Fig. 1. Amplitude of phase perturbations in phase screens with two different atmospheric refractive index structure constants

Fig. 2. Simulation schematic of superimposed OAM optical signal one-dimensional detection scheme

Fig. 3. Schematic structure of two CNN models

Fig. 4. Accuracy of two CNN models in recognizing the same dataset

Fig. 5. Accuracy of the CNN model trained with different training parameters to recognize superimposed OAM optical signals under different turbulence intensity disturbances

Fig. 6. Schematic diagram of superimposed OAM optical signal transmission and recognition experiment

Fig. 7. Photograph of superimposed OAM optical signal transmission and recognition experiment setup

Fig. 8. Link for superimposed OAM optical signal transmission and recognition experiment

Fig. 9. Turbulence strength and intensity distribution of superimposed OAM beams after turbulent perturbation in an outdoor environment

Fig. 10. One-dimensional intensity data of different superimposed OAM beams in an outdoor environment

Fig. 11. Architecture diagram of FPGA deploying CNN

Fig. 12. Timing diagram of data processing on FPGA

Fig. 13. The cross talk matrix for CNN recognition of superimposed OAM optical signals recognition on FPGA at 20∶00

Fig. 14. Recognition accuracy of CNNs deployed on FPGA in an outdoor environment