Transfer Learning for Neural Networks-Based Equalizers in Coherent Optical Systems

Freire, Pedro J.; Abode, Daniel; Prilepsky, Jaroslaw E.; Costa, Nelson; Spinnler, Bernhard; Napoli, Antonio; Turitsyn, Sergei K.

doi:10.1109/jlt.2021.3108006

Cited by 42 publications

(23 citation statements)

References 48 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A shorter section is then trained and reused for the subsequent sections. This method is also known as a transfer learning method, similar to what has been shown in [11]. By stacking of sections, we show in a proof-of concept experiment a total training time reduction of about 67% with a SNR gain that is close to what one would find for a fully trained network.…”

Section: Introductionsupporting

confidence: 62%

Reducing Training Time of Deep Learning Based Digital Backpropagation by Stacking

Bitachon

Eppenberger

Baeuerle³

et al. 2022

IEEE Photon. Technol. Lett.

View full text Add to dashboard Cite

A method for reducing the training time of a deep learning based digital backpropagation (DL-DBP) is presented. The method is based on dividing a link into smaller sections. A smaller section is then compensated by the DL-DBP algorithm and the same trained model is then reapplied to the subsequent sections. We show in a 32 GBd 16QAM 2400 km 5-channel wavelength division multiplexing transmission link experiment that the proposed stacked DL-DBPs provides a 0.41 dB gain with respect to linear compensation scheme. This needs to be compared with a 0.56 dB gain achieved by a non-stacked DL-DBPs compensated scheme for the price of a 203% increase in total training time. Furthermore, it is shown that by only training the last section of the stacked DL-DBP, one can increase the compensation performance to 0.48 dB.

show abstract

Section: Introductionsupporting

confidence: 62%

Reducing Training Time of Deep Learning Based Digital Backpropagation by Stacking

Bitachon

Eppenberger

Baeuerle³

et al. 2022

IEEE Photon. Technol. Lett.

View full text Add to dashboard Cite

show abstract

“…We used the same simulator as described in Refs. 10 , 29 , to generate our training and testing datasets, and the same procedure to training the NN-based equalizer (see “ Numerical setup and neural network model ” subsection in “ Methods ” for more details). In our configuration, the NN is placed at the receiver (Rx) side after the Integrated Coherent Receiver (ICR), Analog-Digital Converter (ADC), and DSP block.…”

Section: Resultsmentioning

confidence: 99%

“…In this case, it may be a point of concern. This issue has already been addressed in previous studies 29 , where it has been demonstrated that using transfer learning can drastically reduce the training time and training data requirements when changes on the transmission setup occur.…”

Section: Methodsmentioning

confidence: 91%

Experimental implementation of a neural network optical channel equalizer in restricted hardware using pruning and quantization

Costa

Freire

Prilepsky

et al. 2022

Sci Rep

Self Cite

View full text Add to dashboard Cite

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. However, this is still a highly challenging problem, mainly due to the computational complexity of the artificial neural networks (NNs) required for the efficient equalization of nonlinear optical channels with large dispersion-induced memory. To implement the NN-based optical channel equalizer in hardware, a substantial complexity reduction is needed, while we have to keep an acceptable performance level of the simplified NN model. In this work, we address the complexity reduction problem by applying pruning and quantization techniques to an NN-based optical channel equalizer. We use an exemplary NN architecture, the multi-layer perceptron (MLP), to mitigate the impairments for 30 GBd 1000 km transmission over a standard single-mode fiber, and demonstrate that it is feasible to reduce the equalizer’s memory by up to 87.12%, and its complexity by up to 78.34%, without noticeable performance degradation. In addition to this, we accurately define the computational complexity of a compressed NN-based equalizer in the digital signal processing (DSP) sense. Further, we examine the impact of using hardware with different CPU and GPU features on the power consumption and latency for the compressed equalizer. We also verify the developed technique experimentally, by implementing the reduced NN equalizer on two standard edge-computing hardware units: Raspberry Pi 4 and Nvidia Jetson Nano, which are used to process the data generated via simulating the signal’s propagation down the optical-fiber system.

show abstract

“…The first one is straightforward, where we retrain the NN from scratch using a different dataset. The second strategy can make use of the pretrained NFT-Net model and utilise domain randomisation and adaptation 68,69 . We believe that after the retraining procedure, the NFT-Net (or some of its modifications, if we find that the capacity of the proposed NN architecture is insufficient to account for some complicated real-world effects) should be capable to account for the spurious soliton emergence and involved noise properties taking place in the realistic optical transmission systems.…”

Section: Discussionmentioning

confidence: 99%

Neural networks for computing and denoising the continuous nonlinear Fourier spectrum in focusing nonlinear Schrödinger equation

Sedov

Freire

Seredin

et al. 2021

Sci Rep

Self Cite

View full text Add to dashboard Cite

We combine the nonlinear Fourier transform (NFT) signal processing with machine learning methods for solving the direct spectral problem associated with the nonlinear Schrödinger equation. The latter is one of the core nonlinear science models emerging in a range of applications. Our focus is on the unexplored problem of computing the continuous nonlinear Fourier spectrum associated with decaying profiles, using a specially-structured deep neural network which we coined NFT-Net. The Bayesian optimisation is utilised to find the optimal neural network architecture. The benefits of using the NFT-Net as compared to the conventional numerical NFT methods becomes evident when we deal with noise-corrupted signals, where the neural networks-based processing results in effective noise suppression. This advantage becomes more pronounced when the noise level is sufficiently high, and we train the neural network on the noise-corrupted field profiles. The maximum restoration quality corresponds to the case where the signal-to-noise ratio of the training data coincides with that of the validation signals. Finally, we also demonstrate that the NFT b-coefficient important for optical communication applications can be recovered with high accuracy and denoised by the neural network with the same architecture.

show abstract

Transfer Learning for Neural Networks-Based Equalizers in Coherent Optical Systems

Cited by 42 publications

References 48 publications

Reducing Training Time of Deep Learning Based Digital Backpropagation by Stacking

Reducing Training Time of Deep Learning Based Digital Backpropagation by Stacking

Experimental implementation of a neural network optical channel equalizer in restricted hardware using pruning and quantization

Neural networks for computing and denoising the continuous nonlinear Fourier spectrum in focusing nonlinear Schrödinger equation

Contact Info

Product

Resources

About