Publication

Abstract : In this letter, we present automated modulation classification (AMC) for reconfigurable intelligent surface (RIS)-assisted multiple input multiple output-orthogonal frequency division multiplexing (MIMO-OFDM) systems under imperfect channel state information (CSI), residual carrier frequency offset (CFO) and symbol time offset (STO) errors. We leverage a triple attention-aided vision transformer (TrpViT) architecture, which uses a vision-centric approach within the transformer network to enhance global information acquisition. The TrpViT is implemented by utilizing three complementary attention mechanisms spatial, dilated, and channel attention in a unique attention block. This unique attention block extracts spatially local features while expanding the scope to capture more comprehensive signal features. The adopted attention mechanisms effectively capture long-range spatial dependencies and channel interactions within input signals by optimizing the model complexity. The performance of the proposed method is compared against existing models and it has been demonstrated that the proposed method accurately classifies higher modulation schemes for RIS-assisted MIMO-OFDM systems. The computational complexity of the proposed model is also compared with the existing state-of-the-art.