Summary

International Conference on Emerging Technologies for Communications

2022

Session Number:S13

Session:

Number:S13-1

Comparison of overfitting characteristics of ANN- and VSTF-based nonlinear equalizers for repeated random bit patterns in optical communication systems

Kai Ikuta,  Jinya Nakamura,  Daisuke Motai,  Moriya Nakamura,  

pp.-

Publication Date:2022/11/29

Online ISSN:2188-5079

DOI:10.34385/proc.72.S13-1

PDF download (739.2KB)

Summary:
Nonlinear equalizers based on the artificial neural network (ANN) and Volttera series transfer function (VSTF) have been studied to compensate for nonlinear waveform distortion caused by optical nonlinearities in optical transmission systems. However, overfitting characteristics of both methods have not been fully clarified. In this paper, we investigated and compared the overfitting of ANN- and VSTF-based nonlinear equalizers for repeated random bit patterns.