Presentation Information

[TuP-B-28]Deep Neural Network Decoding of Redundant-Bit-Free 2× Oversampled Waveform Utilizing Inter-Sample Correlation Underwater Optical Wireless Communication

○Xinyu Wang1,2, Taishu Takatsuki2, Ekkaphol Khansalee2, Md. Khaled Hasan2, Masanori Hanawa2, Zhu Li1 (1.Hangzhou Dianzi Univeristy, 2.University of Yamanashi)

Keywords:

Implementation of digital signal processing subsystems for optical communications,Artificial intelligence and machine learning for optical transmission systems and subsystems

A deep neural network-based decoding method of 2× oversampled waveform samples utilizing inter-sample correlation was investigated, achieving 2.3dB and 1.1dB gains over threshold detection and maximum likelihood estimation in underwater optical wireless communication.