【作者】 宋涛；

【导师】 殷吉昊；

【作者基本信息】 电子科技大学 ， 工程硕士（专业学位）， 2017， 硕士

【摘要】 定位问题通常是指利用一组在空间上作特定分布的观测站点与目标之间构成的相对空间几何关系,产生一组特定的观测量,如到达时间(Time of Arrival,TOA)、到达时差(Time Difference of Arrival,TDOA)、到达角度(Direction of Arrival,DOA)等,实现目标空间位置坐标参数的高精度估计。定位问题一般假设信号从目标出发后沿直线传播直达观测站点,也称为视距传播(Line of Sight,LOS)假设。这一假设在一些实际问题中往往不成立,例如,在蜂窝或无线局域网(Wireless Local Area Networks,WLAN)定位问题中,城市和近郊移动站台或是其它复杂环境下实现视距传播一般困难,大多数情况下信号收发设备之间存在障碍,没有直达路径传播,信号通过反射或者衍射等路径到达接收机,这一现象称为非视距(Non Line of Sight,NLOS)传播。非视距传播是无线定位结果产生偏差的主要原因,有效抑制NLOS产生的偏差是提升无线定位精度的关键所在。本文围绕非直达波偏差抑制算法,研究了如下几方面工作:1.研究NLOS定位模型,分别针对TOA和TDOA下的NLOS定位模型进行说明,同时介绍了基于TOA下的LOS和NLOS基本定位算法,并对两种情况下算法性能进行仿真对比。2.研究Chan的基于残差检测(Residual Test,RT)抑制NLOS偏差定位算法,并对其存在的大量冗余计算进行改进,提出一种简洁的改进RT抑制NLOS偏差定位算法。3.在无线传感器网络下,介绍基于TOA下的两种凸松弛算法(半定松弛(Semi-define Relaxation,SDR)和二阶锥松弛(Second-order Cone Relaxation,SOCR))来抑制NLOS偏差。算法能够充分利用NLOS误差信息,分为两种情况进行分析,提供更为精确的定位性能。同时引出了一种改进的稳健SDR算法。4.将基于TOA的凸松弛算法引入到TDOA系统中,通过对稳健最小二乘(Robust Least Squares,RLS)问题进行凸松弛得到两种稳健定位方法。仿真验证了这两种方法相对于普通凸松弛算法的优越性。更多还原

【Abstract】 Source localization problems usually refer to using a set of sensors distributed in the observation space to establish some relative geometric relationships between a source and measurements(such as Time of Arrival(TOA),time Difference of Arrival(TDOA)and Direction of Arrival(DOA))and then achieve the source location estimate.A source localization problem often assumes that the source signal arrives at a sensor in the so-called line-of-sight(LOS)propagation way.This assumption often collapses in practice.For example,in cellular or wireless local area networks(WLAN)positioning problems,usually there is no direct path propagation(or say line-of-sight propagation)between an emitter and a receiver(e.g.,urban and suburban mobile stations)since many obstacles appear in the propagation channels.The source signals have to arrive at a receiver from the reflection or diffraction paths,which is called non-line of sight(NLOS)propagation.The NLOS propagation is the main reason of enlarging the positioning bias,which should be suppressed to improve the positioning accuracy when we design a positioning method.This study focuses on the bias-reduced localization methods for the NLOS localization problems.The main work includes:1.The NLOS localization models are first introduced using TOA and TDOA measurements respectively.Then some bias-reduced TOA-based localization methods are studied and compared through numerical examples.2.Chan’s RT method is first studied for reducing localization bias that is caused by the NLOS phenomenon.Then we improve it by using a distance residual random variable and a simplified RT method to avoid a large number of redundant calculations.3.Two convex relaxation algorithms(i.e.,SDR and SCOR,respectively)based on TOA measurements are first introduced to suppress the NLOS bias for the localization problems in the wireless sensor network,which can make full use of the NLOS error information and can provide good positioning performance.Then an improved robust algorithm is achieved based on the above SDR algorithm.4.Inspired by the idea of the TOA-based convex relaxation algorithm,two robust TDOA-based positioning methods are obtained through the RLS convex relaxation procedure.Simulation results show the superiority of these two methods over the general convex relaxation algorithms.更多还原