CN112584311B - Indoor three-dimensional space fingerprint positioning method based on WKNN fusion - Google Patents

Indoor three-dimensional space fingerprint positioning method based on WKNN fusion Download PDF

Info

Publication number
CN112584311B
CN112584311B CN202011477793.8A CN202011477793A CN112584311B CN 112584311 B CN112584311 B CN 112584311B CN 202011477793 A CN202011477793 A CN 202011477793A CN 112584311 B CN112584311 B CN 112584311B
Authority
CN
China
Prior art keywords
distance
positioning
fingerprint
wknn
target
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.)
Active
Application number
CN202011477793.8A
Other languages
Chinese (zh)
Other versions
CN112584311A (en
Inventor
李天成
杨浩宇
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Northwestern Polytechnical University
Original Assignee
Northwestern Polytechnical University
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 Northwestern Polytechnical University filed Critical Northwestern Polytechnical University
Priority to CN202011477793.8A priority Critical patent/CN112584311B/en
Publication of CN112584311A publication Critical patent/CN112584311A/en
Application granted granted Critical
Publication of CN112584311B publication Critical patent/CN112584311B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/02—Services making use of location information
    • H04W4/023—Services making use of location information using mutual or relative location information between multiple location based services [LBS] targets or of distance thresholds
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/30—Services specially adapted for particular environments, situations or purposes
    • H04W4/33—Services specially adapted for particular environments, situations or purposes for indoor environments, e.g. buildings
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Position Fixing By Use Of Radio Waves (AREA)
  • Collating Specific Patterns (AREA)

Abstract

The invention discloses an indoor three-dimensional space fingerprint positioning method based on WKNN fusion. Under the condition of not thinning the 3D fingerprint database, K nearest neighbor reference points can be obtained through KNN, and a positioning result is obtained through weighting; and solving another positioning result according to the geometric relation between the coordinates and the distance of the base station in the database, and fusing the two results to realize more accurate and stable positioning precision. The method fully utilizes the position fingerprint information, and has the advantages of simple calculation, high positioning precision and high stability.

