CN101867943A - WLAN indoor tracking method based on particle filtering algorithm - Google Patents

WLAN indoor tracking method based on particle filtering algorithm Download PDF

Info

Publication number
CN101867943A
CN101867943A CN 201010207003 CN201010207003A CN101867943A CN 101867943 A CN101867943 A CN 101867943A CN 201010207003 CN201010207003 CN 201010207003 CN 201010207003 A CN201010207003 A CN 201010207003A CN 101867943 A CN101867943 A CN 101867943A
Authority
CN
China
Prior art keywords
point
particle
coordinates
particles
step
Prior art date
Application number
CN 201010207003
Other languages
Chinese (zh)
Inventor
刘宁庆
刘菁宇
孟维晓
徐玉滨
沙学军
马琳
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
Application filed by 哈尔滨工业大学 filed Critical 哈尔滨工业大学
Priority to CN 201010207003 priority Critical patent/CN101867943A/en
Publication of CN101867943A publication Critical patent/CN101867943A/en

Links

Abstract

The invention provides a WLAN indoor tracking method based on a particle filtering algorithm, and relates to a WLAN indoor tracking method, and solving the problem that the tracking precision is reduced caused by the nolinear model when tracking a dynamic object in the indoor environment. The method comprises the following steps of: building signal intensity probability distribution of an off-line stage with the signal intensity value of a reference point, defining the reference point which has the shortest Euclidean distance with a particle point, taking the reference point as a particle point correcting valve to obtain each particle point and weight values corresponding thereto, and obtaining the estimated value of the position of the user after updating the weight value with a resample arithmetic. The method is suitable for the WLAN indoor tracking.

Description

基于粒子滤波算法的WLAN室内跟踪方法 WLAN indoor particle filter tracking method based on

技术领域 FIELD

[0001] 本发明涉及一种WLAN室内跟踪方法。 [0001] The present invention relates to a method of tracking an indoor WLAN. 背景技术 Background technique

[0002] WLAN网络是计算机网络与无线通信技术相结合的产物。 [0002] WLAN network is a computer network and wireless communication technology product of the combination. 它利用射频传输技术进行数据的传送,为用户提供无线宽带接入服务。 It uses radio transmission technology for data transmission, to provide users with broadband wireless access. WLAN的发展,解决了有线网络布线困难的问题,打破了宽带接入的地域限制,满足了用户移动数据通信的需要,实现数据通信的移动化、漫游化和宽带化。 WLAN development, solves the problem of wired network routing difficulties, breaking the geographical restrictions broadband access, to meet the needs of users of mobile data communications for mobile data communications, roaming and broadband. WLAN的产生和不断发展满足了人们对高效率、高质量、高带宽、低成本无线网络通信的需求。 WLAN generation and evolving to meet the people, demand for low-cost wireless communication network for high-efficiency, high-quality, high-bandwidth. WLAN的产生和不断发展满足了人们对高效率、高质量、高带宽、 低成本无线网络通信的需求。 WLAN generation and evolving to meet the people, demand for low-cost wireless communication network for high-efficiency, high-quality, high-bandwidth. 而在实际使用中,用于定位的用户终端可能是静止的,也有可能是移动的,所以对室内无线跟踪系统的研究在此背景下发展起来。 In actual use, the user terminal may be used to locate stationary, there may be mobile, so the study of indoor wireless tracking system developed in this context. 现有的室内跟踪技术主要有扩展卡尔曼滤波EKF(Extended Kalman Filter)、修正增益的扩展卡尔曼滤波MGEKF(Modified Gain Extended Kalman Filter)和粒子滤波PF(Particle Filter)等,而在其中,粒子滤波由于其精度可以逼近最优估计,而且适用于任何能用状态空间模型表示的非线性系统,受到高度重视,在近些年来得到快速的发展。 Existing technologies include indoor tracking extended Kalman filter EKF (Extended Kalman Filter), a gain correction Extended Kalman Filter MGEKF (Modified Gain Extended Kalman Filter) and a particle filter PF (Particle Filter) and the like, in which the particle filter Because of its accuracy can approximate the optimal estimate, but also apply to any non-linear system can be represented by the state-space model, we are highly valued, rapid development in recent years.

[0003] 粒子滤波算法是利用序列重要性采样的概念近似,用离散的随机样本近似相应的概率密度函数。 [0003] The particle filter using sequential importance sampling concept approximation, approximation with the corresponding probability density function of discrete random samples. 在粒子滤波中,概率密度函数被一系列离散的带权重的样本近似。 In the particle filter, the probability density function is approximated with weights of a series of discrete samples. 随着样本粒子数量的增加,粒子滤波接近于贝叶斯最优估计。 As the sample increases the number of particles, the particles close to the Bayes optimal estimation filter. 粒子滤波技术在非线性、非高斯系统表现出来的优越性,使得其成为在室内定位跟踪领域里人们研究的热点。 Particle filter technology in the non-linear, non-Gaussian system demonstrated superiority that made it a hot research topic in the field of indoor location tracking. 但是,在实际应用中,粒子滤波算法中粒子点的传统选取方法会对整个系统造成很大的负担,且环境适应也较差。 However, in practical applications, the conventional method of selecting particles have particle filter algorithm point of the whole system a great burden, and environment adaptation is also poor.

