Connect public, paid and private patent data with Google Patents Public Datasets

LDE (Linear Discriminant Analysis) algorithm-based WiFi (Wireless Fidelity) indoor locating method

Download PDF

Info

Publication number
CN103079269A
CN103079269A CN 201310029536 CN201310029536A CN103079269A CN 103079269 A CN103079269 A CN 103079269A CN 201310029536 CN201310029536 CN 201310029536 CN 201310029536 A CN201310029536 A CN 201310029536A CN 103079269 A CN103079269 A CN 103079269A
Authority
CN
Grant status
Application
Patent type
Prior art keywords
method
dimension
indoor
radio
locating
Prior art date
Application number
CN 201310029536
Other languages
Chinese (zh)
Inventor
马琳
周才发
徐玉滨
秦丹阳
孟维晓
崔扬
Original Assignee
哈尔滨工业大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date

Links

Abstract

The invention discloses an LDE (Linear Discriminant Analysis) algorithm-based WiFi (Wireless Fidelity) indoor locating method, relating to an indoor locating method and solving the problem of poor location instantaneity of the existing WiFi indoor locating method. The realizing process of the LDE algorithm-based WiFi indoor locating method comprises two stages of an offline stage and an online stage, wherein the offline stage comprises the steps of: constructing a WiFi network, measuring RSS (Received Signal Strength) and constructing a Radio Mao; estimating an intrinsic dimension of the Radio Map by adopting an intrinsic dimension estimating method; carrying out dimension reduction process on the Radio Map by adopting an LDE algorithm to obtain a Radio Map subjected to dimension reduction process and a feature transform matrix, wherein an optimal dimension reduction result and a corresponding feature transform matrix are used as a matching database and a corresponding RSS transform matrix in the online stage. The online stage comprises the steps of: carrying out feature transform on the RSS received in a testing point, matching by adopting a KNN (k-Nearest Neighbor) algorithm and the Radio Map subjected to dimension reduction process to obtain predicted coordinates of the testing point. The invention is suitable for indoor location.

Description

基于LDE算法的WiFi室内定位方法 WiFi-based indoor positioning algorithm method LDE

技术领域 FIELD

[0001] 本发明涉及一种室内定位方法。 [0001] The present invention relates to a method for indoor positioning.

背景技术 Background technique

[0002] 随着无线局域网络在世界范围的飞速发展和移动终端设备的广泛普及,近年来出现了许多室内定位相关的技术和应用。 [0002] With the widespread popularity of wireless local area networks in the world and the rapid development of mobile terminal equipment, in recent years there have been many indoor positioning technology and related applications. 由于多径效应、信号衰减及室内定位环境的复杂性,基于传统的信号传播模型的室内定位方法难以达到高精度的室内定位要求。 Due to multipath effects, attenuation, and the complexity of indoor positioning environment, the conventional method of indoor positioning based on a signal propagation model is difficult to achieve high accuracy indoor positioning requirements. 基于到达时间(Time of Arrival)、到达时间差(Time Difference of Arrival)或到达角度(Angles ofArrival)等定位方法虽然可以基本满足定位精度需求,然而都需要定位终端有额外的硬件设备支持,具有较大局限性,从而导致基于上述几类定位方法的室内定位系统没有得到普及。 Based on the arrival time (Time of Arrival), a time difference (Time Difference of Arrival), or angle of arrival (Angles ofArrival) and other positioning method, although can basically meet the positioning accuracy requirements reach, however, it requires positioning terminal has additional hardware support, with a larger limitations, resulting in not been universal indoor location system based on the above-mentioned types of positioning method.

[0003] 目前,基于WLAN位置指纹(Finger Print)的WiFi室内定位方法得到了广泛应用。 [0003] Currently, WLAN location fingerprint (Finger Print) of WiFi-based indoor positioning method has been widely used. 该方法的网络构建方法成本低廉,其使用2.4GHz ISMdndustrial Science Medicine)公共频段且无需在现有设施之上添加定位测量专用硬件。 The method of constructing the network cost method, using 2.4GHz ISMdndustrial Science Medicine) common frequency band and without adding special hardware positioning measurements on existing facilities. 只需要通过移动终端的无线网卡及相应软件测量接收到的接入点(Access Point, AP)的信号强度(Received SignalStrength, RSS),由此来构建网络信号覆盖图(Radio Map),进而通过匹配算法来预测移动用户所处位置的坐标,或相对位置。 Only received through the wireless network card of the mobile terminal and an access point corresponding to the measurement software (Access Point, AP) signal strength (Received SignalStrength, RSS), thereby constructing a network coverage map (Radio Map), and further by matching algorithm to predict the coordinates of the mobile user's location, or relative position.

[0004] 然而通过该方式建立的Radio Map包含有庞大的数据信息,且随着定位区域扩大,Radio Map可能(依据定位匹配方式及算法选择)呈指数形势增长。 [0004] However, the manner established by Radio Map contains huge data information, and with the location area to expand, Radio Map may grow exponentially situation (based positioning methods and matching algorithm selection). 获得尽可能多的相关数据特征信息对于整个系统来说会提升定位精度,但是处理大量的特征信息增加算法开销,定位算法无法在处理能力有限的移动终端上有效运行,同时某些特征信息可能是对于定位没有作用甚至有负面作用,致使匹配效率降低,从而导致匹配定位算法的实现变得更加复杂,并且定位精度下降。 Get as much relevant data feature information for the whole system will improve location accuracy, but a lot of the feature information processing algorithms to increase overhead, positioning algorithm can not operate effectively on a limited processing capability of the mobile terminal, while some feature information may be for positioning does not work even have a negative effect, resulting in matching efficiency is reduced, resulting in matching to achieve localization algorithm becomes more complex, and positioning accuracy.

