CN107979817A - A kind of mobile terminal two dimension fingerprint positioning method - Google Patents
A kind of mobile terminal two dimension fingerprint positioning method Download PDFInfo
- Publication number
- CN107979817A CN107979817A CN201711195755.1A CN201711195755A CN107979817A CN 107979817 A CN107979817 A CN 107979817A CN 201711195755 A CN201711195755 A CN 201711195755A CN 107979817 A CN107979817 A CN 107979817A
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- sampling
- interpolation
- fingerprint information
- Prior art date
- Legal status (The legal status 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 status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 58
- 238000005070 sampling Methods 0.000 claims abstract description 59
- 238000012549 training Methods 0.000 claims description 15
- 238000012360 testing method Methods 0.000 claims description 14
- 239000013598 vector Substances 0.000 claims description 8
- 238000012545 processing Methods 0.000 claims description 7
- 238000001914 filtration Methods 0.000 claims description 4
- 238000011156 evaluation Methods 0.000 claims description 3
- 230000008859 change Effects 0.000 claims description 2
- 230000000295 complement effect Effects 0.000 claims description 2
- 238000010276 construction Methods 0.000 claims description 2
- 238000007781 pre-processing Methods 0.000 claims description 2
- 238000013480 data collection Methods 0.000 abstract 1
- 230000007123 defense Effects 0.000 abstract 1
- 230000004927 fusion Effects 0.000 abstract 1
- 238000010586 diagram Methods 0.000 description 9
- 230000000694 effects Effects 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 6
- 238000004088 simulation Methods 0.000 description 5
- 238000005259 measurement Methods 0.000 description 4
- 238000000926 separation method Methods 0.000 description 4
- 239000003245 coal Substances 0.000 description 3
- 238000012417 linear regression Methods 0.000 description 3
- 230000004807 localization Effects 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 230000010365 information processing Effects 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 230000035772 mutation Effects 0.000 description 2
- 208000001408 Carbon monoxide poisoning Diseases 0.000 description 1
- 238000007476 Maximum Likelihood Methods 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000009423 ventilation Methods 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W64/00—Locating users or terminals or network equipment for network management purposes, e.g. mobility management
- H04W64/006—Locating users or terminals or network equipment for network management purposes, e.g. mobility management with additional information processing, e.g. for direction or speed determination
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Position Fixing By Use Of Radio Waves (AREA)
Abstract
The present invention relates to a kind of mobile terminal two dimension fingerprint positioning method, is sampled in the location fingerprint information for the sampled point that off-line phase support mobile terminal chooses mine laneway；Sampling point position finger print information after pretreatment is fitted using SVR to the variation function curve in Kriging interpolation；Utilize the location fingerprint information of the non-sample area of Kriging Interpolate estimation method completions；Two-dimentional fingerprint location information database will be constructed after the location fingerprint information fusion of sampled point and interpolation point；The location fingerprint that on-line stage carries out mine worker using constructed database matches positioning.The present invention realizes the purpose for reducing data collection task amount on the premise of mine worker's location fingerprint positioning accuracy is ensured.The application scenarios of the present invention include indoor positioning, civil air defense ground engineering, tunnel and command post etc..
Description
Technical Field
The invention relates to a two-dimensional fingerprint positioning method for a mobile terminal, and belongs to the field of indoor positioning.
Background
With the great popularization of GPS and Beidou navigation and positioning systems, the existing outdoor positioning technology is quite mature. However, in indoor positioning, particularly in underground environments such as mine tunnels, a new positioning technology is required because it is difficult to receive satellite signals satisfying positioning requirements. Underground roadways are dark, long and narrow, and have unsmooth ventilation and moist air, so that various coal mine accidents such as explosion, collapse, carbon monoxide poisoning and the like seriously threaten the life safety of underground workers. When an accident occurs, the positioning precision of underground mine workers directly influences the efficiency and success rate of the development of rescue work. Therefore, the method has great practical significance and application value for accurate positioning of underground mine workers.