发明内容 SUMMARY

[0004] 本发明是为了解决室内环境下对动态目标的跟踪问题中,非线性模型问题导致跟踪精度降的问题,从而提供一种基于粒子滤波算法的WLAN室内跟踪方法。 [0004] The present invention is to solve the problems of indoor environmental dynamic target tracking problem, the problem causes tracking precision nonlinear model drop, thereby providing a particle filter tracking method indoor WLAN Algorithm.

[0005] 基于粒子滤波算法的WLAN室内跟踪方法,它由以下步骤实现: [0005] WLAN indoor particle filter tracking method algorithm, which is implemented by the steps of:

[0006] 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,并在所述室内环境中均勻设置Nkp个参考点; [0006] Step a, is arranged for the indoor environment of the N access point AP, to ensure coverage of the environment at any point by one or more than one access point AP emitted, and uniformly disposed in said chamber a Nkp environment reference point;

[0007] 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,并利用最大似然法计算每个参考点的信号强度先验概率分布; [0007] 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 using the maximum likelihood method to calculate the signal strength of each reference point prior probability distribution;

[0008] 步骤三、在第k时刻下,对描述粒子分布的重要性密度函数进行 [0008] Step three, in the first time k, the description of the importance of the particle distribution density function is

采样,获得队个粒子点ζί,计算第i个粒子点坐标与每个参考点坐标之间的欧几里得距离, 并选择欧几里得距离最小值所对应的参考点坐标作为该粒子点的修正坐标;所述i = 1、2……Ns ; Sampling particles obtain team points ζί, calculates the Euclidean distance between the i-th point coordinates particles and the coordinates of each reference point, the reference point coordinate and selects the minimum Euclidean distance corresponding point of the particles as correction coordinates; the i = 1,2 ...... Ns;

[0009] 步骤四、根据步骤二获得的每个参考点的信号强度先验概率分布和重要性密度函数以及第k-1时刻粒子点的权值计算第k时刻粒子点所对应的权值ωί ;步骤五、根据步骤四获得的第k时刻的粒子点的权值ωί,采用重采样算法对当前时刻的粒子点进行重新采样得到新的粒子点作为当前时刻的粒子点,并替代原有的粒子点,并获得最终的位置估计坐标; 所述最终的位置估计坐标即为第k时刻下待跟踪目标的坐标,然后返回执行步骤三,获得下一时刻的待跟踪目标的坐标; [0009] Step 4 according to the signal strength of each reference point obtained in step two a priori probability density function of the distribution and importance of the weight of the particle and the right point in time k-1 value of the k-th calculation point in time a value corresponding particle ωί ; step 5 points according to the weights of the particles at time k values ​​obtained in step four ωί, resampling algorithm using the particle current time point resampled to obtain new particles as point particles at the current time point, and replace the original the final position of the point of the particle, and obtain an estimated coordinates; position coordinates of the final estimate is the target to be tracked under the coordinate at time k, then returning to step three, the next target to be tracked to obtain the time coordinate;

[0010] N、Ns, Nep为正整数;k为大于或等于1的整数。 [0010] N, Ns, Nep is a positive integer; K is an integer greater than or equal to 1.

[0011] 步骤二中所述利用最大似然法计算每个参考点的信号强度先验概率分布是通过公式: [0011] Step II of the distributed computing each reference point using the maximum likelihood method is the prior probability signal strength by the formula:

[0012] [0012]

RSS值样本采样总数 RSS value of the sample total number of samples

[0013] 获得的,式中,Sj为第u个参考点处接收第j个AP的信号强度值;Coimt(Sj)表示在第U个参考点处接收到信号强度为~的信号数量-J1k为第k时刻第U个参考点的位置坐标; [0013] obtained, wherein, Sj j-th received signal strength value to the AP at the u-th reference points; Coimt (Sj) represents received at the first reference point U to the signal intensity - the number of signals -J1k U coordinates for the location of the reference point of time k;

[0014] [0014]

[0015] 步骤三中所述对重要性密度函数采样获得粒子点<,计算第i个粒子点坐标与每个参考点坐标之间的欧几里得距离是通过公式: [0015] step 3 the importance of the functions of sample points obtained particles <calculates the Euclidean distance between the i-th point coordinates particles and the coordinates of each reference point by the formula:

[0016] [0016]

[0017] 获得的;其中,(X丨,乂)表示第k时刻第i个粒子的坐标;(〜Yj)表示第j个参考点的坐标。 [0017] The obtained; wherein, (X Shu, qe) represent the coordinates of the i-th particle at time k; (~Yj) represents the coordinates of the j-th reference points.