[0005] 当AP的数目增加及定位的参考点(Reference Point)增加时,Radio Map的数据信息增加。 [0005] When the number of the AP and the positioning of the reference point (Reference Point) increases, the data information of the Radio Map. 此时,Radio Map中代表的AP数目的信息表示了数据的维数。 At this time, information representing the number of AP Radio Map showing the dimension of the data. 因此,当AP数目增加,Radio Map就变成了高维数据。 Thus, when increasing the number of AP, Radio Map becomes a high-dimensional data. 为减轻处理高维数据的负担,降维算法是有效的解决方法之一。 To alleviate the burden of processing high-dimensional data, dimension reduction algorithm is an effective solution. 高维数据可能包含很多特征,这些特征都在描述同一个事物,这些特征一定程度上是紧密相连的。 High-dimensional data may contain many features that are describing the same thing, to some extent, these characteristics are closely linked. 如当从各个角度对同一个物体同时拍照时,得到的数据就含有重叠的信息。 Simultaneously such as when photographing the same object from various angles, the resulting data to contain overlapping information. 如果能得到这些数据的一些简化的不重叠的表达,将会极大地提高数据处理运行的效率并一定程度上提高准确度。 Some simplifications can be obtained if the data do not overlap with the expression of these will greatly improve the efficiency of data processing operation and improve the accuracy of a certain extent. 降维算法的目的也正是在于提高高维数据的处理效率。 Dimensionality reduction algorithms aim is also to improve the efficiency of processing high-dimensional data.

[0006] 除了可以简化数据使其能够高效处理外,降维方法还可以实现数据可视化。 [0006] In addition it is possible to simplify the data processing efficiency, the method may further dimension reduction data visualization. 由于很多统计学的和机器学习算法对于最优解的准确性很差,降维的可视化应用可以令用户能够实际看到高维数据的空间结构和算法输出的能力,具有很强的应用价值。 Since many statistical and machine learning algorithms for the optimal solution accuracy is poor, lower-dimensional visualization applications can make the user the ability to be able to actually see the spatial structure and algorithm output of high-dimensional data, it has a strong practical value.

[0007]目前有很多基于不同目的的降维算法,包括有线性与非线性降维算法。 [0007] There are many different purposes dimensionality reduction algorithms, including linear and nonlinear dimensionality reduction algorithm. 其中PCA(Principal Component Analysis)和LDA(Linear Discriminant Analysis)是典型的线性降维算法。 Wherein the PCA (Principal Component Analysis) and LDA (Linear Discriminant Analysis) is a typical linear dimension reduction algorithm. 这一类算法对于具有线性结构的高维数据有着良好的处理结果,但对于非线性结构的高维数据没有好的结果。 This class of algorithms have good processing results, but no good result for the non-linear high-dimensional data structure for high-dimensional data having a linear structure. 典型的非线性降维算法以流形学习(ManifoldLearning)算法。 Typical nonlinear dimensionality reduction manifold learning algorithm (ManifoldLearning) algorithm. 2000年Science杂志上同一期发表了3篇有关于流形学习算法中提出了 In 2000 on the same issue of Science magazine published about three manifold learning algorithm is proposed

2 种经典的流形学习算法:LLE (Local Linear Embedding)及ISOMAP (Isometric Mapping)。 Two kinds of classic manifold learning algorithms: LLE (Local Linear Embedding) and ISOMAP (Isometric Mapping). 由此,各种基于不同的准则的流形学习算法被提出并有一部分流形学习算法应用于图像处理方面。 Accordingly, various manifold learning algorithms based on different criteria have been proposed and a portion of manifold learning algorithms applied to the image processing.

[0008]目前的WiFi室内定位方法存在的Radio Map数据库大、在线定位阶段计算复杂度高难以在移动终端实现、定位的实时性差等问题。 [0008] the presence of the current WiFi indoor positioning method Radio Map database is large, on-line positioning stage high computational complexity difficult to achieve in the mobile terminal, and poor real-time positioning.

发明内容 SUMMARY

[0009] 本发明是为了解决现有的WiFi室内定位方法的定位的实时性差的问题,从而提供一种基于LDE算法的WiFi室内定位方法。 [0009] The present invention is made to solve the problem of poor real-time positioning of a conventional indoor positioning method WiFi, WiFi thereby providing a location based indoor LDE Algorithm.

[0010] 基于LDE算法的WiFi室内定位方法,它由以下步骤实现: [0010] WiFi LDE method of indoor positioning algorithm, which is implemented by the steps of:

[0011] 步骤一、针对室内环境布置N个接入点AP,确保所述环境中任意一点被一个或一个以上的接入点AP发出的信号覆盖,所述N个接入点AP组成WiFi网络;在所述室内环境中均匀设置Nkp个参考点;N和Nkp均为正整数; [0011] Step a, the N access point AP is arranged for indoor environment to ensure coverage of the environment at any point by one or more than one access point AP sent by the access point AP consisting of N WiFi network ; Nkp reference points are evenly arranged in the indoor environment; N are positive integers and Nkp;

[0012] 步骤二、选取一个参考点为坐标原点建立二维直角坐标系,获得Nkp个参考点在该二维直角坐标系中的坐标位置,在离线阶段中在每个参考点上利用信号接收机采集来自每一个接入点AP的信号强度RSS值,并作为该接入点AP的位置特征信息;并根据N个接入点AP的位置特征信息构建室内信号覆盖图Radio Map ; [0012] Step 2. Selection of a reference point to establish the origin of the coordinate two-dimensional rectangular coordinate system, obtained Nkp reference point coordinate position within the two-dimensional rectangular coordinate system, in the offline phase in the received signal, at each reference point machine acquisition signal strength (RSS) from each of the access point AP and the access point AP as the location characteristic information; and build indoor coverage map based on the position information of the N feature access point AP Radio Map;

[0013] 步骤三、采用本征维数估计算法对步骤二获得的室内信号覆盖图Radio Map进行本征维数分析,获得本征维数分析结果; [0013] Step three, using the intrinsic dimension of the indoor coverage estimation algorithm of FIG. Step II was performed Radio Map which intrinsic dimension analysis, intrinsic dimension analysis result obtained;

[0014] 步骤四、根据步骤三获得的本征维数分析结果采用LED算法将室内信号覆盖图Radio Map内的所有参考点降维至本征维数,获得特征变换矩阵,并生成降维后的信号覆盖图Radio Map*; [0014] Step four, LED algorithm according to the intrinsic dimension analysis step III was the result of all the reference points in the indoor coverage FIG Radio Map dimensionality reduction to the intrinsic dimension is obtained wherein the transformation matrix, and generates the dimensionality reduction FIG signal coverage Radio Map *;

