AirFM-DDA: Wireless Foundation Model for 6G Physical Layer

A novel design paradigm for AI-native 6G networks is emerging, fueled by substantial foundation models aimed at physical layer functions. Current wireless models utilize channel state information (CSI) within the space-time-frequency (STF) domain, where multipath components are intertwined and layered, complicating universal channel representation. Additionally, their dependence on global attention leads to significant computational demands. The newly introduced AirFM-DDA model transforms CSI from the STF domain to the Delay-Doppler-Angle (DDA) domain, effectively clarifying multipath components along relevant physical axes. It features a window-based attention mechanism with frame-structure-aware positional encoding (FS-PE) to enhance efficiency. This model targets improvements in physical layer tasks like channel estimation and signal detection for AI-native 6G networks. The research can be found on arXiv with the identifier 2605.00020.