The geographic environment and the electromagnetic environment with complex roadways make positioning of mine workers difficult. Therefore, in combination with the above complexities of underground mine worker location, a number of scholars have recently proposed a series of classical location schemes. A mine worker label positioning system is developed in an initiative position estimation using RFID tag, A least-square advance approach, the main principle of the system is that a card reader deployed in a roadway reads a label carried by a worker to estimate the position of the worker, the structural principle of the system is simple and feasible, the related technology is mature at present, but the position of the worker is estimated to be the area positioning with lower precision by means of the position of the card reader, and the requirement of rescue in accidents on the positioning precision is difficult to meet. Energy-efficiency index localization of smart hand-held devices using Bluetooth discloses: on the basis of a mine Wi-Fi network architecture, a positioning method based on received signal strength RSSI is provided, the RSSI value received by a worker terminal is compared with a signal propagation loss model of an existing Wi-Fi wireless Access Point AP (Access Point, AP) to estimate the distance between the worker terminal and the AP, and then the worker is positioned by methods such as maximum likelihood estimation and the like on the basis of the measured distances from the terminal to 3 or more nearby APs. The "improved TDOA applied person localization system in the same mine" discloses: mine worker location is achieved by measuring the Time Difference of arrival (TDOA) of radio signals transmitted by worker terminals at a plurality of APs. The two methods are widely used in outdoor and indoor positioning, theoretically have higher positioning precision, and have mature technology. However, in a lengthy tunnel environment, multipath effects and standing waves of signal propagation have a large influence on ranging. And the AP is deployed on an approximate straight line in the tunnel, and errors are brought when triangulation is implemented, so that the engineering realization effect of the method is not ideal. A new method is developed in a coal mine underground WLAN positioning method based on an RSS finger model, and a fingerprint positioning technology is provided based on a Wi-Fi network architecture and mainly divided into an off-line stage and an on-line stage. In an off-line stage, performing interval traversal sampling on a target area, and storing RSSI values of a plurality of nearby APs received by each positioning point as position fingerprint information of the positioning point in a special database; and in the online stage, a positioning function is mainly completed, a position fingerprint formed by RSSI of each AP received by a worker terminal at an unknown position is matched with position fingerprint information of a known position in a database by using a matching algorithm, and the position with the highest matching degree with the position fingerprint received by the worker terminal in the database is the current position of the worker. The method does not need distance measurement, effectively solves the problem of error influence on positioning caused by a complex electromagnetic environment of a roadway, but needs a large amount of data sampling for improving the positioning precision, and thus the workload is increased. Therefore, in order to reduce the workload, the method ignores the roadway width during roadway modeling and simplifies the roadway width into a one-dimensional model. However, ignoring the width of the actual lane reduces the accuracy of the positioning to some extent.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a two-dimensional fingerprint positioning method for a mobile terminal, which has the following specific technical scheme:
a two-dimensional fingerprint positioning method for a mobile terminal comprises the following steps:
step one, AP deployment and regional gridding division
When wireless AP deployment is carried out in a positioning area, two-dimensional deployment is adopted, namely, two longitudinal APs are respectively deployed at two sides of a roadway, and the distance between the transversely deployed APs is determined according to the effective coverage radius of the APs in the actual roadway;
step two, sampling stage
The sampling stage comprises selection of sampling points, sampling of sample fingerprint information and processing of fingerprint information data; selecting sampling points by adopting an interval traversal sampling method, and selecting one sampling point every other grid; sampling and storing the sample fingerprint information for multiple times by depending on the position fingerprint information of the selected sampling point by the mobile terminal in an off-line stage; when data acquisition is carried out, the mobile terminal only records signal intensity values sent by 4 adjacent APs in the area as position fingerprint information of the sampling point; the fingerprint information data processing is to perform Gaussian filtering processing on the stored sample fingerprint information to filter out interference items, and the specific method is as follows:
the density function of the acquisition result x satisfies the gaussian distribution as:
wherein mu is the mean value of the data sets collected for multiple times, and sigma is the standard deviation; selecting corresponding RSSI data with the high probability event of 0.6-1 (f) (x) and the geometric mean value of the RSSI values in the range as the final RSSI value of the adopted AP;
step three, interpolation stage and database construction
The interpolation stage comprises interpolation point selection, variogram fitting and Kriging interpolation estimation, wherein the interpolation point is the central point of a grid point which is left by removing sampling points in the sampling stage in the area grid division;
calculating a variation function value between sampling points according to the received signal intensity of one AP, and performing SVR-based variation function curve fitting; estimating the RSSI value of the AP at each interpolation point by using a Kriging interpolation method based on the fitted variation function curve; similarly, the signal strength of other 3 APs received by the sampling point is respectively interpolated and estimated by the SVR-Kriging interpolation estimation method to complement the RSSI value of the 3 APs in each non-sampling area in the two-dimensional grid division; generating the position fingerprint information of the interpolation point by combining the RSSI estimated by the 4 APs at the interpolation point; the position fingerprint information of the sampling point and the interpolation point is fused to construct an information database for two-dimensional fingerprint positioning of the mobile terminal;
step four, on-line positioning
In the on-line positioning stage, a mobile terminal carried by a mine worker firstly acquires the signal intensity of each AP of an unknown position, and then selects the RSSI of 4 required APs to form the position fingerprint information of the position through data preprocessing; and when fingerprint positioning is carried out, matching is carried out on the position fingerprint information of the position where the mine worker is located and the corresponding constructed database, and the position of the most similar fingerprint information in the database is output as the position of the mine worker.
Further, the SVR-based variogram curve fitting step is as follows:
step S1, calculating the distance h of all sample data pairs by using the space variation functioniAnd the corresponding function of variation value gamma (h)i) Composition data set [ h ]i,γ(hi)]；
The spatial variation function is as follows:
wherein h isiRepresenting the vector distance, N, of a pair of sample points in a spatial regionhiRepresenting the distance h between all pairs of sample pointsiThe number of (2);
step S2, randomly extracting 80% of data from the data set to generate a training set T { [ h ]1,γ(h1)],[h2,γ(h2)],…,[hl,γ(hl)]And taking the rest data as a test set;
s3, training the training set T by adopting SVR, and fitting a variation function curve gamma (h);
and step S4, performing performance evaluation on the variation function curve gamma (h) by using the test set generated in the step S2, outputting the variation function curve gamma (h) if the expected requirement is met, and otherwise, modifying the SVR parameters and returning to the step S3 for re-fitting.
