发表机构
Technische Universität Berlin; Huawei Heisenberg Research Center (Munich); Huawei Technologies Co., Ltd., Shanghai, China; Technical University of Munich（柏林工业大学; 华为海森堡研究中心（慕尼黑）; 华为技术有限公司（上海）; 慕尼黑工业大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
提出轨迹引导分词（TGT），结合复杂轨迹分解与非对称注意力，在Wi-Fi感知中实现高效压缩，达到92.83%的最高平均准确率，并支持跨域应用。
AI 中文摘要
Wi-Fi信道状态信息（CSI）可实现非接触式存在检测和手势识别。其高维复值时间序列需要输入表示，在压缩过程中保留信息丰富的时变特征。我们提出轨迹引导分词（TGT），将复杂轨迹分解与非对称注意力相结合，构建紧凑的连续令牌。对于每个天线链路和子载波，正交赫尔默特变换将短的有序时间补丁分解为局部中心和中心轨迹坐标。键从中心轨迹坐标学习，而值保留两个分量。可学习的查询将子载波聚合为频率槽，再融合为时间令牌。从头开始联合训练，TGT与TokenMLP在自采集数据集上所有评估的前端-后端组合中达到最高平均准确率92.83%。在EHUNAM和Widar上的实验进一步支持TGT在跨域存在检测和手势识别中的适用性。
英文摘要
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.
