CN103763769B - Based on the indoor fingerprint positioning method that access point reselection procedure and self-adapting cluster divide - Google Patents
Based on the indoor fingerprint positioning method that access point reselection procedure and self-adapting cluster divide Download PDFInfo
- Publication number
- CN103763769B CN103763769B CN201310733950.0A CN201310733950A CN103763769B CN 103763769 B CN103763769 B CN 103763769B CN 201310733950 A CN201310733950 A CN 201310733950A CN 103763769 B CN103763769 B CN 103763769B
- Authority
- CN
- China
- Prior art keywords
- cluster
- fingerprint
- self
- rss
- optimal
- 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.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 102
- 238000003860 storage Methods 0.000 claims abstract description 19
- 230000004807 localization Effects 0.000 claims abstract description 17
- 230000008569 process Effects 0.000 claims abstract description 15
- 239000013598 vector Substances 0.000 claims description 24
- 230000005484 gravity Effects 0.000 claims description 18
- 230000007246 mechanism Effects 0.000 claims description 11
- 238000005259 measurement Methods 0.000 claims description 5
- 230000004992 fission Effects 0.000 claims description 3
- 230000003044 adaptive effect Effects 0.000 claims description 2
- 238000012790 confirmation Methods 0.000 claims description 2
- 238000013500 data storage Methods 0.000 claims description 2
- 230000000694 effects Effects 0.000 claims description 2
- 230000014759 maintenance of location Effects 0.000 claims description 2
- 238000005303 weighing Methods 0.000 claims description 2
- 238000004364 calculation method Methods 0.000 abstract description 9
- 238000007796 conventional method Methods 0.000 abstract description 7
- 230000008901 benefit Effects 0.000 abstract description 4
- 238000004422 calculation algorithm Methods 0.000 description 9
- 238000005516 engineering process Methods 0.000 description 7
- 238000000691 measurement method Methods 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 238000012935 Averaging Methods 0.000 description 1
- 240000007594 Oryza sativa Species 0.000 description 1
- 235000007164 Oryza sativa Nutrition 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 231100000768 Toxicity label Toxicity 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000010181 polygamy Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 235000009566 rice Nutrition 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
Landscapes
- Mobile Radio Communication Systems (AREA)
- Position Fixing By Use Of Radio Waves (AREA)
Abstract
The invention discloses a kind of indoor fingerprint positioning method divided based on AP reselection procedures and self-adapting cluster, including off-line data collecting stage and On-line matching positioning stage：The off-line data collecting stage divides the method that clusters and grid quantity is compressed with self-adapting cluster, each cluster is selected the most strong AP of stationkeeping ability with AP reselection procedure strategies and gathers, finally give the V RSS fingerprint bases of new definition；On-line matching positioning stage is using the self-adapting cluster division localization method based on V RSS fingerprint bases；It is an advantage of the invention that amount of calculation and computation complexity are significantly less than conventional method, the electric energy that positioning time and mobile terminal to be positioned are consumed in position fixing process can be saved；Present invention efficiently solves amount of storage present in conventional fingerprint indoor positioning process is big, computationally intensive, the uncontrollable problem of positioning precision.
Description
Technical field
The present invention relates to be based on access point（Access Point, abbreviation AP）The interior of reselection procedure and self-adapting cluster division refers to
Line localization method, belongs to the wireless location technology field in wireless communication system.
Background technology
At present, with wireless network widely available and mobile intelligent terminal fast development, location Based service receives
To increasing concern, obtained in fields such as emergency relief, health care, social networks, navigation and monitoring and widely should
With and show huge market prospects.
In open outdoor environment, global positioning system (GPS), property of network auxiliary global global position system (A-GPS)
The outdoor environment location information for meeting various accuracy requirements can be provided with network based positioning system.Indoors in environment, especially
It is hot spot region, such as supermarket, exhibition room, hospital, guild hall, theater, library, prison etc., need of the people to Indoor Location Information
Ask also increasingly stronger, to realize the comprehensive intelligent Services such as navigator fix, context-aware, the monitoring of personnel's goods and materials.But by
Cannot effectively be received in many room areas in gps signal, such as parking lot, hospital, and its positioning precision is relatively low, therefore it is outdoor
Location technology cannot meet the demand of indoor positioning.
Most of existing traditional indoor positioning technologies need extra specialized hardware facility, and use cost is high, covering
Scope is small, limits the popularization of location Based service environment indoors.It is strong based on WLAN (WLAN) and reception signal
The indoor positioning technologies of (RSS) are spent, being based entirely on existing WLAN infrastructure and mobile terminal just can independently realize positioning, nothing
Any extra special equipment is needed, low cost is widely distributed, and the accuracy requirement of most of indoor positioning applications can be met, into
It is the study hotspot of indoor positioning technologies.
WLAN indoor positioning technologies are segmented into two methods：Geometric measurement method and fingerprint location method.Geometric measurement method is
Refer to the position that user is calculated using geometric principle and path loss model.But because the extreme of indoor radio propagation environment is answered
Polygamy, the RSS propagation model precision of predictions of geometric measurement method are often relatively low, and positioning precision is not high, and need to be known a priori by access point
（Access Point, abbreviation AP）Particular location, it is impossible to meet actual demand.Fingerprint location method collection reference point locations
RSS samples, build fingerprint database, and fingerprint matching by RSS samples draws positioning result, fingerprint location method positioning precision
Apparently higher than geometric method, and require no knowledge about the position of AP.
But, in view of increasing user relies primarily on battery powered mobile device, and existing traditional WLAN refers to
Line localization method all has that amount of storage is big, computationally intensive, and the load and high energy consumption of terminal become urgently to be resolved hurrily asking
Topic.
