基于信息熵的WLAN室内定位算法研究

发布时间：2018-12-13 11:14

【摘要】：无线局域网作为宽带有线接入网的补充应用越来越广泛，同时也催生了以无线局域网为基础的各类服务如WLAN室内定位服务等。而基于位置指纹的WLAN室内定位系统以其操作及设备简单等特点而成为研究热点。因而本文将基于位置指纹的WLAN室内定位方法作为主要研究内容，并通过改进该方法提高定位准确度和定位所需时间。基于位置指纹的WLAN室内定位一般分两个阶段：离线阶段Radio Map的建立和在线定位阶段。在离线阶段，通过实测得到参考点的位置信息及相应的RSS值形成Radio Map；在线阶段使用特征匹配算法计算出在线测得数据的物理位置。基于位置指纹的定位算法需要解决两个问题：定位的准确性和时效性。因而本文研究了聚类算法、AP选择算法及Radio Map更新算法。首先，本文分析了现有的基于Radio Map的WLAN室内定位的特点，根据其关键的两个环节即Radio Map的建立及特征匹配算法进行分析。位置指纹的创建方法有两种，即自由空间传播模型法和接收到RSS值的特征值法，本文选用RSS特征值法。RSS值随着时间，天线朝向，参考点位置变化而变化，因而需要选用合理的方法建立Radio Map。特征匹配算法中，包括最简单的最近邻算法、经典的K近邻算法和加权K近邻算法。其次，本文通过分析Radio Map，研究如何对Radio Map进行化简及更新操作。为了定位的时效性，本文首先对Radio Map进行聚类处理，将RadioMap划分为几个小类，然后在每一个小类中使用AP选择算法选择出合适的AP组合用于定位。在聚类算法中，研究了最简单的K均值聚类算法、引入隶属度概念的模糊K均值聚类算法和无需指定初始聚类数的仿射传播聚类算法；在AP选择算法中，研究了随机选择及均值最大选择AP方法、信息熵增益方法和互信息熵方法。最后，为定位的准确性，研究了基于隐马尔科夫模型的Radio Map更新方法，，并使用EM算法对隐马尔科夫模型进行求解。最后，通过在真实环境下的实验仿真，利用特征匹配算法进行定位。对聚类算法、AP选择算法及Radio Map更新算法进行了性能分析，并基于实验环境选择了合适的参数以期达到定位准确度高及定位时间短的特点。
[Abstract]:WLAN is more and more widely used as a supplement to broadband wired access network. At the same time, WLAN services such as WLAN indoor positioning services are given birth to. The WLAN indoor positioning system based on position fingerprint has become a research hotspot because of its simple operation and equipment. Therefore, the WLAN indoor location method based on location fingerprint is taken as the main research content in this paper, and the accuracy and time of location are improved by improving the method. WLAN indoor location based on position fingerprint is generally divided into two stages: the establishment of Radio Map and the online location. In the off-line phase, the position information of the reference point and the corresponding RSS value are measured to form the Radio Map; online phase. The physical position of the on-line measured data is calculated by using the feature matching algorithm. The localization algorithm based on location fingerprint needs to solve two problems: accuracy and timeliness. Therefore, clustering algorithm, AP selection algorithm and Radio Map update algorithm are studied in this paper. Firstly, this paper analyzes the characteristics of existing WLAN indoor location based on Radio Map, and analyzes its two key links, namely, the establishment of Radio Map and the feature matching algorithm. There are two methods to create position fingerprint, that is, free space propagation model method and eigenvalue method that receives RSS value. In this paper, RSS eigenvalue method is used. The RSS value changes with time, antenna orientation and reference point position. Therefore, it is necessary to select a reasonable method to establish Radio Map.. The feature matching algorithms include the simplest nearest neighbor algorithm, the classical K nearest neighbor algorithm and the weighted K nearest neighbor algorithm. Secondly, this paper studies how to simplify and update Radio Map by analyzing Radio Map,. In order to get the timeliness of the localization, the Radio Map is first clustered, the RadioMap is divided into several subclasses, and then the appropriate AP combination is selected by using the AP selection algorithm in each subclass. In the clustering algorithm, the simplest K-means clustering algorithm is studied, the fuzzy K-means clustering algorithm based on membership degree and the affine propagation clustering algorithm without specifying the initial clustering number are introduced. In the AP selection algorithm, the random selection and the mean maximum selection AP method, the information entropy gain method and the mutual information entropy method are studied. Finally, for the accuracy of location, the Radio Map updating method based on Hidden Markov Model is studied, and the EM algorithm is used to solve the Hidden Markov Model. Finally, through the real-time simulation, the feature matching algorithm is used to locate the location. The performance of clustering algorithm, AP selection algorithm and Radio Map update algorithm are analyzed, and the suitable parameters are selected based on the experimental environment in order to achieve the characteristics of high localization accuracy and short localization time.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN925.93

【参考文献】