CN107087256A - A kind of fingerprint cluster method and device based on WiFi indoor positionings - Google Patents
A kind of fingerprint cluster method and device based on WiFi indoor positionings Download PDFInfo
- Publication number
- CN107087256A CN107087256A CN201710160778.2A CN201710160778A CN107087256A CN 107087256 A CN107087256 A CN 107087256A CN 201710160778 A CN201710160778 A CN 201710160778A CN 107087256 A CN107087256 A CN 107087256A
- Authority
- CN
- China
- Prior art keywords
- msub
- mrow
- point
- cluster
- signal intensity
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
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
-
- 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
- G01S5/14—Determining absolute distances from a plurality of spaced points of known location
-
- 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)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The invention provides the fingerprint cluster method and device based on WiFi indoor positionings, wherein, this method includes:S1 stochastical samplings in the target area;S2 sets up the signal intensity profile model of each point according to the Gaussian process of sampled point;S3 selectes multiple sample points in signal intensity profile model, and regard selected sample point as central point;S4 calculates the Euclidean distance between remaining sample point and each central point in signal intensity profile model, and remaining sample point is sorted out to the cluster of the central point nearest apart from oneself respectively;S5 recalculates the central point of each cluster, judges to calculate whether the distance between obtained central point and the central point of selecting are less than predetermined threshold value;S6 completes the fingerprint cluster of the signal intensity profile model for the WAP.It covers the whole plane that fingerprint expands to target area by one-dimensional sample path, and fingerprint is sorted out using clustering algorithm, is born with the fingerprint library searching that this reduces real-time positioning stage.
Description
Technical field
The present invention relates to field of locating technology, more particularly to a kind of fingerprint cluster method and device.
Background technology
With the fast development of development of Mobile Internet technology, location Based service demand therewith more and more extensively, permeate in
The various aspects such as military, business, life.At present, for positioning, people it is most-often used be global positioning system (Global
Positioning System, GPS), it gives user by the signal time difference between multi-satellite and the GPS of user
Positioning, orientation range can be covering the whole world, but satellite-signal is easily blocked by various barriers, i.e., location technology is not particularly suited for
Indoor or built-up occasion, need to carry out indoor positioning using other technological means.
Indoor positioning technologies based on WiFi (Wireless Fidelity, Wireless Fidelity) fingerprint base are obtained in recent years
Extensive concern, the technology mainly by multiple by dividing in advance and collecting multiple nothings in regions with positional information
Received signal strength (Received Signal Strength, RSS) value of line Network Access Point (Access Point, AP),
Set up the indoor positioning fingerprint base of a RSS value corresponding relation being collected into comprising each location point and the location point, with this,
When being positioned to mobile device, it can be matched by finding the RSS values being collected into current mobile device in fingerprint base
Location point, realize positioning.
At present for the research of the indoor locating system based on WiFi, the foundation and implementation of WiFi fingerprint bases are concentrated mainly on
Position two aspects.Wherein, WiFi fingerprint bases are the set of all WiFi fingerprints collected, although fingerprint base is all based on adopting
The fingerprint that sample is obtained, but different processing modes can obtain different types of fingerprint base, the locating effect of realization is different, and it is determined
Locating effect of the WiFi fingerprint location systems in implementation process is determined, so WiFi fingerprint bases is accurate with being reliably very heavy
Want.For fingerprint negligible amounts, more scattered etc. technical problem in WiFi fingerprint bases, although can use and set up continuous model
Mode solve, but in continuous model, typically one position coordinates one signal strength values of correspondence, the number of fingerprint
Amount is excessively huge, undoubtedly can bring larger burden to the retrieval of real-time positioning stage fingerprint base, influences the efficiency positioned in real time.
The content of the invention
In view of the above-mentioned problems, the invention provides a kind of fingerprint cluster method and device based on WiFi indoor positionings, having
Effect solves in the prior art, to set up after continuous model for the WiFi fingerprints of collection, the excessive technology of positioning stage retrieval burden
Problem.
The technical scheme that the present invention is provided is as follows:
A kind of fingerprint cluster method based on WiFi indoor positionings, including:
S1 is in the target area for the selected multiple sampled points of WAP stochastical sampling;
S2 sets up the RSS distributed models of each point according to the Gaussian process of sampled point;
S3 selectes multiple sample points in the RSS distributed models, and regard selected sample point as central point;
S4 calculates the Euclidean distance between remaining sample point and each central point in RSS distributed models, and then by remaining sample
Point is sorted out to the cluster of the central point nearest apart from oneself respectively;
S5 recalculates the central point of each cluster, and judges to calculate obtained central point and selected central point in same cluster
The distance between whether be less than predetermined threshold value, if so, jumping to step S6, otherwise, jump to step S4;
S6 completes the fingerprint cluster of the RSS distributed models for the WAP.
It is further preferred that in step s 4, square Europe in RSS distributed models between remaining sample point and each central point
Formula is apart from V:
Wherein, uiRepresent i-th of central point, SiRepresent the set of remaining sample point in RSS distributed models, xjRepresent RSS points
The coordinate of remaining sample point in cloth model;
And/or, in step s 5, the central point for recalculating each cluster is:
Wherein, miRepresent the quantity for the sample point that i-th of cluster includes, SiRepresent remaining sample point in RSS distributed models
Set, xjRepresent the coordinate of remaining sample point in RSS distributed models.
It is further preferred that specifically including in step s 2:
S21 obtains the degree of correlation between each point signal strength values according to the degree of correlation between sampled point;
S22 obtains RSS distributed models according to the degree of correlation between each point signal strength values;
S23 obtains the corresponding signal strength values in any point in target area according to RSS distributed models, completes this and wirelessly connects
The foundation of access point fingerprint base in the target area.
It is further preferred that in the step s 21:
Degree of correlation k (x between sampled pointi,yj) be:
Degree of correlation cov (x between each point signal strength valuesi,yj) be:
Wherein,Represent the sample variance of sampled point;L represents signal strength space correlation yardstick variable;Represent
Gaussian noise;As i=j, δ (xi,yj) value be 1, be otherwise 0;
In step S22:
According to the degree of correlation between the signal strength values obtained in step S21, each point signal is obtained for sampling dot matrix X
Intensity matrix Y degree of correlation cov (Y):
Wherein, K represents covariance matrix K [i, j]=k (x according to the sampling dot matrix X n × b tried to achievei,yj);
Obtain RSS distributed models:
It is further preferred that in step S23, any point x in target area*Signal strength values y*For:
Wherein,k*For point x*
With sample dot matrix X covariance matrix, wherein, value k (i)=k (x in the covariance matrix*,xi)。
Present invention also offers a kind of fingerprint cluster device based on WiFi indoor positionings, including:
Information receiving module, for receiving multiple sampled points for selected wireless access point sampling;
Model building module, the Gaussian process of the sampled point for being received according to information receiving module sets up each point
RSS distributed models;
Sample point selection module, for selecting multiple sample points in the RSS distributed models set up from model building module, and
It regard selected sample point as central point;
Computing module, for calculating the central point that remaining sample point is selected with sample point selection module in RSS distributed models
Between Euclidean distance, and for recalculating the central point of each cluster according to the categorization results of classifying module;
Classifying module, sorts out remaining sample point in oneself is nearest for the result of calculation according to computing module
The cluster of heart point;
Judge module, for judging that calculating module calculates obtained central point in same cluster and sample point selection module is selected
The distance between central point whether be less than predetermined threshold value.
