基于EGN辅助机器学习的多周期网络规划QoT估计
网络与互联网体系结构
2023-05-15 v1
摘要
光纤网络中快速增长的流量需求要求光路径配置具备灵活性与准确性，为此快速且准确的传输质量(QoT)估计至关重要。本文提出一种满足上述要求的基于机器学习(ML)的QoT估计方法。所提出的梯度提升ML模型使用预计算的每信道自信道干扰值作为具代表性且压缩的特征，以估计灵活栅格网络中的非线性干扰。以增强型高斯噪声(GN)模型仿真为基线，该ML模型取得约0.1 dB的平均绝对信噪比误差，优于GN模型。针对三种不同网络拓扑与复杂度各异的网络规划方法，开展了以ML与GN作为路径计算单元(PCE)进行比较的多周期网络规划研究。结果表明，ML PCE在所有拓扑上均能匹配或略优于GN PCE的性能，同时将网络规划计算时间最多缩短70%。
引用
@article{arxiv.2305.07332,
title = {QoT estimation using EGN-assisted machine learning for multi-period network planning},
author = {Jasper Müller and Sai Kireet Patri and Tobias Fehenberger and Helmut Griesser and Jörg-Peter Elbers and Carmen Mas-Machuca},
journal= {arXiv preprint arXiv:2305.07332},
year = {2023}
}
备注
This work has been partially funded in the framework of the CELTIC-NEXT project AI-NET-PROTECT (Project ID C2019/3-4) by the German Federal Ministry of Education and Research (#16KIS1279K). Carmen Mas-Machuca acknowledges the support by the Federal Ministry of Education and Research of Germany (BMBF) in the programme of "Souver\"an. Digital. Vernetzt." joint project 6G-life (#16KISK002)