Furthermore, the Kriging interpolation estimation method is to change the region variation of interpolation points satisfying the second-order stationary or intrinsic in the region into R (x)0) And m regional variation quantities satisfying the second-order stationary sample in the neighborhood range are R (x)i) (i ═ 1,2, …, m) by comparison with the known R (x)i) The weighted sum of the values can estimate the R (x) to be estimated for the interpolation point0) Namely:
the invention has the beneficial effects that:
according to the invention, interpolation estimation is carried out on the non-sampled area through the provided SVR-Kriging interpolation value on the basis of sparse sampling data, so that the aim of improving the fingerprint positioning precision of mine workers is fulfilled on the premise of reducing the position fingerprint information data acquisition workload. The invention solves the problem of low positioning precision in position fingerprint positioning, and also solves the problem of increased data acquisition workload caused by the establishment of a two-dimensional fingerprint information base, so that the invention has good application prospect in the indoor positioning field of mine worker positioning and the like.
Drawings
FIG. 1 is a schematic view of a roadway unit according to the present invention;
FIG. 2 is a schematic diagram of the AP deployment model according to the present invention;
FIG. 3 is a flowchart of a two-dimensional fingerprint positioning method for a mobile terminal according to the present invention;
FIG. 4 is a schematic diagram of a curve fitted to the AP1 training set according to the present invention;
FIG. 5 is a diagram of the test results of the AP1 test set of the present invention;
FIG. 6 is a QQ plot of the training set fitting results of the present invention;
FIG. 7 is a QQ chart of the test set test results of the present invention;
FIG. 8 is a comparison of the interpolation estimates of the present invention;
fig. 9 is a received signal strength distribution diagram of the AP1 according to the present invention;
fig. 10 is a received signal strength distribution diagram of the AP2 according to the present invention;
fig. 11 is a received signal strength distribution diagram of the AP3 according to the present invention;
fig. 12 is a received signal strength distribution diagram of the AP4 according to the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
In the research of the two-dimensional fingerprint positioning method of the mobile terminal, an underground mine roadway model needs to be established and analyzed. Although the length of the underground mine roadway is hundreds to thousands of meters, and the width is 4-5 m, because of less fork, each section of roadway is in a linear shape even if the direction is changed. Therefore, the tunnel is divided into a plurality of linear tunnel units when the model is built, as shown in fig. 1.
When the wireless AP is deployed in the roadway unit shown in fig. 1, in order to achieve uniform coverage of the wireless AP signal, arrangement deployment is generally adopted according to the effective coverage distance of the AP. Due to the long and narrow lanes, more wireless APs are required. When position fingerprint information is sampled, if the number of APs participating in the sampling at a certain sampling point is not limited, the acquisition dimensionality is high, and the complexity is high. In fact, the greater the number of APs not participating in the location determination, the greater the accuracy of the location fingerprint location. Some weak signal APs at longer distances will adversely affect the positioning accuracy of the algorithm. Therefore, the invention adopts two-dimensional deployment, the AP deployment model is shown in figure 2, namely, two longitudinal APs are respectively deployed at two sides of the roadway, and the distance between the transversely deployed APs is determined according to the effective coverage radius of the APs in the actual roadway environment. In a certain tunnel unit, the acquisition of the position fingerprint information only records the RSSI of 4 APs in the unit.
Without loss of generality, the length of the roadway unit is set to 50m according to the effective coverage radius of the AP. When sampling point position fingerprint data is collected, the mobile terminal only records RSSI values of 4 adjacent APs in the unit where the mobile terminal is located.
As shown in fig. 2, when the location fingerprint information is collected in the cell where the worker 1 is located, only RSSI values of AP1, AP2, AP3 and AP4, which are typically RSSI values of 4 APs having the highest signal strength received by the mobile terminal, are recorded. If the AP serial number corresponding to the RSSI value has individual deviation, the AP is numbered according to a certain sequence, and the AP can be distinguished according to the MAC addresses of other APs which do not have deviation. Because the position fingerprint data acquisition methods in each roadway unit are similar, the roadway unit where the worker 1 is located is selected for research, so that the research is convenient. The fingerprint information for each location may be stored in a database in the format of table 1.
TABLE 1 storage format for fingerprint information of sampling points
The invention mainly aims to complete data acquisition and interpolation estimation of a two-dimensional mine roadway on the premise of reducing workload and ensure the accuracy of position fingerprint positioning. The flow of the two-dimensional fingerprint positioning method of the mobile terminal is shown in fig. 3.
In the off-line stage, in order to facilitate the selection of the sampling points and the interpolation points, uniform square grid division is firstly carried out on the positioning area, the specific grid division size can be adjusted according to the positioning precision required in actual use, and the higher the required precision is, the denser the grid division is. The second is the core technology of the off-line stage, which mainly comprises two stages of sampling and interpolation. Wherein, the sampling stage mainly completes 3 functions of sampling point selection, sample fingerprint information acquisition and fingerprint information processing. And the interpolation stage realizes interpolation estimation of an unstamped grid area, and comprises 3 parts of interpolation point selection, variogram fitting and Kriging interpolation estimation. Finally, the sampled fingerprint information and interpolated estimated fingerprint information are stored in a database for reference in the online positioning phase.
In an online stage, a mobile terminal carried by a mine worker firstly acquires the signal intensity of each AP of an unknown position, and then selects the RSSI of 4 APs meeting the requirements to generate the position fingerprint information of the unknown position. And finally, matching with a database constructed in an off-line stage by using a nearest neighbor matching algorithm and the like, and outputting the position of the fingerprint information which is most similar to the position fingerprint information acquired by the worker mobile terminal in the database as the position of the mine worker.
