Performance study of OAM optical communication systems via multi-layer feature interaction in turbulent atmospheric environments

【摘要】 This work proposes an optical ResNet-transformer robust（OptiRes-TR） model and low-density parity-check Gray（LDPC-Gray） coding for orbital angular momentum-shift keying（OAM-SK） free-space optics（FSO） communications.Original information undergoes LDPC-Gray encoding and 16-level mapping to superposed Laguerre-Gaussian beams,with atmospheric turbulence simulated via power spectrum inversion.The OptiRes-TR model achieves 90.43% classification accuracy under strong turbulence.Combined with probability-assisted LDPC-Gray demodulation,it maintains bit error rates stably between 9.22 × 10^-6 and 3.01 × 10^-3.The OptiRes-TR model exhibits superior convergence and timeliness versus convolutional neural networks（CNNs）.更多还原

【Abstract】 This work proposes an optical ResNet-transformer robust（OptiRes-TR） model and low-density parity-check Gray（LDPC-Gray） coding for orbital angular momentum-shift keying（OAM-SK） free-space optics（FSO） communications.Original information undergoes LDPC-Gray encoding and 16-level mapping to superposed Laguerre-Gaussian beams,with atmospheric turbulence simulated via power spectrum inversion.The OptiRes-TR model achieves 90.43% classification accuracy under strong turbulence.Combined with probability-assisted LDPC-Gray demodulation,it maintains bit error rates stably between 9.22 × 10^-6 and 3.01 × 10^-3.The OptiRes-TR model exhibits superior convergence and timeliness versus convolutional neural networks（CNNs）.更多还原