Track the frontier, from condensed matter to cosmology.
Search papers, labs, and topics across Lattice.
This paper introduces Trajectory-Guided Tokenization (TGT) to compress high-dimensional complex Wi-Fi CSI time series for sensing tasks by preserving temporal variations. TGT decomposes CSI patches into local-center and trajectory coordinates using an orthonormal Helmert transform, with asymmetric attention learning keys from trajectory and values from both components. Jointly trained TGT with TokenMLP achieves state-of-the-art 92.83% mean accuracy on a self-collected dataset and generalizes to cross-domain presence detection and gesture recognition.
Wi-Fi sensing accuracy jumps to 92.83% by decomposing CSI into trajectory-guided tokens that capture complex temporal dynamics.
Wi-Fi channel state information (CSI) enables contactless presence detection and gesture recognition. Its high-dimensional complex-valued time series require input representations that preserve informative temporal variations during compression. We propose Trajectory-Guided Tokenization (TGT), which combines complex trajectory decomposition with asymmetric attention to construct compact continuous tokens. For each antenna link and subcarrier, an orthonormal Helmert transform decomposes short, ordered temporal patches into local-center and centered-trajectory coordinates. Keys are learned from the centered-trajectory coordinates, while values retain both components. Learnable queries aggregate subcarriers into frequency slots, which are fused into temporal tokens. Trained jointly from scratch, TGT with TokenMLP achieves the highest mean accuracy of 92.83% among all evaluated frontend-backend combinations on the self-collected dataset. Experiments on EHUNAM and Widar further support the applicability of TGT to cross-domain presence detection and gesture recognition.
TU Munich