Additionally, traditional WLAN fingerprint positioning method positioning precisions are uncontrollable, it is impossible to realize the user fixed to limited precision
Demand, how to realize that precision is controllable to meet different user to the different demands of required precision, how in WLAN indoor positionings
Amount of storage is saved in method, reduces computation complexity etc. as when previous important technical problem anxious to be resolved.
The content of the invention
It is an object of the invention to provide it is a kind of can overcome above-mentioned technical problem based on access point reselection procedure and adaptive
The indoor fingerprint positioning method for answering cluster to divide, the present invention be on the basis of conventional fingerprint localization method, with reference to new definition from
Dynamic cluster sorting technique and access point（Access Point, abbreviation AP）Reselection procedure strategy, enters to indoor mobile terminal to be positioned
Row positioning, is used to solve that amount of storage present in conventional fingerprint indoor positioning process is big, computationally intensive, positioning precision is uncontrollable
Problem.
Core technical features of the invention and thinking are optimal AP set reselection procedure, base based on geographical position and identification
Divide localization method in the self-adapting cluster of multiway tree.Multiway tree refers to quaternary tree or Octree herein, for plane positioning problem
Will be using quaternary tree, will be using Octree for space orientation problem.
The present invention is comprised the following steps：
Step one：If it is dry lattice to be conventionally evenly dividing area to be targeted（grid）, choose each grid
Geometric center position as a reference point, in each reference point locations all access points that can receive of measurement（Access
Point）Signal intensity, generate received signal strength fingerprint vector（RSS vectors）, set up basis RSS fingerprint bases；
Step 2：With reference to the AP reselection methods based on geographical position and identification, self-adapting cluster division cluster method and
Self-study mechanism, the variable RSS fingerprint bases of generation grid size（V-RSS fingerprint bases）；
Step 3：Mobile terminal to be positioned is moved into certain position, the RSS vectors of real-time reception are measured, repeatedly surveyed
Measure average value and carry out denoising；
Step 4：The RSS vectors that will be measured in step 3 are compared with V-RSS fingerprint bases, with based on multiway tree
Self-adapting cluster division localization method estimates the position of mobile terminal to be positioned；
Step 5：Output positioning result, the positioning result includes cluster numbering, the mobile terminal to be positioned of best match
Location estimation and uncertainty.
The present invention has following innovative characteristicses and beneficial effect：
（1）Optimal AP set reselection procedures based on geographical position and identification；
AP selections are that whole area to be targeted is applied to after once selecting in conventional fingerprint localization method；Or
By clustering algorithms, one clusters（cluster）Gather using one group of optimal AP.
In the present invention, although AP selections are also based on cluster for unit, AP selections will be carried out to each cluster, it is different
It is：First, the present invention is directed to different geographical position（Geographical cluster）Calculate different optimal AP set；Second, optimal AP set
Using identification as weighing criteria during selection；3rd, in the present invention, although submanifold is divided by female cluster, but
The optimal AP set that submanifold is used from female cluster is but probably different, because when optimal AP set is calculated, being based only on this
All reference points in cluster are calculated, that is to say, that female cluster includes different reference points from submanifold, by AP systems of selection,
The optimal AP set for drawing is also likely to be different.
AP reselection procedures mechanism searches out the most strong AP of space separating capacity for the cluster of diverse geographic location, and its benefit is removal
It is smaller to positioning contribution or even play the AP of side effect, both reduced the storage dimension of fingerprint, positioning precision is improve again.
（2）V-RSS fingerprint bases；
Conventional fingerprint localization method uses grid size to be fixed, participates in the AP of positioning and gather constant location fingerprint
Storehouse, its storage mode is with two dimension（Plane positioning）Or it is three-dimensional（Space orientation）Based on matrix；Compared with conventional fingerprint localization method
V-RSS fingerprint bases of the invention have 3 innovative points：Grid（Cluster）Size is not fixed, each grid（Cluster）AP set it is variable,
Using quaternary tree（Plane positioning）Or Octree storage（Space orientation）.
Above innovative point of the invention is to divide the method that clusters and self-study mechanism realization based on AP reselection procedures, self-adapting cluster
's：First, to different clusters, it is chosen in the cluster and separating capacity most strong AP in space is gathered as optimal AP, refers in V-RSS
Line storehouse only needs to store the fingerprint of optimal AP set, and the fingerprint of other AP is abandoned, and different clusters deposit fingerprint correspondence
AP set be incomplete same, it is therefore desirable in fingerprint base record AP numbering.
Secondly, using self-adapting cluster splitting method, whole area to be targeted is divided into cluster not of uniform size, the size of cluster
Depending on all AP the cluster space separating capacity, if certain sub-regions exist strong space separating capacity AP set,
The subregion can be divided into many less clusters, otherwise then the region can only be divided into a few larger cluster or even can only return
It is 1 cluster, its benefit is the reduction of the stored number of fingerprint.During actual location, do not possessing strong space separating capacity
AP subregion, even if splitting thin again, many again fingerprints are stored, for improving positioning precision also without gain.
In addition, identification threshold value is to determine whether current cluster can continue the key of division, the threshold value both can be by artificial root
Set according to practical experience, it is also possible to voluntarily obtained by self-study mechanism, its method is that identification threshold value is set into expected positioning to miss
The function of difference, wireless signal fluctuation characteristic and cluster size, determines the expression formula or numerical value of the function by way of self study
Corresponding relation.
（3）The tuning on-line method of the self-adapting cluster division based on multiway tree of the invention；
Used conventional fingerprint localization method carries out exhaustive contrast more by online real time fingerprint and offline fingerprint, find it is European away from
From（Or weighted euclidean distance）The minimum corresponding position of fingerprint is used as location estimation；By comparison, the present invention is used
Based on quaternary tree（Plane positioning）Or Octree（Space orientation）Search type localization method.