[0018] 步骤三中所述选择欧几里得距离最小值min所对应的参考点坐标作为该粒子点的修正坐标是通过修真公式: [0018] Step three of said selected reference point coordinate minimum value min corresponding to the Euclidean distance as the correction point coordinates of the particles comprehension by the formula:

[0020] 获得的。 [0020] obtained.

[0021] 步骤四中所述根据粒子滤波算法理论,计算第i个粒子点所对应的权值是根据公式: [0021] Fourth step according to the theory of particle filtering algorithm to calculate the i-th particle points corresponding weights are according to the formula:

[0022] [0022]

[0023] 获得的;式中,为第k时刻第i个粒子所对应的权值;~表示正比于关系; I Xi)和;?(Χί 11^1)分别表示粒子点为;^时接收信号样本为Zk时的概率及粒子的转移概率;ω;^为第ki时刻第i个粒子所对应的权值,由于每一步都利用了重采样算法, 因此第kl时刻的权值为1/NS ;Sj为第u个参考点接收到的第j个AP的RSS值。 Receiving ^ time; wherein, for the first time k i particle corresponding weights;; [0023] I ~ represents proportional to the relation; I Xi), and;? (Χί 11 ^ 1) each represent the particle point transition probability signal samples and particles during Zk; ω; ^ i is the weight of the particles corresponding to the ki th time, since each step using a resampling algorithm, and therefore the right time value of kl 1 / NS; Sj is a reference point to the u-th received j-th AP of RSS values.

[0024] 步骤五中所述采用重采样算法对每个粒子进行重新采样得到新的粒子集的表达 Expression [0024] The fifth step of resampling algorithm using resampling for each particle resulting in a new set of particles

式为: Formula:

获得最终的位置估计坐标是通过公式 The final estimated position coordinates are obtained by the formula

获得的。 acquired.

[0025] 有益效果:本发明提出了一种有效确定用户的位置后验信息、抗干扰能力强且环境适应性好的基于粒子滤波的WLAN室内跟踪方法,该方法首先利用参考点的信号强度值建立离线阶段的信号强度概率分布,然后确定与粒子点欧氏距离最小的参考点并将其作为粒子点修正值,进而求出每个粒子点及其对应的权值,在采用重采样算法更新权值后,最终给出用户的位置估计值,该方法充分利用了空间中位置与RSS值的相关性,有效的解决了粒子滤波算法中处理粒子时的难题,减少对系统的负担,同时兼顾WLAN室内跟踪系统的有效性和可靠性,跟踪精度较高。 [0025] Advantageous Effects: The present invention provides an efficient inspection to determine the location information of the user, and the anti-interference ability indoor WLAN particle filter tracking method based on good environmental adaptability, the first method using a reference point in a signal strength value probability signal strength distribution established offline phase, and then determine the correction value the smallest particle point Euclidean distance reference point as point particles, each particle point and then find the corresponding weights, the update algorithm using resampling after the weights, the user is given the final position estimate, which takes full advantage of the spatial correlation of the position of RSS values, effectively solve the problem when the particles are treated in the particle filter, to reduce the burden on the system, taking into account validity and reliability of the WLAN indoor tracking system, tracking higher accuracy.

附图说明 BRIEF DESCRIPTION

[0026] 图1是本发明方法的流程示意图;图2是实施方式一中实施例中所述的实验场景示意图。 [0026] FIG. 1 is a schematic flow diagram of the method of the present invention; FIG. 2 is a scenario described in the experimental embodiment 1 a schematic embodiment. 图3是实施方式一的实例分析中所述的跟踪区域。 FIG 3 is an example embodiment of the analysis of the tracking area. 图4、图5和图6是实施方式一的实例分析中粒子滤波算法的收敛性分析的实验结果。 Figures 4, 5 and 6 are examples of embodiments of convergence analysis of the experimental results of the analysis of a particle filter. 图7、图8和图9是实例分析中的速度为lm/s,粒子数分别为4、16和64情况下的跟踪性能实验结果。 Figures 7, 8 and 9 are examples of the speed of analysis / s, respectively, the number of particles results lm track performance in the case of 4, 16 and 64. 图10、图11和图12是实施方式一的实例分析中速度为2m/s,粒子数分别为4、16和64情况下的跟踪性能实验结 10, FIG. 11 and FIG. 12 is an example embodiment of the assay velocity / s, respectively, the number of particles 2m tracking performance of the experimental results under the case 4, 16 and 64

:^ ο : ^ Ο

具体实施方式 Detailed ways

[0027] 具体实施方式一、结合图1说明本具体实施方式,基于粒子滤波算法的WLAN室内跟踪方法,它由以下步骤实现: [0027] a specific embodiment, described in conjunction with this specific embodiment of FIG. 1 embodiment, WLAN indoor particle filter tracking method algorithm, which is implemented by the steps of:

[0028] 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,并在所述室内环境中均勻设置Nkp个参考点; [0028] Step a, is arranged for the indoor environment of the N access point AP, to ensure coverage of the environment at any point by one or more than one access point AP emitted, and uniformly disposed in said chamber a Nkp environment reference point;

