基于惯性传感器的改进高斯粒子滤波室内定位算法

发布时间：2018-03-31 04:15

本文选题：无线室内定位　切入点：惯性传感器　出处：《新疆大学》2016年硕士论文

【摘要】：在无线室内定位研究中,目前通常使用的是WiFi位置指纹方法,该方法分为离线指纹数据库建立阶段和在线定位阶段,离线阶段首先把楼层划分成网格,将网格中心点定为位置指纹点,逐个采集位置指纹点上对应的若干个热点(AP)的接受信号强度指示值(RSSI),建立该楼层位置指纹数据库。位置指纹法实现的最大难点在于离线指纹库建立阶段,该阶段需要采集所有指纹点的RSSI值,由于多径、衰减、环境干扰等诸多因素,单个指纹点RSSI数据将会不固定,存在波动现象,因此在采集某个指纹点RSSI数据时,一般会在一个指纹点采集多个RSSI值,对采集到的原始数据做均值、分类等方法处理后,作为该位置指纹点的RSSI值存入指纹数据库。一般需要定位的区域覆盖区域普遍较大,每个指纹点都将花费数分钟时间,该阶段将花费大量时间及资源。为了减少离线阶段指纹数据库采集工作,本文提出了利用智能手机内置惯性传感器采集三轴加速度数据,来计算用户在室内运动路径,在经过位置指纹点时记录该点RSSI值,自动建立离线指纹库的方法。若所有内置惯性传感器的手机用户都参与数据采集工作,就可以基本覆盖整个区域,将大大节省离线指纹数据库建立成本,并且还解决了环境变化、AP变动需要重新采集指纹数据的问题。由于手机内置惯性传感器采集的数据存在噪声和干扰等随机噪声,测量结果精度不高,将导致计算出的路径有较大误差,运动距离越远,累积误差越大。因此本文提出了改进的高斯粒子滤波算法,采集到原始加速度数据后,先利用改进的高斯粒子滤波算法对加速度数据滤波,再计算终端运动轨迹。实验仿真表明,利用改进的高斯粒子滤波算法对加速度数据滤波后,计算出的路径较接近实际运动路径,定位精度与传统方法建立的数据库基本一致。但是自动建立指纹数据库的方法将节省大量资源,并且改进的算法具有良好的实时性和鲁棒性。
[Abstract]:In the research of wireless indoor location, WiFi location fingerprint method is usually used at present. The method can be divided into offline fingerprint database establishment stage and online location stage. At the off-line stage, the floors are first divided into grids. The center of the grid is designated as the fingerprint point, The received signal strength indication value of the corresponding hot spot APs on one by one is used to establish the location fingerprint database of this floor. The biggest difficulty of the location fingerprint method is the establishment of the off-line fingerprint database. In this stage, we need to collect the RSSI value of all fingerprint points. Because of many factors, such as multipath, attenuation, environment interference and so on, the RSSI data of a single fingerprint point will not be fixed and fluctuate, so when we collect RSSI data from a certain fingerprint point, In general, multiple RSSI values are collected at a fingerprint point. After processing the original data collected by means of mean, classification and other methods, the RSSI value of the fingerprint point at this position is stored in the fingerprint database. Generally, the area covered by the location is generally larger. Each fingerprint point will take several minutes, which will take a lot of time and resources. In order to reduce the work of fingerprint database acquisition in off-line stage, this paper proposes to collect three-axis acceleration data by using the built-in inertial sensor of smart phone. To calculate the indoor movement path of the user, record the RSSI value of the point when passing through the fingerprint point, and automatically establish the off-line fingerprint database. If all the mobile phone users with the built-in inertial sensor participate in the data collection, Can basically cover the entire area, will greatly reduce the cost of setting up an offline fingerprint database, It also solves the problem that the change of AP requires the re-acquisition of fingerprint data. Because of the random noise such as noise and interference in the data collected by the built-in inertial sensor of mobile phone, the accuracy of the measurement results is not high. Therefore, the improved Gao Si particle filter algorithm is proposed to collect the original acceleration data. The acceleration data is filtered by the improved Gao Si particle filter algorithm, and then the terminal motion trajectory is calculated. The experimental simulation shows that the acceleration data is filtered by the improved Gao Si particle filter algorithm. The calculated path is close to the actual moving path, and the location accuracy is basically the same as the database established by the traditional method, but the method of automatically establishing the fingerprint database will save a lot of resources. And the improved algorithm has good real-time and robustness.
【学位授予单位】：新疆大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TN92

【相似文献】