基于多径空间谱及信号子载波的低成本高鲁棒非绑定式目标定位研究

Prior works in device-free localization (DfL) typically face two key challenges: 1) poor positioning accuracy under multipath effects, and 2）poor robustness using general equipment, and 3) high hardware and deployment cost. In this project, we build a multipath signal spectrum based on amplitude-phase information of the single, and use the statistical characteristics of signals, which caused by the moving target, for fine-grained target localization. Then a phase calibration technique based on nonlinear optimization is proposed to effectively eliminate the instability of the phase of wireless signals, which significantly improves the positioning robustness. In addition, based on our analysis on the relationship between antenna topology and estimation performance of the multipath signal spectrum, we minimize the antenna numbers while satisfying the desired accuracy by using synthetic aperture radar (SAR). By modelling the device-free localization problem as a set of over-determined equations, we can determine locations of the target by solving the nonlinear optimization problem, which helps to reduce the survey cost and avoid model mismatch problem. At last, we implement our system for several representative application scenarios and evaluate its effectiveness under different conditions..In conclusion, the main idea behind our project is to explore the relationship between the signal change and the target location by analyzing how target affects the wireless signal. In addition, we explore the inherent relationship among localization accuracy, usability, robustness and the unstability caused by changing environment. The research findings of this project will provide theoretical support and reference for the behavior monitoring applications in wireless network.

现有非绑定式定位方法面临信号多径效应导致定位精度低、通用设备定位鲁棒性差的挑战和实施代价大的问题。项目旨在借助信号幅度与相位信息构建监测区域多径信号空间谱，利用目标对信号空间谱干扰的分布规律实现高精度定位；利用相位差建立非线性方程，用优化方法实现相位校准与稳定，提升定位鲁棒性；采用合成孔径雷达(SAR)稀疏布阵思想，研究天线拓扑对多径信号空间谱估计精度的影响，在保证信号空间谱估计精度前提下减少阵列天线数量，降低定位成本。建立无需学习的非绑定式定位模型，将定位问题转化为非线性优化求解，解决学习方法勘测成本高及场景变化致模型失用问题。针对典型应用场景，设计非绑定式定位原型系统，验证所提方法有效性。其科学实质是通过目标对多径信号空间谱干扰规律研究，寻求无线信号与目标的时空关联规律，探索定位精度、鲁棒性、可用性和环境时变性间的本质联系，研究成果有望为无线网络行为监测应用提供有价

本项目针对现有非绑定式定位方法面临信号多径效应导致定位精度低、通用设备定位鲁棒性差的挑战和实施代价大等相关问题，研究了在多种约束条件下的被动式目标定位以及大规模感知方法，物联网并发传输等问题。项目有效的解决了传统定位方法认为多径不利的问题，利用多径效应实现了分米级的定位精度且无需离线训练，突破了真实环境下定位算法鲁棒性差，精度提升难的问题。项目利用超材料，通过设计可调超平面LLAMA，并将其集成到通信链路中来实时调整收发端之间的极化匹配状态，从而优化通信链路，提高了现有商用设备的通信质量并增强感知效果。在无线信号感知方面，项目组利用商用设备对人体呼吸进行检测，可以辅助监控呼吸相关疾病，同时利用商用设备对毛笔书写轨迹进行监控，辅助量化毛笔书写过程。该项目的多个成果先后发表在Mobicom,ICDCS,Sensys,Infocom,TON,TMC等高水平国际会议和期刊上，并被MIT,普林斯顿,UIUC等多个国际知名高校引用。相关成果被英国泰晤士报报道，英国每日邮报，参考消息网等媒体广泛报道。本项目的研究对物联网产业发展和升级有着重要意义。