[0029] 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,并利用最大似然法计算每个参考点的信号强度先验概率分布; [0029] 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 using the maximum likelihood method to calculate the signal strength of each reference point prior probability distribution;

[0030] 步骤三、在第k时刻下,对描述粒子分布的重要性密度函数 [0030] Step three, in the first time k, the description of the importance of the particle distribution density function

进行 get on

采样,获得Ns个粒子点<,计算第i个粒子点坐标与每个参考点坐标之间的欧几里得距离, 并选择欧几里得距离最小值所对应的参考点坐标作为该粒子点的修正坐标;所述i = 1、 2……Ns; Sampling points to obtain particles Ns <calculates the Euclidean distance between the i-th point coordinates particles and the coordinates of each reference point, the reference point coordinate and selects the minimum Euclidean distance corresponding point of the particles as correction coordinates; the i = 1, 2 ...... Ns;

[0031] 步骤四、根据步骤二获得的每个参考点的信号强度先验概率分布和重要性密度函数以及第kl时刻粒子点的权值计算第k时刻粒子点所对应的权值; [0031] Step 4 according to the signal strength of each reference point obtained in step two and the prior probability distribution density function and importance weight of particles kl time point k-value calculating weights corresponding to the point in time the particle;

[0032] 步骤五、根据步骤四获得的第k时刻的粒子点的权值fi^,采用重采样算法对当前时刻的粒子点进行重新采样得到新的粒子点作为当前时刻的粒子点,并替代原有的粒子点,并获得最终的位置估计坐标;所述最终的位置估计坐标即为第k时刻下待跟踪目标的坐标,然后返回执行步骤三,获得下一时刻的待跟踪目标的坐标; [0032] Step V. The weight of the particle point in the k-th time step four values ​​obtained fi ^, using resampling algorithm particle current point in time resampled to obtain new particle point particles as points at the current time, and substitute the original point of the particles, and to obtain the coordinates of the final location estimate; estimated coordinate position of the final target to be tracked is the k-th time coordinates, and then returns to step three, the next time is obtained to be the coordinates of the target track;

[0033] N、Ns, Nep为正整数;k为大于或等于1的整数。 [0033] N, Ns, Nep is a positive integer; K is an integer greater than or equal to 1.

[0034] 步骤二中所述利用最大似然法计算每个参考点的信号强度先验概率分布是通过公式: [0034] Step II of the distributed computing each reference point using the maximum likelihood method is the prior probability signal strength by the formula:

[0035] [0035]

[0036] 获得的,式中,Sj为第u个参考点处接收第j个AP的信号强度值;Coimt(Sj)表示在第u个参考点处接收到信号强度为~的信号数量•' rk为第k时刻第u个参考点的位置坐标; [0036] obtained, wherein, Sj received j-th signal strength value AP to the u-th reference points; Coimt (Sj) represents the number of signals received signal strength at the u-th reference point of ~ • ' rk position coordinates of the reference points of the u-th time k;

[0037] j为整数。 [0037] j is an integer.

[0038] 步骤三中所述对重要性密度函数采样获得粒子点J^,计算第i个粒子点坐标与每个参考点坐标之间的欧几里得距离是通过公式: [0038] step 3 the importance of the particle density function obtained by sampling points J ^, calculates the Euclidean distance between the i-th point coordinates particles and the coordinates of each reference point by the formula:

[0039] [0039]

[0040] 获得的;其中,乂)表示第k时刻第i个粒子的坐标;NS为粒子数目;(Xj,yj)表示第j个参考点的坐标。 [0040] obtained; wherein qe) represent the coordinates of the i-th particle at time k; the NS is the number of particles; (Xj, yj) denotes the j-th coordinate of the reference point.

[0041] 步骤三中所述选择欧几里得距离最小值min所对应的参考点坐标作为该粒子点的修正坐标是通过修真公式: [0041] Step three of said selected reference point coordinate minimum value min corresponding to the Euclidean distance as the correction point coordinates of the particles comprehension by the formula:

[0043] 获得的。 [0043] obtained.

[0044] 步骤四中所述根据粒子滤波算法理论,计算第i个粒子点所对应的权值ωί是根据公式: [0044] Fourth step according to the theory of particle filtering algorithm, the i-th particle weight calculation point values ​​corresponding ωί is according to the formula:

[0046] 获得的;式中,< 为第k时刻第i个粒子所对应的权值;~表示正比于关系; Wherein <k for the first time corresponding to the i-th particle weights;; [0046] - indicates obtained is proportional to the relationship;

和I 分别表示粒子点为;^时接收信号样本为Zk时的概率及粒子的 And I represent the point of particle; ^ when the received signal samples and the probability of particles when Zk

转移概率•'ω“为第ki时刻第i个粒子所对应的权值,由于每一步都利用了重采样算法, 因此第kl时刻的权值为1/NS ;Sj为第u个参考点接收到的第j个AP的RSS值。 Transition probability • 'ω "weight value ki of the i-th time corresponding to the particles, since each step using a resampling algorithm, and therefore the right time value of kl 1 / NS; Sj is the u-th received reference points to the j-th AP value of RSS.

