CN107948930B - Indoor positioning optimization method based on position fingerprint algorithm - Google Patents

Indoor positioning optimization method based on position fingerprint algorithm Download PDF

Info

Publication number
CN107948930B
CN107948930B CN201711495301.6A CN201711495301A CN107948930B CN 107948930 B CN107948930 B CN 107948930B CN 201711495301 A CN201711495301 A CN 201711495301A CN 107948930 B CN107948930 B CN 107948930B
Authority
CN
China
Prior art keywords
point
rssi
positioning
neighborhood
position fingerprint
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
CN201711495301.6A
Other languages
Chinese (zh)
Other versions
CN107948930A (en
Inventor
邢建川
董科廷
韩保祯
张易丰
丁志新
康亮
王翔
侯鑫宇
王书琪
邵慧
陈朝阳
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.)
University of Electronic Science and Technology of China
Original Assignee
University of Electronic Science and Technology of China
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 University of Electronic Science and Technology of China filed Critical University of Electronic Science and Technology of China
Priority to CN201711495301.6A priority Critical patent/CN107948930B/en
Publication of CN107948930A publication Critical patent/CN107948930A/en
Application granted granted Critical
Publication of CN107948930B publication Critical patent/CN107948930B/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
    • H04W4/021—Services related to particular areas, e.g. point of interest [POI] services, venue services or geofences
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S11/00—Systems for determining distance or velocity not using reflection or reradiation
    • G01S11/02—Systems for determining distance or velocity not using reflection or reradiation using radio waves
    • G01S11/06—Systems for determining distance or velocity not using reflection or reradiation using radio waves using intensity measurements
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S5/00—Position-fixing by co-ordinating two or more direction or position line determinations; Position-fixing by co-ordinating two or more distance determinations
    • G01S5/02—Position-fixing by co-ordinating two or more direction or position line determinations; Position-fixing by co-ordinating two or more distance determinations using radio waves
    • G01S5/0252—Radio frequency fingerprinting
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S5/00—Position-fixing by co-ordinating two or more direction or position line determinations; Position-fixing by co-ordinating two or more distance determinations
    • G01S5/02—Position-fixing by co-ordinating two or more direction or position line determinations; Position-fixing by co-ordinating two or more distance determinations using radio waves
    • G01S5/12—Position-fixing by co-ordinating two or more direction or position line determinations; Position-fixing by co-ordinating two or more distance determinations using radio waves by co-ordinating position lines of different shape, e.g. hyperbolic, circular, elliptical or radial
    • 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
    • H04W4/02—Services making use of location information
    • H04W4/025—Services making use of location information using location based information parameters

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Position Fixing By Use Of Radio Waves (AREA)

Abstract

The invention discloses an indoor positioning optimization method based on a position fingerprint algorithm. On the basis of the existing position fingerprint positioning method, the invention carries out corresponding optimization processing on signal intensity acquisition, fingerprint library density, fingerprint library precision, positioning efficiency and equipment error, thereby realizing precision improvement and efficiency improvement of the positioning process. In the positioning process, the invention does not need the support of extra hardware, so the positioning cost is lower. Under the conditions of accurate signal intensity measurement and higher precision and density of the position fingerprint database, the invention can provide higher positioning precision.