It is an advantage of the invention that the Euclidean distance of online real time fingerprint and all offline fingerprints need not be calculated, and calculating
Only need to consider the linear space that optimal AP collection is combined into during Euclidean distance, amount of calculation of the invention and computation complexity are significantly low
In conventional method, compared with conventional method, on the premise of it ensure that positioning precision, both reduce needs storage to the present invention
Fingerprint quantity reduces the dimension of fingerprint vector again；Positioning time and mobile terminal to be positioned can also be saved in position fixing process
The electric energy of middle consumption；Because conventional fingerprint localization method can only provide location estimation but location estimation cannot be evaluated, and
The present invention can not only provide location estimation can also provide the uncertainty of location estimation；Present invention efficiently solves tradition
Amount of storage present in fingerprint indoor positioning process is big, computationally intensive, the uncontrollable problem of positioning precision, with stronger practicality
Value and realistic meaning.
Brief description of the drawings
Fig. 1 is overview flow chart of the invention；
Fig. 2 is generation V-RSS fingerprint base exemplary plots during off-line data collecting of the invention；
Fig. 3 is that off-line phase self-adapting cluster of the invention divides the flow chart that clusters；
Fig. 4 is the V-RSS fingerprint base exemplary plots stored with quaternary tree of the invention；
Fig. 5 is on-line stage self-adapting cluster division positioning flow figure of the invention；
Fig. 6 is tuning on-line result exemplary plot of the invention；
Fig. 7 is tuning on-line process example figure of the invention.
Specific embodiment
The present invention will be described in detail with reference to the accompanying drawings and examples.As shown in figure 1, the present invention is divided into offline fingerprint
Two stages of acquisition phase and On-line matching positioning stage.
In the offline fingerprint collecting stage of the invention, be on the basis of traditional RSS fingerprint bases, to be clustered with self-adapting cluster division
Method is compressed to grid quantity, each cluster is selected the most strong AP of stationkeeping ability with AP reselection procedure strategies and gathers, finally
The V-RSS fingerprint bases for newly defining are obtained, on the premise of positioning precision is ensured, both reducing needs the fingerprint quantity of storage, and
The dimension of fingerprint vector is reduced, after dividing self-study mechanism with cluster, above procedure is automatically performed, without manual intervention.
The offline fingerprint collecting stage of the invention mainly completes the foundation of fingerprint collecting and fingerprint base, including sets up basis
RSS fingerprint bases and V-RSS fingerprint bases are set up on this basis；
If it is dry lattice to be first evenly dividing area to be targeted, select the geometric center of each grid as a reference point；So
Taken multiple measurements in each reference point afterwards, store the signal intensity of all AP that can be measured, generated after averaging RSS to
Amount, is stored in basic fingerprint database；
The foundation of V-RSS fingerprint databases, carries out cluster Pre-splitter to area to be targeted first, then selects positioning in the cluster
The most strong AP of ability constitutes the optimal AP set of the cluster, then according to certain principle（For example, identification of the optimal AP set to the region
Whether ability exceedes threshold value）Judge whether the cluster is necessary to continue cluster division, and the related data of the cluster is stored in V-RSS fingerprints
In storehouse.Hereafter, for be necessary continue cluster divide cluster carry out again cluster Pre-splitter, AP reselection procedures, judgement can continue division
Flow, untill all clusters need not all divide again.After the completion of self-adapting cluster division, whole region is divided into size not
Consistent many clusters, these clusters all need not continue to division（Because all AP in the region do not possess segment the cluster
Ability）, the longer sides of the cluster minimum positioning precision that can be considered as the cluster long；V-RSS fingerprint databases of the invention are with quaternary tree
（Plane positioning）Or Octree（Space orientation）Form storage.
On-line matching positioning stage of the invention has broken the method based on the exhaustive matching of traditional RSS fingerprint bases, the present invention
The self-adapting cluster division localization method based on V-RSS fingerprint bases is employed, so as to greatly reduce computation complexity.It is of the invention
On-line matching positioning stage has main steps that selection is most matched with real time fingerprint from V-RSS fingerprint bases cluster, completes space
Positioning.The signal intensity of all AP for being received to mobile terminal to be positioned first carries out real-time sampling and denoising, records RSS
Vector；Then with based on quaternary tree（Plane positioning）Or Octree（Space orientation）The self-adapting cluster location algorithm of search, from
Root cluster starts, and current cluster is divided into four or eight submanifolds, calculates real time fingerprint and four or eight center of gravity fingerprints of submanifold
Euclidean distance, the nearest submanifold of selection Euclidean distance the submanifold continued again as coarse positioning result, then cluster division, it is European away from
From the flow such as calculating and selection, untill current submanifold can not divide again, exported the cluster as positioning result；With the cluster
Geometric center as mobile terminal to be positioned location estimation, it is long as the uncertain of positioning result using the longer sides of the cluster
Degree.Specific embodiment of the invention below will be illustrated by taking plane positioning problem as an example.Here is related in accompanying drawing of the invention
Each variable represented by implication Verbose Listing：
State for convenience, the present invention defines the numbering of cluster first：Define the generally root cluster of area to be targeted, cluster
Numbering is 0.It is 1,2,3,4 by four submanifold number consecutivelies if root cluster can divide, this four clusters are the 1st layer of cluster.It is right again
1st layer of all clusters enter line splitting judgement, by taking cluster 2 as an example（Other are similarly）If cluster 2 can continue division, by four
Submanifold number consecutively is 21,22,23,24, and these clusters are the 2nd layer of cluster；If cluster 2 cannot continue division, in the absence of cluster 21,
22,23,24.The like can obtain the coding rule of all clusters, for example in the figure 7, the cluster q=of red-label position
423, it represents the 2nd the 3rd of submanifold the grandson's cluster of cluster 4, and the cluster is the 3rd layer of cluster.
