基于指纹和AP选择的WLAN室内定位技术研究

发布时间：2018-03-23 12:39

本文选题：WLAN室内定位　切入点：位置指纹　出处：《江苏科技大学》2017年硕士论文

【摘要】：随着数据业务和多媒体业务在社会生活中应用的日益频繁,基于位置的服务(Location-Based Services,LBS)在人们的需求中呈现出明显的上升趋势。尤其是在地下停车场、机场大厅、购物中心等场景,常常需要对室内物体进行精确定位,传统的卫星定位技术采用微波传输,由于多径干扰、非视距传输、信号衰减等原因,信号难以穿过复杂的建筑物外层,无法满足室内定位的要求。随着室内无线信号设施被广泛部署,基于WLAN的室内定位技术成为近年来定位领域中的重要研究方向。本文对WLAN室内定位技术进行了研究,首先分析了WLAN室内定位技术的研究现状与发展趋势,对来自接入点(Access Point,AP)的信号接收强度(Received Signal Strength,RSS)传播特性进行了实地测量研究。在此基础上,对位置指纹特征匹配算法和AP选择算法进行分析,并分别提出改进算法,提高了室内定位精度,具体工作如下:(1)研究RSS信号的传播特性:在实际定位环境下,对RSS信号进行了实地测量与分析。分别以采样时长、人员走动、AP选取、采样点位置为变量,研究RSS概率分布情况。另外,通过设计实验研究了采样网格尺寸和AP数量对定位效果的影响,确定了适用于本文系统的指纹网格边长与最佳AP选择数量。(2)研究指纹特征匹配算法:提出了一种基于贝叶斯算法的改进特征匹配算法。引入区域划分的概念,将定位空间内所有参考点进行聚类分簇,解决了朴素贝叶斯算法仅靠参考点后验概率估算位置的不足。通过仿真实验,验证了改进算法的有效性,2米精度范围内的累积概率,由改进前的60%提升至70%。(3)研究AP选择算法:提出了一种基于信息增益的AP选择算法。在InfoGain算法的基础上,引入邻近AP间的差异性判定,与信息增益值进行加权,共同量化AP的综合可辨识性,并以此为标准选择最具位置辨识能力的AP参与定位计算。通过仿真实验,所提的AP选择算法能够有效去除冗余AP信息;与特征匹配算法联合运用,降低了定位复杂度,与InfoGain算法相比,2米精度范围内的累积概率,由78%提升至85%。
[Abstract]:With the increasing use of data services and multimedia services in social life, location-based services (LBSs) show an obvious upward trend in people's needs, especially in underground parking lots, airport halls, shopping centers, etc. It is often necessary to accurately locate indoor objects. The traditional satellite positioning technology uses microwave transmission. Because of multipath interference, non-line-of-sight transmission, signal attenuation and other reasons, the signal is difficult to pass through the outer layer of complex buildings. The indoor location technology based on WLAN has become an important research direction in the field of localization in recent years with the widespread deployment of indoor wireless signal facilities. In this paper, the indoor positioning technology of WLAN has been studied. Firstly, the research status and development trend of WLAN indoor positioning technology are analyzed, and the propagation characteristics of received Signal received signal from access Point WLAN are studied in the field. The location fingerprint feature matching algorithm and AP selection algorithm are analyzed, and the improved algorithms are proposed to improve the indoor positioning accuracy. The specific work is as follows: 1) study the propagation characteristics of RSS signal: in the actual location environment, The field measurement and analysis of RSS signal are carried out. The probability distribution of RSS is studied by taking the sampling time, the selection of personnel walking AP and the location of sampling points as variables, respectively. The effects of sampling mesh size and AP number on the location effect are studied through design experiments. The fingerprint feature matching algorithm is studied. An improved feature matching algorithm based on Bayesian algorithm is proposed, and the concept of region partition is introduced. By clustering all reference points in the location space, the deficiency of naive Bayes algorithm to estimate the location only by a posteriori probability of reference points is solved. The effectiveness of the improved algorithm is verified by simulation experiments, and the cumulative probability in the precision range of 2 meters is verified. This paper studies the AP selection algorithm from 60% to 70% before the improvement. An AP selection algorithm based on information gain is proposed. Based on the InfoGain algorithm, the difference between adjacent AP is determined and weighted with the information gain. The synthesis identifiability of AP is quantized, and the AP with the most ability of position identification is selected as the standard to participate in the location calculation. Through the simulation experiment, the proposed AP selection algorithm can effectively remove redundant AP information and be used in conjunction with the feature matching algorithm. Compared with the InfoGain algorithm, the cumulative probability in the precision range of 2 meters is reduced from 78% to 85%.
【学位授予单位】：江苏科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN925.93

【参考文献】