It is further preferred that in computing module:
A square Euclidean distance in RSS distributed models between remaining sample point and each central point is:
Wherein, uiRepresent i-th of central point, SiRepresent the set of remaining sample point in RSS distributed models, xjRepresent RSS points
The coordinate of remaining sample point in cloth model;
And/or, it is according to the central point that the categorization results of classifying module recalculate each cluster:
Wherein, miRepresent the quantity for the sample point that i-th of cluster includes, SiRepresent remaining sample point in RSS distributed models
Set, xjRepresent the coordinate of remaining sample point in RSS distributed models.
It is further preferred that the degree of correlation that computing module is additionally operable between the sampled point that is received according to information receiving module is obtained
The degree of correlation between each point signal strength values, the RSS distributed models for being set up according to model building module, which are calculated, obtains mesh
Mark the corresponding signal strength values in any point in region;
Model building module is used to obtain RSS distributed models according to the degree of correlation between each point signal strength values.
It is further preferred that in computing module:
Degree of correlation k (x between sampled pointi,yj) be:
Degree of correlation cov (x between each point signal strength valuesi,yj) be:
Wherein,Represent the sample variance of sampled point;L represents signal strength space correlation yardstick variable;Represent high
This noise;As i=j, δ (xi,yj) value be 1, be otherwise 0;
In model building module:
The degree of correlation between obtained signal strength values is calculated according to computing module, each point is obtained for sampling dot matrix X
Signal intensity matrix Y degree of correlation cov (Y):
Wherein, K represents covariance matrix K [i, j]=k (x according to the sampling dot matrix X n × b tried to achievei,yj);
And then obtain RSS distributed models:
It is further preferred that in computing module:
The RSS distributed models set up according to model building module, any point in target area is calculated according to below equation
x*Signal strength values y*:
Wherein,k*For point x*
With sample dot matrix X covariance matrix, wherein, value k (i)=k (x in the covariance matrix*,xi)。
In the present invention, stochastical sampling is carried out to each indoor WAP, obtains the WiFi fingerprints of limited quantity, and
The RSS distributed models of each point are set up according to the Gaussian process of sampled point, the signal strength values of each point in target area are calculated;With
The sample point in the RSS distributed models of foundation is classified using clustering algorithm afterwards, generation is based on needed for probabilistic model algorithm
Fingerprint base.
Its WiFi fingerprint for obtaining continuous whole plane according to discrete WiFi fingerprints distribution is distributed, especially for random
The region being not directed in sampling process, can equally be covered by this method;In addition, for WiFi fingerprint collectings initial stages,
In the situation of fingerprint collecting negligible amounts, also signal strength model can be set up according to this method, obtain any in target area
The corresponding signal strength values in position (fingerprint value), the two dimension that WiFi fingerprints are expanded into target area by one-dimensional sample path is complete flat
Face covering helps to position there is provided more WiFi fingerprints quantity, and WiFi fingerprints are sorted out using clustering algorithm, carries significantly
While rising the precision of interior WiFi positioning, the fingerprint library searching burden of real-time positioning stage is reduced.
Brief description of the drawings
Below by clearly understandable mode, preferred embodiment is described with reference to the drawings, to above-mentioned characteristic, technical characteristic,
Advantage and its implementation are further described.
Fig. 1 is a kind of embodiment schematic flow sheet of fingerprint cluster method based on WiFi indoor positionings in the present invention;
Fig. 2 is the fingerprint cluster method another embodiment schematic flow sheet based on WiFi indoor positionings in the present invention;
Fig. 3 is the sampling point distributions in an example of the invention for a certain WAP;
The RSS distributed models that Fig. 4 sets up for the present invention according to sampled point as shown in Figure 3;
Fig. 5 is the fingerprint base obtained using fingerprint cluster algorithm in an example of the invention;
Fig. 6 is the fingerprint cluster schematic device based on WiFi indoor positionings in the present invention;
Reference:
100- fingerprint cluster devices, 110- information receiving modules, 120- model building modules, 130- sample point selection moulds
Block, 140- computing modules, 150- classifying modules, 160- judge modules.
Embodiment
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, control is illustrated below
The embodiment of the present invention.It should be evident that drawings in the following description are only some embodiments of the present invention, for
For those of ordinary skill in the art, on the premise of not paying creative work, other can also be obtained according to these accompanying drawings
Accompanying drawing, and obtain other embodiments.
Barogram is widely used in meteorological research, and it can be obtained by arbitrary target position by setting up barogram
Atmospheric pressure value, it is known that correspondingly atmospheric pressure value is also higher in the high place of latitude, apart from the nearlyer RSS of the distance of WAP
Value is also bigger;Temperature and humidity change, and the factor such as air motion can cause the multipath under the fluctuation of air pressure, indoor scene to imitate
Should, the factor such as non-line-of-sight propagation also result in the signal fluctuation phenomenon for not deferring to propagation model.Understand, similar to atmospheric pressure value
Consecutive variations, RSS values are also consecutive variations indoors, i.e., when the RSS values of target area position are undergone mutation, its neighbouring position
The change of same trend can be also presented in the RSS values put.
With this, if the problems such as the WiFi Fingerprint separations existed for current WiFi indoor positionings, negligible amounts, in target area
The RSS distributed models of similar " barogram " can be obtained in domain, it is possible to obtain the approximate RSS of target area optional position to
Amount, although can solve the problem that the very few problem of fingerprint number of samples in current WiFi indoor positionings in this way, but RSS is distributed
Model is a kind of continuous model, if target area scope is very big, and correspondingly fingerprint quantity also can be very big, undoubtedly can be to fixed in real time
Fingerprint library searching brings burden during position.
Based on this, a kind of fingerprint cluster method based on WiFi indoor positionings is provided in the present invention, it is assumed that interior includes
Multiple WAPs, gather WiFi fingerprints, and be directed to for each WAP one by one in the fingerprint base method for building up
The WiFi fingerprints collected obtain RSS distributed models according to Gaussian process, complete building for fingerprint base;Afterwards, cluster is respectively adopted
Algorithm RSS distributed models corresponding to each WAP are clustered.
