Modulation Identification of Underwater Acoustic Communications Signals Based on Generative Adversarial Networks

The modern military needs for information acquisition and processing make the modulation identification of underwater acoustic communication signals become the focus of research. We propose a modulation identification method based on generative adversarial networks (GAN) to increase the robustness of modulation identification for underwater acoustic communication signals. The generator of GAN is trained to enhance the distorted signals and the discriminator is trained to extract features from underwater acoustic communication signals and classify them automatically. This method relies less expertise in signal processing. Simulating experiments are performed to evaluate the performance of the proposed method under multipath fading and additive white gaussian noise (AWGN) channel, and the result shows that the proposed method reaches higher accuracy than using a deep convolution neural network.