Description

Indoor positioning optimization method based on position fingerprint algorithm
Technical field
The invention belongs to the technical field of indoor positioning, and particularly relates to an indoor positioning method based on a position fingerprint algorithm.
Background
The fingerprint positioning algorithm completes positioning through correlation between Received Signal Strength Indication (RSSI) information, and estimates positioning coordinates by matching the RSSI information at a positioning point with the RSSI information of a position fingerprint. The positioning principle of the fingerprint positioning algorithm is shown in fig. 1, and the positioning process is divided into two stages: firstly, an off-line training stage: selecting n points in the area as reference points to acquire the position fingerprints so as to complete the construction of a position fingerprint database, wherein the position fingerprints are composed of position coordinates and RSSI (received signal strength indicator) information at the reference points; II, an online positioning stage: and collecting RSSI information at the positioning point, matching the collected RSSI information with the position fingerprints in the position fingerprint database, and screening out the first K position fingerprints with the highest matching degree to realize position estimation. Common algorithms for location fingerprinting include nearest neighbor, K-neighbor (KNN), and weighted K-neighbor (WKNN).
The nearest neighbor algorithm carries out positioning according to the similarity between samples, and the core idea is that the coordinate of a reference point which is most similar to the RSSI vector at a positioning point is selected from a position fingerprint database to be used as the coordinate of a positioning node, wherein the similarity between the RSSI vectors is judged through the Euclidean distance between the RSSI vectors, and the smaller the Euclidean distance is, the higher the similarity between the two RSSI vectors is. Euclidean distance between RSSI vectors
Figure BDA0001536284760000011
Wherein rssiiRepresenting the signal strength, rssi, of the ith wireless AP (access node) at the positioning nodek,iThe signal strength of the ith wireless AP at the kth reference point in the position fingerprint database is shown, and m represents the number of the wireless APs.
Improvement of K-nearest neighbor algorithm as nearest neighbor algorithmIt is also positioned according to the similarity between nodes. During positioning, the K nearest neighbor algorithm selects K reference points with the highest similarity with the positioning nodes from the position fingerprint library, and calculates the centroid of a polygon formed by the K reference points, wherein the position of the centroid is the position of the positioning node:
Figure BDA0001536284760000012
wherein (x)i,yi) Representing the coordinates of the reference point.
The weighted K-nearest neighbor algorithm is an improvement on the K-nearest neighbor algorithm, and further reduces errors in the positioning process. Compared with the K-nearest neighbor algorithm, after the K reference points with the highest similarity are obtained by the weighted K-nearest neighbor algorithm, the centroid of the polygon formed by the K reference points is not directly calculated, but the coordinate positions of the reference points are weighted according to the similarity between each reference point and the positioning node, and the result of weighted addition is the final positioning result:
Figure BDA0001536284760000013
wherein d isiEuclidean distance (x) representing RSSI vector between nodesi,yi) Representing the coordinates of the reference point. The weighted K neighbor algorithm controls the influence of the reference point on the positioning point through weighting, so that the positioning result is more accurate.
However, in the existing location fingerprint positioning method, only RSSI information acquisition is simply performed to construct a location fingerprint database in the fingerprint acquisition process, and the fingerprint acquisition precision is not enough; when the position fingerprint library is constructed, the position fingerprint library is simply constructed in a manual acquisition mode, and once the density requirement of the position fingerprint library is improved, the construction workload of the position fingerprint library is greatly increased; noise points existing in the position fingerprint database are not removed in any mode, and the fingerprint accuracy in the position fingerprint database is to be further improved; meanwhile, when the position fingerprints are matched, the matching is performed one by one, the time complexity of matching processing is relatively high, in addition, the positioning error caused by equipment difference is not specially processed, and a certain positioning error is caused to the positioning result.
Disclosure of Invention
The invention aims to: aiming at the existing problems, the positioning optimization method for effectively improving the positioning accuracy and the fingerprint matching processing time complexity is provided.
The indoor positioning optimization method based on the position fingerprint algorithm comprises the following steps:
a step of constructing a position fingerprint database:
101: deploying m wireless access devices in an area to be positioned, setting n reference points (for example, performing grid division on an indoor plane graph, and taking grid points as the reference points), and recording position information of each reference point;
102: collecting the received signal strength value rssi of each reference point to each wireless access device for multiple times, and rejecting the currently collected rssi if the currently collected rssi meets the condition that | rssi-mu | is more than 1.96 sigma during each collection, wherein sigma and mu respectively represent the standard deviation and the mean value of Gaussian distribution of the received signal strength;
and carrying out mean value filtering smoothing treatment on the multiple acquisition results to obtain the received signal strength value of each reference point to each wireless access device, and recording the received signal strength value as rssik,iWherein k is a reference point identifier and i is a wireless access device identifier; forming a signal strength vector of the reference point by m groups of received signal strength values of the same reference point, and recording the signal strength vector as RSSIk＝(rssik,1,rssik,2,…,rssik,n) Combining the signal intensity vector and the position information of each reference point into a position fingerprint and storing the position fingerprint into a position fingerprint database;
103: performing density enhancement treatment on the position fingerprint database:
increasing the position fingerprints in the database in a polynomial surface fitting mode;
and then, carrying out neighborhood filtering method on the added database to screen the position fingerprint database noise points: traversing each reference point in the position fingerprint database, taking the current reference point as a central point c, and calculating the central point c and each neighborhood point x based on the P neighborhood of the central point cpAverage similarity between them
Figure BDA0001536284760000021
Wherein P is a preset neighborhood number, and subscript P is a neighborhood point xpIs determined by the point identifier of (a),
Figure BDA0001536284760000022
representing a center point c and a neighborhood point xpEuclidean distance of signal strength vectors between; and combining each neighborhood point xpThe central point c in the P neighborhood is removed to obtain each neighborhood point xpIs/are as follows
Figure BDA0001536284760000031
Each neighborhood point
Figure BDA0001536284760000032
Subscript j is the neighborhood point
Figure BDA0001536284760000033
And then calculates each neighborhood point xpAverage degree of similarity of
Figure BDA0001536284760000034
Wherein
Figure BDA0001536284760000035
Represents point xpAnd point
Figure BDA0001536284760000036
The Euclidean distance of the signal intensity vectors between the adjacent points so as to obtain the mean value of the average similarity of all the adjacent points of the central point c
Figure BDA0001536284760000037
If it is
Figure BDA0001536284760000038
Recording the current center point c as a noise point z;
updating the received signal strength value rssi of the noise point z based on the received signal strength value of the P neighborhood of each noise point zz,iComprises the following steps:
Figure BDA0001536284760000039
wherein
Figure BDA00015362847600000310
Neighborhood points representing noise point z
Figure BDA00015362847600000311
Received signal strength value for ith wireless access device, where j represents a neighborhood point
Figure BDA00015362847600000312
Is determined by the point identifier of (a),
Figure BDA00015362847600000313
