近年来随着高精度定位需求的不断高涨，各种定位技术发展方兴未艾，其中，低功耗蓝牙（BLE）凭借其部署广泛、功耗极低、性价比高的优势而受到诸多青睐。传统的蓝牙定位手段多借助于信号强度指示（RSSI），由于信号强度受环境影响程度大，在室内尤甚，通常只能提供米级定位精度，必须寻求其他定位手段。在最近的蓝牙核心规范v5.1中，基于天线阵列的测角定位方案被引入，依据阵列部署端的不同具体可支持到达角（AOA）和离开角（AOD）测量，其基本原理是同一信号到达不同阵元时相位存在一定差异，此相位差决定于信号到达角和阵列构型，这一定位方案被认为有望突破目前蓝牙定位精度普遍停留在米级的桎梏。 与一般阵列测向不同，蓝牙AOA/AOD方案乃是基于一种分时工作机制。以AOA为例，传统阵列定位系统中接收机配备多变频通道，各阵元同时工作以确保接收同一信号，而蓝牙AOA方案规定接收机仅单通道，各阵元分时接收信号。如此设计的主要原因是受限于蓝牙硬件设备和维持低成本。由此导致的棘手问题是：既然各阵元接收的是不同时刻信号，则难以确保相位关系的正确无误。另外，由于发射信号的极化状态是时变的，而惯常使用的标量阵列不具备极化域信息感知能力，不能应对天线极化失配引发的信号畸变，多径抑制效果差，测向性能不稳定。为克服以上问题，本文设计并实现了一个基于极化敏感阵列（PSA）的蓝牙单基站定位系统。PSA由极化选择特性不尽相同的多阵元组成，可以探测信号来波方向和极化状态信息，具备更精确稳定的测向潜力。本文详尽分析了系统分时接收信号导致的信号频率误差（SFO），以及PSA中普遍存在却难以消除的通道间失配误差（CME），并分别提出了基于最小二乘的SFO补偿算法和基于辅助信源的CME校准方法，消除接收信号的幅相误差。 本文还研究了基于PSA的定位算法，实时定位系统要求兼顾高精度和低时延，鉴于接收信号受来波方向和极化状态多维参数影响，参数估计基于最大似然准则以确保高精度，借助交替迭代优化以确保低时延，得到了精确的三维角度估计，通过蒙特卡洛仿真验证算法性能。此外还推导了PSA的角度估计克拉美罗界（CRLB）并与标量阵列进行对比，并基于CRLB分析得到了一些阵列测向性能的关键结论。最终，分别部署了蓝牙单基站实时AOA/AOD定位系统，实验显示中值定位误差为30/36cm，时延均低于0.1s。
In recent years, with the increasing demand for high precision positioning, various positioning technologies are in the ascence. Among them, Bluetooth Low Energy (BLE) has been favored for the advantages of wide deployment, extremely low power consumption, and high cost-effective. Traditional Bluetooth positioning methods mostly rely on the signal strength indicator (RSSI) yet it can only provides meter-level positioning accuracy since the signal strength is seriously affected by the environment especially in the indoor scene. and other positioning methods must be sought. In the recent Bluetooth Core Specification V5.1, an antenna array based positioning scheme has been introduced. Specifically, it can support Angle of Arrival (AOA) and Angle of Departure (AOD) depending on the deployment of the antenna array. The basic principle is that there exist a phase difference between array elements, which is determined by the signal direction and array configuration. This positioning scheme is expected to break through the shackles of the precision of current Bluetooth positioning methods which generally stays at the meter level. Different from the general array direction finding methods, Bluetooth AOA/AOD scheme is based on a time-sharing mechanism. Taking AOA as an example, the receiver in the traditional array positioning system is equipped with multiple frequency conversion channels. All elements work simultaneously to ensure that they receive the same signal, while the Bluetooth AOA positioning scheme stipulates that the receiver only equipped with a single frequency conversion channel, and all elements receive signals in a time-sharing manner. The main reasons for the design are naturally stem from the Bluetooth hardware limitation and cost control considerations. Unfortunately, advantages would become a liability. The tricky problem is that since each array element receives signals at different times, it is difficult to ensure the correct phase relationship. In addition, the commonly used scalar array is not capable of percepting the time-varying signal polarization state information, and it cannot cope with the signal distortion caused by the antenna polarization mismatch and the multipath interference, the result is unstable direction finding performance. To over the problems, this paper presents the design and implementation of a single anchor BLE localization system based on the polarization sensitive array (PSA). The PSA composed of multiple elements with different polarization selection characteristics, which can detect the direction and the polarization state information of the signal, and excavate the finer-grained information in the waveform, which is capable of providing more stable and accurate positioning results. In this paper, the influence of signal frequency error (SFO) in the time-sharing system and the channel mismatch error (CME) prevail throughout the multiple antenna array are analyzed in detail, and we put forward a least squares based SFO mitigation algorithm as well as an auxiliary source based CME calibration method, respectively. This paper also studies the localization algorithm based on the PSA. Real-time positioning system requires both high precision and low time delay. Considering the signal are influenced by multiple parameters including signal direction and polarization state, parameter estimation based on the maximum likelihood criterion to ensure high precision, and the time delay reduction by means of alternating iterative optimization. To evaluate the performance of the algorithm, the angle estimation accuracy is obtained by Monte Carlo simulation. Moreover, the Cramer lower bound of the PSA is derived and compared with the scalar array. Several key conclusions about the array direction finding performance are obtained by CRLB analysis. Finally, we deploy a real-time single anchor BLE localization system using angular information, which is demonstrated to achieve a median accuracy of 30cm in experiments while the delay less than 0.1s.