[0015] 步骤五、在在线阶段,测量室内环境中欲定位点的信号强度RSS值,并将该信号强度RSS值与步骤四获得的特征变换矩阵相乘,获得信号强度变换值RSS* ; [0015] Step 5 in the online phase, measuring the indoor environment to be setpoint signal strength (RSS), and converts the signal intensity RSS characteristic value obtained in step four matrix multiplication, a signal strength RSS * conversion value;

[0016] 步骤六、采用KNN算法对步骤五获得的信号强度变换值RSS*与步骤四生成的降维后的信号覆盖图Radio Map*进行位置匹配,获得欲定位点的位置坐标,完成欲定位点的室内定位。 [0016] Step 6 using KNN algorithm coverage map the signal intensity conversion value RSS * and Step V was four generating dimensionality reduction Radio Map * position matching, the position coordinates to be the anchor point, to complete the desire positioning indoor positioning points.

[0017] 步骤三中采用本征维数估计算法为特征值估计法。 Three using [0017] Step intrinsic characteristic dimension estimated value estimation algorithm.

[0018] 步骤三中采用本征维数估计算法为包数估计法。 Three using [0018] Step intrinsic dimension estimation algorithm is a packet number estimating method.

[0019] 步骤三中采用本征维数估计算法为测地线最小生成树算法。 Three using [0019] Step intrinsic dimension estimation algorithm geodesic minimum spanning tree algorithm.

[0020] 采用测地线最小生成树算法对步骤二获得的室内信号覆盖图Radio Map进行本征维数分析是通过公式: [0020] The minimum spanning tree algorithm geodesic indoor coverage FIG Step II was performed Radio Map which intrinsic dimension is analyzed by the formula:

1 1

[0021] intrinsic dim = Z [0021] intrinsic dim = Z

I — Cl I - Cl

[0022] 实现的;式中:dintHnsic;dim为本征维数分析结果;a表示最小生成树的线性拟合表达式的斜率。 [0022] implemented; wherein: dintHnsic; dim intrinsic dimension analysis result; A represents the slope of a linear fitting expression minimum spanning tree.

[0023] 步骤四中获得特征变换矩阵与生成降维后的信号覆盖图Radio Map*之间的关系为: [0023] Step four is obtained wherein the transformation between the matrix Map * after generating dimensionality reduction FIG Radio coverage is:

Figure CN103079269AD00061

[0025] 采用KNN算法对步骤五获得的信号强度变换值RSS*与步骤四生成的降维后的信号覆盖图Radio Map*进行位置匹配的方法是通过公式: [0025] The method Map * KNN algorithm for matching the position of the dimensional coverage map down converted signal strength RSS * value obtained in the Step five four Radio generated by the formula:

[0026] (λΛ/)= — [0026] (λΛ /) = -

[0027] 实现的; [0027] implemented;

[0028] 式中:U',y')为欲定位点的坐标,(Xi,Yi)为第i个近邻点的坐标,i为正整数;K为KNN算法中近邻点的总数。 [0028] wherein: U ', y') is the coordinates of the point to be positioned, (Xi, Yi) coordinates of the i-th neighbor points, i is a positive integer; K is the total number of neighbor points in the KNN algorithm.

[0029] 本发明的WiFi室内定位实时性高。 [0029] WiFi indoor positioning invention real-time high. 同时,本发明采用LDE算法将Radio Map降维至本征维数,降低了现有WiFi室内定位方法中存在的Radio Map数据量大,以及降低了在线定位阶段计算复杂度,使其易于在移动终端实现。 Meanwhile, the present invention uses the Radio Map LDE dimension reduction algorithm to the intrinsic dimension, reducing the amount of data existing Radio Map WiFi indoor positioning method exists, and to reduce the computational complexity of the line positioning stage, make it easy to move the terminal implementation.

附图说明 BRIEF DESCRIPTION

[0030] 图1是具体实施方式一中所述的实验场景示意图。 [0030] FIG. 1 is a schematic view of a specific embodiment the experimental scene. 图2是本发明方法的信号流程示意图。 FIG 2 is a signal flow of the method of the present invention. FIG.

具体实施方式 detailed description

[0031] 具体实施方式一、结合图2说明本具体实施方式,基于LDE算法的WiFi室内定位方法,它由以下步骤实现: [0031] a specific embodiment, described in conjunction with FIG. 2 present specific embodiment, WiFi LDE method of indoor positioning algorithm, which is implemented by the steps of:

[0032] 步骤一、针对室内环境布置`N个接入点AP,确保所述环境中任意一点被一个或一个以上的接入点AP发出的信号覆盖,所述N个接入点AP组成WiFi网络;在所述室内环境中均匀设置Nkp个参考点;N和Nkp均为正整数; [0032] Step a, are arranged for the indoor environment `the N access point AP, to ensure coverage of the environment at any point by one or more than one access point AP sent by the N WiFi access points AP Composition network; Nkp reference points uniformly arranged in the indoor environment; N are positive integers and Nkp;

[0033] 步骤二、选取一个参考点为坐标原点建立二维直角坐标系,获得Nkp个参考点在该二维直角坐标系中的坐标位置,在离线阶段中在每个参考点上利用信号接收机采集来自每一个接入点AP的信号强度RSS值,并作为该接入点AP的位置特征信息;并根据N个接入点AP的位置特征信息构建室内信号覆盖图Radio Map ; [0033] Step 2. Selection of a reference point to establish the origin of the coordinate two-dimensional rectangular coordinate system, obtained Nkp reference point coordinate position within the two-dimensional rectangular coordinate system, in the offline phase in the received signal, at each reference point machine acquisition signal strength (RSS) from each of the access point AP and the access point AP as the location characteristic information; and build indoor coverage map based on the position information of the N feature access point AP Radio Map;

[0034] 步骤三、采用本征维数估计算法对步骤二获得的室内信号覆盖图Radio Map进行本征维数分析,获得本征维数分析结果; [0034] Step three, using the intrinsic dimension of the indoor coverage estimation algorithm of FIG. Step II was performed Radio Map which intrinsic dimension analysis, intrinsic dimension analysis result obtained;

