In the coherent long-reach passive optical networks (LRPON), it is crucial to propose cost-effective digital signal processing (DSP) technologies to reduce the overall complexity and power consumption. This paper has proposed a low-complexity chromatic dispersion (CD) estimation scheme based on deep neural networks (DNN) and the error vector magnitude (EVM). To add comparisons, the performances of CD estimation schemes using other two well-known machine learning algorithms including the k-nearest neighbor (KNN) and the decision tree (DT) have also been investigated. The simulation results show that the proposed CD estimation scheme is effective in the coherent LRPON with the quadrature phase shift keying (QPSK) and 16-ary quadrature amplitude modulation (QAM) systems at 14Gbaud rate, 28Gbaud rate and 56Gbaud rate. The comprehensive performances of the DNN outperform those of the KNN and the DT. The mean estimation error of the DNN is less than 20ps/nm within the 100 km access distance in the 28Gbaud QPSK/16QAM systems. What's more, compared with classical methods using the CD scanning and frequent domain equalizers (FDE), the computation complexity of the proposed CD estimation scheme based on the DNN-EVM has been respectively reduced by 72.3 times, 86.7 times and 2.8 times about the amount of multipliers, adders and comparators.

Highlights

In the future fixed-mobile convergent passive optical network (PON), driven by emerging bandwidththirsty network services such as the high-definite video, the virtual/augmented reality and the cloud computing, the increasing capability demand between the optical line terminal (OLT) and the optical network unit (ONU) will be improved from 10Gbit/s to beyond 100Gbit/s [1]
The simulation results show that the proposed chromatic dispersion (CD) estimation scheme is effective in the coherent long-reach passive optical networks (LRPON) with the quadrature phase shift keying (QPSK) and 16-ary quadrature amplitude modulation (QAM) systems at 14Gbaud rate, 28Gbaud rate and 56Gbaud rate
The results show that larger CD estimation range, more advanced coherent systems and the similar estimation accuracy are available in the proposed deep neural networks (DNN)-based approach and it is more suitable for the low-complexity CD estimation for the coherent LRPON

Summary

Introduction

In the future fixed-mobile convergent passive optical network (PON), driven by emerging bandwidththirsty network services such as the high-definite video, the virtual/augmented reality and the cloud computing, the increasing capability demand between the optical line terminal (OLT) and the optical network unit (ONU) will be improved from 10Gbit/s to beyond 100Gbit/s [1]. In the metro and backbone networks, the adaptive CD compensation methods have the large dynamic range and they are generally based on the CD scanning and the FDE, where varied CD values with certain scanning steps are sent into the FDE module until the peak-to-average-power ratio (PAPR), the value of the Godard error function or the parameter extracted from the delay-tap plot of the received signal reach to the minimum value [11]–[13]. These adaptive CD compensation methods are straightforward and effective.

Principal of the Proposed Scheme

Simulation Results and Discussion

The CD Estimation Accuracy Performance

The Computation Complexity Analysis

Conclusions