用于光纤传输均衡的低复杂度卷积神经网络
信号处理
2022-10-12 v1 人工智能
摘要
提出了一种卷积神经网络以减轻光纤传输效应，与替代均衡器相比，可训练参数减少了五倍；且在复杂度相当的情况下，与 DBP 相比，MSE 改善了 3.5 dB。
引用
@article{arxiv.2210.05454,
title = {Low Complexity Convolutional Neural Networks for Equalization in Optical Fiber Transmission},
author = {Mohannad Abu-romoh and Nelson Costa and Antonio Napoli and João Pedro and Yves Jaouën and Mansoor Yousefi},
journal= {arXiv preprint arXiv:2210.05454},
year = {2022}
}
备注
2 pages, 3 figures. Submitted to the OSA Advanced Photonics Congress 2021. Presented in Signal Processing in Photonic Communications (SPPCom) 2021. From the session: Neural Networks Applications for Photonic Systems (SpM5C)