[0047] 步骤五中所述采用重采样算法对每个粒子进行重新采样得到新的粒子集的表达 Expression [0047] The fifth step of resampling algorithm using resampling for each particle resulting in a new set of particles

式为:丨;^丨〜;获得最终的位置估计坐标是通过公式•人获得的。 Formula: Shu; ^ Shu ~; obtain a final location estimate coordinates are obtained by a formula • people.

[0048] 下面举一个实例来进行分析: [0048] Here is an example for analysis:

[0049] 选择的实验场景如图2所示;实施方式一中实施例中所述的实验场景示意图。 [0049] Experimental selected scene shown in Figure 2; experiment scene described in the first embodiment in a schematic embodiment. 该实验场景尺寸面积为66. 43X24. 9m2,高度3m。 The experiment scene area size 66. 43X24. 9m2, the height of 3m. 且拥有19个实验室,1个会议室和1个乒乓球室。 And it has 19 laboratories, a conference room and a table tennis room. 墙的材料是砖块,铝合金窗户和金属门。 The wall material is brick, aluminum windows and metal doors. AP为Linksys WAP54G,固定在2m高度,支持IEEE 802. Ilg 标准,传输速率54Mbps。 AP is a Linksys WAP54G, 2m fixed height support standards IEEE 802. Ilg, the transmission rate 54Mbps. 接收端采用装有Intel PRO/ffireless 3945ABG 无线网卡的ASUS A8F笔记本电脑,接收机离地面1. 2m。 ASUS receiving end with Intel PRO / ffireless 3945ABG wireless card A8F laptop receiver 1. 2m from the ground.

[0050] AP的放置位置及参考点的选取如图3所示,其中标记31为参考点,曲线32为运动路径。 [0050] placement and select the reference point AP is shown in FIG. 3, wherein the mark 31 as a reference point, the curve 32 for the motion path. 该区域共包括65个参考点,且用RPi (i = 1,…,65)表示参考点i ;箭头指示方向为实际运动路径方向。 The reference region includes a total of 65 points, and with RPi (i = 1, ..., 65) represents the reference point I; arrows indicate the direction of the actual movement path direction. 该跟踪区域规则,覆盖性能较好,在该区域的任何位置均可接收到来自AP的信号。 The tracking area rule, good coverage properties, can receive the signal from the AP at any position in the region. 使用NetStumbler信号采集软件,在每个参考点处,进行3分钟的WLAN信号采集,每秒采样两次。 Use NetStumbler signal acquisition software, at each reference point, for 3 minutes WLAN signal acquisition, sampled twice per second.

[0051] 图4、图5和图6给出了粒子滤波算法在WLAN室内环境下的跟踪收敛性分析的实验结果,图4为粒子点与实际运动轨迹点之间的关系图;其中标记41为粒子点,标记42为实际运动轨迹点;图7、图8和图9给出了粒子滤波算法在移动终端的速度为lm/s时,不同的粒子数条件下的实验结果;其中标记71为参考点,标记72为实际运动轨迹点,标记73为粒子滤波跟踪曲线,标记74为粒子点;标记81为参考点,标记82为实际运动轨迹点,标记83为粒子点,标记84为粒子滤波跟踪曲线;标记91为参考点,标记92为粒子点,标记93为实际运动曲线,标记94为粒子滤波跟踪曲线;具体的均方根误差及仿真时间比较如表1所示:[0052] 表1速度为lm/s时的均方根误差及仿真时间比较 [0051] FIG 4, FIG 5 and FIG 6 shows the experimental result of the tracking Convergence analysis in indoor environments WLAN particle filter, FIG 4 is a relationship between the particle trajectory point and the actual point; wherein numerals 41 particle point, numeral 42 is the actual trajectory point; FIG. 7, FIG. 8 and FIG. 9 shows the particle filter when the speed of the mobile terminal of lm / s, the experimental results under different conditions of the number of particles; wherein numerals 71 as a reference point, numeral 72 is the actual trajectory point, marked 73 as a particle filter curve, labeled 74 as a particle point; mark 81 as a reference point, numeral 82 is the actual trajectory point, marked 83 as a particle point, labeled 84 particles filtering the tracking curve; mark 91 as a reference point, the point labeled 92 particles, the actual motion curve labeled 93, numeral 94 is a particle filter tracking curve; RMSE specific simulation time comparison and shown in table 1: [0052] table 1 is the root mean square error and speed of simulation time at lm / s Comparative

[0053] [0053]

