Presentation Information
[WP-F-8]ML PAM4 Pre-emphasis Coefficients Optimization Algorithm for Mitigating Gain Nonlinearity in SOAs
○Boyang Zheng, Nicola Calabretta (Eindhoven Univ. of Technology and Eindhoven Hendrik Casimir Institute)
Keywords:
Machine learning in optical communication networks,Photonic switching systems and networks
This paper proposes a ML algorithm using stochastic gradient descent (SGD) to optimize PAM4 pre-emphasis to mitigate gain nonlinearity in semiconductor optical amplifier (SOA). Experimental results validate its effectiveness, reducing the power penalty from 21 dB to 4.2 dB with 53 GBaud PAM4 data transmission.