representing neighborhood points
Figure BDA00015362847600000314
Weighted value of, and
Figure BDA00015362847600000315
Lz,jneighborhood points representing noise point z
Figure BDA00015362847600000316
The Euclidean distance between the reference point and two coordinate points of the noise point z is obtained, so that a new signal intensity vector of a reference point corresponding to the noise point z is obtained;
104: clustering the position fingerprint database by adopting K-means clustering, judging whether the number of clusters of each type does not exceed a cluster number threshold value, if so, storing the current type cluster, otherwise, continuing to perform the K-means clustering on the obtained type cluster until the number of clusters of the obtained type cluster does not exceed the cluster number threshold value, and thus obtaining a constructed position fingerprint database;
and (3) real-time positioning and matching:
201: collecting m groups of received signal strength values at a point to be positioned to obtain a signal strength vector RSSI of the point to be positionedt；
202: obtaining a matching result of a to-be-positioned point in a position fingerprint database by a weighted K nearest neighbor method based on the difference value:
the signal of a to-be-positioned point is strengthenedDegree vector RSSItMinus the RSSI of each component intTo obtain a difference vector delta RSSIt；
Will RSSItOr delta RSSItSimilarity matching is carried out between the cluster center and each cluster center of the stored clusters (the smaller the Euclidean distance between signal intensity vectors is, the higher the similarity is), and the cluster with the highest similarity is searched;
based on difference vector Δ RSSItSearching the first K position fingerprints with the highest similarity in the cluster with the highest similarity, and performing weighted addition on the position coordinates corresponding to the K position fingerprints by adopting a weighted K neighbor algorithm to obtain the positioning coordinates of the positioning points;
203: and carrying out longitude and latitude coordinate conversion on the positioning relative coordinates to obtain the position information of the to-be-positioned point.
In summary, due to the adoption of the technical scheme, the invention has the beneficial effects that:
(1) fingerprint collection precision promotes: in the existing method, in the fingerprint acquisition process, only RSSI information acquisition is simply carried out to construct a position fingerprint database; according to the invention, Gaussian filtering and mean filtering are carried out on the RSSI information, so that the accuracy of the acquired RSSI information is effectively improved.
(2) Enhancing the density of the position fingerprint database: when the existing method is used for constructing the position fingerprint database, the position fingerprint database is simply constructed in a manual acquisition mode, and once the density requirement of the position fingerprint database is improved, the construction workload of the position fingerprint database is greatly increased; according to the method, a polynomial surface fitting mode is adopted, and under the condition that the density of an original position fingerprint library is small, the number of fingerprints in the database is increased through a data fitting mode, so that the purpose of increasing the density of the position fingerprint library is achieved, and the positioning accuracy of a position fingerprint positioning algorithm is improved.
(3) The method has the advantages that noise points in the position fingerprint database are effectively removed, so that the fingerprint accuracy in the position fingerprint database is higher, and high-accuracy data support is provided for an online positioning stage of a positioning algorithm.
(4) The positioning efficiency is improved: in the existing method, the matching is performed one by one when the position fingerprint database is matched, and the time complexity of the matching process from the position fingerprint database is O (n); in the invention, after the position fingerprint database is subjected to clustering analysis by a K-means clustering algorithm, the time complexity of the positioning process can be reduced to O (logn) by matching the clustering center and then matching the clusters.
(5) Effectively eliminate equipment error: the algorithm of the fingerprint matching process is improved through the difference consistency of the RSSI between different devices, and the positioning error caused by the difference of the devices is effectively reduced.
Drawings
FIG. 1 is a schematic diagram of location fingerprint positioning;
FIG. 2 is a diagram of 8 neighborhood regions;
FIG. 3 is a plan view of a positioning environment;
FIG. 4 is a flow diagram of location fingerprint library construction;
FIG. 5 is a flow chart of online positioning;
fig. 6 is a diagram of RSSI information acquisition;
FIG. 7 is a K-nearest neighbor (KNN) error distribution plot;
FIG. 8 is a weighted K-nearest neighbor (WKNN) error distribution plot;
FIG. 9 is a graph of the error profile of the method of the present invention;
FIG. 10 is a comparison of positioning errors;
FIG. 11 is a graph of cumulative error distribution.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the following embodiments and accompanying drawings.
According to the invention, the positioning precision and the positioning processing efficiency of indoor positioning of the existing position fingerprint algorithm are improved through signal intensity acquisition optimization, position fingerprint library density optimization, position fingerprint library precision optimization, positioning efficiency optimization and equipment error optimization.
In the construction process of the position fingerprint database, because the acquired signal strength information has time-varying property, the method can adopt a filtering method to process the acquired RSSI information during specific implementation so as to filter error information in the RSSI information and improve the accuracy of the acquired information. The distribution of signal intensities measured at the same location conforms to a gaussian distribution model. So the RSSI vector for any reference point is (RSSI)1,rssi2,rssi3,…,rssim) Any component of it (rssi)iI-1, …, m, m representing the number of APs) conforms to a gaussian distribution model, where m represents the number of radio access points, i.e., the number of components of the RSSI vector.
In statistics, the confidence interval represents an estimation interval of the overall parameter value constructed for the sample, and the confidence of the confidence interval is determined by the confidence level. For the gaussian distribution of the signal strength (signal strength value rssi of the collected single AP) at the reference point, the relation between the confidence interval and the confidence level is: p { X1＜rssi＜X21- α, wherein X1Lower bound, X, representing confidence interval2Representing an upper bound of confidence interval, 1- α representing rssi at confidence interval (X)1，X2) The confidence level within.
The confidence interval of the signal strength at the reference point represents the estimation of the sample at the reference point on a certain interval of the overall parameter, and the selection of a reasonable confidence interval can truly reflect the variation range of the signal strength. For the signal intensity outside the range of the confidence interval, the event with the signal intensity is considered to be a small probability event, and meanwhile, for the acquisition of the signal intensity at the reference point, the samples outside the confidence interval exceed the normal error range and can seriously affect the sampling result, so the samples outside the confidence interval are rejected by the method. In the present embodiment, the upper and lower bounds of the confidence interval are set to (μ -1.96 σ, μ +1.96 σ), and samples that are not within the confidence interval range are considered to be coarse errors and are discarded. According to the boundary setting, for the signal intensity value rssi of the collected single AP, when | rssi-mu | is larger than 1.96 sigma, the signal intensity value rssi of the current collection is rejected, wherein sigma represents the standard deviation of Gaussian distribution, and mu represents the mean value of the Gaussian distribution.
Then, the filtered signal intensity is smoothed by means of mean processing, for example, the filtered signal intensity is smoothed by means of mean processing to obtain a smoothed resultFruit
Figure BDA0001536284760000051
And T is the sampling number of any reference point to the same wireless AP, and the obtained signal intensity is added to a position fingerprint database as the signal intensity of the position fingerprint at the reference point after smoothing treatment, so that the accuracy of the signal intensity is improved, and the reliability of the position fingerprint in the construction process of the position fingerprint database is guaranteed.
