基于Wi-Fi感知的室内穿墙场景下无源人体目标检测与定位技术研究

针对Wi-Fi感知技术在室内穿墙场景下难以捕获微弱的目标反射信号以及难以估计目标初始位置的问题，本项目拟利用信道状态信息（Channel State Information, CSI）多径可分辨特性，重构和抵消非目标反射路径CSI，实现干扰信号的有效抑制。在此基础上，提取目标反射路径CSI时频相关性特征和不同子载波之间的相关性特征，利用深度学习网络构建分类器，实现对墙后目标的高精度检测。然后，利用统计方法检验目标反射路径CSI在能量及空域上的分布情况，筛选最优路径，从而实现目标初始位置估计。最后，根据目标反射路径的空域及时域参数，构建空间几何约束模型，实现对墙后目标的高精度被动定位。本项目旨在突破现有Wi-Fi感知技术在微弱反射信号获取和目标初始位置估计方面的瓶颈，促进其在目标入侵检测、智能家居以及人机交互等领域的应用，具有重要的科学意义和经济价值。

Aiming at the problem that it is difficult for Wi-Fi sensing to capture subtle signals reflected by a target and estimate its initial position under indoor through-the-wall scenarios, this project intends to utilize the multipath separability feature of channel state information (CSI) to reconstruct and cancel CSI of a non-target reflection path, so as to effectively suppress interference signals. On this basis, time-frequency correlation characteristics of CSI reflected by the target and correlation characteristics between different subcarriers are extracted, and a classifier is constructed by using deep learning networks to realize a high-precision detection for the target behind the wall. Then, a novel hypothesis test method is proposed to test energy and space distributions for target reflection paths and single out an optimal path, so as to realize the initial position estimation for the target. Finally, according to spatial and temporal parameters of the target reflection path, a spatial geometric constraint model is constructed to realize a high-precision passive localization for the target behind the wall. This project aims to break through bottlenecks of the existing Wi-Fi sensing in terms of acquisition of weak reflected signals and estimation of the initial position, and promote its applications in target intrusion detection, smart home, human-computer interaction and so on, which is of great scientific significance and economic value.