AdaNN: Adaptive Neural Network-Based Equalizer via Online Semi-Supervised Learning

The demand for high speed data transmission has increased rapidly, leading to advanced optical communication techniques. In the past few years, multiple equalizers based on neural network (NN) have been proposed to recover signal from nonlinear distortions. However, previous experiments mainly focused on achieving low bit error rate (BER) on certain dataset with an offline-trained NN, neglecting the generalization ability of NN-based equalizer when the properties of optical link change. The development of efficient online training scheme is urgently needed. In this article, we’ve proposed an adaptive online training scheme for NN-based equalizer. Our scheme can fine-tune NN parameters with only received signals rather than original symbols as reference. Once the system enters online stage, no labeled training sequence needs to be transmitted. By introducing data augmentation and virtual adversarial training, the convergence speed has been accelerated by 4.5 times, compared with decision-directed self-training. The proposed adaptive NN-based equalizer is called “AdaNN”. Its BER has been evaluated under two scenarios: a 56 Gb/s PAM4-modulated VCSEL-MMF optical link (100-m), and a 32 Gbaud 16QAM-modulated Nyquist-WDM system (960-km SSMF). In our experiments, with the help of AdaNN, BER values can be quickly stabilized below 1e-3 after receiving $\text{10}^\text{5}$ unlabeled symbols. AdaNN shows great performance improvement compared with non-adaptive NN and conventional MLSE.

[1]  Timothy J. O'Shea,et al.  Physical Layer Communications System Design Over-the-Air Using Adversarial Networks , 2018, 2018 26th European Signal Processing Conference (EUSIPCO).

[2]  Shin Ishii,et al.  Virtual Adversarial Training: A Regularization Method for Supervised and Semi-Supervised Learning , 2017, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[3]  Lin Sun,et al.  Machine Learning Adaptive Receiver for PAM-4 Modulated Optical Interconnection Based on Silicon Microring Modulator , 2018, Journal of Lightwave Technology.

[4]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[5]  Chia-Chien Wei,et al.  Employing Deep Neural Network for High Speed 4-PAM Optical Interconnect , 2017, 2017 European Conference on Optical Communication (ECOC).

[6]  Geoffrey Ye Li,et al.  Deep Learning Based on Orthogonal Approximate Message Passing for CP-Free OFDM , 2019, ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[7]  Yoram Singer,et al.  Adaptive Subgradient Methods for Online Learning and Stochastic Optimization , 2011, J. Mach. Learn. Res..

[8]  Mérouane Debbah,et al.  Wireless Networks Design in the Era of Deep Learning: Model-Based, AI-Based, or Both? , 2019, IEEE Transactions on Communications.

[9]  Harri Valpola,et al.  Weight-averaged consistency targets improve semi-supervised deep learning results , 2017, ArXiv.

[10]  Chia-Chien Wei,et al.  Convolutional Neural Network based Nonlinear Classifier for 112-Gbps High Speed Optical Link , 2018, 2018 Optical Fiber Communications Conference and Exposition (OFC).

[11]  C. Schow,et al.  A 71-Gb/s NRZ Modulated 850-nm VCSEL-Based Optical Link , 2015, IEEE Photonics Technology Letters.

[12]  Geoffrey E. Hinton,et al.  Learning representations by back-propagating errors , 1986, Nature.

[13]  Nathan Srebro,et al.  The Marginal Value of Adaptive Gradient Methods in Machine Learning , 2017, NIPS.

[14]  Jakob Hoydis,et al.  An Introduction to Deep Learning for the Physical Layer , 2017, IEEE Transactions on Cognitive Communications and Networking.

[15]  Geoffrey E. Hinton,et al.  On the importance of initialization and momentum in deep learning , 2013, ICML.

[16]  Vishnu Raj,et al.  Backpropagating Through the Air: Deep Learning at Physical Layer Without Channel Models , 2018, IEEE Communications Letters.

[17]  Chia-Chien Wei,et al.  81.7% Complexity Reduction of Volterra Nonlinear Equalizer by Adopting L1 Regularization Penalty in an OFDM Long-Reach PON , 2017, 2017 European Conference on Optical Communication (ECOC).

[18]  Jonathon Shlens,et al.  Explaining and Harnessing Adversarial Examples , 2014, ICLR.

[19]  Dong-Hyun Lee,et al.  Pseudo-Label : The Simple and Efficient Semi-Supervised Learning Method for Deep Neural Networks , 2013 .

