CN105044662B - A kind of fingerprint cluster multi-point joint indoor orientation method based on WIFI signal intensity - Google Patents
A kind of fingerprint cluster multi-point joint indoor orientation method based on WIFI signal intensity Download PDFInfo
- Publication number
- CN105044662B CN105044662B CN201510280786.1A CN201510280786A CN105044662B CN 105044662 B CN105044662 B CN 105044662B CN 201510280786 A CN201510280786 A CN 201510280786A CN 105044662 B CN105044662 B CN 105044662B
- Authority
- CN
- China
- Prior art keywords
- fingerprint
- point
- reference point
- class
- vector
- 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
Links
Classifications
-
- 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
- G01S5/0009—Transmission of position information to remote stations
-
- 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
- G01S5/0009—Transmission of position information to remote stations
- G01S5/0018—Transmission from mobile station to base station
- G01S5/0036—Transmission from mobile station to base station of measured values, i.e. measurement on mobile and position calculation on base station
-
- 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
Abstract
A kind of fingerprint cluster multi-point joint indoor orientation method based on WIFI signal intensity, by collecting the received signals fingerprint at reference point, fingerprint classification number and initial cluster center feature vector are determined using a kind of dynamic virtual point initial cluster center selection method based on distance matrix, the fingerprint of fingerprint base is realized into classification by a kind of initial clustering method, then cluster fuzzy partitioning is realized with maximization log-likelihood probability expectation method on this basis, make to overlap between class and class, and obtains the probabilistic model parameter of fingerprint.In positioning, the reference point for participating in positioning is obtained using the signal vector of real-time reception, probabilistic model parameter and category feature fingerprint vector, finally estimates positioning coordinate using a kind of method of multiple reference points alignment by union.The present invention is greatly improved fingerprint matching efficiency and positioning accuracy, reduces the cost of wireless location, had the stronger market competitiveness using fingerprinting localization algorithm after cluster optimization.
Description
Technical field
The present invention relates to wireless communication technology fields, and in particular to the fingerprint cluster multi-point joint under WIFI signal intensity value
The localization method of indoor locating system.
Background technique
WIFI is the wireless network standards based on IEEE802.11 standard, is PC in people's daily life, holds mobile set
The terminals such as standby wirelessly access one of the main way of Internet core net.With the continuous development of wireless technology,
Nowadays almost all covers WIFI access point in the large-scale social scene of living such as market, house, office, and its development becomes
It will definitely finally realize covering all around for living region.It is more convenient using WIFI progress indoor positioning under this trend, and
WIFI signal overcomes GPS signal that can not have the shortcomings that effect spread indoors, and existing WIFI access point is fully utilized, without as blue
Tooth positioning arranges great deal of nodes like that, low in cost without wiring.Fingerprint location technology is a kind of positioning side based on not ranging
Method, typical localization method such as TOA, TDOA, AOA, RSSI etc., since there are serious multiple scattering, above-mentioned ginsengs for indoor environment
Often there is large error in several estimations, positioning performance is often not satisfactory;And this imparametrization localization method of fingerprint location
Without estimating environmental parameter, it is effective against indoor multipath propagation, the precision of indoor positioning is greatly enhanced, generally exists
Fingerprint of the signal for containing environmental error as the reference point is received before positioning in reference point, then will be connect in positioning stage
The signal and fingerprint base received compares, to realize location estimation.
Summary of the invention
Technical problem: the object of the present invention is to provide a kind of movements in the indoor environment of WIFI covering to request positioning
The method and system of user-provided location estimation is realized.It is real by building library, fingerprint cluster and the matched process of tuning on-line offline
It is existing.
Technical solution: to achieve the purpose of the present invention, the present invention relates to a mobile terminals to acquire signal, a service
Device establishes fingerprint base and realizes localization method, while the present invention need to apply under the indoor environment of more WIFI access points, wherein
Server selection software includes that acquisition signal builds library facility and localization method realization function offline.
The invention mainly comprises: it is excellent under a kind of dynamic virtual point initial cluster center selection algorithm based on distance matrix
Change cluster fuzzy partitioning clustering method, one kind being based on class participation multi-point joint dynamic positioning method, and detailed process is as follows:
(1) mode that a reference point is arranged at interval of 1 meter divides area to be targeted in a grid formation, and each lattice point is
1 reference point, divides N number of reference point altogether.
