Secure Communications in Satellite Systems: A Deep Reinforcement Learning-based Intelligent Spectrum Control Framework

With the development of the sixth-generation (6 G) technology, satellite communication has become increasingly critical for providing large-coverage and high-speed service. However, the reliability and security of satellite communication are facing huge threats due to the increase in various interference means and the existence of eavesdroppers. To cope with these problems, an intelligent spectrum control (ISC)-assisted method has been proposed to enhance the security of satellite communication. In this scheme, by applying the Long Short-Term Memory (LSTM) prediction network and the deep reinforcement learning (DRL) decision network, a frequency slot set selection scheme is proposed to obtain an available frequency slot set with a lower interference probability. To further improve the security of the system, a high-performance sequence set is generated through a series of iterations and mapping operations by leveraging block cipher principles. Moreover, we analyzed the signal-to-interference-and-noise ratio (SINR) of the authorized users and eavesdroppers during the data transmission process. Based on this analysis, the reliable transmission probability and secrecy outage probability of the ISC-based system are derived. Simulation results show that the complexity and the ability to prevent being predicted of the sequence outperform those of the widely used sequence with variable parameters. Meanwhile, the performance evaluation verifies that secure transmission can be achieved by applying the proposed ISC-based scheme.