[0035] 步骤四、根据步骤三获得的本征维数分析结果采用LED算法将室内信号覆盖图Radio Map内的所有参考点降维至本征维数,获得特征变换矩阵,并生成降维后的信号覆盖图Radio Map*; [0035] Step four, LED algorithm according to the intrinsic dimension analysis step III was the result of all the reference points in the indoor coverage FIG Radio Map dimensionality reduction to the intrinsic dimension is obtained wherein the transformation matrix, and generates the dimensionality reduction FIG signal coverage Radio Map *;

[0036] 步骤五、在在线阶段,测量室内环境中欲定位点的信号强度RSS值,并将该信号强度RSS值与步骤四获得的特征变换矩阵相乘,获得信号强度变换值RSS* ; [0036] Step 5 in the online phase, measuring the indoor environment to be setpoint signal strength (RSS), and converts the signal intensity RSS characteristic value obtained in step four matrix multiplication, a signal strength RSS * conversion value;

[0037] 步骤六、采用KNN算法对步骤五获得的信号强度变换值RSS*与步骤四生成的降维后的信号覆盖图Radio Map*进行位置匹配,获得欲定位点的位置坐标,完成欲定位点的室内定位。 [0037] Step 6 using KNN algorithm coverage map the signal intensity conversion value RSS * and Step V was four generating dimensionality reduction Radio Map * position matching, the position coordinates to be the anchor point, to complete the desire positioning indoor positioning points.

[0038] 步骤三中采用本征维数估计算法为特征值估计法、包数估计法或测地线最小生成树算法。 Three using [0038] Step intrinsic characteristic dimension estimated value estimation algorithm, or the number of packets estimation geodesic minimum spanning tree algorithm. [0039] 采用测地线最小生成树算法对步骤二获得的室内信号覆盖图Radio Map进行本征维数分析是通过公式: [0039] The minimum spanning tree algorithm geodesic indoor coverage FIG Step II was performed Radio Map which intrinsic dimension is analyzed by the formula:

[0040] [0040]

Figure CN103079269AD00071

[0041] 实现的;式中:dintHnsic;dim为本征维数分析结果;a表示最小生成树的线性拟合表达式的斜率。 [0041] implemented; wherein: dintHnsic; dim intrinsic dimension analysis result; A represents the slope of a linear fitting expression minimum spanning tree.

[0042] 步骤四中获得特征变换矩阵V'与生成降维后的信号覆盖图Radio Map*之间的关系为: [0042] Step four transformation matrix obtained wherein V 'and generates a signal coverage down view of the dimensional relationship between the Radio Map * is:

[0043] Radio Map* = Y'.Radio Map。 [0043] Radio Map * = Y'.Radio Map.

[0044] 采用KNN算法对步骤五获得的信号强度变换值RSS*与步骤四生成的降维后的信号覆盖图Radio Map*进行位置匹配的方法是通过公式: [0044] The method Map * KNN algorithm for matching the position of the dimensional coverage map down converted signal strength RSS * value obtained in the Step five four Radio generated by the formula:

[0045] [0045]

Figure CN103079269AD00072

[0046] 实现的; [0046] implemented;

[0047] 式中:U',y')为欲定位点的坐标,(Xi,Yi)为第i个近邻点的坐标,i为正整数;K为KNN算法中近邻点的总数。 [0047] wherein: U ', y') is the coordinates of the point to be positioned, (Xi, Yi) coordinates of the i-th neighbor points, i is a positive integer; K is the total number of neighbor points in the KNN algorithm.

[0048] 本实施方式中,Radio Map的本征维数的获取通过下述步骤实现: [0048] The embodiment according to the present embodiment, the intrinsic dimension Radio Map which acquisition achieved by the following steps:

[0049] 本征维数是对于高维数据进行本征空间维数及空间重建所需最小的独立变量的个数。 [0049] The intrinsic dimension is the eigenspace dimensions and spatial reconstruction of the required minimum number of independent variables with respect to high-dimensional data. 在具体实际计算中,由于高维数据的本征并不明显,通常不是寻求得到确切的本征维数,而是寻求估计本征维数的可信取值。 In particular the actual calculation, due to the intrinsic high-dimensional data is not obvious, it is not usually seek to get the exact dimensions of the intrinsic, but seek credible estimate of intrinsic value dimension. 具体的说,给定一个来自高维空间的样本,本征维数估计算法的中心任务和重要内容就是通过这些样本数据来确定这个高维结构的本征维数。 Specifically, given a sample from a high-dimensional space, the intrinsic dimension estimate the central task and an important part of the algorithm is to determine the intrinsic dimension of the high-dimensional structure of these sample data.

[0050] Radio Map的本征维数的估计是LDE算法的重要输入参数,这关系到降维的结果是否能够代表Radio Map的高维空间的特征,因此准确有效的本征维数的估计至关重要。 [0050] The estimate of Radio Map of the intrinsic dimension is an important input parameter LDE algorithm, whether it relates to the results of dimensionality reduction can represent features Radio Map of high-dimensional space, so accurate and effective is estimated that the number of intrinsic dimension to critical. 目前,常用本征维数估计算法分为两类:局部估计与全局估计。 At present, the common intrinsic dimension estimation algorithm is divided into two categories: partial estimation and global. 在本专利中,采用全局部算法估计对Radio Map的本征维数进行估计,并作为LDE算法的输入变量。 In this patent, all-local algorithm estimates the number of Radio Map intrinsic dimension is estimated, and LDE algorithm as input variables. 本专利中采用测地距最小生成树算法(Geodesic Minimum Spanning Tree,GMST)对Radio Map的本征维数进行估计。 Geodetic distance using minimum spanning tree algorithm (Geodesic Minimum Spanning Tree, GMST) number of Radio Map intrinsic dimension is estimated in this patent. 下面对GMST算法的理论进行分析。 The following theory GMST algorithm is analyzed.