Description

Indoor three-dimensional space fingerprint positioning method based on WKNN fusion
Technical Field
The invention belongs to the technical field of indoor positioning, and particularly relates to an indoor fingerprint positioning method.
Background
Indoor positioning and navigation have become a research hotspot in the field of location-based services. The indoor fingerprint-based positioning method has been studied deeply at home and abroad, a large number of two-dimensional indoor fingerprint positioning algorithms are provided, and good positioning accuracy is obtained. However, in many three-dimensional indoor application scenarios, such as large-scale multi-layer underground parking lots, shopping malls, etc., the two-dimensional positioning technology is difficult to meet the actual positioning requirements.
Documents "road testing of 3D Indor Localization based on Fingerprint Technology in Wireless Sensor Networks [ 201310 th International Conference on electric Engineering/Electronics, Computer, Telecommunications and Information Technology, Krabi,2013, pp.1-6 ]" and "WiFi-based Indoor three-dimensional Localization Technology research [ D ]. Nanjing postal university, 2018." disclose a KNN Localization method and WKNN (weighted K-Nearest neighbor, weighted K-near neighbor) Localization algorithm based on 3D Indoor fingerprints, respectively. According to the method, the two-dimensional indoor fingerprint positioning is increased to the three-dimensional space, and the two-dimensional positioning algorithm is transferred to the three-dimensional positioning, so that the indoor positioning precision is improved. However, because useful information of the location fingerprint is not fully mined, for example, the KNN algorithm has a low target resolution in a certain area, and WKNN only makes a single K nearest neighbor reference point weighted value, the methods described in the literature have certain volatility and instability, and for this reason, a denser offline fingerprint database needs to be re-established. However, introducing a dimension will result in a large increase in the data size of the off-line fingerprint database, and the use of a sparse fingerprint database can only bring about poor accuracy. Therefore, it is necessary to consider both the data amount of the fingerprint library and the indoor positioning accuracy in the three-dimensional indoor positioning.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention provides an indoor three-dimensional space fingerprint positioning method based on WKNN fusion. Under the condition of not thinning the 3D fingerprint database, K nearest neighbor reference points can be obtained through KNN, and a positioning result is obtained through weighting; and solving another positioning result according to the geometric relation between the coordinates and the distance of the base station in the database, and fusing the two results to realize more accurate and stable positioning precision. The method makes full use of the position fingerprint information, and has the advantages of simple calculation, high positioning precision and high stability.
The technical scheme adopted by the invention for solving the technical problem comprises the following steps:
step 1: establishing a fingerprint off-line information base;
an off-line library building stage: selecting a plurality of reference points in a positioning area, and measuring the received signal strength of all base stations on each reference point; storing the reference point coordinates and the received signal strength into a database together to form a fingerprint database;
and (3) in an online positioning stage: matching the signal intensity information received by the target with information in a fingerprint library by using KNN to obtain K nearest neighbor reference point coordinates:
Figure GDA0003635176030000021
in the formula D p Indicating signal strength distance, OB indicating signal strength information received by the object, NO k Representing k nearest-neighbor reference points matched by KNN algorithmFingerprint information;
step 2: introducing distance weight W to K nearest neighbor reference point coordinates obtained in the step 1 i ：
Figure GDA0003635176030000022
Wherein (x) i ,y i ,z i ) Reference point coordinates representing the ith nearest neighbor, (x, y, z) target coordinates, d i Representing the signal strength distance between the signal strength information received by the target and the fingerprint information of the ith reference point, wherein epsilon is a positive number;
