Optimization of Joint Equalization of High-Speed Signals using Bayesian Machine Learning
The paper elaborates an efficient algorithm for optimization of joint Feed-Forward Equalization (FFE) and Decision Feedback Equalization (DFE) for non-return-to-zero (NRZ) and 4 level pulse amplitude modulation (PAM-4) signals using Bayesian Machine Learning approach previously introduced for NRZ by authors and expanded for PAM-4. A new optimal covariant function and hyper-parameters has been selected for the Bayesian optimization. Cost function for the Bayesian optimization is chosen based on eye height. The proposed method was compared to the conventional Least Mean Square (LMS) method and showed significant improvement. Test cases were performed for several data rates of NRZ and PAM-4 signals with crosstalk and injected jitter. Test results show that the proposed algorithm is the more effective the higher data rates are considered.
N. Dikhaminjia et al., "Optimization of Joint Equalization of High-Speed Signals using Bayesian Machine Learning," Proceedings of the 2021 Joint IEEE International Symposium on EMC/SI/PI, and EMC Europe (2021, Raleigh, NC), pp. 48 - 52, Institute of Electrical and Electronics Engineers (IEEE), Aug 2021.
The definitive version is available at https://doi.org/10.1109/EMC/SI/PI/EMCEurope52599.2021.9559157
2021 IEEE International Joint Electromagnetic Compatibility Signal and Power Integrity and EMC Europe Symposium, EMC/SI/PI/EMC Europe 2021 (2021: Jul. 26-Aug. 13, Raleigh, NC)
Electrical and Computer Engineering
Electromagnetic Compatibility (EMC) Laboratory
Keywords and Phrases
Bayesian Optimization; Cost Function; Covariant Function; DFE; Equalization; FFE; Joint Optimization; Tap Coefficients
International Standard Book Number (ISBN)
Article - Conference proceedings
© 2021 Institute of Electrical and Electronics Engineers (IEEE), All rights reserved.
13 Aug 2021