[20]  Christian Neumeyr,et al.  Experimental performance evaluation of equalization techniques for 56 Gb/s PAM-4 VCSEL-based optical interconnects , 2015, 2015 European Conference on Optical Communication (ECOC).

[21]  Fan Zhang,et al.  Intra-channel nonlinearities mitigation in pseudo-linear coherent QPSK transmission systems via nonlinear electrical equalizer , 2009 .

[22]  Fan Zhang,et al.  Fiber Nonlinearity Mitigation in Single Carrier 400 G and 800 G Nyquist-WDM Systems , 2018, Journal of Lightwave Technology.

[23]  H. Robbins.  A Stochastic Approximation Method , 1951 .

[24]  Stephan ten Brink,et al.  Online Label Recovery for Deep Learning-based Communication through Error Correcting Codes , 2018, 2018 15th International Symposium on Wireless Communication Systems (ISWCS).

[25]  Andreas Leven,et al.  Applying Neural Networks in Optical Communication Systems: Possible Pitfalls , 2017, IEEE Photonics Technology Letters.

[26]  Jeffrey G. Andrews,et al.  Deep Learning-Based Channel Estimation for High-Dimensional Signals , 2019, 1904.09346.

[27]  Zuyuan He,et al.  Threshold-Based Pruned Retraining Volterra Equalization for 100 Gbps/Lane and 100-m Optical Interconnects Based on VCSEL and MMF , 2019, Journal of Lightwave Technology.

[28]  Xiaojin Zhu,et al.  --1 CONTENTS , 2006 .

[29]  J. Kahn,et al.  Compensation of Dispersion and Nonlinear Impairments Using Digital Backpropagation , 2008, Journal of Lightwave Technology.

[30]  Yue Liu,et al.  Experimental study of support vector machine based nonlinear equalizer for VCSEL based optical interconnect , 2018, Optics Communications.

[31]  Timo Aila,et al.  Temporal Ensembling for Semi-Supervised Learning , 2016, ICLR.

[32]  Chrysostomos L. Nikias,et al.  Adaptive equalization for PAM and QAM signals with neural networks , 1991, [1991] Conference Record of the Twenty-Fifth Asilomar Conference on Signals, Systems & Computers.

[33]  Zhangyuan Chen,et al.  Low computationally complex recurrent neural network for high speed optical fiber transmission , 2019, Optics Communications.

[34]  Xiaofeng Hu,et al.  Recurrent Neural Network (RNN) Based End-to-End Nonlinear Management for Symmetrical 50Gbps NRZ PON with 29dB+ Loss Budget , 2018, 2018 European Conference on Optical Communication (ECOC).

[35]  Geoffrey Ye Li,et al.  Deep Learning based End-to-End Wireless Communication Systems with Conditional GAN as Unknown Channel , 2019 .

[36]  Peter A. Andrekson,et al.  94-Gb/s 4-PAM Using an 850-nm VCSEL, Pre-Emphasis, and Receiver Equalization , 2016, IEEE Photonics Technology Letters.

[37]  Bin Dong,et al.  CURE: Curvature Regularization For Missing Data Recovery , 2019, SIAM J. Imaging Sci..

[38]  Geoffrey Ye Li,et al.  Power of Deep Learning for Channel Estimation and Signal Detection in OFDM Systems , 2017, IEEE Wireless Communications Letters.

[39]  Fan Zhang,et al.  High Speed Band-Limited 850-nm VCSEL Link Based on Time-Domain Interference Elimination , 2017, IEEE Photonics Technology Letters.

[40]  Bin Dong,et al.  Nostalgic Adam: Weighing more of the past gradients when designing the adaptive learning rate , 2019, IJCAI.

[41]  Stanley Osher,et al.  Low Dimensional Manifold Model for Image Processing , 2017, SIAM J. Imaging Sci..

[42]  Ronen Dar,et al.  Inter-Channel Nonlinear Interference Noise in WDM Systems: Modeling and Mitigation , 2015, Journal of Lightwave Technology.

[43]  Varghese Antony Thomas,et al.  DSP-Enabled 100 Gb/s PAM-4 VCSEL MMF Links , 2017, Journal of Lightwave Technology.

[44]  Polina Bayvel,et al.  End-to-End Deep Learning of Optical Fiber Communications , 2018, Journal of Lightwave Technology.

[45]  Peter A. Andrekson,et al.  Sensitivity Improvements in an 850-nm VCSEL-Based Link Using a Two-Tap Pre-Emphasis Electronic Filter , 2017, Journal of Lightwave Technology.