and step 3: adopting a Chan algorithm based on TDOA to relocate the target;
step 3-1: and (3) calculating:
Figure GDA0003635176030000023
where d is the actual distance between the target and the base station, d 0 As reference distance, P (d) and P (d) 0 ) Respectively indicate when the target is at a distance d and d from the base station 0 Average received power in time, α represents a path loss parameter; theta represents noise due to shadow fading and fast fading, and theta is a mean of 0, variance
Figure GDA0003635176030000024
Normal random variable of (1);
step 3-2: an expression of the distance d is derived from equation (3):
Figure GDA0003635176030000025
equation (4) is modified:
Figure GDA0003635176030000026
step 3-3: the method for reducing the influence of the noise by adopting the adaptive iterative variance algorithm of the noise specifically comprises the following steps:
the raw noise is defined as:
Figure GDA0003635176030000027
the noise obtained by the adaptive iterative variance algorithm of the noise is
Figure GDA0003635176030000028
Step 3-3-1: order to
Figure GDA0003635176030000029
Let i be 2;
step 3-3-2: computing
Figure GDA0003635176030000031
Step 3-3-3: adding 1 to i;
Figure GDA0003635176030000032
step 3-3-4: repeating the steps 3-3-2 to 3-3-3 until
Figure GDA0003635176030000033
When it is time, the cycle is ended, and
Figure GDA0003635176030000034
Figure GDA0003635176030000035
variance of theta using adaptive iterative variance algorithm with noise
Figure GDA0003635176030000036
Carrying out treatment;
according to the distance d i Corresponding noise thereof
Figure GDA0003635176030000037
The method is obtained by adopting a Chan algorithm in TDOA to carry out positioning calculationA new set of target point coordinates (x, y, z);
and 4, step 4: fusing the positioning results of the step 2 and the step 3, which comprises the following specific steps:
in the X direction:
Figure GDA0003635176030000038
in the formula (I), the compound is shown in the specification,
Figure GDA0003635176030000039
the result of the fusion in the X direction is shown,
Figure GDA00036351760300000310
indicating position information of the WKNN method in the X direction;
Figure GDA00036351760300000311
indicating the position information, variance, of the Chan algorithm in the X direction
Figure GDA00036351760300000312
And
Figure GDA00036351760300000313
obtaining by adopting a statistical method of multiple measurements;
y direction:
Figure GDA00036351760300000314
in the formula (I), the compound is shown in the specification,
Figure GDA00036351760300000315
the result of the fusion in the Y direction is shown,
Figure GDA00036351760300000316
indicating position information of the WKNN method in the Y direction;
Figure GDA00036351760300000317
representing the Chan AlgorithmPosition information in the Y direction, variance
Figure GDA00036351760300000318
And
Figure GDA00036351760300000319
obtaining by adopting a statistical method of multiple measurements;
the Z direction:
Figure GDA00036351760300000320
in the formula (I), the compound is shown in the specification,
Figure GDA00036351760300000321
showing the result of the fusion in the Z direction,
Figure GDA00036351760300000322
indicating position information of the WKNN method in the Z direction;
Figure GDA00036351760300000323
indicating the position information, variance, of the Chan algorithm in the Z direction
Figure GDA00036351760300000324
And
Figure GDA00036351760300000325
and obtaining the data by adopting a statistical method of multiple measurements.
Preferably, in the formula (1), when p is 1, D p Denotes the Manhattan distance, D when p is 2 p Representing the euclidean distance.
Preferably, the epsilon < 0.01.
Preferably, said d 0 ＝1m，α＝2。
By adopting the WKNN fusion-based indoor three-dimensional space fingerprint positioning method, the position fingerprint information is fully mined by adopting an information fusion algorithm, the positioning precision is improved, and the instability in the positioning process is overcome. The method breaks through indoor plane positioning, can position in an indoor three-dimensional space, and is suitable for scenes such as indoor multilayer parking lots, superstores, railway stations and the like.
Drawings
FIG. 1 is a flow chart of the structure of the method of the present invention.
Figure 2 is a structural framework of the method of the invention.
Fig. 3 is a simulation scenario 1 of the method of the present invention.
Fig. 4 is a simulation scenario 2 of the method of the present invention.