[0051] 测地线最小生成树(GMST)估计是基于测地线最小生成树的长度函数强烈依赖于本征维数d的。 [0051] geodesic minimum spanning tree (GMST) estimate is based on a minimum spanning tree of the geodesic length function is strongly dependent on the present intrinsic dimensionality of d. GMST是指定义在数据集X上的近邻曲线的最小生成树。 Refers to the minimum spanning tree is defined GMST neighbor graph on the X data set. GMST的长度函数是在测地线最小生成树中所有边缘对应的欧氏距离之和。 GMST length function is a minimum spanning tree geodesic all edges corresponding to the sum of the Euclidean distance.

[0052] 与ISOMAP相似,GMST估计在数据集X上构造一条近邻曲线G,其中,在X内每一个数据点Xi都和它的k个近邻力,相连接。 [0052] Similar to ISOMAP, GMST an estimated neighbor configuration data set in the curve G X, wherein X in each of the data points Xi are the k nearest neighbors and its force, is connected. 测地线最小生成树T定义为X上的最小曲线,它具 Geodesic minimum spanning tree T is defined as the minimum on the curve X, which has

有长度: There Length:

[0053] [0053]

Figure CN103079269AD00073

[0054] 其中,T是曲线G的所有子树集合,e是树T的一个边缘,ge是边缘e对应的欧氏距离。 [0054] where, T is the curve G of all subtrees set, e is an edge of a tree T, ge edge corresponding to the Euclidean distance e. 在GMST估计中,一些子集JC: I由各种大小m组成,并且子集A的GMST的长度L (A)也需要计算。 In GMST estimation, some subset of the JC: I m from the composition of various sizes, and a subset of GMST A length L (A) can be calculated. [0055] 理论上, [0055] In theory,

Figure CN103079269AD00081

是线性的,从而可以由:y = ax+b这种形式的函数来估计,通过最小二乘法可以估算出变量a和b。 Is linear, it can be made: estimated y = ax + b of this function of the form, can be estimated variables a and b by the least squares method.

[0056] 可以证明:由a的估算值和d=1/1-a能够得到本征维数的估计。 [0056] can be shown that: estimates of a and d = 1/1-a can be estimated intrinsic dimension.

[0057] 由GMST算法给出本征维数d的表达式为式(2)所示。 [0057] algorithm given by GMST intrinsic dimension d expression represented by the formula (2). 本征维数d是LDE算法的另一个重要的输入参数。 Intrinsic dimension d LDE algorithm is another important input parameter.

[0058] d =1/1-a [0058] d = 1/1-a

[0059] 运用LDE算法实现对Radio Map进行降维并获取特征权值矩阵过程通过下述步骤实现: [0059] Application of LDE Radio Map algorithm to reduce the dimension and weight acquisition value matrix process characterized by the following steps implemented:

[0060] LDE算法是基于类间散度及类内散度最大化的一种流形学习算法。 [0060] LDE algorithm is based on the divergence between classes and maximization of a divergence manifold learning algorithms. 在对LDE算法 In the algorithm of LDE

进行理论分析之前对LDE算法给定的输入数据做如下说明:输入高维数据点 Do the following instructions before LDE theoretical analysis of a given algorithm input data: Input high-dimensional data points

Figure CN103079269AD00082

数据点Xi的类标记为Yi ∈{1,2,…,P},其中P表示将高维数据划分为P个子流形,即将输入的高维数据分成P类。 The data points marked with a class Xi Yi ∈ {1,2, ..., P}, where P represents the high-dimensional data into P sub-manifold, forthcoming high-dimensional input data into classes P.

[0061] 将输入的高维数据表示成矩阵的形式:X = [x1; X2,…,Xffl]∈Rnx'从矩阵表示的形式来看,矩阵中的列代表一个高维数据点。 [0061] high-dimensional representation of the input data into a matrix: X = [x1; X2, ..., Xffl] ∈Rnx expressed in the form of a matrix from the 'point of view, a matrix columns represent high-dimensional data points. 下面结合错误! The following combination of error! 未找到引用源。 Reference source not found. 所示的LDE算法流程来对其算法理论进行推导。 LDE algorithm flow shown in its theoretical derivation algorithm.

[0062] 构造邻接图:根据高维数据点的类标记信息及其近邻关系构造无方向图G及G'。 [0062] FIG abutment structure: The tag class high-dimensional data point information and the neighbor configuration relations undirected graph G and G '. 其中近邻关系是采用KNN算法给出的准则,即选择数据点最近的K个点作为其邻居,G表示当Xi与Xj的类标记信息Ji≠yj时且Xp Xj互为K近邻关系;G'示当Xi与Xj的类标记信息关y」时且X1、χ」互为K近邻关系。 Wherein the neighborhood relation is given using KNN algorithm criterion, i.e., to select the nearest data point as its neighbor points K, G represents a Xp Xj and K mutually neighbor relation when Xi and Xj class mark information Ji ≠ yj; G ' It shows when Xi and Xj information on the class mark y and X1, χ "mutual relationship when K neighbors."

[0063] 计算权值矩阵:根据(I)构造的邻接图采用类高斯函数进行权值矩阵的计算。 [0063] calculating the weight matrix: weight matrix is ​​calculated according to the adjacency graph (I) constructed using Gaussian functions. 其表达式为(3)所示。 Which was expressed as (3) of FIG. 公式(3)中&表示近邻点Xi与\之间的权值, Equation (3) represents the weight between & neighbor points Xi and \,

Figure CN103079269AD00083

为近邻点Xi与\之间的范数距离,采用矩阵方式计算范数距离,t为权值归一化参数。 Norm as the distance between \ neighbor points Xi, is calculated using the matrix norm distance, t is the weight normalization parameters.