In the tunnel unit model established by the invention, the width of the tunnel unit is set to be 4m, and the length of the tunnel unit is set to be 50 m. On the premise of not affecting the performance, assuming that the average positioning accuracy requirement reaches 2m, the area is uniformly divided into 25 × 2-dimensional grid areas during grid division, and the APs are deployed at four corners of the area.
1. Sampling phase
And the sampling stage mainly finishes sampling point selection, fingerprint information acquisition and fingerprint information processing. Firstly, in order to prevent the data acquisition amount from being overlarge due to the fact that the number of sampling points is too large, an interval traversal sampling method is adopted, and 25 sampling points are selected at intervals of one grid. Secondly, to avoid measurement errors, multiple samples are taken and stored in the format shown in table 1 during sample fingerprint information acquisition. And finally, performing Gaussian filtering processing on the stored sample fingerprint information.
The data acquired by each sampling point for multiple times are known to be mutually independent and obey Gaussian distribution, and noise errors can be filtered through Gaussian filtering to obtain accurate and stable sampling point RSSI values. Taking the RSSI value of AP1 as an example, the density function of the result x of multiple acquisitions is as follows:
where μ is the mean of the data set and σ is the standard deviation. Selecting corresponding RSSI values within the range of f (x) being more than or equal to 0.6 and less than or equal to 1 (empirical value), and solving the geometric mean value of the RSSI values as the RSSI values of the AP1 acquired by the sampling points.
2. Interpolation stage
The interpolation stage mainly comprises 3 parts of interpolation point selection, variogram fitting and Kriging interpolation estimation. The interpolation points are the remaining 25 grid points of the 50 area grids except the selected sampling points.
2.1Kriging interpolation estimation
The Kriging interpolation method is derived from the geostatistical estimation of mine reserves, and is derived from estimation in various fields nowadays. The main principle is to estimate the characteristic attribute of an interpolation point through the characteristic attribute of a sampling point in the neighborhood range of the interpolation point, and the method is proved to be a linear unbiased estimation method.
The variation of a certain unknown interpolation point region is R (x)0) The area variation of m known sampling points satisfying the second order stationary in the neighborhood range is R (x)i) (i ═ 1,2, …, m). By applying a pair of known m R (x)i) Weighted summation can estimate R (x)0) Namely:
in the present invention, R (x)0) For the interpolation point RSSI value to be estimated, R (x)i) Is the RSSI value, lambda, of a known sample pointiIs m R (x) participating in interpolation estimationi) The weight of (a) is determined,it can be seen that the key to the Kriging interpolation method is λiAnd (4) calculating. By regional variation R (x)i) Satisfying the second order stationary condition:
setting R (x)0) Is estimated as R*(x0) At this time, the interpolation point x0The estimated variance is minimal:
Varmin＝Var[R(x0)-R*(x0)]＝E{[R(x0)-R*(x0)]2}
the conditional extremum is obtained by introducing Lagrange multiplier mu, which can be expressed as:
a Kriging equation set is obtained through formula derivation:
wherein,representing a sample point xiAnd xjThe value of the variation function between. The weight λ can be obtained by solving the above equation setiThus, the key to the problem translates into the solution of the function of variation and the fitting of its curve of variation.
2.2 SVR-based variogram fitting
As a core part of the Kriging interpolation method, the characteristic attribute value of an interpolation point can be deduced by utilizing the rule that the sample data attribute value changes along with the separation distance through a mutation function. The variation function can be expressed as follows:
wherein h isiThe vector distance, also called the separation distance, N, representing a pair of sample points in a spatial regionhiRepresents the separation distance h in all the sampling point pairsiThe number of point pairs of (a). The function value of variation corresponding to the separation distance of all the point pairs in the data set can be calculated according to the space variation function formula, and a function curve gamma (h) can be fitted through the function values of variation, so that the function value of variation between the interpolation point to be estimated and the sampling point which is in the neighborhood and is estimated can be calculated according to the curve, and the weight lambda can be obtainedi。
When the variation function curve is fitted, the existing model is usually selected for least square fitting. The existing variation function models mainly include a spherical model, a Gaussian model, a linear model and the like. However, the fitting method of model substitution usually selects a relevant model according to manual experience, and the model selection lacks a reliable theoretical basis. And the problem that the selected ready-made model can not completely accord with the actual sample data is solved, a plurality of variation function models are often required to be selected for comparison, and the selected model which is relatively consistent can not achieve the optimal effect. In order to solve the problems, the method directly fits the sample variation function value from the perspective of Support Vector Regression (SVR), and overcomes the limitation of substituting the variation function model into the fitting method.
In practice, SVR is mostly linear regression fit, and in the case of variogram fit, the function curve is mostly non-linear fit. Therefore, a non-linear mapping function needs to be introduced to map the sample data into a high-dimensional feature space for linear regression fitting.