It is as shown in Figure 1 a kind of embodiment schematic flow sheet of the fingerprint cluster method, it can be seen that referring at this
Line clustering method includes:S1 is in the target area for the selected multiple sampled points of WAP stochastical sampling;S2 according to
The Gaussian process of sampled point sets up the signal intensity profile model of each point;S3 selectes multiple samples in signal intensity profile model
Point, and it regard selected sample point as central point;S4 calculates the Europe between remaining sample point and each central point in RSS distributed models
Family name's distance, and then remaining sample point is sorted out to the cluster of the central point nearest apart from oneself respectively;S5 recalculates each cluster
Central point, and judge to calculate whether the distance between obtained central point and the central point of selecting are less than default threshold in same cluster
Value, if so, jumping to step S6, otherwise, jumps to step S4;S6 completes the signal intensity profile mould for the WAP
The fingerprint cluster of type.
Above-mentioned embodiment is improved and obtains present embodiment, as shown in Fig. 2 in the present embodiment, the fingerprint
Storehouse method for building up includes:S1 is in the target area for the selected multiple sampled points of WAP stochastical sampling;S21 according to
The degree of correlation between sampled point obtains the degree of correlation between each point signal strength values;S22 is according between each point signal strength values
The degree of correlation obtains signal intensity profile model;S23 obtains any point correspondence in target area according to signal intensity profile model
Signal strength values, complete the foundation of WAP fingerprint base in the target area;S3 is in signal intensity profile model
Multiple sample points are selected, and regard selected sample point as central point;S4 calculate in RSS distributed models remaining sample point with it is each in
Euclidean distance between heart point, and then remaining sample point is sorted out to the cluster of the central point nearest apart from oneself respectively;S5 is again
The central point of each cluster is calculated, and judges the distance between obtained central point and the central point of selecting are calculated in same cluster whether
Less than predetermined threshold value, if so, jumping to step S6, otherwise, step S4 is jumped to;S6 completes the signal for the WAP
The fingerprint cluster of intensity distribution model.
Assuming that Gaussian process has one group of training sample for including noise:
D={ (x1,y1),(x2,y2),...,(xn,yn)}
Then:
yi=f (xi)+ε
Wherein, xiRepresent d dimension spacesIn independent variable sample value, yiDesired value/observed value is represented, noise ε represents equal
Value is that 0, variance isGaussian Profile.
Based on this, in the above-described embodiment, n input is represented from scalar x with n × d matrix X, n observed value
Represented with the matrix Y of n × 1.Specifically, in RSS distributed models, X represents n sampling point position coordinate of target area, and Y is represented
Selected WAP is in the corresponding signal strength values of these sampled points (RSS values).
Gaussian process estimates Posterior distrbutionp by function f and training sample D, its core be between different observed values be
Related, i.e. f (xi) and f (yi) covariance depend on xiWith yiCovariance.In view of Gaussian noise, between sampled point
Covariance function k (xi,yj) be specially:
Degree of correlation cov (x between each point signal strength valuesi,yj) be:
Wherein,Represent the sample variance of sampled point;L represents signal strength space correlation yardstick variable, to determine
Signal intensity correlation is smaller between the speed that correlation rises or falls between 2 points, the bigger expression sampled equidistant points of l;Represent Gaussian noise;As i=j, δ (xi,yj) value be 1, be otherwise 0.
With this, for input position coordinate sample matrix X (sampling dot matrix), each position coordinate points signal intensity square is obtained
Battle array Y degree of correlation cov (Y) be:
Wherein, K represents covariance matrix K [i, j]=k (x according to the sampling dot matrix X n × b tried to achievei,yj).It can send out
Existing signal strength values are Gaussian Profile:RSS distributed models are obtained with this.
As can be seen that whenever there is one group of sampling dot matrix X input from above formula, it is possible to utilize covariance matrix K descriptions
Sampled point is as the degree of correlation between coordinate and signal strength values, with this, for selected WAP, appoints in target area
Anticipate a point x*Signal strength values y*For:
Wherein,k*For point x*
With sample dot matrix X covariance matrix, wherein, value k (i)=k (x in the covariance matrix*,xi)。
In an example, the original sample point of a certain wireless aps in target area is illustrated in figure 3, based on foregoing description
The RSS distributed models set up of fingerprint base method for building up as shown in figure 4, specific, in figs. 3 and 4, transverse and longitudinal coordinate x/y is represented
The coordinate position (two-dimensional coordinate) of sampled point, z-axis represents the corresponding signal strength values of sampled point.
Because RSS distributed models are a kind of continuous models, during in order to mitigate in position fixing process fingerprint library searching burden,
In above-mentioned embodiment, further using clustering method to including sample point cluster.It is known that cluster
(Clustering) refer to lay down a criterion, the basis classified as some object set, these obtained classes are claimed
For cluster (Cluster).It is similar with the object in cluster, the object in different clusters has certain difference under the standard of formulation
The opposite sex.That is, cluster is for the object with some common ground is gathered for a class, larger object is tried one's best with otherness
Separation.Clustering is a kind of statistical analysis technique, applied to pattern-recognition, the market segmentation, Biology seed coating analysis, data space
The various fields such as distribution.
In clustering, in order to describe the degree of closeness between target, or distinctiveness ratio between target is, it is necessary to fixed
Some adopted class statistic amounts are as the quantizating index of clustering, and Euclidean distance is the one kind for being usually utilized to represent distinctiveness ratio
Index.Specifically, for two p dimensional vectors x in spacei=(xi1,xi2,...,xip)TAnd xj=(xj1,xj2,...,xjp)T, then
Euclidean distance between two vectors can be expressed as:
Based on this, in the above-described embodiment, by above-mentioned xiAnd yiThe two-dimensional coordinate of sampling is replaced with, it is poly- using k-means
Class algorithm carries out classification according to the position coordinates of sample point can complete the cluster to sample point in signal intensity profile model, have
Body:
K (K are selected in N number of sample point first in the RSS distributed models of a certain WAP in target area<N it is) individual to make
Centered on point;For remaining sample point in RSS distributed models, its Euclidean distance for arriving each central point is measured, and then by residue
Sample point is sorted out to the cluster of the central point nearest apart from oneself respectively.Specifically, in RSS distributed models remaining sample point with it is each in
A square Euclidean distance V between heart point is:
Wherein, uiRepresent i-th of central point, SiRepresent the set of remaining sample point in RSS distributed models, xjRepresent RSS points
The coordinate of remaining sample point in cloth model.
Sort out after completing, the central point u of each cluster is recalculated according to following formulai:
Wherein, miRepresent the quantity for the sample point that i-th of cluster includes, SiRepresent remaining sample point in RSS distributed models
Set, xjRepresent the coordinate of remaining sample point in RSS distributed models.