FIG. 5 is a result of scene 1 simulation accuracy for the method of the present invention.
FIG. 6 is a result of scene 2 simulation accuracy for the method of the present invention.
Detailed Description
The invention is further illustrated with reference to the following figures and examples.
As shown in fig. 1 and fig. 2, the present invention provides an indoor three-dimensional spatial fingerprint positioning method based on WKNN fusion, which includes the following steps:
step 1: establishing a fingerprint off-line information base;
an off-line library building stage: selecting a plurality of reference points in a positioning area, and measuring the received signal strength of all base stations on each reference point; storing the reference point coordinates and the received signal strength into a database together to form a fingerprint database;
and (3) in an online positioning stage: matching the signal intensity information received by the target with information in a fingerprint library by using KNN to obtain K nearest neighbor reference point coordinates:
Figure GDA0003635176030000041
in the formula, D p Indicating signal strength distance, OB indicating signal strength information received by the object, NO k Fingerprint information of k nearest neighbor reference points matched by the KNN algorithm is represented;
step 2: for the K nearest neighbor reference point coordinates obtained in the step 1, but KNN is a mean value algorithm, namely, several reference points are obtainedIs regarded as the target position, is usually inaccurate, so the WKNN algorithm is adopted to introduce the distance weight W i ：
Figure GDA0003635176030000051
Wherein (x) i ,y i ,z i ) Reference point coordinates representing the ith nearest neighbor, (x, y, z) target coordinates, d i Representing the signal strength distance between the signal strength information received by the target and the fingerprint information of the ith reference point, wherein epsilon is a very small positive number and is used for preventing the divisor in the formula from being zero;
and step 3: adopting a Chan algorithm based on TDOA to relocate the target;
step 3-1: and (3) calculating:
Figure GDA0003635176030000052
where d is the actual distance between the target and the base station, d 0 As reference distance, P (d) and P (d) 0 ) Respectively indicate when the target is at a distance d and d from the base station 0 Average received power in time, α represents a path loss parameter; theta denotes noise due to shadow fading and fast fading, and theta is a mean of 0 and a variance
Figure GDA0003635176030000053
Normal random variable of (1);
step 3-2: an expression of the distance d is derived from equation (3):
Figure GDA0003635176030000054
an unbiased estimator of the distance estimate is used to modify equation (4):
Figure GDA0003635176030000055
step 3-3: the formula (5) can generate larger distance deviation when the signal changes frequently, therefore, the invention provides a noise adaptive iterative variance algorithm, thereby reducing the influence of noise on propagation path evaluation, and the specific steps are as follows:
the raw noise is defined as:
Figure GDA0003635176030000056
the noise obtained by the adaptive iterative variance algorithm of the noise is
Figure GDA0003635176030000057
Step 3-3-1: order to
Figure GDA0003635176030000058
Let i be 2;
step 3-3-2: computing
Figure GDA0003635176030000059
Step 3-3-3: adding 1 to i;
Figure GDA00036351760300000510
step 3-3-4: repeating the steps 3-3-2 to 3-3-3 until
Figure GDA00036351760300000511
When it is time, the cycle is ended, and
Figure GDA00036351760300000512
Figure GDA00036351760300000513
variance of theta using adaptive iterative variance algorithm with noise
Figure GDA00036351760300000514
Carrying out treatment;
according to distance d i And itCorresponding noise
Figure GDA0003635176030000061
Performing positioning calculation by using a Chan algorithm in TDOA to obtain a group of new target point coordinates (x, y, z);
in a three-dimensional space, because the nodes matched with the position fingerprints are enough, the invention provides a TDOA-based distance measurement method for geometric positioning, and distance difference is obtained by differentiating the distances; obtaining a weighted least square result of the position by a hyperbolic positioning method or a trilateral positioning method, wherein higher positioning accuracy can be obtained by a Chan algorithm or a Taylor series expansion method of TDOA in a three-dimensional space;