Without loss of generality, a training set T { (x) containing l samples is given1,y1),(x2,y2),…,(xl,yl)}∈(RnX R). Wherein x isk(k e (1, l)) represents the input vector for the kth training sample, ykIs the corresponding output value. The linear regression function set in the high-dimensional feature space is:
f(x)＝wΦ(x)+b
wherein Φ (x) is the nonlinear mapping function. Introduce an epsilon linear insensitive loss function:
L(f(x),y,ε)＝max(0,|y-f(x)|-ε)
where f (x) is the predicted value of the output and y is the true value of the sample, a relaxation variable ξ is introducedk，The linear support vector regression problem can be expressed as:
where C is a penalty parameter. When solving, the lagrange function factor is used to convert the above problem into its dual form, which is expressed as follows:
α thereink，Is Lagrange multiplier, K (x)k,xj)＝Φ(xk)Φ(xj) Is a kernel function. Suppose thatAndfor an optimal solution, the parameters w, b can then be expressed as:
wherein N isNSVRepresenting the number of support vectors, for sample point xiIf it corresponds toIf not, the sample point is a support vector. Obtaining a regression function:
the kernel function K (x) needs to be determinedkX) and a penalty parameter C. When the variation function curve is fitted, the RBF kernel function has a better effect, and the formula is as follows:
K(xi,xj)＝exp(-σ||xi-xj||2),σ＞0
where σ is a nuclear parameter. Therefore, the parameters σ and C mainly need to be determined.
In summary, the specific steps of the SVR-based variogram fitting are as follows:
step 1), calculating the distance h of all sample data pairs according to a spatial variation function formulaiAnd the corresponding function of variation value gamma (h)i) Composition data set [ h ]i,γ(hi)]；
Step 2), randomly extracting 80% of data from the data set to generate a training set T { [ h ]1,γ(h1)],[h2,γ(h2)],…,[hl,γ(hl)]And (4) taking the rest 20% of data as a test set;
step 3), training the training set T by adopting SVR, and fitting a variation function curve gamma (h);
and 4) performing performance evaluation on the variation function curve gamma (h) by using the test set generated in the step 2, outputting the variation function curve gamma (h) if the expected requirement is met, and otherwise, modifying the SVR parameters and returning to the step 3 for re-fitting.
Simulation experiment
In order to verify the effectiveness and feasibility of the method, 3 experiments are designed based on 3 angles of the measured data in the simulation roadway unit from the analysis of the fitting performance of the variation function curve, the analysis of the SVR-Kriging interpolation estimation performance and the analysis of the two-dimensional fingerprint positioning precision of the mobile terminal.
① analysis of performance of curve fitting of variation function
The performance of the fitting of the variation function is an important index for determining the interpolation estimation performance and the fingerprint positioning accuracy. The variation function values among 25 sampling points are calculated based on the received signal strength of AP1, and a variation function curve fitting based on SVR is performed by using Matlab, and 80% of the variation function values are randomly selected as a training set, and the rest are used as a test set, so that the obtained results are shown in fig. 4 and 5.
For more intuitive measurement, a QQ diagram is drawn for the training set fitting result and the test set testing result, as shown in fig. 6 and 7.
In fig. 6, it can be seen that in the lower left corner region of the diagonal line in the QQ diagram, the point fitting value is located near the expected value, the upper and lower fluctuation is not large, and the point fitting value is basically consistent with the diagonal line, which indicates that the SVR regression fitting model is good in reasonableness. And the upper right corner of the diagonal line is a site with higher significance, and the fitting values exceed the expected values, so that the fitting method is proved to cover the site with higher relevance and has better accuracy. FIG. 7 shows the comparison of the fitting model with the test set, which has a performance substantially identical to that of FIG. 6, and the validity and accuracy of the fitting of SVR are verified again.
To better compare the performance of each fitting method, the root mean square error RMSE and the decision coefficient R of each model are calculated2The results are shown in Table 2.
TABLE 2 comparison of the fitting Performance of the variogram of the different models
It can be seen from Table 2 that the SVR fit RMSE is low, R2Higher performance and certain advantages compared with other models. Therefore, compared with the traditional model fitting method, the SVR-based mutation function fitting method provided by the invention has better effect.
② SVR-Kriging interpolation estimation performance analysis
In order to verify the effectiveness of the SVR-Kriging interpolation algorithm, the invention, the inverse distance weight Interpolation (IDW) method and the general Kriging interpolation method are used to perform interpolation estimation on the 25 interpolation points respectively and compare with the actual measurement values, and the result is shown in fig. 8.
The RMSE of the 3 interpolation estimation methods is calculated separately. Wherein, the SVR-Kriging is 1.2780, the IDW is 3.2647, and the common Kriging interpolation method is 1.7345. Therefore, the SVR-Kriging interpolation estimation method has good performance and certain effectiveness.
In view of the high precision requirement on the position of a worker in a mine environment and the convenience for accurate rescue particularly when an abnormal condition occurs, compared with the traditional method, the SVR-Kriging interpolation estimation method is innovative and improved, and can meet the actual application requirement.
③ location precision analysis of two-dimensional fingerprint of mobile terminal
And the received RSSIs of other 3 APs are respectively used for carrying out interpolation estimation provided by the invention, and a required position fingerprint information database can be finally generated by combining the interpolation result of the AP 1. In order to more intuitively reflect the fingerprint positioning information database, the RSSI and location information results corresponding to each AP are shown in fig. 9 to 12. In the coordinate axes of FIGS. 9-12, the x-axis and y-axis represent two-dimensional coordinates of the mineworker, and the z-axis represents RSSI (dBm) of the AP received at the location. During fingerprint positioning, the RSSI of 4 APs received at the positions of mine workers in the online stage is compared with the databases corresponding to the images in FIGS. 9-12, and the position fingerprint information with the highest matching degree with the fingerprint information collected by the worker mobile terminal in the databases is selected to determine the position coordinates of the workers.