In the location fingerprint positioning algorithm, the greater the density of location fingerprints in the location fingerprint database is, the higher the positioning accuracy of the positioning algorithm is. The fingerprint density of position fingerprint storehouse is decided by the number of sampling points in the unit area, and the number of sampling points is more, and fingerprint density is big more, consequently, can improve the fingerprint density of position fingerprint storehouse through the number of sampling points in the increase position fingerprint storehouse to the realization is to the optimization of position fingerprint positioning algorithm, however, along with the promotion of position fingerprint density in the position fingerprint storehouse, the number of sampling points in the position fingerprint storehouse will increase in a large number, and its work load of establishing will be exponential growth.
On the basis of not increasing the workload of constructing the position fingerprint database, the invention adopts a data fitting mode to increase the fingerprint density of the position fingerprint database, fits new position fingerprint information according to the known position fingerprint information and adds the new position fingerprint information into the position fingerprint database, thereby achieving the purpose of increasing the density of the position fingerprint database. In the specific implementation, according to the signal intensity value corresponding to the fingerprint of the known position, a polynomial surface fitting mode is adopted to fit the distribution curved surface of the wireless signal intensity in the space, the signal intensity value of a new sampling point is calculated according to the fitted curved surface equation, and the coordinates and the signal intensity value of the new sampling point are combined to complete the construction of a group of new position fingerprints. A large number of new position fingerprints can be calculated through the fitted signal intensity distribution surface equation and are added into the position fingerprint library, and therefore the fingerprint density of the position fingerprint library is improved.
When the density of the position fingerprint database is improved by adopting a polynomial surface fitting mode, certain noise points are introduced into the position fingerprint database because certain difference possibly exists between a fitting result and a true value, and therefore the position fingerprint database with the improved density needs to be denoised, so that the accuracy of the position fingerprint in the position fingerprint is improved, and accurate data support is provided for a position fingerprint positioning algorithm. Because the distribution of the signal intensity in the space is continuous and similar, the method adopts a neighborhood mean value filtering algorithm to carry out denoising processing on the position fingerprint database. The core idea of the neighborhood mean filtering algorithm is to use the mean value of the points in the adjacent region of the noise point to replace the noise point, thereby achieving the purpose of removing the noise point. The neighborhood mean filtering algorithm generally involves the selection problem of neighborhoods, and commonly comprises 4 neighborhoods, 8 neighborhoods and the like, and when the position fingerprint library is processed, the 8 neighborhoods are preferably selected. When the neighborhood mean filtering algorithm is used for denoising the position fingerprint database, the judgment of noise points is the key of the denoising algorithm. According to the continuity and the proximity similarity of the wireless signal intensity distribution in the space, the invention judges the noise point according to the similarity of the position fingerprints between the proximity points. The position fingerprints between adjacent points in the space have high similarity, and the similarity between the position fingerprints is reduced when noise points exist between the adjacent points, wherein the similarity between the position fingerprints between the adjacent points is measured by the Euclidean distance between RSSI vectors, and the shorter the Euclidean distance is, the higher the similarity between the position fingerprints is.
In the neighborhood mean filtering algorithm, the rule of noise point judgment is as follows: when the average similarity between the central point and its neighboring points is greater than the average similarity between the neighboring points and their corresponding neighboring points, the central point is considered as a noise point, as shown in fig. 2, the point to be processed is the central point c, and 8 neighboring points of the central point are represented by x1～x8The corresponding points represent that the average similarity between the central point and its neighborhood points is:
Figure BDA0001536284760000061
where P represents the number of neighborhood points of c (i.e., for 8 fields, P is 8),
Figure BDA0001536284760000062
denotes c points and xiEuclidean distance of RSSI vectors between points.
The average similarity between the neighborhood point of the center point and the neighborhood point corresponding to the neighborhood point is as follows:
Figure BDA0001536284760000063
wherein the content of the first and second substances,
Figure BDA0001536284760000071
denotes xiNeighborhood points corresponding to the points
Figure BDA0001536284760000072
Number of (2), wherein the neighborhood points
Figure BDA0001536284760000073
In which the point c is not included in the graph,
Figure BDA0001536284760000074
denotes xiDot sum
Figure BDA0001536284760000075
Euclidean distance of RSSI vectors between points.
After the average similarity between all neighborhood points of the central point and the corresponding neighborhood points is obtained, carrying out mean processing on all obtained results to obtain the mean value of the average similarity of all neighborhood points of c:
Figure BDA0001536284760000076
when in use
Figure BDA0001536284760000077
And in time, considering the point c as a noise point, marking as a noise point z, and calculating a signal intensity value at the noise point by adopting an inverse distance weighted interpolation method:
Figure BDA0001536284760000078
wherein the neighborhood point weight of the noise point z
Figure BDA0001536284760000079
i＝1,…,m，Lz,jNeighborhood point x representing noise point zz,jThe Euclidean distance between two coordinate points of the noise point z is obtained, so that a new RSSI vector RSSI of the noise point is obtainedz＝(rssiz,1,rssiz,2,…,rssiz,m)。
The inverse distance weighting interpolation method reflects the influence of different neighborhood points on the central point in a weighting mode according to the position relation between the central point and the neighborhood points, thereby more accurately calculating the interpolation value of the central point. Compared with common mean processing, the inverse distance weighted interpolation method accurately controls the influence of the neighborhood point on the central interpolation point in a weighting mode, so that the calculation result is more accurate. After the position fingerprint library is subjected to denoising processing, the position fingerprint has high accuracy, and the position fingerprint library provides accurate data support for positioning the position fingerprint library in an online positioning stage.
When the location fingerprint positioning algorithm is used for online positioning, the measured RSSI vectors need to be matched with the RSSI vectors corresponding to the location fingerprints in the location fingerprint database one by one. Wherein the similarity between the RSSI vectors is measured by the euclidean distance between the RSSI vectors. The greater the position fingerprint density of the position fingerprint database is, the higher the positioning accuracy of the positioning algorithm is. Therefore, in order to improve the positioning accuracy of the location fingerprint positioning algorithm, a large amount of location fingerprint information is usually stored in the location fingerprint database, and matching the location fingerprints one by one may result in a long time spent in the positioning process, thereby affecting the positioning efficiency of the location fingerprint positioning algorithm. For a large amount of fingerprint information in the position fingerprint database, the invention introduces the idea of clustering to perform clustering division on the position fingerprint database so as to improve the matching efficiency of the position fingerprint positioning algorithm and further improve the positioning efficiency of the positioning algorithm. In an off-line training stage, a clustering algorithm divides all position fingerprints in a position fingerprint database into N clusters, and if a