Enhanced chaotic communication with machine learning

Abstract

Communication with chaotic signals holds a significant position in the field of secure communication and has consistently been research hotspot. While representative chaotic communication frameworks are all based on the deployment of robust synchronization or complex correlators, they pose considerable challenges to practical applications. In this work, a machine-learning-based framework is proposed for the chaotic shift keying scheme, which is robust against noise deterioration. Specifically, we adopt the reservoir computing technique with noise training schema to enhance the robustness of the entire communication process. Overall, the novel structure we propose fully leverages the predictive capabilities of neural networks, providing a new perspective for machine learning in the field of chaotic communication and significantly improving the accuracy of existing technologies.