Deep Leaning-Based Channel Estimation for RIS-Assisted Full-Duplex MIMO System With Hardware Impairments

This letter proposes a deep learning-based channel estimation method for reconfigurable intelligent surface (RIS)-assisted multiple-input multiple-output (MIMO) full-duplex (FD) systems under hardware impairments (HI). Unlike traditional model-based estimators that rely on ideal assumptions and require complex analytical models for non-idealities, the proposed approach leverages data-driven learning to capture nonlinear distortions. Learning directly from data allows the model to adapt to practical impairments without explicit modeling, resulting in enhanced robustness and low-latency inference. Simulation results demonstrate that the proposed method achieves superior estimation performance, highlighting its suitability for real-time deployment in hardware-impaired environments.