large amount of position fingerprint data still exist in the divided clusters, further clustering the clusters until the number of the position fingerprints in the subdivided clusters is relatively small, namely the position fingerprints do not exceed a preset number threshold; and in the online positioning stage of the position fingerprint positioning algorithm, matching the RSSI vector of the positioning point with the position fingerprint of the center of each cluster, and selecting the cluster with the highest similarity with the positioning point. And for the screened clusters, if the clusters are further clustered and divided, continuously matching the clustering centers divided by the clusters by adopting the method until the screened clusters are not further subdivided, matching the position fingerprints in the screened clusters one by one, selecting the first K position fingerprints with the highest matching degree, and weighting the corresponding coordinates to obtain a positioning result. After the clustering idea is used for optimizing the position fingerprint database, the complexity of the positioning time in the online positioning stage can be reduced from o (n) to o (logn), so that the efficiency of the positioning algorithm is remarkably improved. After the clustering idea is introduced into the location fingerprint positioning algorithm, although the clustering process of the location fingerprint database consumes a large amount of time resources and computing resources, the clustering process is completed by the server where the location fingerprint database is located in the off-line training stage, so that the clustering process does not affect the positioning efficiency in the on-line positioning stage.
In the clustering algorithm, the clustering algorithm based on the division comprises a K-means clustering algorithm, a Clarans clustering algorithm and the like. In the specific embodiment, a K-means clustering algorithm is adopted, and the core idea of clustering is to use the mean value of the K clustering division schemes to represent various corresponding samples through continuous iteration of the K clustering division schemes, so that the obtained overall error is minimum. In the location fingerprint positioning algorithm, the K-means clustering algorithm is used for classifying the location fingerprint database, so that the matching times among RSSI vectors in the positioning process can be effectively reduced, the matching process is accelerated, and the purpose of improving the positioning efficiency is achieved. After the acquisition, fitting and denoising of the position fingerprint database are completed, clustering division can be performed by using a K-means clustering algorithm:
(1) and randomly selecting K samples from the position fingerprint database as initial central points of K clusters.
(2) And calculating Euclidean distances between each sample and K cluster central points in the position fingerprint database, and dividing the samples into the cluster with the shortest Euclidean distance, wherein the Euclidean distance in the algorithm is the Euclidean distance between RSSI vectors.
(3) And recalculating the central points of the K clusters by using a mean value method according to the position fingerprint samples classified into the clusters.
(4) And (3) repeating the steps (1), (2) and (3) until iteration converges (the position of the central point is not changed or iterated to a specified number of times, and the like), thereby completing the K division of the position fingerprint database.
The position fingerprint library relates to the collection of RSSI information by using a terminal in an off-line training stage and an on-line positioning stage of a position fingerprint positioning algorithm, but the intelligent terminals used by different users in different stages may have differences, the RSSI information collected by the wireless network cards of different terminals in the same position may have great differences, and meanwhile, the antenna gains of different wireless network cards can also cause the difference of measurement results. Therefore, when different types of intelligent terminals are used for positioning, the positioning result can be changed due to different intelligent terminals, and the positioning accuracy is affected. For the technical problem, in the specific embodiment, the positioning is performed by using a weighted K nearest neighbor algorithm based on the RSSI difference, and compared with a common weighted positioning algorithm, the algorithm can eliminate positioning errors caused by differences of devices, thereby effectively improving the accuracy and reliability of the positioning algorithm.
The weighted K neighbor algorithm based on the RSSI difference is an improvement of the weighted K neighbor algorithm, and the main difference is embodied in the processing of an RSSI vector, namely, the RSSI acquired by a terminal is subjected to difference processing, and each component in the RSSI vector is subtracted by the minimum component rsi in the RSSI vectorminObtaining Δ RSSI ═ RSSI1-rssimin,rssi2-rssimin,rssi3-rssimin,…,rssin-rssimin) Because the difference values among the components in the RSSI collected by the heterogeneous terminals are kept consistent, the delta RSSI obtained by the heterogeneous terminals at the same positioning node is kept consistent, and the error caused by the difference of the terminals can be effectively eliminated by performing the position fingerprint positioning based on the delta RSSI. When the weighted K nearest neighbor algorithm based on the RSSI difference value is used for matching the position fingerprints, the delta RSSI vector is used for matching, and the positions are screened outThe first K (in the invention, the value of K in the K-means clustering algorithm has no correlation with K here, and only refers to a preset constant value) position fingerprints with the highest similarity in the fingerprint library are weighted and added by adopting a weighted K neighbor algorithm to the position coordinates corresponding to the K position fingerprints, and finally the positioning coordinates of the positioning points are obtained.
Examples
This embodiment is at a height of about 84m2The indoor environment, wherein 4 APs are deployed in the indoor environment, and 25 points are selected as reference points in the indoor environment to construct a location fingerprint database, the distance between adjacent reference points is 2m, referring to fig. 3, the deployment locations of the 4 APs are shown in fig. 3, which includes two TP _ L INK T L-WR 886N routers, one TP _ L INK T L-WR 740N router and one Tenda 811R router
In order to implement real-time positioning processing on any positioning point, a position fingerprint database is first constructed in an offline state, see fig. 4, and the specific process is as follows:
(1) and setting a position fingerprint acquisition scheme.
In this embodiment, a wireless notebook computer equipped with a network adapter is used as a signal strength acquisition terminal, and software such as WirelessMon, SPSS Statistics, MAT L AB is used for RSSI information acquisition and data processing, where WirelessMon is used for acquiring RSSI information of 4 APs, SPSS Statistics is used for completing K-means clustering of a location fingerprint library, and MAT L AB is used for performing polynomial surface fitting on a distribution surface of signal strength.
(2) And acquiring n groups of RSSI information, wherein each group comprises signal strength values of m different APs, and obtaining n groups of RSSI vectors.
RSSI information (n is the number of preset reference points) of 4 indoor AP nodes is collected by WirelessMon software, and the process of collecting the RSSI information by using the WirelessMon software is shown in figure 6, wherein the X axis represents time, and the Y axis represents signal intensity, and the unit is dBm.
(3) The coarse error is removed by gaussian filtering.
As can be seen from FIG. 6, the rssi values collected fluctuate between-60 dBm and-80 dBm, and it can be seen that the signal intensity is time-varying, so that it is subjected to Gaussian filtering: and when the acquired single rssi value meets | rssi-mu | larger than 1.96 sigma, rejecting the signal intensity value acquired this time.
(4) The mean filter smoothes the RSSI information.
And (4) performing mean value smoothing on the RSSI information obtained in the step (3).
(5) Combining the RSSI information and the location information into a location fingerprint: (x)k,yk):(rssik,1,rssik,2,…,rssik,n) (ii) a And adds it to the location fingerprint repository.
(6) And (5) repeating the steps (2) to (5) until the reference point acquisition work is finished.
(7) Data fitting increases the location fingerprint library density.
