【摘要】 针对载波频偏（CFO）严重干扰Wi-Fi感知所依赖的信道状态信息（CSI）精度，构成其性能提升的核心瓶颈这一难题，提出一种面向Wi-Fi CSI感知场景的CFO精确估计方法。该方法通过构建精确定时的测量帧序列消除接收端相位的2π模糊性，然后利用测量帧间隔的变化量和相位差的变化量进行线性拟合，并设计多层感知机机制优化拟合结果，得到CFO粗估计。最后采用超分辨算法对粗估计进行搜索，实现CFO的精确估计。实验结果表明，所提方法可在工作频偏为±100 kHz的商用网卡上实现对CFO的精确估计，其平均误差仅为17.0327 Hz，应用于CFO校准后，Wi-Fi感知的测距和到达角精度均得到显著提升。更多还原

【Abstract】 To address the problem that carrier frequency offset(CFO) severely interferes with the accuracy of channel state information(CSI) required for Wi-Fi sensing, constituting the critical bottleneck for performance improvement, an accurate CFO estimation method for Wi-Fi CSI sensing scenarios was proposed. In this method, the 2π phase ambiguity at the receiver was eliminated by constructing a precisely timed sequence of measurement frames. The variations in measurement frame intervals and phase differences were then utilized for linear fitting, and a multi-layer perceptron mechanism was designed to optimize the fitting results, thereby obtaining a coarse CFO estimation. Finally, a super-resolution algorithm was employed to perform a search on the coarse estimate to achieve accurate CFO estimation. Experimental results show that the proposed method achieves highly accurate CFO estimation on commercial network interface cards operating with a frequency offset of ±100 kHz, with an average error of only 17.0327 Hz. When applied to CFO calibration, both ranging accuracy and angle-of-arrival accuracy in Wi-Fi sensing are significantly improved.更多还原