Orbital angular momentum underwater wireless optical communication system based on convolutional neural network

In this work, a novel 16-ary orbital angular momentum shift keying (OAM-SK) underwater wireless optical communication (UWOC) system based on convolutional neural network (CNN) demodulator and Gerchberg-Saxton CNN (GS-CNN) beam generator is proposed. The bit error rate (BER) performance of the proposed UWOC system with different turbulence intensity, transmission distance, and relative intensity of temperature and salinity is further investigated. By comparing with the results from the UWOC system based on GS beam generator, it is revealed that the BER performance can be improved obviously for the proposed OAM-SK UWOC system combining the CNN demodulator and GS-CNN beam generator.