The method comprises the steps of adopting a Curve Fitting tool box of MAT L AB to carry out polynomial surface Fitting on each component in RSSI vectors in a position fingerprint library, enabling the decision coefficient (R square) of a Fitting surface to be 0.85, obtaining a surface equation of signal intensity distribution after Fitting a signal intensity distribution surface graph of each component in the RSSI vectors, calculating RSSI information corresponding to different positions in the embodiment according to the surface equation, combining coordinates of different positions with the corresponding RSSI information to fit new position fingerprint data, and completing the increase work of the number of fingerprints in the position fingerprint library by constructing and adding a plurality of groups of position fingerprint data into the position fingerprint library so as to improve the fingerprint density of the position fingerprint library.
(8) And (4) screening the position fingerprint database noise points by using a neighborhood filtering method.
Setting neighborhood mode 8 neighborhood, traversing each reference point in the position fingerprint database, taking the current reference point as a central point c, and calculating the average similarity between the central point c and the neighborhood points thereof
Figure BDA0001536284760000102
And all neighborhood points of the center point cMean of average similarity
Figure BDA0001536284760000103
If it is
Figure BDA0001536284760000104
The current center point c is marked as noise z.
(9) And updating the RSSI information of the noise point by an inverse distance weighted interpolation method.
(10) And classifying the position fingerprint database by the K mean distance so as to obtain the constructed position fingerprint database.
The method comprises the steps of carrying out K-means clustering on 79 groups of position fingerprints in a position fingerprint database by using SPSS statics as a tool, and dividing all the position fingerprints into 8 clusters, wherein the number of cluster members in each cluster is shown in table 1, and the cluster center of each cluster is shown in table 2.
TABLE 1
Clustering 1 2 3 4 5 6 7 8
Number of 18 10 7 15 9 6 6 8
TABLE 2
Figure BDA0001536284760000101
Figure BDA0001536284760000111
After the location fingerprint database is constructed, the location coordinate information of the location point can be calculated and output in real time based on the RSSI information collected by any location point, see fig. 5, which includes the following processes:
(1) and acquiring m sets of RSSI values at the point to be positioned to obtain an RSSI vector of the point to be positioned.
In this embodiment, 51 points are arbitrarily selected from the to-be-positioned area as to-be-positioned points (test points) to test the positioning performance of the invention.
Firstly, measuring relative coordinates between 51 points to be positioned and a position fingerprint database reference point by using a ranging tool to obtain real relative coordinates of the 51 points to be positioned. Then, using a WirelessMon acquisition tool to acquire the RSSI information of 51 points to be located, so as to obtain the RSSI vector of each point to be located.
(2) And smoothing the RSSI vector of the to-be-positioned point by adopting a mean filtering method.
(3) And matching the position fingerprint library by a weighted K nearest neighbor method based on the difference.
(4) And obtaining the relative position coordinates of the to-be-positioned point.
(5) And carrying out longitude and latitude coordinate conversion on the positioning relative coordinates to obtain the position information of the to-be-positioned point.
In order to verify the positioning thinking of the invention, the precision analysis of the positioning optimization scheme is completed by comparing the calculated relative position coordinates with the actually measured relative position coordinates.
(1) And (5) analyzing a positioning error distribution surface graph.
Fig. 7, 8 and 9 are error distribution surface diagrams of an unoptimized K-nearest neighbor algorithm, an unoptimized weighted K-nearest neighbor algorithm and an optimized positioning scheme of the present invention, respectively, in the present embodiment, wherein the positioning error is measured by an absolute error. In the figure, the X axis and the Y axis represent position coordinates in cm, and the Z axis represents the magnitude of the error distance in cm. Fig. 7, 8 and 9 respectively show the distribution of errors in different positioning methods, and the analysis of the three diagrams yields that: the optimized positioning scheme has small overall positioning error fluctuation, the overall error is maintained at a relatively low level, and the positioning precision of the optimized positioning scheme is better compared with the unoptimized K neighbor algorithm and the unoptimized weighted K neighbor algorithm.
(2) And analyzing the positioning error comparison graph.
Fig. 10 is a graph of error comparison between an unoptimized K-nearest neighbor algorithm, an unoptimized weighted K-nearest neighbor algorithm, and an optimized positioning scheme of the present invention, wherein the X-axis represents the number of positioning points used for testing, and the Y-axis represents the positioning errors corresponding to the positioning points. From the analysis in the figure, the maximum error of the optimized positioning scheme is smaller than that of the two non-optimized positioning algorithms. Meanwhile, mean value processing is carried out on the positioning error data of the three positioning modes to obtain that the average positioning error of the unoptimized K neighbor algorithm is 252.86 centimeters, the average positioning error of the unoptimized weighted K neighbor algorithm is 251.67 centimeters, and the average positioning error of the optimized positioning scheme is 196.57 centimeters, so that compared with the unoptimized K neighbor algorithm and the unoptimized weighted K neighbor algorithm, the positioning accuracy of the optimized positioning scheme is respectively improved by 22.26 percent and 21.89 percent.
(3) Cumulative error profile analysis
FIG. 11 is a graph of cumulative error distribution between an unoptimized K-nearest neighbor algorithm, an unoptimized weighted K-nearest neighbor algorithm, and an optimized positioning scheme of the present invention, wherein the X-axis represents the error distance and the Y-axis represents the cumulative probability corresponding to the error distance. From the analysis in the figure, in the positioning results of the optimized positioning scheme, 49.02% of the results fall within the error range of 200 cm, 88.24% of the results fall within the error range of 300 cm, and 100% of the results fall within the error range of 400 cm, compared with the optimized positioning scheme, the data of the unoptimized K-neighbor algorithm is 33.34%, 68.63% and 84.31, the data of the unoptimized weighted K-neighbor algorithm is 31.38%, 64.71% and 86.27%, and the analysis data can obtain that the overall positioning error of the optimized positioning scheme is better than that of the two unoptimized positioning algorithms.
(4) Localization efficiency analysis
In table 3, the positioning efficiency of the positioning algorithm before and after clustering and partitioning the position fingerprint database is compared, the data in the analysis table can be obtained, after clustering and partitioning the position fingerprint database, the positioning efficiency of the positioning algorithm is much higher than that before clustering and partitioning, and the advantage of the positioning efficiency is more obvious under the condition that the scale of the position fingerprint database is larger.
Figure BDA0001536284760000121
In conclusion, the invention starts from five aspects of signal intensity acquisition, position fingerprint library density, position fingerprint library precision, positioning efficiency and positioning error, realizes the optimization of a position fingerprint positioning algorithm, and completes the double optimization of the positioning precision and the positioning efficiency. The experimental result shows that the average positioning error of the optimized positioning scheme is 196.57 cm, and compared with the common K neighbor algorithm and the weighted K neighbor algorithm, the positioning precision of the optimized positioning scheme is improved by 22.26% and 21.89%; compared with a location fingerprint positioning algorithm without clustering in a location fingerprint database, the optimized positioning scheme reduces the time complexity of the positioning process from O (n) to O (logn), and has great advantages in the aspect of positioning efficiency under the condition of large scale of the location fingerprint database.
While the invention has been described with reference to specific embodiments, any feature disclosed in this specification may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise; all of the disclosed features, or all of the method or process steps, may be combined in any combination, except mutually exclusive features and/or steps.