and 4, step 4: fusing the positioning results of the step 2 and the step 3, which comprises the following specific steps:
and respectively obtaining a WKNN positioning result and a multilateral positioning result based on TDOA through the 3 steps, regarding the two positioning results as two different information sources in fusion positioning, and respectively fusing the three dimensions of X, Y and Z through an information fusion formula.
In the X direction:
Figure GDA0003635176030000062
in the formula (I), the compound is shown in the specification,
Figure GDA0003635176030000063
the result of the fusion in the X direction is shown,
Figure GDA0003635176030000064
indicating position information of the WKNN method in the X direction;
Figure GDA0003635176030000065
indicating the position information, variance, of the Chan algorithm in the X direction
Figure GDA0003635176030000066
And
Figure GDA0003635176030000067
obtaining by adopting a statistical method of multiple measurements;
y direction:
Figure GDA0003635176030000068
in the formula (I), the compound is shown in the specification,
Figure GDA0003635176030000069
the result of the fusion in the Y direction is shown,
Figure GDA00036351760300000610
indicating position information of the WKNN method in the Y direction;
Figure GDA00036351760300000611
indicating the position information, variance, of the Chan algorithm in the Y direction
Figure GDA00036351760300000612
And
Figure GDA00036351760300000613
obtaining by adopting a statistical method of multiple measurements;
the Z direction:
Figure GDA00036351760300000614
in the formula (I), the compound is shown in the specification,
Figure GDA00036351760300000615
showing the result of the fusion in the Z direction,
Figure GDA00036351760300000616
indicating position information of the WKNN method in the Z direction;
Figure GDA00036351760300000617
indicating the position information, variance, of the Chan algorithm in the Z direction
Figure GDA00036351760300000618
And
Figure GDA00036351760300000619
and obtaining the data by adopting a statistical method of multiple measurements.
Preferably, in the formula (1), when p is 1, D p Denotes the Manhattan distance, D when p is 2 p Representing the euclidean distance.
Preferably, the epsilon < 0.01.
Preferably, said d 0 ＝1m，α＝2。
The specific embodiment is as follows:
step 1, setting a simulation scene.
Simulation scenario set to [500,500,500 ]]m 3 Under the three-dimensional coordinate system of (1):
scene one: as shown in fig. 3, the target is located at [420,420,420] meters, the target is located by using a location fingerprint method and using KNN, WKNN algorithms and a Chan algorithm based on TDOA, the location results obtained by the two algorithms are fused, and the result is included in a scene-location table.
TABLE 1 scene-location table
True position of target [420,420,420]
Position of KNN [412.5,412.5,412.5]
WKNN position [415.67,415.67,415.67]
Chan algorithm position [425.24,412.22,418.31]
Fusion site [423.32,412.91,417.78]
On the basis of the above steps, 1000 times of target positioning is carried out, the distance from the positioning result to the true value in various methods is used as an independent variable, the percentage falling into an error circle is used as a dependent variable, and a Cumulative Distribution Function (CDF) diagram related to the error is made.
Scene two: as shown in fig. 4, the target moves arbitrarily in the positioning space, the positioning and tracking are performed by using the position fingerprint KNN, the position fingerprint WKNN, and the geometric positioning Chan method, the two methods, WKNN and Chan, are fused, and the positioning effect is evaluated.
And 2, constructing a fingerprint database. 8 vertexes of the cube are selected in the space as each reference point of one target to be positioned.
The first stage is as follows: in the positioning area, 20 × 20 × 20 reference points are selected according to a regular shape, that is, every 25m of each coordinate is recorded as a fingerprint reference point, and the received signal strength of all APs is measured on each reference point. The reference point coordinates and the received signal strength are stored in a database together to form a fingerprint database with 8000 reference points. The mapping relation between the positions in the fingerprint database and the fingerprint signals is stored in a relation table.
And a second stage: and (3) calculating Euclidean distance information of the target and each reference point by using the formula (1), and matching the Euclidean distance information with the signal information in the database by KNN.