(2) terminal for holding loading server positioning software successively receives the letter from M WIFI access point in N number of reference point
Number intensity vector ri, it is stored in fingerprint database.
(3) distance matrix calculating is carried out according to fingerprint base vector of samples, fingerprint classification is realized according to clustering algorithm, obtains class
The class distribution matrix of feature vector and reference point received signals fingerprint.
(4) mobile subscriber to be positioned obtains current WIFI list and corresponding signal value automatically, sends positioning to server and asks
It asks, and sends WIFI signal vector.
(5) server matches after obtaining signal vector to be positioned with all kinds of feature vectors, obtains class label.According to class label
Selected local positioning region, estimated location to be positioned is calculated using local locating method.
(6) estimated location is returned to mobile subscriber to be positioned by server.
The utility model has the advantages that the invention has the following advantages that
A. reduce equipment investment.It is reduced into using existing WIFI access point without being redeployed to localizing environment
This;
B. location efficiency is improved.Without measuring to environmental parameter, reduce positioning previous work;
C. positioning accuracy is increased.The analysis and processing to fingerprint characteristic are strengthened, fingerprint vector and position are efficiently extracted
The feature of confidence breath, greatly improves positioning accuracy;
D. a kind of new algorithm of fingerprint location is proposed.By the analysis to fingerprint cluster process, so that building in library most offline
Latter step cluster boundary can fuzzy partitioning, a kind of new location algorithm is proposed in this principle, matching algorithm carries out when to positioning
Optimization, and consider the select permeability of reference point, it is optimal location algorithm.
Detailed description of the invention
Fingerprint cluster multi-point joint indoor orientation method flow chart of the Fig. 1 based on WIFI signal intensity.
Fig. 2 is built the library stage offline, and handheld terminal acquires signal schematic representation in each reference point.
Fig. 3 positioning stage schematic diagram.
Initial hardening point of Fig. 4 lower reference point distribution map of classifying.S1 represents localization region.
Reference point distribution map under Fig. 5 fuzzy partitioning is classified.S2 represents localization region.
Reference point distribution map in one kind under Fig. 6 fuzzy partitioning.S3 represents localization region.
Specific embodiment
It is as follows for the analysis and implementation process of fingerprint clustering method in (3):
1, a kind of dynamic virtual point initial cluster center selection method based on distance matrix
According to process (1) mode, the terminal for holding loading server positioning software is successively come from the reception of N number of reference point
The signal strength vector r of M WIFI access pointi, it is stored in fingerprint database.riVector is made of M signal strength component, i.e.,
It is with reference to point set
U=u | u≤N, u ∈ N*}
Since a certain WIFI signal value of pointing out can float up and down at any time in practice, so should be adopted in same reference point
Collect multiple signal, i.e., acquire T signal strength vector in each reference point, the signal strength vector of acquisition is embodied as:
T sampling filter is handled, is averaged and is stored in fingerprint database as fingerprint, then fingerprint is at each reference point
Establish reference point initial range matrix DN×N:
Wherein, apart from element DI, j=| | ri-rj||2, DI, j=DJ, i, which is symmetrical matrix.Based on apart from square
The dynamic virtual point initial cluster center selection method specific embodiment of battle array is as follows:
1) for Distance matrix DN×NIf apart from element DI, j< ε, then by original with reference to i in point set U, j point deletion.
2) addition represents i, and the received signals fingerprint of virtual reference point v, the v point of j isUpdate set U=U ∪ vg-
{ i, j }, g=1,2 ....
3) current reference point number N is updated, Distance matrix D is updatedN×N, the dimension of distance matrix has reduced at this time, repeats
1), 2), if generating without new virtual point, ε=ε+θ is updated, is repeated 1), 2).
4) it repeats 1), 2), 3), if ε > γ.