[0054] 图10、图11和图12给出了粒子滤波算法在移动终端的速度为2m/s时,不同的粒子数条件下的实验结果,其中,标记101为粒子点,标记102为参考点,标记103为粒子滤波跟踪曲线,标记104为实际运动曲线;标记111为实际运动曲线,标记112为参考点,标记113为粒子点,标记114为粒子滤波跟踪曲线;标记121为参考点,标记122为粒子点,标记123为实际运动曲线,标记124为粒子滤波跟踪曲线。 [0054] FIG. 10, FIG. 11 and FIG. 12 shows the particle filter when the speed of the mobile terminal is 2m / s, the experimental results under different conditions of the number of particles, wherein the particles are labeled as point 101, the reference mark 102 point, numeral 103 is a particle filter curve, labeled 104 is the actual motion profile; numeral 111 is the actual motion profile, labeled 112 as a reference point, numeral 113 particle point, numeral 114 is a particle filter tracking curve; labeled 121 as a reference point, 122 is a point labeled particles, numeral 123 is the actual motion profile, numeral 124 is a particle filter tracking curve. 具体的均方根误差及仿真时间比较如表2所示。 DETAILED RMS error and simulation time comparison shown in Table 2.

[0055] 表2速度为2m/s时的均方根误差及仿真时间比较 [0055] Table 2 is the root mean square error and speed of simulation time at 2m / s Comparative

[0056] [0056]

[0057] 显然,在粒子数选取适当的情况下,基于粒子滤波算法的WLAN室内跟踪方法在跟踪误差及对系统造成的负担方面都具有极大的优势。 [0057] Obviously, the number of particles in selecting appropriate, WLAN indoor particle filter tracking method based on the tracking error, and the burden on the system has a great advantage.

Claims (6)