The present invention is comprised the following steps：
Step one：If it is dry lattice to be conventionally evenly dividing area to be targeted（grid）, choose each grid
Geometric center position as a reference point, in each reference point locations all access points that can receive of measurement（Access
Point, AP）Signal intensity, generate received signal strength fingerprint vector（RSS vectors）, set up basis RSS fingerprint bases；By i-th
The Finger-print labelling method of individual reference point is si=(si,1,si,2…si,M）, i=1,2 ... C；Then basis RSS fingerprint bases are
Step 2：With reference to the method that clusters that the AP reselection methods based on geographical position and identification, self-adapting cluster divide
And self-study mechanism, the variable RSS fingerprint bases of generation grid size（V-RSS fingerprint bases）；
Fig. 2 illustrates the specific steps how area to be targeted generates V-RSS fingerprint bases.Fig. 3 then specifically illustrates each
Layer grid is the flow for how dividing and determining generation, and each step is illustrated according to Fig. 3；
（0）When initial, set one and store the empty stack of cluster to be divided, and current cluster is set to root cluster；
（1）If current cluster is q, Pre-splitter is done to q, as shown in Fig. 2 virtual cross is made to cluster q divide equally, after Pre-splitter
To four submanifold q1,q2,q3,q4；
（2）Calculate the center of gravity fingerprint vector of each submanifold（Abbreviation center of gravity fingerprint）；With submanifold q1As a example by, its center of gravity fingerprint rq1=
(rq1,1,rq1,2...rq1,M), wherein rq1,jRepresent j-th AP in q1Average received signal strength in submanifold；
（3）AP reselection procedures；AP reselection procedures refer to that current cluster needs to re-start AP selections after changing, and are finally different geographical positions
The cluster put defines the optimal AP set of oneself；The system of selection of optimal AP set is that selection stationkeeping ability is most strong from AP set
Some AP, the purpose is to filter, stationkeeping ability is poor, the AP of stability difference reduces the calculating of tuning on-line to improve positioning precision
Amount；The existing many algorithms of AP selections, the foundation that the present invention select using identification as AP, identification be defined as between cluster distance and
The ratio of cluster internal variance, chooses the maximum some AP of identification and constitutes optimal AP set；In the algorithm, the definition of identification is drawn
Fisher criterion functions are entered；Fisher criterions are a kind of Multielement statistical analysis methods for differentiating individual generic, and it can
Separating degree between two classifications is quantitatively described；Its definition is：
Wherein, J (Y) is more big, and separating degree is bigger between illustrating two classes；And the criterion can be generalized between multiple classes
Separating degree is described；It is applied in AP algorithms of the invention, tq,jRepresent identifications of j-th AP in cluster q, tq,jMore it is big then
Represent four submanifold qs of j-th AP in cluster q1, q2, q3, q4Between discrimination it is bigger（Separating capacity to cluster q is stronger）；More than
AP selection algorithms are only for example, and the inventive method can also use other AP selection algorithms；
Select concretely comprising the following steps for optimal AP set：
（31）First calculate the center of gravity fingerprint and cluster internal variance of submanifold；
With cluster q1In as a example by j-th AP, its center of gravity fingerprint
Its cluster internal variance
R can similarly be obtainedq2,j,rq3,j,rq4,j,
（32）To cluster q, the identification of j-th AP is defined as：
Identification vector tq=(tq,1, tq,2..., tq,M)；
（33）To tq,jSorted by descending order, take out top n and constitute optimal AP set, corresponding AP numberings
There is vector vqIn；
（34）Respectively obtain four cluster fingerprints of submanifold.For example, submanifold q1Cluster fingerprint is defined as referring to from rq1In select vq
Corresponding fingerprint, obtains the submanifold q after AP is selected1Center of gravity fingerprint uq1, uq1Can be expressed as：
Can similarly obtainWithDifference be that the latter includes all AP in cluster q1Fingerprint, its dimension
It is M, and the former has only included the fingerprint of optimal AP set, its dimension is N, N<M；
（4）Defined function Γ（tq）, for judging that current cluster could divide：
Wherein ε is that can divide thresholding（Identification threshold value）, both can rule of thumb by manually setting, also can be by self-study mechanism
Automatically obtain, its method is the function that identification threshold value is set to expected position error, wireless signal fluctuation characteristic and cluster size,
The expression or numerical value corresponding relation of the function are determined by way of self study.
（5）Data storage：By cluster numbering q, discriminant function Γ（tq）Value, the optimal AP set v of current clusterqStore V-RSS
In fingerprint base；
If Γ（tq）=1, jump to step（6）；
If Γ（tq）=0, jump to step（7）；
（6）Line splitting is entered to cluster q；Actually to the confirmation process of cluster Pre-splitter, cross is carried out to cluster q and is divided equally, obtained
To four submanifold q1,q2,q3,q4, store vqIn the center of gravity fingerprint u that four submanifolds are producedq1,uq2,uq3,uq4, by cluster q2,q3,q4Pressure
In stacking, current cluster is set to q1, jump to step（1）；
（7）Current cluster q can not divide again, and a new cluster is taken out in return stack as current cluster, jump to step（1）；If
The empty then whole cluster fission process of stack terminates, and the generation of V-RSS fingerprint bases is finished.
V-RSS fingerprint bases will be stored in quaternary tree mode, and Fig. 4 gives an example.