Claims (3)

1. The indoor positioning optimization method based on the position fingerprint algorithm is characterized by comprising the following steps:
a step of constructing a position fingerprint database:
101: deploying m wireless access devices in an area to be positioned, setting n reference points, and recording position information of each reference point;
102: collecting the received signal strength value rssi of each reference point to each wireless access device for multiple times, and rejecting the currently collected rssi if the currently collected rssi meets the condition that | rssi-mu | is more than 1.96 sigma during each collection, wherein sigma and mu respectively represent the standard deviation and the mean value of Gaussian distribution of the received signal strength;
and carrying out mean value filtering smoothing treatment on the multiple acquisition results to obtain the received signal strength value of each reference point to each wireless access device, and recording the received signal strength value as rssik,iWherein k is a reference point identifier and i is a wireless access device identifier; forming a signal strength vector of the reference point by m groups of received signal strength values of the same reference point, and recording the signal strength vector as RSSIk＝(rssik,1,rssik,2,…,rssik,n) Combining the signal intensity vector and the position information of each reference point into a position fingerprint and storing the position fingerprint into a position fingerprint database;
103: performing density enhancement treatment on the position fingerprint database:
increasing the location fingerprints in the database by means of polynomial surface fitting:
after Fitting out a signal intensity distribution surface graph of each component in the signal intensity vectors, acquiring a surface equation of signal intensity distribution, calculating to obtain signal intensity vectors corresponding to different positions according to the surface equation, combining the coordinates of the different positions with the corresponding signal intensity vectors, Fitting out a new position fingerprint and adding the new position fingerprint into a position fingerprint library;
and then, carrying out neighborhood filtering method on the added database to screen the position fingerprint database noise points: traversing each reference point in the position fingerprint database, taking the current reference point as a central point c, and calculating the central point c and each neighborhood point x based on the P neighborhood of the central point cpAverage similarity between them
Figure FDA0002402719990000011
Wherein P is a preset neighborhood number, and subscript P is a neighborhood point xpIs determined by the point identifier of (a),
Figure FDA0002402719990000012
representing a center point c and a neighborhood point xpEuclidean distance of signal strength vectors between; and combining each neighborhood point xpThe central point c in the P neighborhood is removed to obtain each neighborhood point xpIs/are as follows
Figure FDA0002402719990000013
Each neighborhood point
Figure FDA0002402719990000014
Subscript j is the neighborhood point
Figure FDA0002402719990000015
And then calculates each neighborhood point xpAverage degree of similarity of
Figure FDA0002402719990000016
Wherein
Figure FDA0002402719990000017
Represents point xpAnd point
Figure FDA0002402719990000018
The Euclidean distance of the signal intensity vectors between the adjacent points so as to obtain the mean value of the average similarity of all the adjacent points of the central point c
Figure FDA0002402719990000019
If it is
Figure FDA00024027199900000110
Recording the current center point c as a noise point z;
updating the received signal strength value rssi of the noise point z based on the received signal strength value of the P neighborhood of each noise point zz,iComprises the following steps:
Figure FDA0002402719990000021
wherein
Figure FDA0002402719990000022
Neighborhood points representing noise point z
Figure FDA0002402719990000023
Received signal strength value for ith wireless access device, where j represents a neighborhood point
Figure FDA0002402719990000024
Is determined by the point identifier of (a),
Figure FDA0002402719990000025
representing neighborhood points
Figure FDA0002402719990000026
Weighted value of, and
Figure FDA0002402719990000027
Lz,jneighborhood points representing noise point z
Figure FDA0002402719990000028
The Euclidean distance between the reference point and two coordinate points of the noise point z is obtained, so that a new signal intensity vector of a reference point corresponding to the noise point z is obtained;
104: clustering the position fingerprint database by adopting K-means clustering, judging whether the number of clusters of each type does not exceed a cluster number threshold value, if so, storing the current type cluster, otherwise, continuing to perform the K-means clustering on the obtained type cluster until the number of clusters of the obtained type cluster does not exceed the cluster number threshold value, and thus obtaining a constructed position fingerprint database;
and (3) real-time positioning and matching:
201: collecting m groups of received signal strength values at a point to be positioned to obtain a signal strength vector RSSI of the point to be positionedt；
202: obtaining a matching result of a to-be-positioned point in a position fingerprint database by a weighted K nearest neighbor method based on the difference value:
the RSSI of the signal strength vector of the point to be locatedtMinus the RSSI of each component intTo obtain a difference vector delta RSSIt；
Will delta RSSItCarrying out similarity matching with each cluster center of the stored clusters, and searching the cluster with the highest similarity;
wherein the similarity match is: the smaller the Euclidean distance between the signal intensity vectors is, the higher the similarity is;
based on difference vector Δ RSSItSearching the first K position fingerprints with the highest similarity in the cluster with the highest similarity, and performing weighted addition on the position coordinates corresponding to the K position fingerprints by adopting a weighted K neighbor algorithm to obtain the positioning coordinates of the positioning points;
203: and carrying out longitude and latitude coordinate conversion on the positioning relative coordinates to obtain the position information of the to-be-positioned point.
2. The method of claim 1, wherein the neighborhood approach is preferably 8 neighborhoods.
3. The method of claim 1 or 2, wherein the reference points are: and carrying out grid division on the indoor plane graph, and taking the grid points as reference points.
CN201711495301.6A 2017-12-31 2017-12-31 Indoor positioning optimization method based on position fingerprint algorithm Active CN107948930B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711495301.6A CN107948930B (en) 2017-12-31 2017-12-31 Indoor positioning optimization method based on position fingerprint algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711495301.6A CN107948930B (en) 2017-12-31 2017-12-31 Indoor positioning optimization method based on position fingerprint algorithm