Figure CN103079269AD00084

[0065] 计算嵌入结果:根据LDE算法的目标——最大化类间散度地同时最小化类内散度。 [0065] Calculation results embedding: LDE algorithm according to the target - to maximize between-class scatter while minimizing the degree of within-class scatter. 散度采用表示同类数据点及不同类的范数距离表示。 Divergence expressed using the same norm of data points from different classes and expressed. 由LDE算法的目标可以得出其相应的优化目标函数,如式所示。 The algorithm can be derived target LDE its corresponding objective function, as shown in the formula. [0066] [0066]

Figure CN103079269AD00091

[0067] 根据式(4)给出的优化目标函数作以下分析: [0067] The formula (4) given objective function for the following analysis:

[0068] 根据矩阵范数的计算式:IMII = YAi,计算式表示为矩阵A的矩阵范数的计算方 [0068] The norm of the matrix calculation formula: IMII YAi, = norm calculation formula for calculating the number of square matrix A

法,计算式给出的方法与矩阵的迹的计算式一致,即: Calculating equation trace method, a method of calculating the matrix given consistent, namely:

Figure CN103079269AD00092

由此式(4)可以表示为矩阵的迹的计算方式: Thereby the formula (4) may be expressed as a trace of a matrix is ​​calculated:

[0069] [0069]

Figure CN103079269AD00093

[0070] 式(5)可以简化为: [0070] Formula (5) can be simplified as:

[0071] [0071]

Figure CN103079269AD00094

[0072]由矩阵迹的计算的标量性质及权值元素均为实数,可以将式(6)简化为: Simplified [0072] The scalar properties and weight matrix calculation trace elements are real numbers, may be of formula (6):

[0073] [0073]

Figure CN103079269AD00095

[0074] 根据简单的数学关系,可以将式(7)简化为: [0074] According to a simple mathematical relationship can be simplified formula (7):

[0075] J(V) = 2tr{VT[X(D/ -Ψ )XT]V} (8) [0075] J (V) = 2tr {VT [X (D / -Ψ) XT] V} (8)

[0076] 式⑶中:X为输入数据,λ,V为特征值与特征向量,W,Ψ分别为G及Gi对应的权值矩阵,D及D'均为对角阵,其对角元素可以由式(9)表示。 [0076] ⑶ formula: X is the input data, λ, V is the eigenvalue and eigenvector, W, Ψ Gi and G respectively corresponding weight matrix, D and D 'are diagonal matrices, the diagonal elements of which may be represented by the formula (9).

[0077] [0077]

Figure CN103079269AD00096

[0078] 根据式⑶的推导方式,同理可以将⑷中的约束条件写成如式⑶相似的形式,由此,可以将(4)表示为如下形式: [0078] The derivation of Formula ⑶, empathy ⑷ constraints may be written in a similar form as ⑶ formula, whereby (4) can be expressed as follows:

[0079] [0079]

Figure CN103079269AD00097

[0080] 对式(10)应用拉格朗日(Lagrange)乘数法,可以得出式(11)所示: [0080] The formula (10) Lagrange (LAGRANGE) multiplier method, can be obtained by formula (11):

[0081] X(D' -r )Χτν = λ X(DW)XTv (11) [0081] X (D '-r) Χτν = λ X (DW) XTv (11)

[0082] 对式(11)进行广义特征值分解,得出其特征值分解的特征值及特征向量,表示为:λ =IIApX2,..., λ η]τ,其对应的特征向量为:V = [V1, ν2,…,νη]τ。 [0082] The formula (11) is generalized eigenvalue decomposition, eigenvalues ​​and eigenvectors obtained which eigenvalue decomposition, it is expressed as: λ = IIApX2, ..., λ η] τ, feature vectors corresponding to: V = [V1, ν2, ..., νη] τ. 取前d个最大的特征值对应的特征向量构成变换矩阵V= [Vl,v2,-,vd]0由LDE算法的输出数据变换方法可以得出,降维后数据为: D take the first largest eigenvalue corresponding eigenvectors transform matrix V = [Vl, v2, -, vd] 0 can be derived from the output of the data conversion method LDE algorithm, reducing the data dimension:

[0083] Zi = V1Xi (12)[0084] 式(12)中,Zi表示输入高维数据点Xi变换后的低维输出数据。 In [0083] Zi = V1Xi (12) [0084] Formula (12), Zi represents the input of the low-dimensional output data points Xi transform high-dimensional data.

[0085] 上述的分析是根据LDE算法流程给出的理论分析及说明。 [0085] The above-described assay is based on the theoretical analysis and the algorithm flow LDE described.

[0086] 由式(5)〜(11)给出LDE算法的理论推导。 [0086] Theory given by formula (5) to (11) LDE derivation algorithm. 通过LDE算法可以得出降维后的信号覆盖图及特征变换矩阵,分别记为Radio Map*和V'。 Algorithm can be derived by LDE signals and overlay characteristics after dimensionality reduction transformation matrix, denoted as Radio Map * and V '.

[0087] 在线定位阶段对RSS及KNN匹配定位通过下述步骤实现: [0087] RSS-line positioning stage and KNN Targeting achieved by the following steps:

[0088] 结合图2的在线阶段所示的流程图对具体实施方式四进行详细说明。 [0088] The specific embodiments described in detail in conjunction with four flowchart shown in FIG. 2 of the online phase. 在线阶段,测试点处接收的RSS= [AP1, AP2,…,APn],与特征变换矩阵V'相乘,从而得出降维后的RSS' = [AP1 ,AP2,…APd],其中d表示本征维数。 Online phase, at the test point the received RSS = [AP1, AP2, ..., APn], wherein the transform matrix V 'is multiplied, so as to arrive after the dimensionality reduction RSS' = [AP1, AP2, ... APd], where d It represents the intrinsic dimension. 再采用KNN算法实现RSS^与Radio Map*的匹配。 Then using KNN algorithm RSS ^ matches the Radio Map *. 采用与RSS'最近的K个参考点的坐标的平均值作为测试点(X' , j'),其表达式为: Using the RSS 'average of the coordinates of the K nearest reference point as a test point (X', j '), which is expressed as:

[0089] [0089]

Figure CN103079269AD00101

[0090] AP在具体的室内布置及实验过程如下例所示:图1示为哈尔滨工业大学科学园2A栋12层的平面图示意,基于WiFi的室内定位系统就是基于该实验环境下建立。 [0090] AP chamber and arranged in a particular experiment the following example: FIG. 1 is shown Harbin Institute of Technology Science Park, Building 12 layer 2A a schematic plan view, WiFi indoor positioning system is based on the establishment of the experimental environment. 在实验环境中,总共布置27个AP,AP离房间地面高度为2米。 In the experimental environment, a total of 27 are arranged AP, AP from the room on the ground height of 2 meters. 在离线阶段,在联想V450笔记本上安装NetStumbler软件,在所有参考点的四个不同的方位上连续采样记录AP的100个RSS值,以及AP的相关信息。 In the offline stage, NetStumbler software installed on the Lenovo V450 notebook, continuous sampling records information about the AP 100 RSS values, as well as the AP in four different directions all reference points. 将所有的采样点的物理坐标及相应的物理坐标及RSS值存储为定位过程所调用的数据,建立Radio Map。 The physical coordinates all the sample points and the coordinates and the corresponding physical RSS value is stored as the positioning data of the called procedure, the establishment of Radio Map. 在实验环境共有900个参考点,其采样密度为0.5米X0.5米。 A total of 900 reference points in the experimental environment, the sampling density is 0.5 m X0.5 m. Radio Map作为LDE算法的输入参数及本征维数估计算法的输入数据。 Radio Map which algorithm input data as an input parameter estimation algorithm and LDE present intrinsic dimensionality.

