Research on simulation method of atmospheric turbulence effects based on generative adversarial network

In view of the problem that the atmospheric turbulence phase screen generated by the traditional numerical simulation method has insufficient low frequency, this paper proposes a phase screen generation method based on Generative Adversarial Network (GAN). The advantages of this method include its strong robustness to atmospheric turbulence, low model depth, few parameters, and low complexity. Compared with the traditional numerical simulation method, it is found that the GAN proposed in this paper not only compensates for the low-frequency deficiency of the traditional numerical simulation method but also speeds up the generation speed of the phase screen. Simultaneously, the trained phase screen model can be directly used for phase screen generation.