And 3, weighting the matched 8 reference point position fingerprints. Introducing the weight of the distance to approach the true value of the target more, including:
KNN algorithm:
Figure GDA0003635176030000081
WKNN algorithm:
Figure GDA0003635176030000082
KNN is a mean value algorithm, namely the mean value of 8 reference points is taken as the target position, in order to improve the precision and reduce the cost, a WKNN algorithm can be adopted, the reciprocal of the distance is taken as a weight factor, namely the weight of which reference point is higher when the reference point is close to, and epsilon is 0.01.
And 4, reversely solving the distance according to the signal strength, and positioning the target again through a Chan algorithm based on TDOA.
Calculating the distance from a target to an anchor node by adopting an unbiased estimator of the signal pair distance, and taking the difference of the distances as the distance difference value of the TDOA:
the Chan algorithm is a method of weighting least squares twice on a target, and in the first weighting, independent variables and long target distance are assumed to obtain a primary target position solution. And (3) obtaining a variance matrix by using the distance approximation between the solution and each anchor node to replace a true value, obtaining a process solution by using a first weighted least square algorithm, constructing a relation matrix because the actual independent variables meet the distance constraint, namely are not independent, obtaining a final solution by using a second weighted least square method, and obtaining an accurate target position solution by reverse solution.
And 5, fusing the WKNN positioning result of the fingerprint and the positioning result of the Chan algorithm.
In this embodiment, the standard deviation σ is taken W ＝5，σ T In practice, the variance should be obtained statistically.
And 6, comparing the performance indexes.
To illustrate the superiority of the present invention, for scene 1, the cumulative distribution function is selected as the comparison index; for scene 2, the position root mean square error and the position average root mean square error are selected as comparison indicators. The comparison method is a plurality of positioning methods mentioned by the invention, such as a fingerprint KNN, a fingerprint WKNN, a Chan algorithm based on TDOA, and a final fusion result of the invention. And (3) taking the accumulated distribution function and Root Mean Square Error (RMSE) as comparison indexes, and defining the target position RMSE at the time t as follows:
Figure GDA0003635176030000083
in the formula, x t ，y t And z t Is the position of the target at time t,
Figure GDA0003635176030000091
and
Figure GDA0003635176030000092
is the target location estimated by various methods.
Fig. 5 shows the results of 1000 monte carlo simulations of a target located at the MS ═ 420,420,420, according to fig. 5:
the convergence of the fusion WKNN algorithm is fastest, and the error is basically within 15 m; the error is basically within 25m after the WKNN algorithm and the Chan algorithm are carried out; the KNN algorithm is located at 25 × 25 × 25m by itself 3 The target in the area is regarded as the center point of the cube, so that the position resolution of the target is low, the jump phenomenon exists in the cumulative distribution function, and the positioning accuracy is the worst.
FIG. 6 is a graph of pairs 500X 500m 3 The target moves randomly and 1000 monte carlo simulations are performed, and the results can be obtained according to fig. 6:
when the traditional KNN algorithm is used for real-time positioning, the target resolution is low, namely the KNN is positioned at 25 multiplied by 25m for the target 3 All will target the center point of this cube; while WKNN can adjust the position by weight within the above-mentioned area range, the highest accuracy is only near the reference point and the center, and the accuracy of the rest of the area is still relatively poor. The positioning accuracy of KNN and WKNN is very unstable in the whole positioning process. The KNN error is about 4-18 m, the WKNN error is about 4-11 m, but the fluctuation is large. The positioning of the Chan algorithm based on the TDOA is relatively stable, the fluctuation is less, but the precision is basically maintained at about 10m, and the fusion result can improve the precision and reduce the instability of WKNN, and the precision is maintained at about 6 m.