Claims (7)

1.基于LDE算法的WiFi室内定位方法,其特征是:它由以下步骤实现: 步骤一、针对室内环境布置N个接入点AP,确保所述环境中任意一点被一个或一个以上的接入点AP发出的信号覆盖,所述N个接入点AP组成WiFi网络;在所述室内环境中均匀设置Nkp个参考点;N和Nkp均为正整数; 步骤二、选取一个参考点为坐标原点建立二维直角坐标系,获得Nkp个参考点在该二维直角坐标系中的坐标位置,在离线阶段中在每个参考点上利用信号接收机采集来自每一个接入点AP的信号强度RSS值,并作为该接入点AP的位置特征信息;并根据N个接入点AP的位置特征信息构建室内信号覆盖图Radio Map ; 步骤三、采用本征维数估计算法对步骤二获得的室内信号覆盖图Radio Map进行本征维数分析,获得本征维数分析结果; 步骤四、根据步骤三获得的本征维数分析结果采用LED算法将室内信号覆盖图 1. WiFi LDE method of indoor positioning algorithm, characterized in that: it is implemented by the following steps: Step 1, the N access point AP is arranged for indoor environment to ensure that the environment is any point of access to one or more coverage point AP sent by the access point AP consisting of N WiFi network; Nkp reference points uniformly arranged in the indoor environment; Nkp and N are positive integers; step 2. selection of a reference point for the coordinate origin establishing a two-dimensional Cartesian coordinate system, obtained Nkp reference point coordinate position in the two-dimensional Cartesian coordinate system, the signal strength RSS acquired from each of the access point AP using the signal receiver at each reference point in the offline phase value, as the position of the access point AP feature information; and build indoor coverage FIG Radio Map according to the position wherein the N access point AP information; step three, using the intrinsic dimension estimation algorithm obtained in step two of the chamber FIG Radio Map coverage for intrinsic dimension analysis to give intrinsic dimension analysis result; step four, LED indoor coverage algorithm according to the intrinsic FIG step three dimensional analysis of the results obtained RadioMap内的所有参考点降维至本征维数,获得特征变换矩阵,并生成降维后的信号覆盖图Radio Map*; 步骤五、在在线阶段,测量室内环境中欲定位点的信号强度RSS值,并将该信号强度RSS值与步骤四获得的特征变换矩阵相乘,获得信号强度变换值RSS* ; 步骤六、采用KNN算法对步骤五获得的信号强度变换值RSS*与步骤四生成的降维后的信号覆盖图Radio Map*进行位置匹配,获得欲定位点的位置坐标,完成欲定位点的室内定位。 All reference point within RadioMap dimensionality reduction to the intrinsic dimension is obtained wherein the transformation matrix, and generates a down signal overlay dimensionality Radio Map *; Step 5 in the online phase, measuring the indoor environment to be setpoint signal strength RSS value, and converts the signal intensity RSS characteristic value obtained in step four matrix multiplication, a signal strength RSS * conversion value; step 6 using KNN algorithm for converting the signal strength RSS * obtained in step five with four values ​​generated in step FIG coverage the Radio Map * dimensionality reduction matched position, the position coordinates of the point to be positioned, to complete the room to be positioned anchor points.
2.根据权利要求1所述的基于LDE算法的WiFi室内定位方法,其特征在于步骤三中采用本征维数估计算法为特征值估计法。 The LDE WiFi indoor positioning algorithm based on the method of claim 1, wherein in step three dimension using the intrinsic characteristic estimation value estimation algorithm.
3.根据权利要求1所述的基于LDE算法的WiFi室内定位方法,其特征在于步骤三中采用本征维数估计算法为包数估计法。 The LDE WiFi indoor positioning algorithm based on the method of claim 1, wherein in step three using the intrinsic dimension estimation algorithm packet number estimating method.
4.根据权利要求1所述的基于LDE算法的WiFi室内定位方法,其特征在于步骤三中采用本征维数估计算法为测地线最小生成树算法。 The LDE WiFi indoor positioning algorithm based on the method of claim 1, wherein in step three using the intrinsic dimension estimation algorithm geodesic minimum spanning tree algorithm.
5.根据权利要求4所述的基于LDE算法的WiFi室内定位方法,其特征在于采用测地线最小生成树算法对步骤二获得的室内信号覆盖图Radio Map进行本征维数分析是通过公式: The WiFi location based indoor LDE algorithm according to claim 4, characterized in that a minimum spanning tree algorithm geodesic indoor coverage FIG Step II was performed Radio Map which intrinsic dimension is analyzed by the formula:
Figure CN103079269AC00021
实现的;式中:dintHnsic;dim为本征维数分析结果;&表示最小生成树的线性拟合表达式的斜率。 Achieved; wherein: dintHnsic; dim intrinsic dimension analysis result; & represents the slope of the linear fit of the minimum spanning tree expressions.
6.根据权利要求1所述的基于LDE算法的WiFi室内定位方法,其特征在于步骤四中获得特征变换矩阵V'与生成降维后的信号覆盖图Radio Map*之间的关系为: Radio Map* = Y'.Radio Map。 The LDE WiFi indoor positioning algorithm based on the method of claim 1, wherein in step IV was characterized transformation matrix V 'Map * the relation between overlay down signal generating dimensionality of Radio: Radio Map * = Y'.Radio Map.
7.根据权利要求1所述的基于LDE算法的WiFi室内定位方法,其特征在于采用KNN算法对步骤五获得的信号强度变换值RSS*与步骤四生成的降维后的信号覆盖图RadioMap*进行位置匹配的方法是通过公式: The WiFi location based indoor LDE algorithm according to claim 1, characterized in that a KNN algorithm coverage map conversion value after the signal intensity RSS * with four Step V was generated for dimension reduction RadioMap * the method of matched positions by the formula:
Figure CN103079269AC00031
实现的; 式中:(Χ',I')为欲定位点的坐标,(Xi, Yi)为第i个近邻点的坐标,i为正整数;κ为KNN算法中近邻点的总数。 Achieved; wherein: (Χ ', I') is the coordinates of the point to be positioned, (Xi, Yi) coordinates of the i-th neighbor points, i is a positive integer; [kappa] Total KNN algorithm neighbor points is.
CN 201310029536 2013-01-25 2013-01-25 LDE (Linear Discriminant Analysis) algorithm-based WiFi (Wireless Fidelity) indoor locating method CN103079269A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201310029536 CN103079269A (en) 2013-01-25 2013-01-25 LDE (Linear Discriminant Analysis) algorithm-based WiFi (Wireless Fidelity) indoor locating method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201310029536 CN103079269A (en) 2013-01-25 2013-01-25 LDE (Linear Discriminant Analysis) algorithm-based WiFi (Wireless Fidelity) indoor locating method

