High-capacity and high-fidelity turbid underwater optical communications via partially coherent fractional vortex beams

A high-capacity and high-fidelity turbid Underwater Optical Communication (UOC) protocol is proposed and experimentally validated, leveraging partially coherent fractional vortex beams for encoding and a Convolutional Neural Network (CNN) for decoding. The protocol achieves high capacity through multiplexing of four coherence lengths, sixteen fractional Orbital Angular Momentum (OAM) modes with 0.25 intervals, and four radial indices, generating 256 distinct intensities of beam for 256-ary encoding. High-fidelity transmission is enabled by the anti-scattering property of partially coherent beams and CNN-based feature recognition, the transmission accuracy above 99.57[Formula: see text] across all turbidity levels. This work demonstrates the feasibility of integrating partially coherent beam with neural networks for turbid UOC, providing a novel solution for future research on achieving superior-performance underwater optical communication in turbid conditions based on low-coherence beams.