Secure Multi-Point Coordinated Beamforming using Deep Learning in 5G and Beyond Networks

In 5G and beyond networks, a critical need exists for a rapid, energy-efficient, and secure beam selection process. This study introduces a secure multi-point coordinated beamforming approach based on deep learning. It prioritizes beam pairs between a transmitter and a legitimate user, with the goal of optimizing the user's signal strength while ensuring that the eavesdropper's signal strength remains below a predefined threshold. Instead of exhaustive search, the method focuses on a limited set of top-performing beam pairs, resulting in reduced communication overhead and energy consumption. The scheme's performance is assessed using statistical system-level variables. Numerical results indicate a 75% reduction in signaling overhead, with 87.41% accuracy in selecting the best beam pair and achieving 99.62% of the desired signal strength. In terms of security, the method enhances secure communication probability by 70.4%, compared to the system without security constraints.