Claims (4)

1. An indoor three-dimensional space fingerprint positioning method based on WKNN fusion is characterized by comprising the following steps:
step 1: establishing a fingerprint off-line information base;
an off-line library building stage: selecting a plurality of reference points in a positioning area, and measuring the received signal strength of all base stations on each reference point; storing the reference point coordinates and the received signal strength into a database together to form a fingerprint database;
and (3) in an online positioning stage: matching the signal intensity information received by the target with information in a fingerprint library by using KNN to obtain K nearest neighbor reference point coordinates:
Figure FDA0003635176020000011
in the formula, D p Indicating signal strength distance, OB indicating signal strength information received by the object, NO k Fingerprint information of k nearest neighbor reference points matched by the KNN algorithm is represented;
step 2: introducing distance weight W to K nearest neighbor reference point coordinates obtained in the step 1 i ：
Figure FDA0003635176020000012
Wherein (x) i ,y i ,z i ) Reference point coordinates representing the ith nearest neighbor, (x, y, z) target coordinates, d i Representing the signal strength distance between the signal strength information received by the target and the fingerprint information of the ith reference point, wherein epsilon is a positive number;
and step 3: adopting a Chan algorithm based on TDOA to relocate the target;
step 3-1: and (3) calculating:
Figure FDA0003635176020000013
where d is the actual distance between the target and the base station, d 0 As reference distance, P (d) and P (d) 0 ) Respectively indicate when the target is at a distance d and d from the base station 0 Average received power in time, α represents a path loss parameter; theta denotes noise due to shadow fading and fast fading, and theta is a mean of 0 and a variance
Figure FDA0003635176020000014
Normal random variable of (1);
step 3-2: an expression of the distance d is derived from equation (3):
Figure FDA0003635176020000015
equation (4) is modified:
Figure FDA0003635176020000016
step 3-3: the method for reducing the influence of the noise by adopting the adaptive iterative variance algorithm of the noise specifically comprises the following steps:
the raw noise is defined as:
Figure FDA0003635176020000017
the noise obtained by the adaptive iterative variance algorithm of the noise is
Figure FDA0003635176020000018
Step 3-3-1: order to
Figure FDA0003635176020000021
Let i be 2;
step 3-3-2: computing
Figure FDA0003635176020000022
Step 3-3-3: adding 1 to i;
Figure FDA0003635176020000023
step 3-3-4: repeating the steps 3-3-2 to 3-3-3 until
Figure FDA0003635176020000024
When it is time, the cycle is ended, and
Figure FDA0003635176020000025
Figure FDA0003635176020000026
variance of theta using adaptive iterative variance algorithm with noise
Figure FDA0003635176020000027
Carrying out treatment;
according to the distance d i Corresponding noise thereof
Figure FDA0003635176020000028
Performing positioning calculation by using a Chan algorithm in TDOA to obtain a group of new target point coordinates (x, y, z);
and 4, step 4: fusing the positioning results of the step 2 and the step 3, which comprises the following specific steps:
in the X direction:
Figure FDA0003635176020000029
in the formula (I), the compound is shown in the specification,
Figure FDA00036351760200000210
the result of the fusion in the X direction is shown,
Figure FDA00036351760200000211
indicating the position of the WKNN method in the X directionInformation;
Figure FDA00036351760200000212
indicating the position information, variance, of the Chan algorithm in the X direction
Figure FDA00036351760200000213
And with
Figure FDA00036351760200000214
Obtaining by adopting a statistical method of multiple measurements;
y direction:
Figure FDA00036351760200000215
in the formula (I), the compound is shown in the specification,
Figure FDA00036351760200000216
the result of the fusion in the Y direction is shown,
Figure FDA00036351760200000217
indicating position information of the WKNN method in the Y direction;
Figure FDA00036351760200000218
indicating the position information, variance, of the Chan algorithm in the Y direction
Figure FDA00036351760200000219
And
Figure FDA00036351760200000220
obtaining by adopting a statistical method of multiple measurements;
the Z direction:
Figure FDA00036351760200000221
in the formula (I), the compound is shown in the specification,
Figure FDA00036351760200000222
showing the result of the fusion in the Z direction,
Figure FDA00036351760200000223
indicating position information of the WKNN method in the Z direction;
Figure FDA00036351760200000224
indicating the position information, variance, of the Chan algorithm in the Z direction
Figure FDA00036351760200000225
And
Figure FDA00036351760200000226
and obtaining the data by adopting a statistical method of multiple measurements.
2. The WKNN fusion-based indoor three-dimensional spatial fingerprint positioning method as claimed in claim 1, wherein in the formula (1), when p is 1, D is p Denotes the Manhattan distance, D when p is 2 p Representing the euclidean distance.
3. The WKNN fusion-based indoor three-dimensional spatial fingerprint positioning method as claimed in claim 1, wherein epsilon < 0.01.
4. The WKNN fusion-based indoor three-dimensional space fingerprint positioning method as claimed in claim 1, wherein d is 0 ＝1m，α＝2。
CN202011477793.8A 2020-12-15 2020-12-15 Indoor three-dimensional space fingerprint positioning method based on WKNN fusion Active CN112584311B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011477793.8A CN112584311B (en) 2020-12-15 2020-12-15 Indoor three-dimensional space fingerprint positioning method based on WKNN fusion

