Full metadata record
| DC Field | Value | Language |
|---|---|---|
| dc.contributor | Department of Electrical Engineering | en_US |
| dc.contributor.advisor | Lau, Pak Tao Alan (EE) | en_US |
| dc.creator | Fan, Qirui | - |
| dc.identifier.uri | https://theses.lib.polyu.edu.hk/handle/200/11752 | - |
| dc.language | English | en_US |
| dc.publisher | Hong Kong Polytechnic University | en_US |
| dc.rights | All rights reserved | en_US |
| dc.title | Digital signal processing for long-haul optical communications using deep learning | en_US |
| dcterms.abstract | In long-haul optical communication systems, compensating nonlinear effects through digital signal processing (DSP) is difficult due to intractable interactions between Kerr nonlinearity, chromatic dispersion (CD) and amplified spontaneous emission (ASE) noise from inline amplifiers. Here, we show the use of deep learning methods to optimize standard digital back propagation (DBP) as a deep neural network (DNN) with interleaving linear and nonlinear operations for fiber nonlinearity compensation in practical single-channel and polarization division multiplexed wavelength division multiplexed experiments. We show improved performance compared to state-of-the-art DSP algorithms and additionally, the optimized DNN-based DBP parameters exhibit a mathematical structure which guides us to further analyze the noise statistics offiber nonlinearity compensation. This machine learning-inspired analysis reveals that ASE noise and incomplete CD compensation of the Kerr nonlinear term produce extra distortions that accumulates along the DBP stages. Therefore, the best DSP should balance between suppressing these distortions and inverting the fiber propagation effects, and such trade-off shifts across different DBP stages in a quantifiable manner. Instead of the common 'black-box' approach to intractable problems, our work shows how machine learning can be a complementary tool to human analytical thinking and help advance theoretical understandings in disciplines such as optics. | en_US |
| dcterms.abstract | An additional hurdle in seamlessly incorporating Machine Learning into optical communications are dynamic transmission impairments such as polarization effects and carrier phase noise that corrupt the training data, making it hard to define cost function for neural network (NN) training, difficult to integrate ML with standard adaptive DSP which result in suboptimal performance or require impractical training methodologies. We show how the adaptive DSP blocks can be treated as extra stateful NN layers and combined with the main NN for so that standard backpropagation-like training algorithm in ML can be applied. In this case, the adaptive filter taps are viewed as NN states which are updated in the forward pass of the backpropagation. We study the combined training of linear and nonlinear parameters in the DBP algorithm for fiber nonlinearity compensation (named generalized DBP (GDBP) hereafter), residual impairments, polarization effects, frequency offsets and carrier phase noise compensation filters as a single NN in a 7-channel polarization multiplexed (PM)-16QAM transmission experiment over 1125 km. We derived the complete set of backpropagation-like gradients and state update equations for the static and dynamic parameters of the whole NN and proposed and open-sourced a JAX-based coding framework for their easy and practical implementation. GDBP is more generalized than other DBP variants proposed in literature and for a given total number of steps, the GDBP is the first experimental demonstration of optimal single-channel DBP based-fiber nonlinearity compensation algorithm. In addition, for complexity constrained situations with shortened filter taps, GDBP enabled a 1 dB performance improvement over other DBP variants. The proposed architechture viewing conventional DSP as stateful NN introduces a potentially transformative technology leading to the development of new data-driven and physics-informed learning machines. Overall, the thesis work provides new tools for betting integration of ML in long-haul optical communications, achieve better transmission performance and shed new insights into the nature of noise and fiber nonlinearity interactions and their mitigation in long-haul systems. | en_US |
| dcterms.extent | ix, 100 pages : color illustrations | en_US |
| dcterms.isPartOf | PolyU Electronic Theses | en_US |
| dcterms.issued | 2022 | en_US |
| dcterms.educationalLevel | Ph.D. | en_US |
| dcterms.educationalLevel | All Doctorate | en_US |
| dcterms.LCSH | Signal processing -- Digital techniques | en_US |
| dcterms.LCSH | Optical communications | en_US |
| dcterms.LCSH | Deep learning (Machine learning) | en_US |
| dcterms.LCSH | Hong Kong Polytechnic University -- Dissertations | en_US |
| dcterms.accessRights | open access | en_US |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/11752