Transformer-based Distributed Machine Learning for Downlink Channel Estimation in RIS-Aided Networks

Reconfigurable Intelligent Surface (RIS)-aided networks increase spectral efficiency and enhance coverage in wireless networks; however, accurate DLCE (downlink channel estimation) in such networks is a challenging task due to the massive pilot overhead. This paper proposes a novel transformer-based distributed machine learning (DML) approach for DLCE in RIS-aided networks. By leveraging a custom self-attention mechanism and DML architecture, this approach effectively captures complex spatial and temporal correlations in the wireless channel. Our results demonstrate better channel estimation accuracy of atleast 67%, 93% and 84% in DLCE convergence, loss and normalized mean square error compared to other neural networks and baseline models.