Research on anti-turbulence methods for vortex beam communication based on FSRA-ConvNet and SFHformer

Vortex light exploits modal orthogonality and orbital angular momentum (OAM) to improve spectral efficiency and system capacity in free-space optical (FSO) communication. However, atmospheric turbulence induces random refractive-index fluctuations that distort the beam wavefront, leading to mode dispersion and crosstalk. These effects hinder accurate OAM mode detection at the receiver and remain a major limitation to communication reliability. This paper proposes an SFHformer network as an adaptive demodulator for OAM mode demodulation. In combination with low-density parity-check (LDPC) encoding and decoding, the performance of normalized min-sum (MS), normalized min-sum scaled (NMS), and offset min-sum (OMS) algorithms coupled with spatial-frequency hybrid transformer (SFHformer) demodulation is evaluated, and an SFHformer-NMS demodulation framework is constructed, achieving a significant reduction in bit error rate. Meanwhile, to mitigate wavefront distortion induced by atmospheric turbulence, an FSRA-ConvNet intensity restoration network is proposed. Built on UNet, this network incorporates the FSRA module to improve multi-scale information utilization and introduces the Converse2D module to enhance feature fusion and local structural detail recovery. Systematic experiments demonstrate that the proposed framework maintains high-quality signal transmission under extreme turbulence conditions.