5) referring to point set U at this time is initial cluster center set.Corresponding received signals fingerprint is initial cluster center
Characteristic fingerprint vector { mj}={ m1, m2..., mK}。
2, the method for preliminary fingerprint cluster
Initial cluster center { m is established in the dynamic virtual point initial cluster center selection method based on distance matrixk
| k=1,2..., K }, class number K.Next will according to initial cluster center by the received signals fingerprint of all reference points according between it
Initial range matrix DN×NIt is divided into K regional area, that is, realizes and the higher received signals fingerprint of similarity is divided into one kind.Point
The target that class is met is:
Wherein hypotaxis degree ρikValue indicate received signals fingerprint riWhether k-th class, d are belonged toikIndicate riIn initial clustering
Heart mkDistance, then:
Realize the specific implementation process of fingerprint cluster are as follows:
1) successively calculating reference point received signals fingerprint ri(i=1,2 ..., N) and cluster centre mk(k=1,2 ..., K) away from
From working as dik=minl∈K(dil) when, enable ρik=1, i.e. riBelong to class k, otherwise ρij=0, i.e. riIt is not belonging to class k.Record is about ρik
Subordinate matrix QN×K。
2) regional area k (kth class) reference point set omega is establishedk(k ∈ K), according to subordinate matrix QN×KIf ρik=1, then
Reference point i ∈ Ωk.Calculate new distance center
If 3) | mk(new)-mk| > τ then uses mk(new)Instead of the distance center m of last iterationk, repeat 1), 2).In 2)
Update Ωk(k∈K)。
If 4) | mk(new)-mk|≤τ:, then cluster centre mkConvergence stops repeating.
5) m at this timek(mew)For the final characteristic fingerprint of regional area k, reference point set omegak(k ∈ K) is regional area k's
Final classification refers to point set.
3, it maximizes log-likelihood probability expectation method and realizes cluster fuzzy partitioning
Since there is hard plot, the i.e. degree of membership of certain reference point one kind in clustering algorithm in 2 in actual operation
Non-zero i.e. 1.And degree of membership is not much different for two class of reference point on class line of demarcation, if it is utterly distributed
Into certain one kind, undesirable similar degree in the class target as big as possible is unfavorable to classifying.By establishing likelihood probability density
Distributed model can be realized the fuzzy partitioning of cluster, determine degree of membership according to probability, then it is in the borderline reference point of class, it is right
Two classes or more multiclass have degree of membership, and value describes the classification situation energy of reference point with this probability between 0 to 1
It is enough properer more effectively to realize class object.Fingerprint classification boundary fuzzy partitioning is executed by two steps, first is that finding out eligible
The expectation of probability density function, second is that asking the parameter value maximized under conditional probability density function expectation maximum, including degree of membership
Probability.Output for fingerprint cluster: category feature fingerprint vector mjAnd reference point clusters array:
By the input as cluster fuzzy partitioning algorithm.Since gaussian probability distribution function is more preferably fitted various probability point
Cloth model, so herein, gauss hybrid models are used for each class reference point signal distributions model, then in whole region
Signal distributions model is in reference point are as follows:
For k-th of local class localization region signal distributions model are as follows:
Wherein gaussian probability is distributedForm are as follows:
For a reference point uiThe attribute having includes (ri, bi), wherein riFor fingerprint vector, biFor participation vector,
bi=(ΥΥ1, Υi2..., ΥiK),Υk∈ (0,1) indicates whether current reference point received signals fingerprint belongs to k-th of class.biNothing
Method directly passes through observation and obtains, the complete likelihood function under model are as follows:
It is directly very high using complete likelihood function solution computation complexity, logarithm can be taken to the likelihood function to this, it will
Multiplication switchs to addition to reduce complexity, convenient for solving.Then log-likelihood function are as follows:
The expectation of log-likelihood function are as follows:
Wherein j-th of observation signal rjTo the participation Υ of k-th of local class localization region signal distributions modeljk, can lead to
Cross its conditional expectation approximate evaluation are as follows:
It is therefore seen that the participation of reference point is numerically equal to the fingerprint vector of the reference point by k-th of local class positioning
The specified posterior probability of regional signal distributed model.