The process of generation V-RSS fingerprint bases is exemplified below.As shown in Fig. 2 being with the fission process of the 1st layer of cluster 4
Example：
（1）Cluster 4 generates four submanifolds 41,42,43,44 by Pre-splitter；
（2）J-th AP is respectively r in the location fingerprint and variance of four submanifolds41,j, r42,j, r43,j, r44,jWith It is possible thereby to calculate its identification t in cluster 44,j；
（3）Assuming that 10 signals of AP are occurred in that in space altogether, it is contemplated that computation complexity needs therefrom to select 4 AP ginsengs
Reorder position（That is M=10, N=4）, to t4,j(j=1,2 ... 10) sort, and take out maximum 4 and gather as optimal AP；Assuming that identification
The AP that degree is taken the first four place is respectively the 7th, 9,1,5 AP, then v4=(7,9,1,5), u4=(r4,1,r4,5,r4,7,r4,9), the two numbers
According to being stored into the node 4 of V-RSS fingerprint bases.
（4）Calculate the average identification of optimal AP set, the Γ if it is more than threshold epsilon（t4）=1, otherwise Γ（t4）=0；It is false
If Γ（t4）=1, then the Pre-splitter of cluster 4 be identified effectively, formed four new clusters 41,42,43 and 44.
The flow for carrying out similar cluster 4 successively with 44 to cluster 41,42,43, can more be refined to whole area to be targeted
Segmentation.It is worth noting that, judging whether the foundation that can divide is Γ to cluster 4（t4）, the AP that it is used is region in all AP
Most strong first four of segmentation ability, i.e., (7,9,1,5)；And judge whether the foundation that can divide is Γ to cluster 41（t41）, what it was used
AP is not necessarily identical with (7,9,1,5), and this is AP reselection procedures.
Calculate the amount of storage of off-line phase finger print data：
By taking the V-RSS fingerprint bases in Fig. 2 as an example, C=16*16=256 in figure, M=10, N=4, V-RSS fingerprint base have the 1st
Layer cluster 1, the 2nd layer of cluster 10, the 3rd layer of cluster 4, wherein the 4th layer of cluster 16, each cluster need 1 cluster numbering of storage, 1 division
Judgment variables and 4 fingerprints, total storage capacity is：（1+10+4+16）*（1+1+4）=186；And the amount of storage of tradition RSS fingerprint bases
For：256*（1+1+10）=3072, the fingerprint amount of storage of the inventive method and the ratio of conventional method can be calculated：
ξ=186/3072≈6%
Step 3：Mobile terminal to be positioned is moved into certain position, the RSS vectors of real-time reception are measured：
w=(w1,…,wj,…wM)
Step 4：The RSS vectors that will be measured in step 3 are compared with V-RSS fingerprint bases, with based on multiway tree
Self-adapting cluster division localization method estimates the position of mobile terminal to be positioned.Step 4 is illustrated in conjunction with Fig. 5.For
Facilitate expression, cluster is represented with the node in quaternary tree, i.e., current present node λkIt is equivalent to current cluster q.
（1）From root node λk, k=0 starts；
（2）The data of V-RSS fingerprint library storages are read, present node λ is judgedkWhether division can be continued：If can be with
Continue to divide, continue executing with step（3）, otherwise skip to step（7）；
（3）Four child nodes of next layer of present node are designated as θ=1,2,3,4；Read per height from V-RSS fingerprint bases
The corresponding center of gravity fingerprint u of nodeθ,θ∈[1,4]；
（4）The real time fingerprint of mobile terminal to be positioned is obtained after the optimal AP Resource selections of present node
（5）Found out from four child nodes withEuclidean distance is nearest, and its numbering is designated as λk+1；
λk+1=arg min||dθ||,θ∈[1,4]；
Euclidean distance formula：
（6）λk+1As present node, step is skipped to（2）；
（7）The corresponding cluster of present node is best match cluster, and output cluster information is used as positioning result, including cluster numbering
（Node serial number）With the size of cluster.
Step 5：Output positioning result, the positioning result includes cluster numbering, the mobile terminal to be positioned of best match
Estimated location and uncertainty, specific embodiment is：Using the geometric center of best match cluster as mobile end to be positioned
The location estimation at end, using the longer sides of the best match cluster uncertainty as positioning result long.
Step 4 and step 5 is exemplified below.
As shown in Figure 7, it is assumed that the position that mobile terminal to be positioned is marked in figure, and area to be targeted is a length of side
It is the square area of 16m, the flow of location algorithm is explained with reference to Dynamic Graph.
（1）Current cluster is root cluster, it is assumed that root cluster can be split into four submanifolds 1,2,3,4, can be with from V-RSS fingerprint bases
Read the optimal AP set v of root cluster0And four center of gravity fingerprint u of submanifold1、u2、u3And u4；
（2）If it is w that mobile terminal to be positioned measures the fingerprint vector for obtaining in real time, v is selected from w0Middle AP is corresponding
Fingerprint, the real time fingerprint vector after composition AP reselection procedures
（3）CalculateWith u1、u2、u3And u4Euclidean distance, ifWith u4Euclidean distance it is minimum, then judge to be positioned
Mobile terminal in cluster 4；
（4）Assuming that cluster 4 can continue division, four submanifolds 41,42,43,44 are obtained, can read from V-RSS fingerprint bases
Go out the optimal AP set v of cluster 44And four center of gravity fingerprint u of submanifold41、u42、u43And u44；V is selected from w4Middle AP is corresponding
Fingerprint, the real time fingerprint vector after composition AP reselection proceduresCalculateWith u41、u42、u43And u44Euclidean distance, ifWith
u42Euclidean distance it is minimum, then judge mobile terminal to be positioned in cluster 42；
（5）Previous step is repeated, continues to judge mobile terminal to be positioned in cluster 423；Next cluster 423 has been found
Can not divide, therefore the geometric center of mobile terminal location to be positioned to cluster 423, its coordinate is（9,5）；In view of cluster
423 length of side is 2 meters, therefore positioning uncertainty is taken as 2 meters, and final positioning result is：Q=423, P=(9 ± 1,5 ± 1)
Rice.