  1. 基于粒子滤波算法的WLAN室内跟踪方法,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,并在所述室内环境中均匀设置NRP个参考点;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NRP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,并利用最大似然法计算每个参考点的信号强度先验概率分布;步骤三、在第k时刻下,对描述粒子分布的重要性密度函数进行采样,获得Ns个粒子点,计算第i个粒子点坐标与每个参考点坐标之间的欧几里得距离,并选择欧几里得距离最小值所对应的参考点坐标作为该粒子点的修正坐标;所述i=1、2……Ns;步骤四、根据步骤二获得 WLAN indoor particle filter tracking method 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 any point of the environment by one or more than one access point given coverage AP, and uniformly set the reference point in the NRP indoor environment; step 2. selection of a reference point to establish the origin of the coordinate two-dimensional rectangular coordinates, to obtain reference points NRP in the two-dimensional Cartesian coordinate system coordinate position, in the offline phase using a signal receiver at each reference point acquisition signal strength (RSS) from each of the access point AP, and calculates the distribution of each reference point using the maximum likelihood method of signal strength of a priori probabilities ; step three, in the first time k, the description of the particle distribution density function of the importance of sampling, to obtain the Ns particles points to calculate the Euclidean distance between the i-th point coordinates particles and the coordinates of each reference point, and select the reference point coordinate corresponding to the minimum Euclidean distance as the correction point coordinates of the particles; the i = 1,2 ...... Ns; step 4 obtained according to step two 每个参考点的信号强度先验概率分布和重要性密度函数以及第k-1时刻粒子点的权值计算第k时刻粒子点所对应的权值步骤五、根据步骤四获得的第k时刻的粒子点的权值,采用重采样算法对当前时刻的粒子点进行重新采样得到新的粒子点作为当前时刻的粒子点,并替代原有的粒子点,并获得最终的位置估计坐标;所述最终的位置估计坐标即为第k时刻下待跟踪目标的坐标;然后返回执行步骤三,获得下一时刻的待跟踪目标的坐标;N、Ns、NRP为正整数;k为大于或等于1的整数。 Each signal strength reference point prior probability distribution density function and importance weights of the k-1 and time point to calculate the weight of particles of particle point in time k corresponding value in step five, the first time k obtained according to step four particle weights point, resampling algorithm using the particle current time point resampled to obtain new particles as point particles at the current time point, and replace the original point of the particles, and to obtain the coordinates of the final location estimate; the final is the coordinates of the location estimate at time k to be under the tracking target coordinates; and then returns to step three, obtain the coordinates of the target to be tracked next time; N, Ns, NRP is a positive integer; k is an integer greater than or equal to 1 . FSA00000162485300011.tif,FSA00000162485300012.tif,FSA00000162485300013.tif,FSA00000162485300014.tif FSA00000162485300011.tif, FSA00000162485300012.tif, FSA00000162485300013.tif, FSA00000162485300014.tif
  2. 2.根据权利要求1所述的基于粒子滤波算法的WLAN室内跟踪方法,其特征在于步骤二中所述利用最大似然法计算每个参考点的信号强度先验概率分布是通过公式:/ .χ count值样本釆样总数获得的,式中,Sj为第u个参考点处接收第j个AP的信号强度值;COimt(Sj)表示在第u个参考点处接收到信号强度为~的信号数量·4为第k时刻第u个参考点的位置坐标; j = 1、2……NKP。 2. The method of claim WLAN indoor particle filter tracking algorithm, wherein said two step profile 1 is calculated by an equation using the maximum of the likelihood of each reference point of the signal strength of a priori probability law: /. χ count value of the sample preclude comp total number obtained by formula, Sj received signal strength value of the j-th AP for the u-th reference points; COimt (Sj) indicates that the received signal strength at the u reference point of ~ · number of signal u 4 position coordinates of a first reference point at time k; j = 1,2 ...... NKP.
  3. 3.根据权利要求1所述的基于粒子滤波算法的WLAN室内跟踪方法,其特征在于步骤三中所述对重要性密度函数采样获得粒子点Ιί,计算第i个粒子点坐标与每个参考点坐标之间的欧几里得距离是通过公式:获得的;其中,(4,乂)表示第k时刻第i个粒子的坐标;(Xj,Yj)表示第j个参考点的坐标。 The indoor WLAN based particle filter tracking method according to claim 1, wherein said sampling step three of the importance of particle density function obtained point Ιί, calculating the i-th point coordinates particles each reference point Euclidean distance between the coordinates is obtained by the formula: obtained; wherein (4, qe) represent the coordinates of the i-th particle at time k; (Xj, Yj) represents the coordinates of the j-th reference points.
  4. 4.根据权利要求1所述的基于粒子滤波算法的WLAN室内跟踪方法,其特征在于步骤三中所述选择欧几里得距离最小值min所对应的参考点坐标作为该粒子点的修正坐标是通过修真公式: The indoor WLAN based particle filter tracking method according to claim 1, wherein in the step of selecting the reference point coordinate three Euclidean distance corresponding to the minimum min as the correction point coordinates of the particles is comprehension by the formula: 获得的。 acquired.
  5. 5.根据权利要求1所述的基于粒子滤波算法的WLAN室内跟踪方法,其特征在于步骤四中所述根据粒子滤波算法理论,计算第i个粒子点所对应的权值是根据公式: The indoor WLAN based particle filter tracking method according to claim 1, wherein the Step 4 according to the theory of particle filtering algorithm to calculate the i-th particle points corresponding weights are according to the formula: 获得的;式中,<为第k时刻第i个粒子所对应的权值;~表示正比于关系; Wherein <k for the first time corresponding to the i-th particle weights;; ~ obtained is proportional to the relationship expressed; 分别表示粒子点为;^时接收信号样本为Zk时的概率及粒子的转移概率;为第k-1时刻第i个粒子所对应的权值,由于每一步都利用了重采样算法, 因此第k-Ι时刻的权值为1/NS ;Sj为第u个参考点接收到的第j个AP的RSS值。 Denote particles points; ^ received signal samples transition probability and particles during Zk; of the weight of the i th particle values ​​corresponding to k-1 in time, since each step utilizing a resampling algorithm, so the first right k-Ι time point is 1 / NS; Sj u for the first reference point to the j-th received AP of RSS values.
  6. 6.根据权利要求1所述的基于粒子滤波算法的WLAN室内跟踪方法,其特征在于步骤五中所述采用重采样算法对每个粒子进行重新采样得到新的粒子集的表达式为 6. The tracking method as claimed in claim indoor WLAN particle filter algorithm, characterized in that said fifth step of resampling algorithm using resampling for each particle obtained expression for a new set of particles of claim 1 ;获得最终的位置估计坐标是通过公式: ; Coordinate position to obtain the final estimate by the formula: 获得的。 acquired.
CN 201010207003 2010-06-23 2010-06-23 WLAN indoor tracking method based on particle filtering algorithm CN101867943A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201010207003 CN101867943A (en) 2010-06-23 2010-06-23 WLAN indoor tracking method based on particle filtering algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201010207003 CN101867943A (en) 2010-06-23 2010-06-23 WLAN indoor tracking method based on particle filtering algorithm

Publications (1)

Publication Number Publication Date
CN101867943A true CN101867943A (en) 2010-10-20

Family

ID=42959450

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201010207003 CN101867943A (en) 2010-06-23 2010-06-23 WLAN indoor tracking method based on particle filtering algorithm

Country Status (1)

Country Link
CN (1) CN101867943A (en)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103618997A (en) * 2013-11-22 2014-03-05 北京邮电大学 Indoor positioning method and device based on signal intensity probability
CN104197932A (en) * 2014-09-26 2014-12-10 哈尔滨工业大学 Praxeology-based position tracking method
CN104375117A (en) * 2013-08-12 2015-02-25 无锡知谷网络科技有限公司 Target locating method and system
CN104793182A (en) * 2015-04-21 2015-07-22 东南大学 Indoor positioning method based on particle filtering under condition of non-Gaussian noises
CN105891865A (en) * 2016-03-28 2016-08-24 南京工程学院 Markov-chain-Monte-Carlo-based particle filter positioning method
CN106056632A (en) * 2016-06-07 2016-10-26 西北工业大学 Radar target tracking method based on dynamic resampling particle filtering
CN107613560A (en) * 2010-11-19 2018-01-19 高通股份有限公司 Radio station it is self-positioning

