no code implementations • 4 Jul 2023 • Sasipim Srivallapanondh, Pedro J. Freire, Ashraful Alam, Nelson Costa, Bernhard Spinnler, Antonio Napoli, Egor Sedov, Sergei K. Turitsyn, Jaroslaw E. Prilepsky
For the first time, multi-task learning is proposed to improve the flexibility of NN-based equalizers in coherent systems.
no code implementations • 16 May 2023 • Sasipim Srivallapanondh, Pedro J. Freire, Antonio Napoli, Sergei K. Turitsyn, Jaroslaw E. Prilepsky
To reduce the complexity of the hardware implementation of neural network-based optical channel equalizers, we demonstrate that the performance of the biLSTM equalizer with approximated activation functions is close to that of the original model.
no code implementations • 9 Dec 2022 • Pedro J. Freire, Sasipim Srivallapanondh, Michael Anderson, Bernhard Spinnler, Thomas Bex, Tobias A. Eriksson, Antonio Napoli, Wolfgang Schairer, Nelson Costa, Michaela Blott, Sergei K. Turitsyn, Jaroslaw E. Prilepsky
The main results are divided into three parts: a performance comparison, an analysis of how activation functions are implemented, and a report on the complexity of the hardware.
no code implementations • 8 Dec 2022 • Sasipim Srivallapanondh, Pedro J. Freire, Bernhard Spinnler, Nelson Costa, Antonio Napoli, Sergei K. Turitsyn, Jaroslaw E. Prilepsky
To circumvent the non-parallelizability of recurrent neural network-based equalizers, we propose knowledge distillation to recast the RNN into a parallelizable feedforward structure.
no code implementations • 26 Aug 2022 • Pedro J. Freire, Antonio Napoli, Diego Arguello Ron, Bernhard Spinnler, Michael Anderson, Wolfgang Schairer, Thomas Bex, Nelson Costa, Sergei K. Turitsyn, Jaroslaw E. Prilepsky
In this work, we propose and evaluate a Bayesian optimization-assisted compression, in which the hyperparameters of the compression are chosen to simultaneously reduce complexity and improve performance.
no code implementations • 24 Jun 2022 • Pedro J. Freire, Michael Anderson, Bernhard Spinnler, Thomas Bex, Jaroslaw E. Prilepsky, Tobias A. Eriksson, Nelson Costa, Wolfgang Schairer, Michaela Blott, Antonio Napoli, Sergei K. Turitsyn
For the first time, recurrent and feedforward neural network-based equalizers for nonlinearity compensation are implemented in an FPGA, with a level of complexity comparable to that of a dispersion equalizer.
no code implementations • 24 Jun 2022 • Pedro Freire, Sasipim Srivallapanondh, Antonio Napoli, Jaroslaw E. Prilepsky, Sergei K. Turitsyn
We provide and link four software-to-hardware complexity measures, defining how the different complexity metrics relate to the layers' hyper-parameters.
no code implementations • 5 Apr 2022 • Vladislav Neskorniuk, Andrea Carnio, Domenico Marsella, Sergei K. Turitsyn, Jaroslaw E. Prilepsky, Vahid Aref
Autoencoder-based deep learning is applied to jointly optimize geometric and probabilistic constellation shaping for optical coherent communication.
no code implementations • 25 Feb 2022 • Pedro J. Freire, Bernhard Spinnler, Daniel Abode, Jaroslaw E. Prilepsky, Abdallah A. I. Ali, Nelson Costa, Wolfgang Schairer, Antonio Napoli, Andrew D. Ellis, Sergei K. Turitsyn
We introduce the domain adaptation and randomization approach for calibrating neural network-based equalizers for real transmissions, using synthetic data.
no code implementations • 30 Sep 2021 • Pedro J. Freire, Antonio Napoli, Bernhard Spinnler, Nelson Costa, Sergei K. Turitsyn, Jaroslaw E. Prilepsky
This paper performs a detailed, multi-faceted analysis of key challenges and common design caveats related to the development of efficient neural networks (NN) nonlinear channel equalizers in coherent optical communication systems.
no code implementations • 28 Sep 2021 • Pedro J. Freire, Jaroslaw E. Prilepsky, Yevhenii Osadchuk, Sergei K. Turitsyn, Vahid Aref
We examine here what type of predictive modelling, classification, or regression, using neural networks (NN), fits better the task of soft-demapping based post-processing in coherent optical communications, where the transmission channel is nonlinear and dispersive.
no code implementations • 17 Sep 2021 • Pedro J. Freire, Yevhenii Osadchuk, Antonio Napoli, Bernhard Spinnler, Wolfgang Schairer, Nelson Costa, Jaroslaw E. Prilepsky, Sergei K. Turitsyn
Addressing the neural network-based optical channel equalizers, we quantify the trade-off between their performance and complexity by carrying out the comparative analysis of several neural network architectures, presenting the results for TWC and SSMF set-ups.
no code implementations • 15 Sep 2021 • Diego R. Arguello, Pedro J. Freire, Jaroslaw E. Prilepsky, Antonio Napoli, Morteza Kamalian-Kopae, Sergei K. Turitsyn
The deployment of artificial neural networks-based optical channel equalizers on edge-computing devices is critically important for the next generation of optical communication systems.
no code implementations • 26 Jul 2021 • Vladislav Neskorniuk, Andrea Carnio, Vinod Bajaj, Domenico Marsella, Sergei K. Turitsyn, Jaroslaw E. Prilepsky, Vahid Aref
We present a novel end-to-end autoencoder-based learning for coherent optical communications using a "parallelizable" perturbative channel model.
no code implementations • 24 Jun 2021 • Pedro J. Freire, Daniel Abode, Jaroslaw E. Prilepsky, Sergei K. Turitsyn
Transfer learning is proposed to adapt an NN-based nonlinear equalizer across different launch powers and modulation formats using a 450km TWC-fiber transmission.
no code implementations • 24 Jun 2021 • Pedro J. Freire, Yevhenii Osadchuk, Bernhard Spinnler, Wolfgang Schairer, Antonio Napoli, Nelson Costa, Jaroslaw E. Prilepsky, Sergei K. Turitsyn
We propose a convolutional-recurrent channel equalizer and experimentally demonstrate 1dB Q-factor improvement both in single-channel and 96 x WDM, DP-16QAM transmission over 450km of TWC fiber.
no code implementations • 11 Apr 2021 • Pedro J. Freire, Daniel Abode, Jaroslaw E. Prilepsky, Nelson Costa, Bernhard Spinnler, Antonio Napoli, Sergei K. Turitsyn
We evaluate the capability of transfer learning to adapt the NN to changes in the launch power, modulation format, symbol rate, or even fiber plants (different fiber types and lengths).
no code implementations • 15 Mar 2021 • Pedro J. Freire, Yevhenii Osadchuk, Bernhard Spinnler, Antonio Napoli, Wolfgang Schairer, Nelson Costa, Jaroslaw E. Prilepsky, Sergei K. Turitsyn
We present the results of the comparative analysis of the performance versus complexity for several types of artificial neural networks (NNs) used for nonlinear channel equalization in coherent optical communication systems.