Orbital angular momentum shift-keying communication through atmospheric turbulence and high scattering channel using a deep learning decoder

To increase the data rate and the channel capacity, orbital angular momentum shift keying (OAM-SK) has been studied extensively for free space optical (FSO) communication. However, the beams carrying OAM are disturbed into speckle patterns via the atmospheric turbulence (AT) and high scattering (HS) channel, which causes difficulty in decoding. Herein, we propose an AT and HS channel model to simulate the propagation of OAM beams, showing that the formed speckle size depends on the topological charge l of the OAM mode, which means the OAM mode can be recognized from the receiving speckle patterns. Proof-of-concept​ experiments of an OAM-SK system using a deep learning decoder are implemented. The results show that under dynamic AT with the refractive index structure constant (the AT strength, C_n²=1×10⁻¹⁶K²m^−2/3) and static HS (a ground glass), we achieve a hybrid eight-mode OAM-SK, where the decoding accuracy of the encoding set l=1,2,3,4,5,6,7,8 reaches >0.95 within a 500 m transmission distance. Moreover, increasing the encoding mode interval can improve the decoding accuracy; when the AT strength increases to C_n²=1×10⁻¹⁴K²m^−2/3 and the scattering media becomes dynamic, the accuracy of the encoding set l=1,4,7,10,13,16,19,22 with a mode interval of 3 can still achieve >0.94 at 100 m. Furthermore, error-free image transmission can be achieved in the OAM-SK system under AT strengths of C_n²=1×10⁻¹⁶K²m^−2/3 and 1×10⁻¹⁵K²m^−2/3 and HS at 100 m. This work provides a potential approach for FSO communication based on OAM modes under harsh AT and HS channels.