基于Wi-Fi的室内定位技术和系统研究

发布时间：2018-12-11 19:27

【摘要】：随着室内定位技术的迅速发展,基于红外线、超声波、蓝牙、超宽带、Wi-Fi(Wireless Fidelity)等的定位技术迅速成为学术和应用研究的热点。基于Wi-Fi的室内定位在定位精度、稳健性、安全性和复杂度等方面有着自身的优势,该技术充分利用了现有的成熟硬件平台和无线网络接入点,能在智能终端上以应用程序的形式实现高精度的定位服务,因而基于Wi-Fi的室内定位研究备受关注。本文对基于Wi-Fi的室内定位技术进行深入探讨,主要研究接收信号预处理技术,离线阶段的位置指纹库构建技术和在线阶段的室内定位算法三个方面。首先,为得到稳定的位置指纹数据,提高定位系统的稳定性,根据室内Wi-Fi信号的传播模型和接收信号特征,研究基于卡尔曼滤波的信号强度预处理方法,使用对数谱域抑制信号多径效应的方法,对接收信号进行预处理。其次,针对指纹库构建代价高的问题,提出数据插值方法,研究矩阵填充的方法,把基于SVT算法的矩阵填充应用于低秩位置指纹库的重建;提出使用密度峰值快速搜索聚类技术对位置指纹地图进行分类,并与K-means聚类和仿射传播聚类对比,对位置指纹地图进行分类预处理。最后,研究在线定位方法,提出基于信号传播模型的接收信号强度填充的定位方法,对基于位置指纹匹配的近邻定位算法、贝叶斯定位算法和压缩感知定位算法进行详细研究。此外,本文设计了一个基于Wi-Fi的室内定位实验系统原型,该系统使用卡尔曼滤波和对数谱域抑制多径效应的方法对接收信号进行预处理,采用密度峰值快速搜索聚类实现位置指纹库分块处理,使用贝叶斯算法和压缩感知算法进行在线定位。在实际室内环境中完成参考点的布署,实现室内区域的位置指纹地图构建,进行定位实验。实验结果显示,该定位系统具有较高的定位精度,定位误差在2米以内的累积概率为85%,同时能够以较少的计算量实现稳健的定位。
[Abstract]:With the rapid development of indoor positioning technology, the localization technology based on infrared, ultrasonic, Bluetooth, ultra-wideband, Wi-Fi (Wireless Fidelity) and so on has become a hot spot in academic and applied research. Indoor positioning based on Wi-Fi has its own advantages in positioning accuracy, robustness, security and complexity. This technology makes full use of existing mature hardware platforms and wireless network access points. The research of indoor location based on Wi-Fi is paid more attention to because it can realize the high precision location service in the form of application program on the intelligent terminal. In this paper, the indoor location technology based on Wi-Fi is deeply discussed, including the pre-processing technology of receiving signal, the construction technology of position fingerprint database in off-line stage and the indoor location algorithm in on-line stage. Firstly, in order to obtain the stable position fingerprint data and improve the stability of the positioning system, according to the propagation model of indoor Wi-Fi signal and the characteristics of the received signal, the signal intensity preprocessing method based on Kalman filter is studied. The received signal is preprocessed by using logarithmic spectral domain to suppress multipath effect. Secondly, aiming at the high cost of constructing fingerprint database, a data interpolation method is put forward, and the method of matrix filling is studied. The matrix filling based on SVT algorithm is applied to the reconstruction of low-rank position fingerprint database. In this paper, the fast searching clustering technique of peak density is used to classify the location fingerprint map, and compared with the K-means clustering and affine propagation clustering, the location fingerprint map is preprocessed. Finally, the on-line localization method is studied, and the location method based on signal propagation model is proposed. The nearest neighbor location algorithm based on location fingerprint matching, Bayesian location algorithm and compressed sensing location algorithm are studied in detail. In addition, a prototype of indoor positioning experiment system based on Wi-Fi is designed in this paper. The system uses Kalman filter and logarithmic spectral domain to pre-process the received signal. The location fingerprint database is partitioned by fast searching clustering with peak density, and online location is carried out by using Bayesian algorithm and compression perception algorithm. In the actual indoor environment, the reference points are deployed, the location fingerprint map of the indoor area is constructed, and the location experiment is carried out. The experimental results show that the positioning system has high positioning accuracy, the cumulative probability of positioning error within 2 meters is 85um, and the robust positioning can be achieved with less calculation.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TN92

【参考文献】