Tailoring optical vortex beams based on diffractive neural networks for high-quality ring-core fiber coupling

Simultaneously transmitting orthogonal optical vortex (OV) beams loaded with different signals in the same optical fiber is a promising technique to cope with the rapidly growing demand for communication capacity. To achieve high-capacity fiber-optic communications based on high-order OV beams, reasonable tailoring of OV beams and their high-quality alignment to fiber cores are of great significance. Here, we propose and experimentally demonstrate an approach to tailoring OV beams based on diffractive neural networks, so that the OV beams entering the fiber are as similar as possible to the fiber-supported OV beams. In addition, a visualized system is also established to monitor the coupling process for high-quality alignment between OV beams and fiber cores. Benefiting from these proposed techniques, coupling loss of different OV mode groups is measured to be no more than 2.1 dB, with an average reduction of 1.09 dB and the lowest reduction of 0.72 dB. Our work may pave the way for optical communication and sensing applications based on fiber-guided OV beams.