Afterwards according to calculate obtained central point judge to calculate in same cluster obtained central point and the central point selected it
Between distance whether be less than predetermined threshold value, if so, complete for the WAP RSS distributed models fingerprint cluster;It is no
Then, remaining sample point in RSS distributed models is remeasured, to the Euclidean distance of new central point, remaining sample point to be sorted out respectively
To the cluster of the central point (central point that calculating obtain) nearest apart from oneself, and the central point of each cluster is recalculated, until same
The distance between obtained central point and the central point of selecting are calculated in cluster and is less than predetermined threshold value uthres, i.e.,
Classification terminates, wherein, uiRepresent i-th of central point, ui-1Represent the i-th -1 central point.It is noted that we are to pre- here
If threshold value uthresOccurrence do not limit, the factor such as number of times of iteration is related in its density to sampled point, cluster process, root
Set according to actual conditions, such as can be set as 5cm (centimetre), 10cm even more many.
Sample point in RSS distributed models is classified as by K cluster by k-means algorithms with this, for each cluster, by this
All sample point coordinate unifications are an identical coordinate in individual cluster, and the coordinate is specially the coordinate of the central point of the cluster.
That is, after cluster, each cluster only correspond to a position coordinates (centre coordinate of cluster), and position coordinates correspondence
The corresponding RSS vectors set of all sample points in the cluster.
After cluster is completed, the fingerprint base positioned in real time based on probabilistic model algorithm is obtained, in an example, if base
Fingerprint base is obtained in above-mentioned clustering algorithm as shown in figure 5, wherein, including positional information 1 and positional information 2, for positional information 1
For, the probability that correspondence RSS averages 1 are -90dBm (decibel milli X) in MAC1 (correspondence one WAP) be probability 11,
The probability that the probability that RSS averages 1 are -89dBm is probability 12, RSS averages 1 are -88dBm is probability 13;Correspondence RSS is equal in MAC2
The probability that the probability that value 2 is -90dBm is probability 21, RSS averages 2 are -89dBm is that probability 22, RSS averages 2 are the general of -88dBm
Rate is probability 23;The probability that correspondence RSS averages 3 are -90dBm in MAC3 is probability 31, the probability that RSS averages 3 are -89dBm is
Probability 32, the probability that RSS averages 3 are -88dBm are probability 33, by that analogy, are entered in real-time positioning stage based on the probabilistic model
Row positioning.
The fingerprint cluster schematic device based on WiFi indoor positionings that the present invention is provided is illustrated in figure 6, can from figure
To find out, include in the fingerprint cluster device 100:Information receiving module 110, model building module 120, sample point selection mould
Block 130, computing module 140, classifying module 150 and judge module 160, wherein, model building module 120 receives mould with information
Block 110 is connected, and sample point selection module 130 is connected with model building module 120, computing module 140 respectively with sample point selection
Module 130, model building module 120 and classifying module 150 are connected, judge module 160 respectively with sample point selection module 130
Connected with computing module 140.
In the course of the work, have selected one do not set up the WAP of fingerprint base after, in target area for selected
The multiple sampled points of WAP stochastical sampling, and each sampled point is sent into fingerprint cluster device 100.Fingerprint cluster is filled
The information receiving module 110 in 100 is put after each sampled point is received, model building module 120 is according to the Gauss of sampled point
Process sets up the RSS distributed models of each point;Afterwards, sample point selection module 130 selectes multiple samples in RSS distributed models
Point, and it regard selected sample point as central point;Afterwards, computing module 140 calculate in RSS distributed models remaining sample point with it is each
Euclidean distance between central point, and then classifying module 150 sorts out remaining sample point to the center nearest apart from oneself respectively
The cluster of point;Afterwards, computing module 140 recalculates the central point of each cluster, and passes through, if so, then completing wirelessly to connect for this
The fingerprint cluster of the RSS distributed models of access point otherwise, computing module 140 recalculate in RSS distributed models remaining sample point with
Euclidean distance between new central point, is reclassified by classifying module 150, recalculates new cluster again afterwards
Central point, until judge module 160 judge to calculate in same cluster obtained central point and the preceding central point once calculated it
Between distance be less than predetermined threshold value.
For further, above-mentioned computing module 140 is additionally operable between the sampled point according to the reception of information receiving module 110
The degree of correlation obtain the degree of correlation between each point signal strength values, the signal intensity profile set up according to model building module 120
Model calculates and obtains the corresponding signal strength values in any point in target area;Model building module 120 is strong according to each point signal
The degree of correlation between angle value obtains signal intensity profile model.
Assuming that Gaussian process has one group of training sample for including noise:
D={ (x1,y1),(x2,y2),...,(xn,yn)}
Then:
yi=f (xi)+ε
Wherein, xiRepresent d dimension spacesIn independent variable sample value, yiDesired value/observed value is represented, noise ε represents equal
Value is that 0, variance isGaussian Profile.
Based on this, in the present embodiment, n input is represented from scalar x with n × d matrix X, n observed value is used
The matrix Y of n × 1 is represented.Specifically, in RSS distributed models, X represents n sampling point position coordinate of target area, and Y represents choosing
Fixed WAP is in the corresponding signal strength values of these sampled points (RSS values).
Gaussian process estimates Posterior distrbutionp by function f and training sample D, its core be between different observed values be
Related, i.e. f (xi) and f (yi) covariance depend on xiWith yiCovariance.In view of Gaussian noise, between sampled point
Covariance function k (xi,yj) be specially:
Degree of correlation cov (x between each point signal strength valuesi,yj) be:
Wherein,Represent the sample variance of sampled point;L represents signal strength space correlation yardstick variable, to determine
Signal intensity correlation is smaller between the speed that correlation rises or falls between 2 points, the bigger expression sampled equidistant points of l;Represent Gaussian noise;As i=j, δ (xi,yj) value be 1, be otherwise 0.
With this, for input position coordinate sample matrix X (sampling dot matrix), each position coordinate points signal intensity square is obtained
Battle array Y degree of correlation cov (Y) be:
Wherein, K represents covariance matrix K [i, j]=k (x according to the sampling dot matrix X n × b tried to achievei,yj).It can send out
Existing signal strength values are Gaussian Profile:RSS distributed models are obtained with this.
