An improvement on the CNN-based OAM Demodulator via Conditional Generative Adversarial Networks

In the paper, an Orbital Angular Momentum (OAM) demodulation method based on Conditional Generative Adversarial Networks(CGAN) is proposed to improve the accuracy of Convolutional Neural Networks (CNN) based demodulator. We train a CGAN on a limited data set, and the discriminator in CGAN is fine-tuned as a new classifier for OAM demodulation. Our numerical simulations demonstrate that the proposed method can improve the accuracy of OAM demodulator from 93.56% to 98.36% over 400-m free-space link when the turbulence strength $C_n^2$ equals 4×10−13 m−2/3.