Presentation Information
[TuP-B-9]Low-Complexity and Rapid-Adaptive Neural Network Equalizer based on SkipNet for High-Speed Optical IM/DD Communication System
○Chengxi Wang1, Zhongya Li1, An Yan1, Junhao Zhao1, Yingjun Zhou1, Jianyang Shi1, Nan Chi1, Zhixue He2, Junwen Zhang1 (1.Key Laboratory of Information Science of Electromagnetic Waves(MoE), Fudan University, 2.Pengcheng Laboratory(PCL))
Keywords:
Artificial intelligence and machine learning for optical transmission systems and subsystems,Digital signal processing algorithms for optical communications
We propose and demonstrate a low-complexity and rapid-adaptive neural network equalizer based on SkipNet for high-speed optical IM/DD system, achieving 300- and 270-Gbps data transmission under B2B and 1-km fiber with improved sensitivity and data-rates.