The parameter vector Φ of built gaussian probability distributed modelk={ αk, mk, Sk 2, to solve parameter, master mould is converted
Form it is expected for log-likelihood, in order to make new iterative model it is expected (6) maximum, to m in (6)kAnd Sk 2Local derviation is sought respectively:
Based on maximizing, log-likelihood probability expectation fuzzy partitioning specific implementation process is as follows:
1) output of fingerprint cluster, i.e. the characteristic fingerprint vector { m of regional area k are utilizedk| k ∈ K }, the ginseng of regional area k
Examination point set omegak(k ∈ K) and its corresponding reference point fingerprint vector { rj|j∈ΩkGenerate gauss hybrid models initial ginseng
Number enablesThe sub-model parameter of k class is calculated by formula (8), (9), (10):
2) j-th of observation signal r is updated according to formula (7)jTo the ginseng of k-th of local class localization region signal distributions model
With degree
The wherein received signals fingerprint r of reference point jjIt is subordinate to the posterior probability of kth class Gauss sub-model are as follows:
The received signals fingerprint r of reference point jjPosterior probability to the Gauss sub-models of all classes and are as follows:
3) according to updated participationThe sub-model parameter of k class is recalculated with formula (8), (9), (10):
4) repeat 2), 3), until
5) stop parameter calculating, algorithm terminates.The participation of K class of N number of reference point at this timeAnd
The parameter Φ of k Gauss sub-modelk(k ∈ K) tends to stablize.Record the participation vector of N number of reference pointThe parameter of k Gauss sub-model
It is as follows with implementation process for the method analysis for selecting local positioning region according to class label in process (5):
1, the method that local positioning region is selected according to class label
The purpose of this method is that the signal vector obtained according to terminal when positioning is quickly found out terminal institute in whole region
Regional area, positioned with the reference point terminal near terminal, and widespread practice is in whole region to every
A reference point matched signal one by one, efficiency is very low, and biggish matching error caused by being easy because of signal floating error, to letter
Number stability requirement it is very high, the method for the present invention according to before propose fingerprint classification method, can by mobile terminal obtain signal
It is matched with all kinds of characteristic fingerprints, which is divided into local class, is positioned by the reference point terminal in local class, is subtracted significantly
Oligodactyly line matching times improve location efficiency and speed.Specific implementation method is as follows:
1) it is r that handheld terminal, which obtains the WIFI signal vector of current anchor point,location.It is calculated separately according to formula (7)
rlocationTo the participation of k classWherein Gauss sub-model parameter is by based on maximization log-likelihood probability expectation
Soft stroke of output ΦkIt specifies, then current anchor point signal vector rlocationParticipation vector
2) according to blocationIn the global participation vector of acquisitionMiddle selection
Participate in the reference point of positioning.Specific method is: for blocation, establish localization region collection
According to ΘlocationIt is rightRetrieval, will be only to localization region collection
ΘlocationMiddle class k has the reference point of participation to find out, and establishes location reference point collection
For as follows using the analysis of local locating method and implementation process in process (5):
1, it is based on class participation multi-point joint localization method
Select local positioning region according to class label, from selected process can be seen that received signals fingerprint from these reference points with
Positioning signal vector category feature having the same, these reference points are close with locating point position from geographical location, then can be with
The position of anchor point is estimated by these reference points, the matching way of comprehensive multiple reference points divides matching error risk,
It can be avoided the high matching error risk generated by single reference point.If so for selected each reference point using flat
Equal point of reference, then being inequitable for reference point higher for matching degree, so should be using based on matching journey
The method of degree assigns the high reference point of matching degree to higher point of reference, and suitably reduces the relatively poor ginseng of matching degree
The point of reference of examination point.Since signal strength and distance are in logarithm non-linear relation in practice, so preferably using negative exponential function
To embody participation.