In order to verify the positioning accuracy of the invention, the data collected by 10 position points are randomly selected from the data collected by the terminal carried by a worker in the on-line positioning stage, and the data is subjected to positioning simulation through a log weight LANDMARC algorithm in a matching algorithm and a two-dimensional position fingerprint database established by the invention. Meanwhile, simulation is respectively carried out by using an RSSI signal propagation model estimation method (comparison 1) provided in Energy-efficiency index localization of small and large devices using Bluetooth and a one-dimensional position fingerprint positioning method (comparison 2) provided in an RSS finger mode-based underground coal mine WLAN positioning method under the same simulation environment, and comparison results are shown in a table 3.
TABLE 3 comparison of positioning effects of three positioning algorithms
Table 3 shows that the average error of the present invention is 1.297m, which is better than the methods proposed by comparison 1 and comparison 2, and the designed average positioning accuracy requirement of 2m is achieved, thus proving the effectiveness of the present invention in positioning position fingerprints.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.
Claims (3)
1. A two-dimensional fingerprint positioning method for a mobile terminal is characterized by comprising the following steps:
step one, AP deployment and regional gridding division
When wireless AP deployment is carried out in a positioning area, two-dimensional deployment is adopted, namely, two longitudinal APs are respectively deployed at two sides of a roadway, and the distance between the transversely deployed APs is determined according to the effective coverage radius of the APs in the actual roadway;
step two, sampling stage
The sampling stage comprises selection of sampling points, sampling of sample fingerprint information and processing of fingerprint information data; selecting sampling points by adopting an interval traversal sampling method, and selecting one sampling point every other grid; sampling and storing the sample fingerprint information for multiple times by depending on the position fingerprint information of the selected sampling point by the mobile terminal in an off-line stage; when data acquisition is carried out, the mobile terminal only records signal intensity values sent by 4 adjacent APs in the area as position fingerprint information of the sampling point; the fingerprint information data processing is to perform Gaussian filtering processing on the stored sample fingerprint information to filter out interference items, and the specific method is as follows:
the density function of the acquisition result x satisfies the gaussian distribution as:
<mrow> <mi>f</mi> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mn>1</mn> <mrow> <msqrt> <mrow> <mn>2</mn> <mi>&pi;</mi> </mrow> </msqrt> <mi>&sigma;</mi> </mrow> </mfrac> <mi>exp</mi> <mo>&lsqb;</mo> <mo>-</mo> <mfrac> <msup> <mrow> <mo>(</mo> <mi>x</mi> <mo>-</mo> <mi>&mu;</mi> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mrow> <mn>2</mn> <msup> <mi>&sigma;</mi> <mn>2</mn> </msup> </mrow> </mfrac> <mo>&rsqb;</mo> </mrow>
wherein mu is the mean value of the data sets collected for multiple times, and sigma is the standard deviation; selecting corresponding RSSI data with the high probability event of 0.6-1 (f) (x) and the geometric mean value of the RSSI values in the range as the final RSSI value of the adopted AP;
step three, interpolation stage and database construction
The interpolation stage comprises interpolation point selection, variogram fitting and Kriging interpolation estimation, wherein the interpolation point is the central point of a grid point which is left by removing sampling points in the sampling stage in the area grid division;
calculating a variation function value between sampling points according to the received signal intensity of one AP, and performing SVR-based variation function curve fitting; estimating the RSSI value of the AP at each interpolation point by using a Kriging interpolation method based on the fitted variation function curve; similarly, the signal strength of other 3 APs received by the sampling point is respectively interpolated and estimated by the SVR-Kriging interpolation estimation method to complement the RSSI value of the 3 APs in each non-sampling area in the two-dimensional grid division; generating the position fingerprint information of the interpolation point by combining the RSSI estimated by the 4 APs at the interpolation point; the position fingerprint information of the sampling point and the interpolation point is fused to construct an information database for two-dimensional fingerprint positioning of the mobile terminal;
step four, on-line positioning
In the on-line positioning stage, a mobile terminal carried by a mine worker firstly acquires the signal intensity of each AP of an unknown position, and then selects the RSSI of 4 required APs to form the position fingerprint information of the position through data preprocessing; and when fingerprint positioning is carried out, matching is carried out on the position fingerprint information of the position where the mobile terminal is located and the corresponding constructed database, and the position of the most similar fingerprint information in the database is output as the position of the mine worker.
2. The two-dimensional fingerprint positioning method for the mobile terminal according to claim 1, wherein: the SVR-based variogram curve fitting steps are as follows:
step S1, calculating the distance h of all sample data pairs by using the space variation functioniAnd the corresponding function of variation value gamma (h)i) Composition data set [ h ]i,γ(hi)]；
The spatial variation function is as follows:
<mrow> <mi>&gamma;</mi> <mrow> <mo>(</mo> <msub> <mi>h</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mn>1</mn> <mrow> <mn>2</mn> <msub> <mi>N</mi> <mrow> <mi>h</mi> <mi>i</mi> </mrow> </msub> </mrow> </mfrac> <munderover> <mo>&Sigma;</mo> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <msub> <mi>N</mi> <mrow> <mi>h</mi> <mi>i</mi> </mrow> </msub> </munderover> <msup> <mrow> <mo>&lsqb;</mo> <mi>R</mi> <mrow> <mo>(</mo> <msub> <mi>x</mi> <mi>j</mi> </msub> <mo>+</mo> <msub> <mi>h</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>-</mo> <mi>R</mi> <mrow> <mo>(</mo> <msub> <mi>x</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> <mo>&rsqb;</mo> </mrow> <mn>2</mn> </msup> </mrow>
wherein h isiRepresenting the vector distance, N, of a pair of sample points in a spatial regionhiRepresenting the distance h between all pairs of sample pointsiThe number of (2);
step S2, randomly extracting 80% of data from the data set to generate a training set T { [ h ]1,γ(h1)],[h2,γ(h2)],…,[hl,γ(hl)]And taking the rest data as a test set;
s3, training the training set T by adopting SVR, and fitting a variation function curve gamma (h);
and step S4, performing performance evaluation on the variation function curve gamma (h) by using the test set generated in the step S2, outputting the variation function curve gamma (h) if the expected requirement is met, and otherwise, modifying the SVR parameters and returning to the step S3 for re-fitting.