As can be seen that whenever there is one group of sampling dot matrix X input from above formula, it is possible to utilize covariance matrix K descriptions
Sampled point is as the degree of correlation between coordinate and signal strength values, with this, for selected WAP, appoints in target area
Anticipate a point x*Signal strength values y*For:
Wherein,k*For point x*
With sample dot matrix X covariance matrix, wherein, value k (i)=k (x in the covariance matrix*,xi)。
Because RSS distributed models are a kind of continuous models, during in order to mitigate in position fixing process fingerprint library searching burden,
In present embodiment, further using clustering method to including sample point cluster, and use Euclidean distance conduct
Measurement index.Specifically, for two p dimensional vectors x in spacei=(xi1,xi2,...,xip)TAnd xj=(xj1,xj2,...,xjp
)T, then the Euclidean distance between two vectors can be expressed as:
Based on this, in the present embodiment, by above-mentioned xiAnd yiThe two-dimensional coordinate of sampling is replaced with, is clustered using k-means
Algorithm carries out classification according to the position coordinates of sample point can complete the cluster to sample point in signal intensity profile model, have
Body:
K (K are selected in N number of sample point first in the RSS distributed models of a certain WAP in target area<N it is) individual to make
Centered on point;For remaining sample point in RSS distributed models, its Euclidean distance for arriving each central point is measured, and then by residue
Sample point is sorted out to the cluster of the central point nearest apart from oneself respectively.Specifically, in RSS distributed models remaining sample point with it is each in
A square Euclidean distance V between heart point is:
Wherein, uiRepresent i-th of central point, SiRepresent the set of remaining sample point in RSS distributed models, xjRepresent RSS points
The coordinate of remaining sample point in cloth model.
Sort out after completing, the central point u of each cluster is recalculated according to following formulai:
Wherein, miRepresent the quantity for the sample point that i-th of cluster includes, SiRepresent remaining sample point in RSS distributed models
Set, xjRepresent the coordinate of remaining sample point in RSS distributed models.
Afterwards according to calculate obtained central point judge to calculate in same cluster obtained central point and the central point selected it
Between distance whether be less than predetermined threshold value, if so, complete for the WAP RSS distributed models fingerprint cluster;It is no
Then, remaining sample point in RSS distributed models is remeasured, to the Euclidean distance of new central point, remaining sample point to be sorted out respectively
To the cluster of the central point (central point that calculating obtain) nearest apart from oneself, and the central point of each cluster is recalculated, until same
The distance between obtained central point and the central point of selecting are calculated in cluster and is less than predetermined threshold value uthres, i.e.,
Classification terminates, wherein, uiRepresent i-th of central point, ui-1Represent the i-th -1 central point.It is noted that we are to pre- here
If threshold value uthresOccurrence do not limit, the factor such as number of times of iteration is related in its density to sampled point, cluster process, root
Set according to actual conditions, such as can be set as 5cm (centimetre), 10cm even more many.
Sample point in RSS distributed models is classified as by K cluster by k-means algorithms with this, for each cluster, by this
All sample point coordinate unifications are an identical coordinate in individual cluster, and the coordinate is specially the coordinate of the cluster central point.Also
It is to say, after cluster, each cluster only correspond to a position coordinates (centre coordinate of cluster), and the position coordinates is to should
The corresponding RSS vectors set of all sample points, after cluster is completed, obtains the finger positioned in real time based on probabilistic model algorithm in cluster
Line storehouse.
It should be noted that above-described embodiment can independent assortment as needed.Described above is only the preferred of the present invention
Embodiment, it is noted that for those skilled in the art, is not departing from the premise of the principle of the invention
Under, some improvements and modifications can also be made, these improvements and modifications also should be regarded as protection scope of the present invention.
Claims (10)
1. a kind of fingerprint cluster method based on WiFi indoor positionings, it is characterised in that the fingerprint cluster method includes:
S1 is in the target area for the selected multiple sampled points of WAP stochastical sampling;
S2 sets up the signal intensity profile model of each point according to the Gaussian process of sampled point;
S3 selectes multiple sample points in the signal intensity profile model, and regard selected sample point as central point;
S4 calculates the Euclidean distance between remaining sample point and each central point in signal intensity profile model, and then by remaining sample
Point is sorted out to the cluster of the central point nearest apart from oneself respectively;
S5 recalculates the central point of each cluster, and judges to calculate in same cluster between obtained central point and selected central point
Distance whether be less than predetermined threshold value, if so, jumping to step S6, otherwise, jump to step S4;
S6 completes the fingerprint cluster of the signal intensity profile model for the WAP.
2. fingerprint cluster method as claimed in claim 1, it is characterised in that
In step s 4, square Euclidean distance V in signal intensity profile model between remaining sample point and each central point is:
<mrow>
<mi>V</mi>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>K</mi>
</munderover>
<munder>
<mo>&Sigma;</mo>
<mrow>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>&Element;</mo>
<msub>
<mi>S</mi>
<mi>i</mi>
</msub>
</mrow>
</munder>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>-</mo>
<msub>
<mi>u</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
Wherein, uiRepresent i-th of central point, SiRepresent the set of remaining sample point in signal intensity profile model, xjRepresent signal
The coordinate of remaining sample point in intensity distribution model;
And/or, in step s 5, the central point for recalculating each cluster is:
<mrow>
<msub>
<mi>u</mi>
<mi>i</mi>
</msub>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<msub>
<mi>m</mi>
<mi>i</mi>
</msub>
</mfrac>
<munder>
<mo>&Sigma;</mo>
<mrow>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>&Element;</mo>
<msub>
<mi>S</mi>
<mi>i</mi>
</msub>
</mrow>
</munder>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
</mrow>
Wherein, miRepresent the quantity for the sample point that i-th of cluster includes, SiRepresent remaining sample point in signal intensity profile model
Set, xjRepresent the coordinate of remaining sample point in signal intensity profile model.
3. fingerprint cluster method as claimed in claim 1 or 2, it is characterised in that specifically include in step s 2:
S21 obtains the degree of correlation between each point signal strength values according to the degree of correlation between sampled point;
S22 obtains signal intensity profile model according to the degree of correlation between each point signal strength values;
S23 obtains the corresponding signal strength values in any point in target area according to signal intensity profile model, completes this wireless
The foundation of access point fingerprint base in the target area.