Publications (2)

Publication Number Publication Date
CN107948930A CN107948930A (en) 2018-04-20
CN107948930B true CN107948930B (en) 2020-07-17

Family

ID=61938193

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711495301.6A Active CN107948930B (en) 2017-12-31 2017-12-31 Indoor positioning optimization method based on position fingerprint algorithm

Country Status (1)

Country Link
CN (1) CN107948930B (en)

Families Citing this family (27)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108596271B (en) * 2018-05-09 2023-05-26 中国平安人寿保险股份有限公司 Evaluation method and device of fingerprint construction algorithm, storage medium and terminal
CN108650626A (en) * 2018-05-18 2018-10-12 华南师范大学 A kind of fingerprinting localization algorithm based on Thiessen polygon
CN109001674B (en) * 2018-05-31 2022-09-02 中国矿业大学 WiFi fingerprint information rapid acquisition and positioning method based on continuous video sequence
CN109059919B (en) * 2018-06-08 2020-05-19 华中科技大学 Indoor positioning method based on crowdsourcing sample weighted surface fitting
WO2020022953A1 (en) * 2018-07-26 2020-01-30 Singapore Telecommunications Limited System and method for identifying an internet of things (iot) device based on a distributed fingerprinting solution
CN108983254A (en) * 2018-08-08 2018-12-11 中国科学院电子学研究所 Two-dimentional range unit and distance measuring method based on light stream sensor
CN109116343B (en) * 2018-09-11 2022-06-03 中北大学 Filtering method for mobile terminal receiving signal intensity
CN109587627B (en) * 2018-12-12 2021-04-06 嘉兴学院 Indoor positioning method for improving terminal heterogeneity problem based on RSSI
CN109739830B (en) * 2019-02-28 2021-01-26 电子科技大学 Position fingerprint database rapid construction method based on crowdsourcing data
CN111654843B (en) * 2019-03-04 2024-04-30 深圳光启空间技术有限公司 Method and system for automatically updating fingerprint database, wifi positioning method and system
CN110166930A (en) * 2019-04-03 2019-08-23 华中科技大学 A kind of indoor orientation method and system based on WiFi signal
CN109738863A (en) * 2019-04-08 2019-05-10 江西师范大学 A kind of WiFi fingerprint indoor positioning algorithms and system merging confidence level
CN110049447B (en) * 2019-04-12 2021-02-05 桂林电子科技大学 Position information-based partnership analysis method
CN110300372A (en) * 2019-07-11 2019-10-01 桂林电子科技大学 A kind of WIFI indoor orientation method based on location fingerprint
CN110381580B (en) * 2019-07-25 2021-03-26 上海开域信息科技有限公司 WiFi positioning method based on proportion optimization
CN111198365A (en) * 2020-01-16 2020-05-26 东方红卫星移动通信有限公司 Indoor positioning method based on radio frequency signal
CN111405461B (en) * 2020-03-16 2021-10-08 南京邮电大学 Wireless indoor positioning method for optimizing equal-interval fingerprint sampling number
CN111953937B (en) * 2020-07-31 2022-11-08 云洲(盐城)创新科技有限公司 Drowning person lifesaving system and drowning person lifesaving method
CN112040397B (en) * 2020-08-13 2023-01-24 西北师范大学 CSI indoor fingerprint positioning method based on adaptive Kalman filtering
CN112055308B (en) * 2020-08-21 2024-02-27 中通服咨询设计研究院有限公司 Multi-layer high-robustness fingerprint positioning method
CN112261578B (en) * 2020-10-21 2023-04-14 南京工业大学 Indoor fingerprint positioning method based on mode filtering
CN113079466B (en) * 2020-10-21 2022-04-26 中移(上海)信息通信科技有限公司 Fingerprint database construction method, device, equipment and computer storage medium
CN112437485B (en) * 2020-10-29 2021-11-12 北京邮电大学 Positioning method and device of fingerprint space interpolation method based on neural network
CN112511985A (en) * 2020-12-21 2021-03-16 迪爱斯信息技术股份有限公司 Alarm position positioning method, system, computer equipment and storage medium
CN113075648B (en) * 2021-03-19 2024-05-17 中国舰船研究设计中心 Clustering and filtering method for unmanned cluster target positioning information
CN114630274B (en) * 2022-03-31 2023-03-14 大连理工大学 Precision estimation method for fingerprint positioning
CN114845388B (en) * 2022-05-17 2023-02-28 电子科技大学 Indoor positioning method for position fingerprint of sub-direction entropy weighting WKNN

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105960021A (en) * 2016-07-07 2016-09-21 济南东朔微电子有限公司 Improved position fingerprint indoor positioning method
CN106093852A (en) * 2016-05-27 2016-11-09 东华大学 A kind of method improving WiFi fingerprint location precision and efficiency
CN106646338A (en) * 2016-12-07 2017-05-10 华南理工大学 Rapidly accurate indoor location method

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7433693B2 (en) * 2004-10-27 2008-10-07 Qualcomm Incorporated Location-sensitive calibration data
KR101901039B1 (en) * 2013-12-26 2018-09-20 인텔 코포레이션 Method and apparatus for cross device automatic calibration

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106093852A (en) * 2016-05-27 2016-11-09 东华大学 A kind of method improving WiFi fingerprint location precision and efficiency
CN105960021A (en) * 2016-07-07 2016-09-21 济南东朔微电子有限公司 Improved position fingerprint indoor positioning method
CN106646338A (en) * 2016-12-07 2017-05-10 华南理工大学 Rapidly accurate indoor location method

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
基于WiFi位置指纹的室内定位算法的研究与优化;毛勤;《万方学术》;20161110;第2.2.2节和第3.1节 *
基于克里金空间插值的位置指纹数据库建立算法;刘志建 等;《计算机应用研究》;20161031;第33卷(第10期);摘要 *
基于边界过滤和邻域均值滤波的室内定位算法;郭昕刚 等;《长春工业大学学报》;20171031;第38卷(第5期);第1.2节,第2节和第3节 *

Also Published As

Publication number Publication date
CN107948930A (en) 2018-04-20

Similar Documents

Publication Publication Date Title
CN107948930B (en) Indoor positioning optimization method based on position fingerprint algorithm
CN106646338B (en) A kind of quickly accurate indoor orientation method
CN105223546B (en) Indoor orientation method based on received signal strength and reference point locations double focusing class
CN112147573A (en) Passive positioning method based on amplitude and phase information of CSI (channel State information)
CN112135248B (en) WIFI fingerprint positioning method based on K-means optimal estimation
CN109189876B (en) Data processing method and device
CN107677989B (en) A kind of indoor location localization method carrying out RSSI removal noise based on RSSI maximum value
CN110728322A (en) Data classification method and related equipment
CN112203324B (en) MR positioning method and device based on position fingerprint database
CN105657653B (en) Indoor positioning method based on fingerprint data compression
CN113079466B (en) Fingerprint database construction method, device, equipment and computer storage medium
CN111669702A (en) Coarse and fine matching positioning method based on WIFI address and field intensity
CN111767273B (en) Data intelligent detection method and device based on improved SOM algorithm
CN116980824B (en) Lightweight weighting integrated learning indoor CSI positioning method
CN111639712A (en) Positioning method and system based on density peak clustering and gradient lifting algorithm
CN108445443B (en) KNN-based fingerprint point clustering method
CN116842818A (en) Structural plane attitude grouping method based on pelican optimization algorithm
CN106658533B (en) Airspace flow characteristic extraction method considering user distribution
CN108462940A (en) A kind of indoor orientation method based on Wi-Fi hotspot
Tao et al. Wi-Fi fingerprint database refinement method and performance analysis
CN114092545A (en) Self-adaptive grid searching method suitable for spherical target fitting
CN112399366A (en) Indoor positioning method based on Hankel matrix and WKNN variance extraction
CN114501520A (en) Network anomaly detection method and device and network equipment
CN114646313A (en) User track positioning method, electronic equipment and computer storage medium
CN111629321A (en) Median Kalman filtering positioning method based on outlier stripping

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