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2006002458A1 (en) * 2004-07-07 2006-01-12 Nariste Networks Pty Ltd Location-enabled security services in wireless network
US20090238406A1 (en) * 2006-09-29 2009-09-24 Thomson Licensing Dynamic state estimation
CN101576616A (en) * 2008-05-06 2009-11-11 广州香港科大研究开发有限公司 Indoor positioning system based on RFID technology

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2006002458A1 (en) * 2004-07-07 2006-01-12 Nariste Networks Pty Ltd Location-enabled security services in wireless network
US20090238406A1 (en) * 2006-09-29 2009-09-24 Thomson Licensing Dynamic state estimation
CN101576616A (en) * 2008-05-06 2009-11-11 广州香港科大研究开发有限公司 Indoor positioning system based on RFID technology

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
《传感技术学报》 20090331 罗海勇,李锦涛,赵方,林以明 基于均值漂移和联合粒子滤波的移动节点定位算法 全文 1-6 第22卷, 第3期 *
《计算机工程与应用》 20100101 宋越明,于宏毅 采用概率密度分布和粒子滤波的室内跟踪算法 正文第1-4页,图2 1-6 第46卷, 第1期 *

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107613560A (en) * 2010-11-19 2018-01-19 高通股份有限公司 Radio station it is self-positioning
CN104375117A (en) * 2013-08-12 2015-02-25 无锡知谷网络科技有限公司 Target locating method and system
CN104375117B (en) * 2013-08-12 2018-05-04 无锡知谷网络科技有限公司 Object localization method and system
CN103618997A (en) * 2013-11-22 2014-03-05 北京邮电大学 Indoor positioning method and device based on signal intensity probability
CN103618997B (en) * 2013-11-22 2017-02-08 北京邮电大学 Indoor positioning method and device based on signal intensity probability
CN104197932A (en) * 2014-09-26 2014-12-10 哈尔滨工业大学 Praxeology-based position tracking method
CN104197932B (en) * 2014-09-26 2017-02-15 哈尔滨工业大学 Praxeology-based position tracking method
CN104793182B (en) * 2015-04-21 2017-05-03 东南大学 Indoor positioning method based on particle filtering under condition of non-Gaussian noises
CN104793182A (en) * 2015-04-21 2015-07-22 东南大学 Indoor positioning method based on particle filtering under condition of non-Gaussian noises
CN105891865B (en) * 2016-03-28 2018-09-25 南京工程学院 Based on Markov chain Monte-Carlo particle filtering localization method
CN105891865A (en) * 2016-03-28 2016-08-24 南京工程学院 Markov-chain-Monte-Carlo-based particle filter positioning method
CN106056632A (en) * 2016-06-07 2016-10-26 西北工业大学 Radar target tracking method based on dynamic resampling particle filtering

Similar Documents

Publication Publication Date Title
KR101354649B1 (en) Binning venues into categories based on propagation characteristics
He et al. CC-KF: Enhanced TOA performance in multipath and NLOS indoor extreme environment
KR20150035745A (en) System, method and computer program for dynamic generation of a radio map
US20080161011A1 (en) Method enabling indoor local positioning and movement tracking in wifi capable mobile terminals
JP5657798B2 (en) Method and apparatus related to the use of position estimation of a mobile terminal within a structure
Chen et al. Bayesian fusion for indoor positioning using bluetooth fingerprints
US9677890B2 (en) Positioning and mapping based on virtual landmarks
CN101778399B (en) Method for optimizing WLAN (Wireless Local Area Network) indoor ANN (Artificial Neural Network) positioning based on FCM (fuzzy C-mean) and least-squares curve surface fitting methods
JP2016513311A (en) Location-based process monitoring
CN102802260B (en) WLAN indoor positioning method based on matrix correlation
CN101547048A (en) Indoor positioning method based on wireless sensor network
CN102209386A (en) Indoor wireless positioning method and device
CN103547938A (en) Measurements and information gathering in a wireless network environment
JP2015535935A (en) Techniques for faster initial positioning times
CN101657014B (en) WLAN indoor positioning method based on nearest neighbor probability method
Cheng et al. Seamless outdoor/indoor navigation with WIFI/GPS aided low cost Inertial Navigation System
US8954267B2 (en) Mobile device positioning
CN101808400A (en) System and method for efficiently populating access point database
CN103200520A (en) Mobile terminal fast and precise positioning method using Wi-Fi
CN105393139B (en) GPS and WLAN hybrid position determines
CN102638889A (en) Indoor wireless terminal positioning method based on Bayes compression sensing
JP6279480B2 (en) Method and apparatus for use in selecting transmission device used in positioning function
US20140274116A1 (en) Mobile device location estimation
US9179265B2 (en) Reducing location search space
CN103384358A (en) Indoor positioning method based on virtual reality and WIFI space field strength

Legal Events

Date Code Title Description
C06 Publication
C10 Request of examination as to substance
C02 Deemed withdrawal of patent application after publication (patent law 2001)