3. The two-dimensional fingerprint positioning method for the mobile terminal according to claim 1, wherein: the Kriging interpolation estimateThe method is to change the region variation of interpolation points satisfying the second order stationary or intrinsic in the region into R (x)0) And m regional variation quantities satisfying the second-order stationary sample in the neighborhood range are R (x)i) (i ═ 1,2, …, m) by comparison with the known R (x)i) The weighted sum of the values can estimate the R (x) to be estimated for the interpolation point0) Namely:
<mrow> <mi>R</mi> <mrow> <mo>(</mo> <msub> <mi>x</mi> <mn>0</mn> </msub> <mo>)</mo> </mrow> <mo>=</mo> <munderover> <mo>&Sigma;</mo> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>m</mi> </munderover> <msub> <mi>&lambda;</mi> <mi>i</mi> </msub> <mi>R</mi> <mrow> <mo>(</mo> <msub> <mi>x</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>.</mo> </mrow>
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711195755.1A CN107979817A (en) | 2017-11-24 | 2017-11-24 | A kind of mobile terminal two dimension fingerprint positioning method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711195755.1A CN107979817A (en) | 2017-11-24 | 2017-11-24 | A kind of mobile terminal two dimension fingerprint positioning method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107979817A true CN107979817A (en) | 2018-05-01 |
Family
ID=62011762
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711195755.1A Pending CN107979817A (en) | 2017-11-24 | 2017-11-24 | A kind of mobile terminal two dimension fingerprint positioning method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107979817A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109143161A (en) * | 2018-09-30 | 2019-01-04 | 电子科技大学 | High-precision indoor orientation method based on mixed-fingerprint Environmental Evaluation Model |
| CN111405509A (en) * | 2020-04-01 | 2020-07-10 | 杭州十域科技有限公司 | Fingerprint positioning algorithm for reducing acquisition cost |
| CN111447549A (en) * | 2019-12-31 | 2020-07-24 | 华东理工大学 | Non-uniform UWB positioning error set network construction method and positioning error modeling method |
| CN112799040A (en) * | 2021-01-25 | 2021-05-14 | 华中科技大学鄂州工业技术研究院 | Method, robot, terminal and storage medium for a robot to search for a target |
| CN113382470A (en) * | 2021-06-08 | 2021-09-10 | 嘉兴霏云信息科技有限公司 | Method for greatly reducing fingerprint acquisition workload in wireless fingerprint positioning |
| CN113747355A (en) * | 2021-09-03 | 2021-12-03 | 中国华能集团清洁能源技术研究院有限公司 | WiFi positioning method, device, equipment and medium in power plant |
| CN114710742A (en) * | 2022-02-28 | 2022-07-05 | 盐城师范学院 | Indoor positioning method for constructing fingerprint map based on multi-chain interpolation |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130196685A1 (en) * | 2008-10-06 | 2013-08-01 | Root Wireless, Inc. | Web server and method for hosting a web page for presenting location based user quality data related to a communication network |
| CN105205114A (en) * | 2015-09-06 | 2015-12-30 | 重庆邮电大学 | Wi-Fi (wireless fidelity) positioning fingerprint database construction method based on image processing |
| CN106028290A (en) * | 2016-05-06 | 2016-10-12 | 浙江工业大学 | Kriging-based multi-dimensional vector fingerprint location method for WSN |
| CN106211327A (en) * | 2016-09-18 | 2016-12-07 | 中山大学 | A kind of method automatically generating location fingerprint data |
| CN106767828A (en) * | 2016-12-29 | 2017-05-31 | 南京邮电大学 | A kind of mobile phone indoor positioning solution |
| CN106815551A (en) * | 2016-12-08 | 2017-06-09 | 新疆农业大学 | A kind of optimization method of the variation function parameter fitting of forest inventory control |
| CN107179525A (en) * | 2017-07-21 | 2017-09-19 | 南京邮电大学 | A kind of location fingerprint construction method of the Kriging regression based on Thiessen polygon |
-
2017
- 2017-11-24 CN CN201711195755.1A patent/CN107979817A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130196685A1 (en) * | 2008-10-06 | 2013-08-01 | Root Wireless, Inc. | Web server and method for hosting a web page for presenting location based user quality data related to a communication network |
| CN105205114A (en) * | 2015-09-06 | 2015-12-30 | 重庆邮电大学 | Wi-Fi (wireless fidelity) positioning fingerprint database construction method based on image processing |
| CN106028290A (en) * | 2016-05-06 | 2016-10-12 | 浙江工业大学 | Kriging-based multi-dimensional vector fingerprint location method for WSN |
| CN106211327A (en) * | 2016-09-18 | 2016-12-07 | 中山大学 | A kind of method automatically generating location fingerprint data |
| CN106815551A (en) * | 2016-12-08 | 2017-06-09 | 新疆农业大学 | A kind of optimization method of the variation function parameter fitting of forest inventory control |
