Learning-Based VLC with Optical Reflecting Intelligent Surface Against Eavesdropping

The broadcast nature of visible light communication (VLC) inherently compromises data confidentiality in wireless transmissions. To address this vulnerability, we propose a learning-driven framework that leverages optical reflecting intelligent surface (ORIS) for multi-user secure VLC systems. Our approach optimizes both LED transmitter beamforming pattern and ORIS unit orientations, thereby enhancing secrecy capacity, signal quality, and energy efficiency. A novel reinforcement learning (RL)-based ORIS orientation scheme is introduced, employing rank correlation to reduce complexity in high-dimensional action spaces. Simulation results demonstrate that the proposed RL-based scheme significantly outperforms conventional methods in anti-eavesdropping performance, proving its efficacy in dynamic VLC environments.