4. fingerprint cluster method as claimed in claim 3, it is characterised in that
In the step s 21:
Degree of correlation k (x between sampled pointi,yj) be:
<mrow>
<mi>k</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msup>
<msub>
<mi>&sigma;</mi>
<mi>f</mi>
</msub>
<mn>2</mn>
</msup>
<mi>exp</mi>
<mo>&lsqb;</mo>
<mfrac>
<mrow>
<mo>-</mo>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>-</mo>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
<mrow>
<mn>2</mn>
<msup>
<mi>l</mi>
<mn>2</mn>
</msup>
</mrow>
</mfrac>
<mo>&rsqb;</mo>
</mrow>
Degree of correlation cov (x between each point signal strength valuesi,yj) be:
<mrow>
<mi>cov</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>y</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<msub>
<mi>y</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mi>k</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>+</mo>
<msubsup>
<mi>&sigma;</mi>
<mi>s</mi>
<mn>2</mn>
</msubsup>
<mi>&delta;</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
Wherein,Represent the sample variance of sampled point;L represents signal strength space correlation yardstick variable;Represent that Gauss makes an uproar
Sound;As i=j, δ (xi,yj) value be 1, be otherwise 0;
In step S22:
According to the degree of correlation between the signal strength values obtained in step S21, each point signal intensity is obtained for sampling dot matrix X
Matrix Y degree of correlation cov (Y):
<mrow>
<mi>cov</mi>
<mrow>
<mo>(</mo>
<mi>Y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mi>K</mi>
<mo>+</mo>
<msubsup>
<mi>&sigma;</mi>
<mi>s</mi>
<mn>2</mn>
</msubsup>
<mi>I</mi>
</mrow>
Wherein, K represents covariance matrix K [i, j]=k (x according to the sampling dot matrix X n × b tried to achievei,yj);
Obtain signal intensity profile model:
5. fingerprint base method for building up as claimed in claim 4, it is characterised in that any one in target area in step S23
Point x*Signal strength values y*For:
<mrow>
<mi>p</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>y</mi>
<mo>*</mo>
</msub>
<mo>|</mo>
<msub>
<mi>x</mi>
<mo>*</mo>
</msub>
<mo>,</mo>
<mi>X</mi>
<mo>,</mo>
<mi>Y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mi>N</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>y</mi>
<mo>*</mo>
</msub>
<mo>;</mo>
<msub>
<mi>&mu;</mi>
<msub>
<mi>x</mi>
<mo>*</mo>
</msub>
</msub>
<mo>,</mo>
<msubsup>
<mi>&sigma;</mi>
<msub>
<mi>x</mi>
<mo>*</mo>
</msub>
<mn>2</mn>
</msubsup>
<mo>+</mo>
<msubsup>
<mi>&sigma;</mi>
<mi>s</mi>
<mn>2</mn>
</msubsup>
<mo>)</mo>
</mrow>
</mrow>
Wherein,k*For point x*With adopting
Sampling point matrix X covariance matrix, wherein, value k (i)=k (x in the covariance matrix*,xi)。
6. a kind of fingerprint cluster device based on WiFi indoor positionings, it is characterised in that the fingerprint cluster device includes:
Information receiving module, for receiving multiple sampled points for selected wireless access point sampling;
Model building module, the signal that the Gaussian process of the sampled point for being received according to information receiving module sets up each point is strong
Spend distributed model;
Sample point selection module, for selecting multiple sample points in the signal intensity profile model set up from model building module,
And it regard selected sample point as central point;
Computing module, for calculating the central point that remaining sample point is selected with sample point selection module in signal intensity profile model
Between Euclidean distance, and for recalculating the central point of each cluster according to the categorization results of classifying module;
Classifying module, sorts out remaining sample point to the central point nearest apart from oneself for the result of calculation according to computing module
Cluster;
Judge module, for judge to calculate in same cluster module calculate obtained central point and sample point selection module it is selected in
Whether the distance between heart point is less than predetermined threshold value.
7. fingerprint cluster device as claimed in claim 5, it is characterised in that in computing module:
A square Euclidean distance in signal intensity profile model between remaining sample point and each central point is:
<mrow>
<mi>V</mi>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>K</mi>
</munderover>
<munder>
<mo>&Sigma;</mo>
<mrow>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>&Element;</mo>
<msub>
<mi>S</mi>
<mi>i</mi>
</msub>
</mrow>
</munder>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>-</mo>
<msub>
<mi>u</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
Wherein, uiRepresent i-th of central point, SiRepresent the set of remaining sample point in signal intensity profile model, xjRepresent signal
The coordinate of remaining sample point in intensity distribution model;
And/or, it is according to the central point that the categorization results of classifying module recalculate each cluster:
<mrow>
<msub>
<mi>u</mi>
<mi>i</mi>
</msub>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<msub>
<mi>m</mi>
<mi>i</mi>
</msub>
</mfrac>
<munder>
<mo>&Sigma;</mo>
<mrow>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>&Element;</mo>
<msub>
<mi>S</mi>
<mi>i</mi>
</msub>
</mrow>
</munder>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
</mrow>
Wherein, miRepresent the quantity for the sample point that i-th of cluster includes, SiRepresent remaining sample point in signal intensity profile model
Set, xjRepresent the coordinate of remaining sample point in signal intensity profile model.
8. fingerprint cluster device as claimed in claims 6 or 7, it is characterised in that
The degree of correlation that computing module is additionally operable between the sampled point that is received according to information receiving module obtains each point signal strength values
Between the degree of correlation, calculated for the signal intensity profile model set up according to model building module and obtain any in target area
Any corresponding signal strength values;
Model building module is used to obtain signal intensity profile model according to the degree of correlation between each point signal strength values.
