Utilizing a tripartite generative adversarial network, a covert communication technique

Abstract

To jointly optimize the transmission covertness and the demodulation accuracy of the covert message, a novel tripartite generative adversarial network (TripartiteGAN) and a covert communication method based on TripartiteGAN were developed.The method's effectiveness was examined.In particular, the amplitude and phase of an input modulated covert data were adjusted using TripartiteGAN so that, for the public user, the distribution of the generated covert signal superposing the overt signal approximated that of the overt signal.The suggested approach might function with an ideal warden that doesn't require manual detection threshold setting or knowledge of the sender's transmit power parameters.