Channel Estimation for Stacked Intelligent Metasurface-Aided Network using Deep Learning

The next generation of wireless communication systems is poised to integrate advanced technologies such as holographic multiple input, multiple output (HMIMO) systems and stacked intelligent metasurfaces (SIMs). These innovations aim to provide dynamic electromagnetic wave control while reducing hardware complexity. In SIM-assisted systems, the number of meta-atoms per SIM layer often exceeds the number of antennas at the base station (BS), making accurate channel estimation a significant challenge. Traditional model-based channel estimation methods often struggle with the high dimensionality and dynamic variability of such environments. To address this hurdle, we propose a deep learning (DL)-based uplink channel estimation framework that learns complex, nonlinear relationships between the received pilots and the channel responses. The proposed method significantly outperforms classical estimators in terms of accuracy and robustness. The simulation results validate its effectiveness under various propagation scenarios.