It is the schematic diagram of mobile terminal seeking best match cluster to be positioned that Fig. 6 is；Quaternary tree refers to the storage shape of V-RSS
Formula, optimal match point is calculated using quaternary tree；The process both can be by the intelligent mobile terminal to be positioned with respective handling ability
Voluntarily complete, can also be completed by AP by after terminal to report AP to be positioned.
The computation complexity of tuning on-line：
The amount of calculation of tuning on-line is mainly consumed and calculated in the Euclidean distance of location fingerprint and real time fingerprint, separately below
From the theoretical and contrast of two angle calculation method of the present invention of example and conventional fingerprint localization method on computation complexity.
If conventional fingerprint localization method finds the nearest fingerprint positions of Euclidean distance, computation complexity using the method for exhaustion
O (C*M) is should be, the method for the present invention only needs to compare in the fingerprint of portion with part AP, maximum computational complexity
Should be o (4N*log4C), the ratio of computation complexity can be calculated：
By taking above-mentioned example as an example, C=16*16=256, M=10, N=4, the Euclidean distance calculation times of conventional method are 256*
10=2560 times, if conventional method also completes position fixing process only with 4 AP, calculation times are dropped to 256*4=1024 times；
The Euclidean distance calculation times of the method for the present invention are（4+4+4）* 4=48 times, can calculate：
η=48/1024≈5%
I.e. amount of calculation of the method for the present invention in the tuning on-line stage is the 5% of conventional method amount of calculation, i.e., side of the invention
Method greatlys save the electric energy of mobile terminal consumption to be positioned in positioning time and position fixing process.
For convenience of description, launch to describe as example with plane positioning problem in specification of the invention, but the present invention
Can be used for space orientation problem, only the positioning region in this specification need to be changed to located space, grid is changed to three-dimensional grid, four
Fork tree is changed to Octree, and the grid of step one is changed to three-dimensional grid, the quaternary tree in step 2 and step 4 is changed into Octree,
Other steps, flow and algorithm all same.
The above, specific embodiment only of the invention, but protection scope of the present invention is not limited thereto, and it is any
In scope disclosed by the invention, the change or replacement that can be readily occurred in should all be contained those familiar with the art
Lid is within the scope of the invention as claimed.
Claims (6)
1. the indoor fingerprint positioning method for being divided based on access point reselection procedure and self-adapting cluster, it is characterised in that including following step
Suddenly：
Step one：If it is dry lattice to be conventionally evenly dividing area to be targeted, the geometric center of each grid is chosen
Position as a reference point, the signal intensity of all access points that can be received in the measurement of each reference point locations, generation receives letter
Number intensity fingerprint vector, i.e. RSS vectors, sets up basis RSS fingerprint bases；
Step 2：Divide cluster method and self-study with reference to the AP reselection methods based on geographical position and identification, self-adapting cluster
Habit mechanism, the variable RSS fingerprint bases of generation grid size, i.e. V-RSS fingerprint bases；
(0) when initial, set one and store the empty stack of cluster to be divided, and current cluster is set to root cluster；
(1) it is q to set current cluster, and Pre-splitter is done to q, makees virtual cross to cluster q and divides equally, and four submanifold q are obtained after Pre-splitter1,
q2,q3,q4；
(2) the center of gravity fingerprint vector of each submanifold is calculated；With submanifold q1As a example by, its center of gravity fingerprint rq1=(rq1,1,rq1,2…rq1,M),
Wherein rq1,jRepresent j-th AP in q1Average received signal strength in submanifold；
(3) AP reselection procedures, current cluster needs to re-start AP selections after changing, and the final cluster for diverse geographic location defines oneself
Optimal AP set；Select concretely comprising the following steps for optimal AP set：
(31) the center of gravity fingerprint and cluster internal variance of submanifold are first calculated；
With cluster q1In as a example by j-th AP, its center of gravity fingerprintSi,jRepresent j-th AP in i-th reference point
Location fingerprint；Cq1Represent cluster q1Middle reference is counted out；
Its cluster internal variance
R can similarly be obtainedq2,j,rq3,j,rq4,j,
(32) to cluster q, the identification of j-th AP is defined as：
Identification vector tq=(tq,1, tq,2..., tq,M)；
(33) to tq,jSorted by descending order, take out top n and constitute optimal AP set, corresponding AP numbered and is existed
Vector vqIn；
(34) four cluster fingerprints of submanifold are respectively obtained；For example, submanifold q1Cluster fingerprint be defined as refer to fromIn select vqIt is corresponding
Fingerprint, obtains the submanifold q after AP is selected1Center of gravity fingerprintCan be expressed as：
Can similarly obtain WithDifference be that the latter includes all AP in cluster q1Fingerprint, its dimension be M,
And the former has only included the fingerprint of optimal AP set, its dimension is N, N<M；
(4) defined function Γ (tq), for judging that current cluster could divide：
