• Technical Conference: 

    09 – 14 May 2021

  • Exhibition: 

    10 – 14 May 2021

Home > Schedule > Post Session

Post Session


STh1F

Network Management and Machine Learning

Presider: Francesco Da Ros, DTU Fotonik

Authentication Required

Access to this content is only available to registered attendees. Registered attendees, please login now to access the content.

Presentations

Physical Layer Security Management Through Machine Learning (STh1F.1)
Presenter: Marija Furdek, Chalmers Tekniska Högskola

We examine the applicability, accuracy and scalability of supervised, unsupervised and semi-supervised learning techniques in diagnosing optical layer security breaches, and discuss how to deploy and incorporate these techniques into network management architecture and procedures.

Authors:Marija Furdek, Chalmers Tekniska Högskola / Carlos Natalino, Chalmers Tekniska Högskola

  Paper

Deep Learning Assisted Pre-Carrier Phase Recovery EVM Estimation for Coherent Transmission Systems (STh1F.2)
Presenter: Yuchuan Fan, KTH Royal Institute of Technology

We exploit deep supervised learning and amplitude histograms of coherent optical signals captured before carrier phase recovery (CPR) to perform time-sensitive and accurate error vector magnitude (EVM) estimation for 32 Gbaud mQAM signal monitoring purposes.

Authors:Yuchuan Fan, KTH Royal Institute of Technology / Aleksejs Udalcovs, Research Institutes of Sweden / Xiaodan Pang, KTH Royal Institute of Technology / Carlos Natalino, Chalmers University of Technology / Richard Schatz, KTH Royal Institute of Technology / Marija Furdek, Chalmers University of Technology / Sergei Popov, KTH Royal Institute of Technology / Oskars Ozolins, KTH Royal Institute of Technology

  Paper

Performance-Enhanced Amplified O-Band WDM Transmission Using Machine Learning Based Equalization (STh1F.3)
Presenter: Yang Hong, University of Southampton

We investigate the performance of a machine learning-based equalization in an amplified 4×50-Gb/s O-band WDM system. The results show that the scheme offers significant receiver sensitivity improvements over decision-feedback equalization, especially at more dispersive wavelengths.

Authors:Yang Hong, University of Southampton / Stavros Deligiannidis, University of West Attica / Natsupa Taengnoi, University of Southampton / Kyle Bottrill, University of Southampton / Naresh Thipparapu, University of Southampton / Yu Wang, University of Southampton / Jayanta Sahu, University of Southampton / David Richardson, University of Southampton / Charis Mesaritakis, University of the Aegean / Adonis Bogris, University of West Attica / Periklis Petropoulos, University of Southampton

  Paper

Maxwell-Boltzmann PMF Design Using Machine Learning for Reconfigurable Optical Fiber Networks (STh1F.4)
Presenter: Henrik Hansen, DTU Fotonik

A neural network is used to predict the optimal Maxwell-Boltzmann probabilistic constellation shaping for a nonlinear channel with inline dispersion-compensation. The network uses only system parameters available at the transmitter and thus requires no feedback

Authors:Henrik Hansen, DTU Fotonik / Metodi Yankov, DTU Fotonik / Leif Oxenløwe, DTU Fotonik / Søren Forchhammer, DTU Fotonik

  Paper

Dual-Learning Based Neural Networks for Short-Reach Optical Communications (STh1F.5)
Presenter: Hao Chen, South China University of Technology

We propose a novel dual-learning (DL) based NN that operates in a self-regularization manner to improve the generalization ability and show in 28Gbaud PAM-4 experiments that this method outperforms filtering solutions and conventional NN methods.

Authors:Hao Chen, South China University of Technology / Xing Liu, South China University of Technology / Zhaoquan Fan, South China University of Technology / Chengju Hu, South China University of Technology / Jian Zhao, South China University of Technology

  Paper

SOAs and Digital Linearization in Optical Networks -- a Stochastic Investigation (STh1F.6)
Presenter: Jacqueline SIME, Ecole Nationale d'Ingénieurs de Brest, Lab-STICC, CNRS, UMR 6285

Digital predistortion has recently spurred interest in photonics. In this paper, the authors perform a sensitivity analysis of three digital predistortion algorithms and demonstrate an increase in performance and, in some cases, robustness to uncertainties.

Authors:Jacqueline SIME, Ecole Nationale d'Ingénieurs de Brest, Lab-STICC, CNRS, UMR 6285 / Pascal Morel, Ecole Nationale d'Ingénieurs de Brest, Lab-STICC, CNRS, UMR 6285 / Igor Stievano, Politecnico di Torino, Dipartimento di Elettronica e Telecomunicazioni / Mihai Telescu, Univ Brest, Lab-STICC, CNRS, UMR 6285 / Noël Tanguy, Univ Brest, Lab-STICC, CNRS, UMR 6285 / Stéphane Azou, Ecole Nationale d'Ingénieurs de Brest, Lab-STICC, CNRS, UMR 6285

  Paper

Experimental Demonstration of Remotely Controlled and Powered Optical Switching Based on Laser-Delivered Bias and Control Signals (STh1F.7)
Presenter: Ahmad Fallahpour, University of Southern California

We experimentally demonstrate a remotely controlled and powered optical switch. The data channel, switching control signal and optical power are wavelength multiplexed in a 7.66 km SMF and transmitted to the remote switch. We transmit a 20 Gbaud QPSK signal that is switched at 2 Mb/s rate.

Authors:Ahmad Fallahpour, University of Southern California / Amir Minoofar, University of Southern California / Fatemeh Alishahi, University of Southern California / kaiheng zou, University of Southern California / Samer Idres, University of Southern California / Hossein Hashemi, University of Southern California / Jonathan Habif, University of Southern California / Moshe Tur, Tel Aviv University / Alan Eli Willner, University of Southern California

  Paper