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011477793.8A CN112584311B (en) 2020-12-15 2020-12-15 Indoor three-dimensional space fingerprint positioning method based on WKNN fusion

Publications (2)

Publication Number Publication Date
CN112584311A CN112584311A (en) 2021-03-30
CN112584311B true CN112584311B (en) 2022-08-05

Family

ID=75135099

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011477793.8A Active CN112584311B (en) 2020-12-15 2020-12-15 Indoor three-dimensional space fingerprint positioning method based on WKNN fusion

Country Status (1)

Country Link
CN (1) CN112584311B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113115224B (en) * 2021-04-06 2021-11-09 中移(上海)信息通信科技有限公司 Indoor positioning method and indoor positioning device
CN114513746B (en) * 2021-12-17 2024-04-26 南京邮电大学 Indoor positioning method integrating triple vision matching model and multi-base station regression model
CN115905787B (en) * 2022-10-21 2023-09-29 盐城师范学院 High-precision indoor positioning method based on fuzzy migration learning model
CN116975588B (en) * 2023-09-22 2023-12-19 太原理工大学 Fault diagnosis method for belt conveyor

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109951807A (en) * 2019-02-27 2019-06-28 西北师范大学 Fusion RSS and CSI indoor orientation method based on WiFi signal
CN110636437A (en) * 2019-09-05 2019-12-31 哈尔滨工程大学 Fingerprint pair-based auxiliary positioning method

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104501796B (en) * 2014-12-16 2017-03-08 重庆邮电大学 A kind of interior WLAN/MEMS merges across floor 3-dimensional localization method
CN106941718B (en) * 2017-04-07 2020-05-26 南京邮电大学 Mixed indoor positioning method based on signal subspace fingerprint database

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109951807A (en) * 2019-02-27 2019-06-28 西北师范大学 Fusion RSS and CSI indoor orientation method based on WiFi signal
CN110636437A (en) * 2019-09-05 2019-12-31 哈尔滨工程大学 Fingerprint pair-based auxiliary positioning method

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
一种基于Wi-Fi信号的位置指纹定位方法;罗旭等;《信息通信》;20190415(第04期);全文 *
复杂室内环境下的混合定位技术研究;缪辉辉;《中国优秀硕士学位论文全文数据库》;20170228;全文 *

Also Published As

Publication number Publication date
CN112584311A (en) 2021-03-30

Similar Documents

Publication Publication Date Title
CN112584311B (en) Indoor three-dimensional space fingerprint positioning method based on WKNN fusion
CN104703143B (en) A kind of indoor orientation method based on WIFI signal intensity
CN103402258B (en) Wi-Fi (Wireless Fidelity)-based indoor positioning system and method
CN109511085B (en) UWB fingerprint positioning method based on MeanShift and weighted k nearest neighbor algorithm
CN108307301B (en) Indoor positioning method based on RSSI ranging and track similarity
CN109672973B (en) Indoor positioning fusion method based on strongest AP
CN104076327B (en) Continuous positioning method based on search space reduction
CN110673089B (en) Positioning method based on arrival time under unknown line-of-sight and non-line-of-sight distribution condition
CN108882149B (en) NLOS compensation positioning method of distance correlation probability
CN110730418A (en) Indoor three-dimensional positioning improvement algorithm based on least square method
CN109348416B (en) Fingerprint indoor positioning method based on binary k-means
Zhong et al. WiFi indoor localization based on K-means
CN105807253A (en) Multi-station redundant information positioning technology without need of prior information
Zhang et al. Research on D2D co-localization algorithm based on clustering filtering
CN112083375A (en) Cooperative positioning algorithm based on position fingerprints and Taylor
CN111263295A (en) WLAN indoor positioning method and device
Diao et al. Enhancing trilateration localization by adaptive selecting distances
CN108810840B (en) Node selection method based on EFIM and distance cooperation in cooperative positioning
CN111314896A (en) Secondary construction method of environment self-adaptive fingerprint database
Yi Exploring and Evaluating the Key Technologies in TDOA-Based Indoor Positioning
Zhong et al. A Wifi positioning method based on stack auto encoder
Liang et al. A fast distributed indoor positioning based on received signal strength index
CN113794986A (en) RSSI probability distribution-based weighted positioning method
Pen et al. An Improved Weighted K-Nearest Neighbor Positioning Method in Buildings for Power Distribution Room
CN104853436A (en) A topology constraints based distributed processing wireless sensing network node flipped and fuzzy locator

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
GR01 Patent grant
GR01 Patent grant