| CN106767828A (en) * | 2016-12-29 | 2017-05-31 | 南京邮电大学 | A kind of mobile phone indoor positioning solution |
| CN107179525A (en) * | 2017-07-21 | 2017-09-19 | 南京邮电大学 | A kind of location fingerprint construction method of the Kriging regression based on Thiessen polygon |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109143161A (en) * | 2018-09-30 | 2019-01-04 | 电子科技大学 | High-precision indoor orientation method based on mixed-fingerprint Environmental Evaluation Model |
| CN109143161B (en) * | 2018-09-30 | 2023-01-10 | 电子科技大学 | High-precision indoor positioning method based on mixed fingerprint quality evaluation model |
| CN111447549A (en) * | 2019-12-31 | 2020-07-24 | 华东理工大学 | Non-uniform UWB positioning error set network construction method and positioning error modeling method |
| CN111447549B (en) * | 2019-12-31 | 2021-06-15 | 华东理工大学 | Non-uniform UWB positioning error set network construction method and positioning error modeling method |
| CN111405509A (en) * | 2020-04-01 | 2020-07-10 | 杭州十域科技有限公司 | Fingerprint positioning algorithm for reducing acquisition cost |
| CN112799040A (en) * | 2021-01-25 | 2021-05-14 | 华中科技大学鄂州工业技术研究院 | Method, robot, terminal and storage medium for a robot to search for a target |
| CN112799040B (en) * | 2021-01-25 | 2023-10-13 | 华中科技大学鄂州工业技术研究院 | Method for searching target by robot, terminal and storage medium |
| CN113382470A (en) * | 2021-06-08 | 2021-09-10 | 嘉兴霏云信息科技有限公司 | Method for greatly reducing fingerprint acquisition workload in wireless fingerprint positioning |
| CN113382470B (en) * | 2021-06-08 | 2022-05-24 | 嘉兴霏云信息科技有限公司 | Method for greatly reducing fingerprint acquisition workload in wireless fingerprint positioning |
| CN113747355A (en) * | 2021-09-03 | 2021-12-03 | 中国华能集团清洁能源技术研究院有限公司 | WiFi positioning method, device, equipment and medium in power plant |
| CN114710742A (en) * | 2022-02-28 | 2022-07-05 | 盐城师范学院 | Indoor positioning method for constructing fingerprint map based on multi-chain interpolation |
| CN114710742B (en) * | 2022-02-28 | 2025-03-04 | 盐城师范学院 | An indoor positioning method based on multi-chain interpolation to construct fingerprint map |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107979817A (en) | A kind of mobile terminal two dimension fingerprint positioning method | |
| Zuo et al. | Multi-phase fingerprint map based on interpolation for indoor localization using iBeacons | |
| CN108107461B (en) | Indoor and outdoor positioning seamless switching method based on mobile terminal | |
| CN103220777A (en) | Mobile device positioning system | |
| CN104635203B (en) | Radio interference source direction-finding and positioning method based on particle filter algorithm | |
| CN104703143A (en) | Indoor positioning method based on WIFI signal strength | |
| CN107392875A (en) | A kind of cloud data denoising method based on the division of k neighbours domain | |
| CN103747419B (en) | A kind of indoor orientation method based on signal strength difference and dynamic linear interpolation | |
| CN105338498A (en) | Construction method for fingerprint database in WiFi indoor positioning system | |
| CN103139907A (en) | Indoor wireless positioning method by utilizing fingerprint technique | |
| CN104038901A (en) | Indoor positioning method for reducing fingerprint data acquisition workload | |
| CN104519571A (en) | Indoor positioning method based on RSS (Received Signal Strength) | |
| CN107179525A (en) | A kind of location fingerprint construction method of the Kriging regression based on Thiessen polygon | |
| CN113301503B (en) | Weak coverage area detection method and device | |
| CN105163385A (en) | Localization algorithm based on sector overlapping area of clustering analysis | |
| CN104507097A (en) | Semi-supervised training method based on WiFi (wireless fidelity) position fingerprints | |
| CN105916202A (en) | Probabilistic WiFi indoor positioning fingerprint database construction method | |
| CN112040405A (en) | An indoor localization method based on kernel extreme learning machine and particle filter | |
| CN112367614A (en) | LSTM-based Wi-Fi and geomagnetic field fusion indoor positioning algorithm | |
| CN103913720B (en) | A kind of indoor orientation method towards nonspecific Wi-Fi equipment | |
| Hou et al. | Positioning by floors based on WiFi fingerprint | |
| Assayag et al. | Indoor positioning system using synthetic training and data fusion | |
| CN114916059A (en) | WiFi fingerprint sparse map extension method based on interval random logarithm shadow model | |
| Arai et al. | Color radiomap interpolation for efficient fingerprint wifi-based indoor location estimation | |
| Wang et al. | A bluetooth location method based on kNN algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180501 |
|
| RJ01 | Rejection of invention patent application after publication |