Neural Network-Based Compensation Techniques for Nonlinear Distortions in Fibre-Optic Links
Description
Optical signals experience nonlinear distortions - so called Kerr nonlinearities - when propagating through a fiber. This makes signal processing more complicated than in linear systems. Multiple ways to cope with this have been proposed in the past, including, e.g., digital backpropagation (DBP), Volterra equalizers or phase conjugation (see [1] for a good overview of conventional methods). Unfortunately, these methods are often unfeasible.
Due to their inherent nonlinear characteristics, neural networks might be a suitable solution for compensating Kerr nonlinearities. They might be used, e.g, as a demapper [2] or as a more efficient alternative to DBP [3]. A good overview of usage of neural networks for nonlinearity compensation is given in [4, Sec. Vb] . This paper also provides a good introduction to machine learning techniques relevant for optical communications.
In this seminar, the student is expected to research neural network based compensation techniques for fibre nonlinearities (e.g., one of the ones briefly described in [4, Sec. Vb]) and present either one in detail or multiple ones as an overview.
Prerequisites
Basic knowledge of optical communications (Optical Communication Systems EI5075, Digital Signal Processing for Optical Communication Systems EI71067 or comparable) is required. Some experience with machine learning is helpful, but papers like [4] can alternatively provide a good introduction to machine learning, as well.
If you are interested, feel free to just contact me via email. We can then discuss the details and fit the topic to your interests.
-----------------------------------------------------------------------------------------------------------------------------------------------
[1] A. Amari, O. A. Dobre, R. Venkatesan, O. S. S. Kumar, P. Ciblat and Y. Jaouën, "A Survey on Fiber Nonlinearity Compensation for 400 Gb/s and Beyond Optical Communication Systems," in IEEE Communications Surveys & Tutorials, vol. 19, no. 4, pp. 3097-3113, Fourthquarter 2017, doi: 10.1109/COMST.2017.2719958.
[2] M. Schaedler, S. Calabrò, F. Pittalà, C. Bluemm, M. Kuschnerov and S. Pachnicke, "Neural Network-Based Soft-Demapping for Nonlinear Channels," 2020 Optical Fiber Communications Conference and Exhibition (OFC), San Diego, CA, USA, 2020, pp. 1-3.
[3] C. Häger and H. D. Pfister, "Nonlinear Interference Mitigation via Deep Neural Networks," 2018 Optical Fiber Communications Conference and Exposition (OFC), San Diego, CA, USA, 2018, pp. 1-3.
[4] F. N. Khan, Q. Fan, C. Lu and A. P. T. Lau, "An Optical Communication's Perspective on Machine Learning and Its Applications," in Journal of Lightwave Technology, vol. 37, no. 2, pp. 493-516, 15 Jan.15, 2019, doi: 10.1109/JLT.2019.2897313.