Attention:The NSF Public Access Repository (PAR) system and access will be unavailable from 11:00 PM ET on Thursday, November 12th until 12:00 AM ET on Friday, November 13th due to maintenance. We apologize for the inconvenience.


Title: Multi-span optical power spectrum prediction using cascaded learning with one-shot end-to-end measurement
Scalable methods for optical transmission performance prediction using machine learning (ML) are studied in metro reconfigurable optical add-drop multiplexer (ROADM) networks. A cascaded learning framework is introduced to encompass the use of cascaded component models for end-to-end (E2E) optical path prediction augmented with different combinations of E2E performance data and models. Additional E2E optical path data and models are used to reduce the prediction error accumulation in the cascade. Off-line training (pre-trained prior to deployment) and transfer learning are used for component-level erbium-doped fiber amplifier (EDFA) gain models to ensure scalability. Considering channel power prediction, we show that the data collection process of the pre-trained EDFA model can be reduced to only 5% of the original training set using transfer learning. We evaluate the proposed method under three different topologies with field deployed fibers and achieve a mean absolute error of 0.16 dB with a single (one-shot) E2E measurement on the deployed 6-span system with 12 EDFAs.  more » « less
Award ID(s):
1827923 2211944 2330333
PAR ID:
10608639
Author(s) / Creator(s):
; ; ; ;
Publisher / Repository:
IEEE
Date Published:
Journal Name:
Journal of Optical Communications and Networking
Volume:
17
Issue:
1
ISSN:
1943-0620
Page Range / eLocation ID:
A23
Subject(s) / Keyword(s):
Sponsoring Org:
National Science Foundation
More Like this
  1. Scalable and efficient methods for predicting optical transmission performance using machine learning (ML) and cascaded learning (CL) are investigated in metro multi-span optical networks. We extend the CL framework beyond multi-span power spectrum prediction to enable accurate estimation of optical signal-to-noise (OSNR) and generalized signal-to-noise (GSNR) under dynamic channel loading, integrating cascaded component models for end-to-end (E2E) optical link performance prediction. To mitigate error accumulation inherent in cascaded predictions, additional E2E optical link measurements and models are incorporated. To ensure scalability, component-level models for erbium-doped fiber amplifier (EDFA) gain and noise figure are pre-trained prior to deployment. We demonstrate that the data collection effort for pre-training the EDFA model can be reduced to only 5.5% of the original training set through transfer learning. Furthermore, Gaussian noise (GN)-based analytical models are leveraged to assist in the training of ML-based models for fiber loss and nonlinear impairments. The proposed approach is evaluated under four distinct optical link configurations. On a five-span system comprising six EDFAs, the method achieves a mean absolute error of 0.22 and 0.13 dB for OSNR prediction of background channels and GSNR prediction for 400 GbE channels, respectively. 
    more » « less
  2. Optical networks satisfy high bandwidth and low latency requirements for telecommunication networks and data center interconnection. To improve network resource utilization, machine learning (ML) is used to accurately model optical amplifiers such as erbium-doped fiber amplifiers (EDFAs), which impact end-to-end system performance such as quality of transmission. However, a comprehensive measurement dataset is required for ML to accurately predict an EDFA’s wavelength-dependent gain. We present an open dataset consisting of 202,752 gain spectrum measurements collected from 16 commercial-grade reconfigurable optical add–drop multiplexer (ROADM) booster and pre-amplifier EDFAs under varying gain settings and diverse channel-loading configurations over 2,785 hours in total, with a total dataset size of 3.1 GB. With this EDFA dataset, we implemented component-level deep-neural-network-based EDFA models and use transfer learning (TL) to transfer the EDFA model among 16 ROADM EDFAs, which achieve less than 0.18/0.24 dB mean absolute error for booster/pre-amplifier gain prediction using only 0.5% of the full target training set. We also showed that TL reduces the EDFA data collection requirements on a new gain setting or a different type of EDFA on the same ROADM. 
    more » « less
  3. We implement a cascaded learning framework using component-level EDFA models for optical power spectrum prediction in multi-span networks, achieving a mean absolute error of 0.17 dB across 6 spans and 12 EDFAs with only one-shot measurement. 
    more » « less
  4. We implement a cascaded learning framework leveraging three different EDFA and fiber component models for OSNR and GSNR prediction, achieving MAEs of 0.20 and 0.14 dB over a 5-span network under dynamic channel loading. 
    more » « less
  5. Accurate modeling of the gain spectrum in erbium-doped fiber amplifiers (EDFAs) is essential for optimizing optical network performance, particularly as networks evolve toward multi-vendor solutions. In this work, we propose a generalized few-shot transfer learning architecture based on a semi-supervised self-normalizing neural network (SS-NN) that leverages internal EDFA features—such as VOA input/output power and attenuation—to improve gain spectrum prediction. Our SS-NN model employs a two-phase training strategy comprising unsupervised pre-training with noise-augmented measurements and supervised fine-tuning with a custom-weighted MSE loss. Furthermore, we extend the framework with transfer learning (TL) techniques that enable both homogeneous (same-feature space) and heterogeneous (different-feature sets) model adaptation across booster, pre-amplifier, and ILA EDFAs. To address feature mismatches in heterogeneous TL, we incorporate a covariance matching loss to align second-order feature statistics between the source and target domains. Extensive experiments conducted across 26 EDFAs in the COSMOS and Open Ireland testbeds demonstrate that the proposed approach significantly reduces the number of measurement requirements on the system while achieving lower mean absolute errors and improved error distributions compared to benchmark methods. 
    more » « less