Publications (1)

Publication Number Publication Date
CN103079269A true true CN103079269A (en) 2013-05-01

Family

ID=48155648

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201310029536 CN103079269A (en) 2013-01-25 2013-01-25 LDE (Linear Discriminant Analysis) algorithm-based WiFi (Wireless Fidelity) indoor locating method

Country Status (1)

Country Link
CN (1) CN103079269A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103648106A (en) * 2013-12-31 2014-03-19 哈尔滨工业大学 WiFi indoor positioning method of semi-supervised manifold learning based on category matching
CN104424276A (en) * 2013-08-30 2015-03-18 中国电信集团公司 Method and device for self-updating fingerprint database based on manifold learning

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110164522A1 (en) * 2006-05-08 2011-07-07 Skyhook Wireless, Inc. Estimation of Position Using WLAN Access Point Radio Propagation Characteristics in a WLAN Positioning System
CN102348160A (en) * 2011-07-15 2012-02-08 中国电信股份有限公司 Multimode signal-based positioning method, system and positioning platform
CN102419180A (en) * 2011-09-02 2012-04-18 无锡智感星际科技有限公司 Indoor positioning method based on inertial navigation system and WIFI (wireless fidelity)

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110164522A1 (en) * 2006-05-08 2011-07-07 Skyhook Wireless, Inc. Estimation of Position Using WLAN Access Point Radio Propagation Characteristics in a WLAN Positioning System
CN102348160A (en) * 2011-07-15 2012-02-08 中国电信股份有限公司 Multimode signal-based positioning method, system and positioning platform
CN102419180A (en) * 2011-09-02 2012-04-18 无锡智感星际科技有限公司 Indoor positioning method based on inertial navigation system and WIFI (wireless fidelity)

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
邓志安: "《基于学习算法的WLAN室内定位技术研究》", 《哈尔滨工业大学》, 1 June 2012 (2012-06-01) *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104424276A (en) * 2013-08-30 2015-03-18 中国电信集团公司 Method and device for self-updating fingerprint database based on manifold learning
CN104424276B (en) * 2013-08-30 2017-12-01 中国电信集团公司 Manifold learning based fingerprint database from the update method and apparatus
CN103648106A (en) * 2013-12-31 2014-03-19 哈尔滨工业大学 WiFi indoor positioning method of semi-supervised manifold learning based on category matching
CN103648106B (en) * 2013-12-31 2017-03-22 哈尔滨工业大学 WiFi indoor positioning method for matching categories based on semi-supervised manifold learning

Similar Documents

Publication Publication Date Title
Segal et al. Generalized-icp.
US20090148051A1 (en) Correlatability analysis for sparse alignment
Singer et al. Vector diffusion maps and the connection Laplacian
Zhang et al. Aircraft recognition in infrared image using wavelet moment invariants
Broadwater et al. Hybrid detectors for subpixel targets
Slama et al. Accurate 3D action recognition using learning on the Grassmann manifold
CN101751695A (en) Estimating method of main curvature and main direction of point cloud data
France et al. Two-way multidimensional scaling: A review
Zhang et al. Endmember extraction of hyperspectral remote sensing images based on the ant colony optimization (ACO) algorithm
CN102802260A (en) WLAN indoor positioning method based on matrix correlation
Xiang et al. Embedding new data points for manifold learning via coordinate propagation
Xiao et al. Geometric characterization and clustering of graphs using heat kernel embeddings
CN101853506A (en) High optical spectrum image end member extraction method based on optimized search strategy
CN103476118A (en) WLAN indoor location fingerprint positioning method used for real-time monitoring
Liu et al. Visual homing from scale with an uncalibrated omnidirectional camera
Li et al. Vanishing point detection using cascaded 1D Hough Transform from single images
Sun et al. Adaptive localization through transfer learning in indoor wi-fi environment
Zhang et al. COCKTAIL: An RF-based hybrid approach for indoor localization
CN102638889A (en) Indoor wireless terminal positioning method based on Bayes compression sensing
Ash et al. On the relative and absolute positioning errors in self-localization systems
Teslic et al. Nonlinear system identification by Gustafson–Kessel fuzzy clustering and supervised local model network learning for the drug absorption spectra process
Będkowski et al. Real time 3D localization and mapping for USAR robotic application
Zhao et al. Optimal sensor placement for target localisation and tracking in 2D and 3D
Pan et al. Accurate and low-cost location estimation using kernels
Kong et al. Optimizing the spatio-temporal distribution of cyber-physical systems for environment abstraction

Legal Events

Date Code Title Description
C06 Publication
C10 Entry into substantive examination
C02 Deemed withdrawal of patent application after publication (patent law 2001)