The coordinate of anchor point is estimated are as follows:
Wherein n is location reference point number, kmIt is reference point m to the point of reference of anchor point, calculation method is:
It is as follows based on class participation multi-point joint localization method process:
1) the point of reference k that location reference point concentrates n reference point is calculated separately by formula (15)m。
2) x coordinate of anchor point is calculated separately by formula (13), (14)With y-coordinate km。
3) estimation of anchor point coordinate is completed
Claims (1)
1. a kind of fingerprint cluster multi-point joint indoor orientation method based on WIFI signal intensity, which is characterized in that including acquisition
Signal builds library facility offline and location algorithm realizes that function, detailed process are as follows:
(1) mode that a reference point is arranged at interval of 1 meter divides area to be targeted in a grid formation, and each lattice point is 1
Reference point divides N number of reference point altogether;
(2) terminal for holding loading server positioning software is successively strong in signal of N number of reference point reception from M WIFI access point
Spend vector ri, it is stored in fingerprint database;
(3) distance matrix calculating is carried out according to fingerprint base vector of samples, fingerprint classification is realized according to clustering algorithm, obtains category feature
The class distribution matrix of vector and reference point received signals fingerprint;
(4) mobile subscriber to be positioned obtains current WIFI list and corresponding signal value automatically, sends Location Request to server, and
Send WIFI signal vector;
(5) server matches after obtaining signal vector to be positioned with all kinds of feature vectors, obtains class label, selected according to class label
Local positioning region calculates estimated location to be positioned using local locating method;
(6) estimated location is returned to mobile subscriber to be positioned by server;
For the dynamic virtual point initial cluster center selection algorithm based on distance matrix used by fingerprint classification in step (3)
Under optimization cluster fuzzy partitioning method include three parts: the dynamic virtual point initial cluster center selecting party based on distance matrix
Method, the method for preliminary fingerprint cluster maximize log-likelihood probability expectation method realization cluster fuzzy partitioning method, and specific method is such as
Under:
1) a kind of dynamic virtual point initial cluster center selection method based on distance matrix,
A1. according to process (1) the model split reference point after, hold the terminal of loading server positioning software successively in N number of ginseng
Examination point receives the signal strength vector r from M WIFI access pointi, it is stored in fingerprint database;riBy M signal strength component structure
At that is,
It is with reference to point set
U=u | u≤N, u ∈ N*}
Each reference point acquires T signal strength vector, and the signal strength vector of acquisition is embodied as:
T sampling filter is handled, is averaged and is stored in fingerprint database as fingerprint, then fingerprint at each reference point are as follows:
Establish reference point initial range matrix DN×N:
Wherein, apart from element DI, j=| | ri-rj||2, DI, j=DJ, i, which is symmetrical matrix;
B1. for Distance matrix DN×NIf apart from element DI, j< ε, by original with reference to i in point set U, j point deletion;
C1. addition represents i, and the received signals fingerprint of virtual reference point v, the v point of j isUpdate set U=U ∪ vgI,
J }, g=1,2 ...;
D1. current reference point number N is updated, Distance matrix D is updatedN×N, the dimension of distance matrix has reduced at this time, repetition a1,
B1 updates ε=ε+θ if generating without new virtual point;
E1. b1, c1, d1 are repeated, if ε > γ, stops iteration;
F1. referring to point set U at this time is initial cluster center set;Corresponding received signals fingerprint is initial cluster center feature
Fingerprint vector { mj}={ m1, m2..., mK};
2) initial cluster center { m is established in the dynamic virtual point initial cluster center selection method based on distance matrixk|k
=1,2..., K }, class number K;Next will according to initial cluster center by the received signals fingerprint of all reference points according between it
Initial range matrix DN×NIt is divided into K regional area, that is, realizes and the higher received signals fingerprint of similarity is divided into one kind;
The method of preliminary fingerprint cluster is as follows:
A2. successively calculating reference point received signals fingerprint riWith cluster centre mkDistance dik, acquisition position where i is represented, k represents poly-
Class center position, whenWhen, enable ρik=1, i.e. riBelong to class k, otherwise ρik=0, i.e. riIt is not belonging to class
k;Record is about ρikSubordinate matrix QN×K;
B2. regional area k reference point set omega is establishedk, k expression kth class, k ∈ K;According to subordinate matrix QN×KIf ρik=1, then
Reference point i ∈ Ωk;Calculate new cluster centre
If c2. | mk(new)-mk| > τ: then use mk(new)Instead of the cluster centre m of last iterationk, repeat a2, b2;In b2 more
New Ωk, k ∈ K;
If d2. | mk(new)-mk|≤τ, then cluster centre mkConvergence stops repeating;
E2. m at this timek(new)For the final characteristic fingerprint of regional area k, reference point set omegak, k ∈ K is the final of regional area k
Classification refers to point set;
3) method that log-likelihood probability expectation method realizes cluster fuzzy partitioning is maximized,
It can be realized the fuzzy partitioning of cluster by establishing likelihood probability Density Distribution model, degree of membership determined according to probability;It presses
Two steps execute, first is that finding out the expectation of eligible probability density function, maximize conditional probability density function expectation second is that asking
Parameter value under maximum, including degree of membership probability, signal distributions model is in whole region internal reference examination point are as follows:
For k-th of local class localization region signal distributions model are as follows:
Wherein gaussian probability is distributedForm are as follows:
For a reference point uiThe attribute having includes (ri, bi), wherein riFor fingerprint vector, biFor participation vector, bi=
(γi1..., γik..., γiK), γik∈ (0,1) indicates whether current reference point received signals fingerprint belongs to k-th of class;biNothing
Method directly passes through observation and obtains, the complete likelihood function under model are as follows:
Logarithm is taken to the likelihood function, then log-likelihood function is converted are as follows:
The expectation of log-likelihood function are as follows:
Wherein j-th of received signals fingerprint rjTo the participation γ of k-th of local class localization region signal distributions modeljk, it can be passed through
Conditional expectation approximate evaluation are as follows:
The participation of reference point is numerically equal to the fingerprint vector of the reference point by k-th of local class localization region signal distributions
The specified posterior probability of model;
The parameter vector Φ of built gaussian probability distributed modelk={ αk, mk, sk 2, log-likelihood expectation shape is converted by master mould
Formula, to m in (6)kAnd sk 2Local derviation is sought respectively:
Process is as follows:
A3. the output of fingerprint cluster, i.e. the characteristic fingerprint vector { m of regional area k are utilizedk| k ∈ K }, the reference point of regional area k
Set omegakAnd its corresponding reference point fingerprint vector { rj|j∈ΩkGenerate gauss hybrid models initial parameter, k ∈ K;It enablesThe Gaussian distribution model parameter of k class is calculated by formula (8), (9), (10):
B3. j-th of received signals fingerprint r is updated according to formula (7)jParticipation to k-th of local class localization region signal distributions model
Degree;
The wherein received signals fingerprint r of reference point jjIt is subordinate to the posterior probability of kth class Gaussian Profile model are as follows:
The received signals fingerprint r of reference point jjPosterior probability to the Gaussian distribution models of all classes and are as follows:
C3. according to updated participationThe Gaussian distribution model parameter of k class is recalculated with formula (8), (9), (10):
D3. b3, c3 are repeated, until
E3. stop parameter calculating, algorithm terminates;The participation of K class of N number of reference point at this timeAnd the ginseng of k Gaussian distribution model
Number ΦkTend to stablize, j ∈ N, k ∈ K;Record the participation vector of N number of reference point
The parameter of k Gaussian distribution model
It include that local positioning region is selected according to class label for local locating method in step (5), then according to class participation
It is as follows that multi-point joint carries out localization method:
1) local positioning region is selected according to class label,
Process is as follows:
A4. it is r that handheld terminal, which obtains the WIFI signal vector of current anchor point,location;R is calculated separately according to formula (7)location