9. fingerprint cluster device as claimed in claim 8, it is characterised in that
In computing module:
Degree of correlation k (x between sampled pointi,yj) be:
<mrow>
<mi>k</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msup>
<msub>
<mi>&sigma;</mi>
<mi>f</mi>
</msub>
<mn>2</mn>
</msup>
<mi>exp</mi>
<mo>&lsqb;</mo>
<mfrac>
<mrow>
<mo>-</mo>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>-</mo>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
<mrow>
<mn>2</mn>
<msup>
<mi>l</mi>
<mn>2</mn>
</msup>
</mrow>
</mfrac>
<mo>&rsqb;</mo>
</mrow>
Degree of correlation cov (x between each point signal strength valuesi,yj) be:
<mrow>
<mi>cov</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>y</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<msub>
<mi>y</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mi>k</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>+</mo>
<msubsup>
<mi>&sigma;</mi>
<mi>s</mi>
<mn>2</mn>
</msubsup>
<mi>&delta;</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<msub>
<mi>x</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
Wherein,Represent the sample variance of sampled point;L represents signal strength space correlation yardstick variable;Represent that Gauss makes an uproar
Sound;As i=j, δ (xi,yj) value be 1, be otherwise 0;
In model building module:
The degree of correlation between obtained signal strength values is calculated according to computing module, each point signal is obtained for sampling dot matrix X
Intensity matrix Y degree of correlation cov (Y):
<mrow>
<mi>cov</mi>
<mrow>
<mo>(</mo>
<mi>Y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mi>K</mi>
<mo>+</mo>
<msubsup>
<mi>&sigma;</mi>
<mi>s</mi>
<mn>2</mn>
</msubsup>
<mi>I</mi>
</mrow>
Wherein, K represents covariance matrix K [i, j]=k (x according to the sampling dot matrix X n × b tried to achievei,yj);
And then obtain signal intensity profile model:
10. fingerprint cluster device as claimed in claim 9, it is characterised in that in computing module:
The signal intensity profile model set up according to model building module, any point in target area is calculated according to below equation
x*Signal strength values y*:
<mrow>
<mi>p</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>y</mi>
<mo>*</mo>
</msub>
<mo>|</mo>
<msub>
<mi>x</mi>
<mo>*</mo>
</msub>
<mo>,</mo>
<mi>X</mi>
<mo>,</mo>
<mi>Y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mi>N</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>y</mi>
<mo>*</mo>
</msub>
<mo>;</mo>
<msub>
<mi>&mu;</mi>
<msub>
<mi>x</mi>
<mo>*</mo>
</msub>
</msub>
<mo>,</mo>
<msubsup>
<mi>&sigma;</mi>
<msub>
<mi>x</mi>
<mo>*</mo>
</msub>
<mn>2</mn>
</msubsup>
<mo>+</mo>
<msubsup>
<mi>&sigma;</mi>
<mi>s</mi>
<mn>2</mn>
</msubsup>
<mo>)</mo>
</mrow>
</mrow>
Wherein,k*For point x*With adopting
Sampling point matrix X covariance matrix, wherein, value k (i)=k (x in the covariance matrix*,xi)。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710160778.2A CN107087256A (en) | 2017-03-17 | 2017-03-17 | A kind of fingerprint cluster method and device based on WiFi indoor positionings |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710160778.2A CN107087256A (en) | 2017-03-17 | 2017-03-17 | A kind of fingerprint cluster method and device based on WiFi indoor positionings |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107087256A true CN107087256A (en) | 2017-08-22 |
Family
ID=59614555
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710160778.2A Pending CN107087256A (en) | 2017-03-17 | 2017-03-17 | A kind of fingerprint cluster method and device based on WiFi indoor positionings |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107087256A (en) |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108089148A (en) * | 2017-12-14 | 2018-05-29 | 电子科技大学 | A kind of passive track-corelation direction cross positioning method based on time difference information |
CN108519578A (en) * | 2018-03-23 | 2018-09-11 | 天津大学 | A kind of indoor positioning fingerprint base construction method based on intelligent perception |
CN109451428A (en) * | 2018-12-26 | 2019-03-08 | 哈尔滨工业大学 | A kind of chamber height localization method based on WiFi cluster |
CN109525337A (en) * | 2017-09-20 | 2019-03-26 | 腾讯科技(深圳)有限公司 | WiFi fingerprint acquisition methods, device, storage medium and equipment |
CN110719566A (en) * | 2019-10-18 | 2020-01-21 | 安徽农业大学 | WiFi clustering algorithm-based agricultural machinery signal-free positioning method |
CN110933631A (en) * | 2019-12-03 | 2020-03-27 | 浙江科技学院 | Indoor positioning method based on WIFI position fingerprint |
CN111126573A (en) * | 2019-12-27 | 2020-05-08 | 深圳力维智联技术有限公司 | Model distillation improvement method and device based on individual learning and storage medium |
CN111385757A (en) * | 2020-03-20 | 2020-07-07 | 浙江大学城市学院 | Technology for eliminating heterogeneity of indoor positioning WiFi fingerprint database equipment |
CN111757258A (en) * | 2020-07-06 | 2020-10-09 | 江南大学 | Self-adaptive positioning fingerprint database construction method under complex indoor signal environment |
CN112015743A (en) * | 2020-05-28 | 2020-12-01 | 广州杰赛科技股份有限公司 | Indoor positioning system fingerprint database construction method and device |
CN113079466A (en) * | 2020-10-21 | 2021-07-06 | 中移(上海)信息通信科技有限公司 | Fingerprint database construction method, device, equipment and computer storage medium |
CN114269006A (en) * | 2021-12-24 | 2022-04-01 | 河海大学 | Indoor AP cluster selection method and device based on information gain rate |
CN114424635A (en) * | 2020-06-01 | 2022-04-29 | 蜂图志科技控股有限公司 | Signal map construction method, device, equipment and readable storage medium |
Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104185275A (en) * | 2014-09-10 | 2014-12-03 | 北京航空航天大学 | Indoor positioning method based on WLAN |
CN104507097A (en) * | 2014-12-19 | 2015-04-08 | 上海交通大学 | Semi-supervised training method based on WiFi (wireless fidelity) position fingerprints |
CN104853317A (en) * | 2014-11-24 | 2015-08-19 | 北京航空航天大学 | WiFi indoor positioning fingerprint database construction and update method |
US20150237480A1 (en) * | 2014-02-14 | 2015-08-20 | Industrial Technology Research Institute | Indoor positioning method, indoor positioning system, and computer-readable medium |
CN105044662A (en) * | 2015-05-27 | 2015-11-11 | 南京邮电大学 | Fingerprint clustering multi-point joint indoor positioning method based on WIFI signal intensity |
CN105120433A (en) * | 2015-08-19 | 2015-12-02 | 上海交通大学 | WLAN indoor positioning method based on continuous sampling and fuzzy clustering |
CN105589064A (en) * | 2016-01-08 | 2016-05-18 | 重庆邮电大学 | Rapid establishing and dynamic updating system and method for WLAN position fingerprint database |
CN105636201A (en) * | 2016-03-14 | 2016-06-01 | 中国人民解放军国防科学技术大学 | Indoor positioning method based on sparse signal fingerprint database |
CN105792230A (en) * | 2016-03-11 | 2016-07-20 | 上海交通大学 | RSS-fingerprint-based access point and scaling point deployment method for indoor positioning system |
WO2016165459A1 (en) * | 2015-09-23 | 2016-10-20 | 中兴通讯股份有限公司 | Method and device for indoor positioning |
CN106054128A (en) * | 2016-05-26 | 2016-10-26 | 天津城建大学 | Indoor navigation method and indoor navigation method device |
-
2017
- 2017-03-17 CN CN201710160778.2A patent/CN107087256A/en active Pending