Wherein ε rule of thumb by manually setting can also can both be automatically obtained, its method for that can divide thresholding by self-study mechanism
It is the function that identification threshold value is set to expected position error, wireless signal fluctuation characteristic and cluster size, by the side of self study
Formula determines the expression or numerical value corresponding relation of the function；
(5) data storage：By cluster numbering q, discriminant function Γ (tq) value, the optimal AP set v of current clusterqStore V-RSS fingerprints
In storehouse；
If Γ (tq)=1, jumps to step (6)；
If Γ (tq)=0, jumps to step (7)；
(6) line splitting is entered to cluster q；Actually to the confirmation process of cluster Pre-splitter, cross is carried out to cluster q and is divided equally, obtain four
Individual submanifold q1,q2,q3,q4, store vqIn the center of gravity fingerprint u that four submanifolds are producedq1,uq2,uq3,uq4, by cluster q2,q3,q4Press-in stack
In, current cluster is set to q1, jump to step (1)；
(7) current cluster q can not divide again, and a new cluster is taken out in return stack as current cluster, jump to step (1)；If stack is empty
Then whole cluster fission process terminates, and the generation of V-RSS fingerprint bases is finished；
Step 3：Mobile terminal to be positioned is moved into certain position, the RSS vectors of real-time reception are measured, repeatedly measurement takes
Average value simultaneously carries out denoising；
Step 4：The RSS vectors that will be measured in step 3 are compared with V-RSS fingerprint bases, with based on the adaptive of multiway tree
Cluster is answered to divide the position that localization method estimates mobile terminal to be positioned；Cluster, i.e. present node are represented with the node in quaternary tree
λkIt is equivalent to current cluster q；
(1-1) is from root node λk, k=0 starts；
(1-2) reads the data of V-RSS fingerprint library storages, judges present node λkWhether division can be continued：If can continue
Division, continues executing with step (1-3), otherwise skips to step (1-7)；
Four child nodes of next layer of (1-3) present node are designated as θ=1,2,3,4；Read per height section from V-RSS fingerprint bases
The corresponding center of gravity fingerprint u of pointθ, θ ∈ [Isosorbide-5-Nitrae]；
The real time fingerprint of mobile terminal (1-4) to be positioned is obtained after the optimal AP Resource selections of present node
(1-5) found out from four child nodes withEuclidean distance is nearest, and its numbering is designated as λk+1；
λk+1=arg min | | dθ| |, θ ∈ [Isosorbide-5-Nitrae]；
Euclidean distance formula：
(1-6)λk+1As present node, step (1-2) is skipped to；
The corresponding cluster of (1-7) present node is best match cluster, and output cluster information is used as positioning result, including cluster numbering and cluster
Size；
Step 5：Output positioning result, the positioning result includes cluster numbering, the position of mobile terminal to be positioned of best match
Put estimation and uncertainty.
2. the indoor fingerprint positioning method divided based on access point reselection procedure and self-adapting cluster according to claim 1, its
It is characterised by, accesses point selection and be although also based on cluster for unit, will carries out AP selections to each cluster, but unlike：The
One, calculate different optimal AP set for different geographical position；Second, identification conduct is used during optimal AP Resource selections
Weighing criteria；3rd, although submanifold is divided by female cluster, but the optimal AP set that submanifold is used with female cluster but may
It is different, because when optimal AP set is calculated, all reference points being based only in the cluster are calculated, that is,
Say, female cluster includes different reference points from submanifold, by AP systems of selection, the optimal AP set for drawing is also likely to be different
's.
3. the indoor fingerprint positioning method divided based on access point reselection procedure and self-adapting cluster according to claim 1, its
It is characterised by, access point AP reselection procedures mechanism searches out the most strong AP of space separating capacity for the cluster of diverse geographic location, and its is good
Place is to eliminate smaller to positioning contribution or even a side effect access point AP.
4. the indoor fingerprint positioning method divided based on access point reselection procedure and self-adapting cluster according to claim 1, its
It is characterised by, first, to different clusters, is chosen in the cluster and separating capacity most strong AP in space is gathered as optimal AP,
V-RSS fingerprint bases only need to store the fingerprint of optimal AP set, and the fingerprint of other AP is abandoned, and different clusters deposit finger
The corresponding AP set of line is incomplete same, it is therefore desirable to the numbering of AP is recorded in fingerprint base.
5. the indoor fingerprint positioning method divided based on access point reselection procedure and self-adapting cluster according to claim 1, its
It is characterised by, using self-adapting cluster splitting method, whole area to be targeted is divided into cluster not of uniform size, the size of cluster depends on
In all AP the cluster space separating capacity, if certain sub-regions exist strong space separating capacity AP set, the son
Region can be divided into many less clusters, otherwise then the region can only be divided into a few larger cluster or even can only be classified as 1
Individual cluster, to reduce the stored number of fingerprint.