To the participation of k classWherein Gaussian distribution model parameter is realized by maximization log-likelihood probability expectation method
The output for clustering fuzzy partitioning is specified, then current anchor point signal vector rlocationParticipation vector
B4. according to blocationIn the global participation vector of acquisitionMiddle selection participates in the ginseng of positioning
Examination point;Specific method is: for blocation, establish localization region collectionAccording to
ΘlocationIt is rightRetrieval, will be only to localization region collection ΘlocationMiddle class k has the reference of participation
Point is found out, and location reference point collection is established
2) it is positioned based on class participation multi-point joint,
After selecting local positioning region according to class label, the position of anchor point, the seat of anchor point are estimated by these reference points
Mark estimation are as follows:
Wherein n is location reference point number, kmIt is reference point m to the point of reference of anchor point, calculation method is:
Process is as follows:
A5. the point of reference k that location reference point concentrates n reference point is calculated separately by formula (15)m;
B5. the x coordinate of anchor point is calculated separately by formula (13), (14)With y-coordinate
C5. the estimation of anchor point coordinate is completed
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510280786.1A CN105044662B (en) | 2015-05-27 | 2015-05-27 | A kind of fingerprint cluster multi-point joint indoor orientation method based on WIFI signal intensity |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510280786.1A CN105044662B (en) | 2015-05-27 | 2015-05-27 | A kind of fingerprint cluster multi-point joint indoor orientation method based on WIFI signal intensity |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105044662A CN105044662A (en) | 2015-11-11 |
CN105044662B true CN105044662B (en) | 2019-03-01 |
Family
ID=54451339
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510280786.1A Active CN105044662B (en) | 2015-05-27 | 2015-05-27 | A kind of fingerprint cluster multi-point joint indoor orientation method based on WIFI signal intensity |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105044662B (en) |
Families Citing this family (26)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105372628A (en) * | 2015-11-19 | 2016-03-02 | 上海雅丰信息科技有限公司 | Wi-Fi-based indoor positioning navigation method |
CN106060779A (en) * | 2016-07-18 | 2016-10-26 | 北京方位捷讯科技有限公司 | Fingerprint feature matching method and device |
CN108243495A (en) * | 2016-12-23 | 2018-07-03 | 亿阳信通股份有限公司 | A kind of location fingerprint database building method, device and method of locating terminal |
CN107087256A (en) * | 2017-03-17 | 2017-08-22 | 上海斐讯数据通信技术有限公司 | A kind of fingerprint cluster method and device based on WiFi indoor positionings |
CN107087276A (en) * | 2017-03-17 | 2017-08-22 | 上海斐讯数据通信技术有限公司 | A kind of fingerprint base method for building up and device based on WiFi indoor positionings |
CN107027148B (en) * | 2017-04-13 | 2020-04-14 | 哈尔滨工业大学 | Radio Map classification positioning method based on UE speed |
CN107222925A (en) * | 2017-07-31 | 2017-09-29 | 广东工业大学 | A kind of node positioning method based on cluster optimization |
CN107517446A (en) * | 2017-09-28 | 2017-12-26 | 知谷(上海)网络科技有限公司 | Indoor orientation method and device based on Wi Fi focuses |
CN108345661B (en) * | 2018-01-31 | 2020-04-28 | 华南理工大学 | Wi-Fi clustering method and system based on large-scale Embedding technology |
CN109068274B (en) * | 2018-09-30 | 2020-06-16 | 电子科技大学 | Method for positioning complex indoor environment target under assistance of fine-grained fingerprint quality |
CN109444938B (en) * | 2018-10-23 | 2023-08-11 | 努比亚技术有限公司 | Composite positioning method of mobile terminal, mobile terminal and readable storage medium |
CN111385805B (en) * | 2018-12-29 | 2022-08-19 | 中兴通讯股份有限公司 | Method and device for generating radio frequency fingerprint information base and positioning grids |
US10761200B1 (en) | 2019-02-27 | 2020-09-01 | Osram Gmbh | Method for evaluating positioning parameters and system |
CN110060102B (en) * | 2019-04-18 | 2022-05-03 | 重庆邮电大学 | Bias label learning-based method for predicting positioning big data of shops where users are located |
CN110366100B (en) * | 2019-07-17 | 2020-11-10 | 京信通信系统(中国)有限公司 | Terminal positioning method, positioning device, readable storage medium and terminal equipment |
CN110519692B (en) * | 2019-09-12 | 2020-10-02 | 中南大学 | Positioning and partitioning method based on Bayes-k mean clustering |
CN111239716B (en) * | 2020-01-21 | 2022-05-06 | 蔡小雨 | Multi-WIFI rapid positioning method and device |
CN111464937B (en) * | 2020-03-23 | 2021-06-22 | 北京邮电大学 | Positioning method and device based on multipath error compensation |