Patent Citations (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20150237480A1 (en) * | 2014-02-14 | 2015-08-20 | Industrial Technology Research Institute | Indoor positioning method, indoor positioning system, and computer-readable medium |
CN104185275A (en) * | 2014-09-10 | 2014-12-03 | 北京航空航天大学 | Indoor positioning method based on WLAN |
CN104853317A (en) * | 2014-11-24 | 2015-08-19 | 北京航空航天大学 | WiFi indoor positioning fingerprint database construction and update method |
CN104507097A (en) * | 2014-12-19 | 2015-04-08 | 上海交通大学 | Semi-supervised training method based on WiFi (wireless fidelity) position fingerprints |
CN105044662A (en) * | 2015-05-27 | 2015-11-11 | 南京邮电大学 | Fingerprint clustering multi-point joint indoor positioning method based on WIFI signal intensity |
CN105120433A (en) * | 2015-08-19 | 2015-12-02 | 上海交通大学 | WLAN indoor positioning method based on continuous sampling and fuzzy clustering |
WO2016165459A1 (en) * | 2015-09-23 | 2016-10-20 | 中兴通讯股份有限公司 | Method and device for indoor positioning |
CN105589064A (en) * | 2016-01-08 | 2016-05-18 | 重庆邮电大学 | Rapid establishing and dynamic updating system and method for WLAN position fingerprint database |
CN105792230A (en) * | 2016-03-11 | 2016-07-20 | 上海交通大学 | RSS-fingerprint-based access point and scaling point deployment method for indoor positioning system |
CN105636201A (en) * | 2016-03-14 | 2016-06-01 | 中国人民解放军国防科学技术大学 | Indoor positioning method based on sparse signal fingerprint database |
CN106054128A (en) * | 2016-05-26 | 2016-10-26 | 天津城建大学 | Indoor navigation method and indoor navigation method device |
Cited By (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109525337A (en) * | 2017-09-20 | 2019-03-26 | 腾讯科技(深圳)有限公司 | WiFi fingerprint acquisition methods, device, storage medium and equipment |
CN108089148B (en) * | 2017-12-14 | 2019-08-30 | 电子科技大学 | A kind of passive track-corelation direction cross positioning method based on time difference information |
CN108089148A (en) * | 2017-12-14 | 2018-05-29 | 电子科技大学 | A kind of passive track-corelation direction cross positioning method based on time difference information |
CN108519578A (en) * | 2018-03-23 | 2018-09-11 | 天津大学 | A kind of indoor positioning fingerprint base construction method based on intelligent perception |
CN109451428A (en) * | 2018-12-26 | 2019-03-08 | 哈尔滨工业大学 | A kind of chamber height localization method based on WiFi cluster |
CN110719566A (en) * | 2019-10-18 | 2020-01-21 | 安徽农业大学 | WiFi clustering algorithm-based agricultural machinery signal-free positioning method |
CN110933631A (en) * | 2019-12-03 | 2020-03-27 | 浙江科技学院 | Indoor positioning method based on WIFI position fingerprint |
CN111126573B (en) * | 2019-12-27 | 2023-06-09 | 深圳力维智联技术有限公司 | Model distillation improvement method, device and storage medium based on individual learning |
CN111126573A (en) * | 2019-12-27 | 2020-05-08 | 深圳力维智联技术有限公司 | Model distillation improvement method and device based on individual learning and storage medium |
CN111385757A (en) * | 2020-03-20 | 2020-07-07 | 浙江大学城市学院 | Technology for eliminating heterogeneity of indoor positioning WiFi fingerprint database equipment |
CN112015743A (en) * | 2020-05-28 | 2020-12-01 | 广州杰赛科技股份有限公司 | Indoor positioning system fingerprint database construction method and device |
CN114424635A (en) * | 2020-06-01 | 2022-04-29 | 蜂图志科技控股有限公司 | Signal map construction method, device, equipment and readable storage medium |
CN111757258B (en) * | 2020-07-06 | 2021-05-14 | 江南大学 | Self-adaptive positioning fingerprint database construction method under complex indoor signal environment |
CN111757258A (en) * | 2020-07-06 | 2020-10-09 | 江南大学 | Self-adaptive positioning fingerprint database construction method under complex indoor signal environment |
CN113079466A (en) * | 2020-10-21 | 2021-07-06 | 中移(上海)信息通信科技有限公司 | Fingerprint database construction method, device, equipment and computer storage medium |
CN114269006A (en) * | 2021-12-24 | 2022-04-01 | 河海大学 | Indoor AP cluster selection method and device based on information gain rate |
CN114269006B (en) * | 2021-12-24 | 2023-04-14 | 河海大学 | Indoor AP cluster selection method and device based on information gain rate |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107087256A (en) | A kind of fingerprint cluster method and device based on WiFi indoor positionings | |
CN108696932B (en) | Outdoor fingerprint positioning method using CSI multipath and machine learning | |
CN107071743B (en) | Rapid KNN indoor WiFi positioning method based on random forest | |
Honkavirta et al. | A comparative survey of WLAN location fingerprinting methods | |
KR102116824B1 (en) | Positioning system based on deep learnin and construction method thereof | |
CN109275095B (en) | Bluetooth-based indoor positioning system, positioning equipment and positioning method | |
CN110049549B (en) | WiFi fingerprint-based multi-fusion indoor positioning method and system | |
CN111479231B (en) | Indoor fingerprint positioning method for millimeter wave large-scale MIMO system | |
CN103916820B (en) | Wireless indoor location method based on access point stability | |
CN106093852A (en) | A kind of method improving WiFi fingerprint location precision and efficiency | |
CN106851571B (en) | Decision tree-based rapid KNN indoor WiFi positioning method | |
CN101620270B (en) | Wireless location method based on cluster-fusion | |
WO2019062734A1 (en) | Indoor positioning method and device based on wi-fi hot spots | |
CN112887902B (en) | Indoor positioning method of WiFi fingerprint based on Gaussian clustering and hybrid measurement | |
CN112135248B (en) | WIFI fingerprint positioning method based on K-means optimal estimation | |
CN104602342B (en) | A kind of high efficiency indoor localization method based on iBeacon equipment | |
CN106851573A (en) | Joint weighting k nearest neighbor indoor orientation method based on log path loss model | |
CN106102163B (en) | WLAN fingerprint positioning method based on RSS linear correlation Yu secondary weighted centroid algorithm | |
CN106131959A (en) | A kind of dual-positioning method divided based on Wi Fi signal space | |
CN112367614B (en) | LSTM-based Wi-Fi and geomagnetic field fusion indoor positioning algorithm | |
CN108924756A (en) | Indoor orientation method based on WiFi double frequency-band | |
CN109640262B (en) | Positioning method, system, equipment and storage medium based on mixed fingerprints | |
CN103258001B (en) | A kind of radio frequency map that embeds algorithm based on local linear is without supervised classification method | |
Ning et al. | Outdoor location estimation using received signal strength-based fingerprinting | |
CN112637950B (en) | Fingerprint positioning method based on angle similarity |
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 | ||
TA01 | Transfer of patent application right |
Effective date of registration: 20201202 Address after: Room 10242, No. 260, Jiangshu Road, Xixing street, Binjiang District, Hangzhou City, Zhejiang Province Applicant after: Hangzhou Jiji Intellectual Property Operation Co., Ltd Address before: 201616 Shanghai city Songjiang District Sixian Road No. 3666 Applicant before: Phicomm (Shanghai) Co.,Ltd. |
|
TA01 | Transfer of patent application right |