Secure Transmission with Artificial Noise and Machine Learning in Low Earth Orbit Satellite Networks

Low Earth orbit (LEO) satellite communications face significant risks of eavesdropping in applications ranging from broadcasting to military operations. This paper investi-gates the potential of integrating machine learning techniques, specifically support vector machine (SVM)-based scheduling, with physical layer security (PLS) techniques in LEO satellite networks. Our proposed method leverages artificial noise (AN) and SVM to improve secrecy performance. Extensive simulations demonstrate that the proposed scheme closely matches optimal performance while being computationally efficient, significantly outperforming random scheduling.