CN111464938B (en) * | 2020-03-30 | 2021-04-23 | 滴图(北京)科技有限公司 | Positioning method, positioning device, electronic equipment and computer readable storage medium |
CN111623466B (en) * | 2020-05-08 | 2021-06-15 | 东南大学 | Method for determining air supply volume of air conditioner of large-space office building based on personnel distribution |
CN111669702A (en) * | 2020-06-04 | 2020-09-15 | 中国矿业大学 | Coarse and fine matching positioning method based on WIFI address and field intensity |
US11914059B2 (en) | 2020-12-15 | 2024-02-27 | Nokia Technologies Oy | Enhanced fingerprint positioning |
CN112734176B (en) * | 2020-12-28 | 2022-04-08 | 中创三优(北京)科技有限公司 | Charging station building method and device, terminal equipment and readable storage medium |
CN112866899B (en) * | 2021-01-04 | 2022-07-05 | 上海航天测控通信研究所 | Man-machine cooperation oriented lunar communication and navigation integration realization method and device |
CN115134815B (en) * | 2022-06-08 | 2024-02-09 | 深圳市三旺通信股份有限公司 | Wireless AP deployment optimization method, system, equipment and storage medium |
WO2024047378A1 (en) * | 2022-08-30 | 2024-03-07 | Wiliot, LTD. | Determining collective location of low energy wireless tags |
Citations (2)
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 |
CN104427610A (en) * | 2013-08-28 | 2015-03-18 | 中国电信集团公司 | Wi-Fi indoor positioning method and Wi-Fi indoor positioning server |
-
2015
- 2015-05-27 CN CN201510280786.1A patent/CN105044662B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104427610A (en) * | 2013-08-28 | 2015-03-18 | 中国电信集团公司 | Wi-Fi indoor positioning method and Wi-Fi indoor positioning server |
CN104185275A (en) * | 2014-09-10 | 2014-12-03 | 北京航空航天大学 | Indoor positioning method based on WLAN |
Non-Patent Citations (3)
Title |
---|
"Enhanced Fingerprinting in WLAN-based Indoor Positioning using Hybrid Search Techniques";Ayah Abusara et al.;《Communications,Signal Processing,and their Applications(ICCSPA),2015 International Conference on》;20150219;第1-6页 |
"基于改进K-means聚类算法的室内WLAN定位研究";陈望 等;《激光杂志》;20140731;第35卷(第7期);第11-14页 |
"基于聚类的快速Wi-Fi定位算法";刘兴川 等;《计算机工程》;20110430;第37卷(第8期);第285-287页 |
Also Published As
Publication number | Publication date |
---|---|
CN105044662A (en) | 2015-11-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105044662B (en) | A kind of fingerprint cluster multi-point joint indoor orientation method based on WIFI signal intensity | |
Song et al. | A novel convolutional neural network based indoor localization framework with WiFi fingerprinting | |
CN106851573B (en) | Log path loss model-based joint weighting K nearest neighbor indoor positioning method | |
CN102098780B (en) | Positioning method and device | |
Zhang et al. | A comprehensive study of bluetooth fingerprinting-based algorithms for localization | |
CN105813194B (en) | Indoor orientation method based on fingerprint database secondary correction | |
CN104602342B (en) | A kind of high efficiency indoor localization method based on iBeacon equipment | |
CN108627798B (en) | WLAN indoor positioning algorithm based on linear discriminant analysis and gradient lifting tree | |
CN106658704A (en) | Positioning method and system of starting point of indoor positioning | |
CN106912105A (en) | 3-D positioning method based on PSO_BP neutral nets | |
CN102638889A (en) | Indoor wireless terminal positioning method based on Bayes compression sensing | |
CN107703480A (en) | Mixed kernel function indoor orientation method based on machine learning | |
Ren et al. | A novel clustering algorithm for Wi-Fi indoor positioning | |
CN109327797A (en) | Mobile robot indoor locating system based on WiFi network signal | |
CN104540221A (en) | WLAN indoor positioning method based on semi-supervised SDE algorithm | |
CN107607122B (en) | The building of location fingerprint library and dynamic updating method towards indoor positioning | |
CN103634902A (en) | Novel indoor positioning method based on fingerprint cluster | |
CN109640262A (en) | A kind of localization method and system, equipment, storage medium based on mixed-fingerprint | |
CN109068274A (en) | A kind of complex indoor environment object localization method under fine granularity fingerprint quality auxiliary | |
CN110049441B (en) | WiFi indoor positioning method based on deep ensemble learning | |
CN109541537B (en) | Universal indoor positioning method based on ranging | |
CN111263295B (en) | WLAN indoor positioning method and device | |
CN107124761B (en) | Cellular network wireless positioning method fusing PSO and SS-ELM | |
CN106922017A (en) | Localization method and terminal | |
He et al. | Research on indoor and outdoor comprehensive positioning technology based on multi-source information assistance |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |