In this paper, we investigate the potential of applying machine learning (ML) technique for nonlinearity compensation in subcarrier multiplexing (SCM) systems. We propose an extension of the learned digital backpropagation (LDBP) method, which was originally developed for single carrier systems, to effectively deal with the effects of both self-subcarrier and cross-subcarrier nonlinearities in a 64-quadrature amplitude modulation dual-polarization system with 32 Gbaud transmission. The performance of our proposed approach is evaluated and compared with non-ML approaches in the literature. The outcomes demonstrate that our approach outperforms the SCM-DBP method by 0.3 dB. The results of this study demonstrate the potential of using ML techniques to improve the transmission performance of SCM systems which also encourages the further exploration of ML techniques for nonlinearity compensation in SCM systems.
|Title of host publication||2023 23rd International Conference on Transparent Optical Networks (ICTON)|
|Number of pages||4|
|Publication status||Published online - 8 Aug 2023|
|Event||2023 23rd International Conference on Transparent Optical Networks : ICTON - Bucharest, Romania|
Duration: 2 Jul 2023 → 6 Jul 2023
|Name||2023 23rd International Conference on Transparent Optical Networks (ICTON)|
|Publisher||IEEE Control Society|
|Conference||2023 23rd International Conference on Transparent Optical Networks|
|Period||2/07/23 → 6/07/23|
Bibliographical noteFunding Information:
This work has been supported by Huawei Technologies Canada.
© 2023 IEEE.
- machine learning
- nonlinearity compensation
- subcarrier multiplexing