6. the indoor fingerprint positioning method divided based on access point reselection procedure and self-adapting cluster according to claim 1, its
It is characterised by, identification threshold value is to determine whether current cluster can continue the key of division, and the threshold value can either be by manually according to reality
Border experience is set, it is also possible to voluntarily obtained by self-study mechanism, its method is that identification threshold value is set into expected position error, nothing
The function of line signal fluctuation feature and cluster size, determines that expression formula or the numerical value correspondence of the function are closed by way of self study
System.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310733950.0A CN103763769B (en) | 2013-12-26 | 2013-12-26 | Based on the indoor fingerprint positioning method that access point reselection procedure and self-adapting cluster divide |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310733950.0A CN103763769B (en) | 2013-12-26 | 2013-12-26 | Based on the indoor fingerprint positioning method that access point reselection procedure and self-adapting cluster divide |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103763769A CN103763769A (en) | 2014-04-30 |
| CN103763769B true CN103763769B (en) | 2017-07-04 |
Family
ID=50530910
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310733950.0A Expired - Fee Related CN103763769B (en) | 2013-12-26 | 2013-12-26 | Based on the indoor fingerprint positioning method that access point reselection procedure and self-adapting cluster divide |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103763769B (en) |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103997783B (en) * | 2014-05-14 | 2017-03-29 | 无锡北邮感知技术产业研究院有限公司 | A kind of outdoor cluster match localization method and device |
| KR101846831B1 (en) * | 2014-06-30 | 2018-04-09 | 후아웨이 테크놀러지 컴퍼니 리미티드 | Method and apparatus for indoor positioning |
| CN105120476B (en) * | 2015-07-20 | 2018-10-16 | 北京邮电大学 | Based on cochannel interference compensation and the modified indoor orientation method of RSS fingerprints |
| CN105657736A (en) * | 2016-03-22 | 2016-06-08 | 清华大学 | Method and device for starting access points of wireless local area network |
| CN107295538B (en) * | 2016-03-30 | 2022-03-18 | 日本电气株式会社 | Positioning reliability calculation method, positioning method using reliability and positioning instrument |
| CN105916113B (en) * | 2016-06-01 | 2019-04-12 | 北京邮电大学 | A kind of localization method and device |
| CN108107407A (en) * | 2017-12-18 | 2018-06-01 | 黑龙江大学 | A kind of three-dimensional sound localization method based on spatially distributed microphone |
| CN110392425B (en) * | 2019-08-20 | 2020-03-17 | 核芯互联科技(青岛)有限公司 | Indoor positioning method, device and system and storage medium |
| CN111757249B (en) * | 2020-05-21 | 2022-05-03 | 重庆邮电大学 | Multi-feature fuzzy mapping access point optimization method based on information gain ratio |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101657014A (en) * | 2009-09-11 | 2010-02-24 | 哈尔滨工业大学 | WLAN indoor positioning method based on nearest neighbor probability method |
| CN102170697A (en) * | 2011-04-06 | 2011-08-31 | 北京邮电大学 | Indoor positioning method and device |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| SG157355A1 (en) * | 2004-10-29 | 2009-12-29 | Skyhook Wireless Inc | Location beacon database and server, method of building location beacon database, and location based service using same |
| US8660577B2 (en) * | 2009-12-04 | 2014-02-25 | Nokia Corporation | Method and apparatus for on-device positioning using compressed fingerprint archives |
-
2013
- 2013-12-26 CN CN201310733950.0A patent/CN103763769B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101657014A (en) * | 2009-09-11 | 2010-02-24 | 哈尔滨工业大学 | WLAN indoor positioning method based on nearest neighbor probability method |
| CN102170697A (en) * | 2011-04-06 | 2011-08-31 | 北京邮电大学 | Indoor positioning method and device |
Non-Patent Citations (4)
| Title |
|---|
| MinWang，ChunkaiZhang.ResidualRanking: a robust Access-Point selection strategy for indoor location tracking.《Proceedings of the 2009 IEEE International Conference on Systems, Man, and Cybernetics》.2009, * |
| Qingyuan Zhao, Sheng Zhang, Xingchuan Liu, and Xiaokang Lin.An Effective Preprocessing Scheme for WLAN-Based Fingerprint Positioning Systems.《IEEE》.2010, * |
| 一种基于AP ID 过滤的WLAN 位置指纹定位算法;赵庆贺， 邓 平， 陈 佳;《通信技术》;20121231;全文 * |
| 都伊林.一种模糊聚类KNN 位置指纹定位算法.《网络与通信》.2012, * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103763769A (en) | 2014-04-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103763769B (en) | Based on the indoor fingerprint positioning method that access point reselection procedure and self-adapting cluster divide | |
| Song et al. | A novel convolutional neural network based indoor localization framework with WiFi fingerprinting | |
| Song et al. | Cnnloc: Deep-learning based indoor localization with wifi fingerprinting | |
| CN103476118B (en) | A kind of WLAN indoor location fingerprint positioning method for monitoring in real time | |
| CN108540929B (en) | Indoor fingerprint positioning method based on RSSI signal strength sequencing | |
| CN103747419B (en) | A kind of indoor orientation method based on signal strength difference and dynamic linear interpolation | |
| CN109275095A (en) | A kind of indoor locating system based on bluetooth, positioning device and localization method | |
| CN110213710A (en) | A kind of high-performance indoor orientation method, indoor locating system based on random forest | |
| CN105682048B (en) | Fingerprint positioning method in Subspace Matching room based on PCA under cellular network environment | |
| CN109151750A (en) | A kind of LTE indoor positioning floor method of discrimination based on Recognition with Recurrent Neural Network model | |
| CN108882172A (en) | A kind of prediction technique of the indoor moving track data based on HMM model | |
| CN110401977A (en) | A kind of more floor indoor orientation methods returning more Classification and Identification devices based on Softmax | |
| CN106793085A (en) | Fingerprint positioning method based on normality assumption inspection | |
| CN110351660A (en) | A kind of bluetooth indoor orientation method based on two-step fingerprint matching framework | |
| CN113596989A (en) | Indoor positioning method and system for intelligent workshop | |
| CN108810799A (en) | A kind of more floor indoor orientation methods and system based on linear discriminant analysis | |
| Liu et al. | An indoor WLAN location algorithm based on fingerprint database processing | |
| CN113453148B (en) | Indoor position fingerprint positioning method combining deep learning and weighted K-neighbor algorithm | |
| Wang et al. | Optimal target tracking based on dynamic fingerprint in indoor wireless network | |
| Maaloul et al. | Bluetooth beacons based indoor positioning in a shopping malls using machine learning | |
| CN109541537A (en) | A kind of pervasive indoor orientation method based on ranging | |
| Chen et al. | Deep neural network based on feature fusion for indoor wireless localization | |
| CN108521631A (en) | A kind of mobile AP recognition methods towards indoor positioning | |
| Yan et al. | Indoor position method of industrial robot based on wifi fingerprint position technology | |
| CN108519579B (en) | WiFi fingerprint positioning method for analyzing optimal AP based on interval overlapping degree |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170704 |