CN106792465B - A kind of indoor fingerprint map constructing method based on crowdsourcing fingerprint - Google Patents
A kind of indoor fingerprint map constructing method based on crowdsourcing fingerprint Download PDFInfo
- Publication number
- CN106792465B CN106792465B CN201611219485.9A CN201611219485A CN106792465B CN 106792465 B CN106792465 B CN 106792465B CN 201611219485 A CN201611219485 A CN 201611219485A CN 106792465 B CN106792465 B CN 106792465B
- Authority
- CN
- China
- Prior art keywords
- fingerprint
- grid
- crowdsourcing
- fitting
- supported collection
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/02—Services making use of location information
- H04W4/029—Location-based management or tracking services
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/29—Geographical information databases
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/02—Services making use of location information
- H04W4/021—Services related to particular areas, e.g. point of interest [POI] services, venue services or geofences
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/80—Services using short range communication, e.g. near-field communication [NFC], radio-frequency identification [RFID] or low energy communication
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Remote Sensing (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Collating Specific Patterns (AREA)
- Position Fixing By Use Of Radio Waves (AREA)
Abstract
A kind of indoor fingerprint map constructing method based on crowdsourcing fingerprint, belong to the indoor positioning technologies based on fingerprint, solve the problems, such as it is existing construct that crowdsourcing fingerprint positions mark inaccuracy, dimension present in indoor fingerprint map are different and positional accuracy caused by being unevenly distributed is not high and computationally intensive based on crowdsourcing fingerprint, for communicating and radio network technique field.The present invention includes grid division step, obtains crowdsourcing fingerprint step, crowdsourcing fingerprint splitting step, crowdsourcing fingerprint quantity judgment step, AP screening step, directly construction grid fingerprint step and fingerprint surface fitting step.The present invention reduces fingerprint collecting workload, be conducive to improve positional accuracy, while reducing fingerprint comparison workload.
Description
Technical field
The invention belongs to the indoor positioning technologies based on fingerprint, and in particular to a kind of indoor fingerprint based on crowdsourcing fingerprint
Figure construction method, for communicating and radio network technique field.
Background technique
With the continuous development of mobile Internet, people are continuously increased indoor location-based information service.It is indoor
Location-based information service mainly includes parking stall lookup, logistics management, merchandise promotion information Push Service etc..At present
Multiple indoor location technology is developed, specifically includes that WiFi positioning, bluetooth positioning, infrared positioning etc..Wherein based on WiFi's
Indoor positioning technologies have obtained extensive research, and reason is that basis is greatly saved in the WiFi signal of large scale deployment
The cost of facility, and present smart phone can detect WiFi signal.
Indoor positioning technologies based on WiFi are broadly divided into two classes: the location technology based on ranging and the positioning based on fingerprint
Technology.Location technology based on ranging mainly uses the polygon location technology based on propagation model ranging.However, in complicated room
In interior environment, due to the signal non line of sight caused by complicated indoor arrangement and signal reflex scattering etc. such as separating and stopping pass
Broadcast so that propagation model parameter estimation is very inaccurate so that distance estimations error is larger, cause positioning performance significantly under
Drop.
The research emphasis for being currently based on fingerprint location technology is to construct fingerprint based on WiFi signal and establish fingerprint map.
Location technology based on fingerprint is mainly made of two stages: offline fingerprint Map building stage and online equipment positioning stage.
The offline fingerprint Map building stage usually measures the fingerprint in known location by professional survey crew, and with constructing indoor fingerprint
Figure.Online equipment positioning stage mainly compares the reference fingerprint in mobile phone real time fingerprint and fingerprint map, with side based on probability
Method or method based on similarity mode are positioned.Location technology based on fingerprint can be divided into again traditional to be surveyed based on scene
The location technology of survey and location technology based on crowdsourcing fingerprint.Traditional location technology based on on-site land survey is by specially surveying
Amount personnel carry out signal measurement in the reference point of known coordinate to construct indoor fingerprint map.Although this technology has higher
Positional accuracy, but this technology need profession survey crew, time and effort consuming.
In recent years, some scholars propose the indoor positioning technologies based on crowdsourcing fingerprint, and thought is mainly using to one
As crowd carry out it is conscious or arbitrarily measurement be used as fingerprint source, avoid the process of special messenger's on-site land survey, moreover it is possible to keep
The frequent update of crowdsourcing data, adapts to the variation of environment.But the indoor positioning technologies based on crowdsourcing fingerprint bring some new
Problem specifically includes that (1) crowdsourcing fingerprint positions mark inaccuracy.Either in such a way that crowdsourcing user actively marks or
Using the passive notation methods of path matching, compared to the measurement of on-site land survey professional, the mark of crowdsourcing fingerprint positions all can
There are biggish errors；(2) crowdsourcing fingerprint dimension is different.It, may be in some functional areas since WiFi signal propagation distance is limited
Domain can receive more access point (AP, Access Point) signal and be merely able to receive less AP in other positions
Signal, in addition, the quantity for the AP signal that different user is received using different brands mobile phone is not also identical, to cause to receive
Crowdsourcing fingerprint there is different AP set and length；(3) crowdsourcing fingerprint is unevenly distributed.Due to the limitation and use of indoor arrangement
Family acquires the uncontrolled of behavior, so that the crowdsourcing fingerprint positions for causing reception to obtain are unevenly distributed, some positions easily reached
The place for being not easy to reach there may be more crowdsourcing fingerprint contains only less crowdsourcing fingerprint or even some places not to be had at all
There is crowdsourcing fingerprint.
Summary of the invention
The present invention provides a kind of indoor fingerprint map constructing method based on crowdsourcing fingerprint, solves existing based on crowdsourcing fingerprint
Positioning caused by constructing crowdsourcing fingerprint positions mark inaccuracy, dimension difference present in indoor fingerprint map and being unevenly distributed is quasi-
The not high and computationally intensive problem of exactness.
A kind of indoor fingerprint map constructing method based on crowdsourcing fingerprint provided by the present invention, including grid division step
Suddenly, obtain crowdsourcing fingerprint step, crowdsourcing fingerprint splitting step, crowdsourcing fingerprint quantity judgment step, AP screening step, directly construct
Grid fingerprint step and fingerprint surface fitting step:
(1) grid division step:
Plane right-angle coordinate is established, planar target region is divided into what K was not overlapped each other according to its physical structure
Functional area, and be the of substantially equal grid of size by each functional regional division, wherein k-th of functional regional division is J
(k) the grid centre coordinate of a grid, j-th of grid is denoted as (Xj, Yj), j=1,2 ..., J (k), k=1,2 ..., K；K is positive
Integer；
(2) crowdsourcing fingerprint step is obtained:
The existing WiFi signal in target area is acquired by way of run trace, frequency acquisition is 0.5~2Hz, walking
Walkable region, the quantity of run trace should reach numf >=0.5 in target area, numf in the target area for track limitation
It include crowdsourcing fingerprint quantity for every square metre；Every run trace near linear and the walking that remains a constant speed substantially, and record every
Starting point coordinate (the x of run traceb, yb) and terminating point coordinate (xe, ye), the fingerprint set l of the run tracetraj: ltraj=<
(xb, yb), (xe, ye), (f1, f2..., fT)>；
Wherein, t-th of crowdsourcing fingerprint f in run tracet:T is to adopt in run trace
The crowdsourcing fingerprint quantity of sample, rt2For the signal strength for receiving the 2nd AP in t-th of crowdsourcing fingerprint, analogize below；NfFor t
The quantity of AP is received in a crowdsourcing fingerprint；According to the distribution situation of walkable region, the length of track is indefinite；According to not
Walking with pedestrian, rate is different, and crowdsourcing fingerprint quantity is different on track；Due to AP limited coverage area, received in each crowdsourcing fingerprint
To the number of AP be unequal；
To each run trace, passing through starting point coordinate (xb, yb) and terminating point coordinate (xe, ye) line segment on, by row
Crowdsourcing fingerprint quantity in track is walked, evenly distributes an approximate coordinate to each crowdsourcing fingerprint；In k-th of functional area, i-th many
Packet fingerprint fiCoordinate be denoted as (xi, yi), i=1,2 ... I (k), I (k) they are the number in k-th of functional area comprising crowdsourcing fingerprint
Mesh；
(3) crowdsourcing fingerprint splitting step:
In same functional area, i-th of crowdsourcing fingerprint f is calculatediTo the Euclidean distance at the grid center of j-th of grid
dij:
Judge whether dij≤ R is, to split Probability pijBy i-th of crowdsourcing fingerprint fiThe grid that j-th of grid is added refers to
Line supported collection GFjIn, otherwise, in grid fingerprint supported collection GFjIn be added without i-th of crowdsourcing fingerprint fi, it is formulated are as follows:dij≤ R, 90cm≤fractionation threshold value R≤150cm；
It is by the reason that a crowdsourcing fingerprint is split in multiple grids in this step: on the one hand, is received based on crowdsourcing mode
Collect obtained fingerprint position information mark inaccuracy；On the other hand, even if part crowdsourcing fingerprint position information is labeled in correctly
It is special in the signal that the signal value that certain position receives can represent its adjacent smaller area according to signal propagation characteristics in grid
Sign；
(4) crowdsourcing fingerprint quantity judgment step:
Judge in target area whether numf >=2, be to carry out step (5)；Otherwise step (6) are carried out；
Even if numf >=2, since crowdsourcing fingerprint is unevenly distributed, still crowdsourcing is not present in some regional areas that cannot be reached
Fingerprint obtains grid fingerprint using step (5) and step (7)；As 0.5≤numf < 2, obtained using step (6) and step (7)
To grid fingerprint；
(5) AP screening step:
Grid each for same functional area judges the crowdsourcing fingerprint whether split in its grid fingerprint supported collection
Number >=Q, screen threshold value Q >=3, be that AP screening then is carried out to the grid fingerprint supported collection, and construct grid fingerprint；Otherwise into
Row step (7)；
(6) grid fingerprint step is directly constructed:
Grid each for same functional area judges the crowdsourcing obtained in its grid fingerprint supported collection with the presence or absence of fractionation
Fingerprint exists and then directly constructs grid fingerprint to the grid；Otherwise step (7) are carried out；
(7) fingerprint surface fitting step:
For corresponding j-th of the grid of grid fingerprint supported collection, grid centre coordinate is (Xj, Yj), it is neighbouring based on its
Grid fingerprint constructs the wireless of a local continuous using fingerprint surface fitting for each AP that neighbouring grid fingerprint includes
Electric map, the computation grid centre coordinate (X in the radio mapj, Yj) at each AP fitted signal intensity value, and will
Its grid fingerprint as the grid；It is to be obtained by AP screening step or directly construction grid fingerprint step adjacent to grid fingerprint
The grid fingerprint of the neighbouring grid arrived.
In the step (3), Probability p is splitij=1, or
Wherein, dij≤ R, L are i-th of crowdsourcing fingerprint according to threshold value R is split, and are split to the quantity of neighbouring grid.
The indoor fingerprint map constructing method, it is characterised in that:
In the step (5), for j-th of grid, for its grid fingerprint supported collection GFjIt carries out AP screening and constructs grid
Lattice fingerprint includes following sub-step:
(5.1) it will be split to a crowdsourcing fingerprint of M (j) therein, a M (j) × N (j) dimension RSS is constructed and (receive signal
Intensity, Received Signal Strength) matrix RM(j)×N(j):
Wherein, rmnReceive the signal strength of n-th of AP for m-th of crowdsourcing fingerprint, m=1 ..., M (j), n=1 ..., N
(j), M (j) and N (j) respectively indicates the number of the number of crowdsourcing fingerprint and the AP received in the grid fingerprint supported collection；
If not receiving the signal of n-th of AP in m-th of crowdsourcing fingerprint, r is enabledmn=0；Since the crowdsourcing in grid refers to
Line includes different AP set and length is different, therefore matrix RM(j)×N(j)For sparse matrix；
(5.2) RSS matrix R is calculatedM(j)×N(j)The variance V of the non-zero values matrix unit of middle each columnn:
Wherein,Indicate the mean value for the signal strength that n-th of AP is received in the grid, VnAs n-th of AP is in phase
Answer the variance of the signal value received in grid；|Wn| representing matrix RM(j)×N(j)In the n-th column nonzero element number；
(5.3) judge whether Vn≤ σ, 0≤variance threshold values σ≤0.5 are then in RSS matrix RM(j)×N(j)In, reject VnInstitute
Otherwise corresponding column retain corresponding column；Obtain the new RSS matrix of M (j) × N1 (j) dimension
Wherein, N1 (j) is the AP number rejected after corresponding AP, N1 (j) < N (j)；
(5.4) the grid fingerprint of j-th of grid is constructedWherein, it is obtained in grid fingerprint
A-th of AP signal value
Wherein, PmjIndicate the fractionation probability m-th of crowdsourcing fingerprint being added in the grid fingerprint supported collection of j-th of grid.
The reason of rejecting AP in this way in sub-step (5.3) be, on the one hand, the crowdsourcing for being split to same grid refers to
Line from different physical locations and time for receiving it is different, according to signal propagation characteristic, the signal of same AP in grid
Intensity value should be theoretically not completely equivalent；If institute comprising the AP of the signal strength indication of some AP in a grid
Have essentially equal in crowdsourcing fingerprint, i.e., required obtained variance yields is smaller, then it is assumed that the AP in this grid without distinguish can, because
This is rejected；On the other hand, if some AP is only contained in one or fewer number of crowdsourcing fingerprint in grid, i.e., required
Obtained variance yields is smaller, then it is assumed that AP receptance in this grid is lower, it may be possible to provide inaccurate location information, even
May be noise signal, therefore rejected, thus the purpose of AP screening be reject the AP and receptance of not separating capacity compared with
Low AP.
The indoor fingerprint map constructing method, which is characterized in that in step (6), for j-th of grid, direct structure
Making grid fingerprint includes following sub-step:
(6.1) its its grid fingerprint supported collection GF will be split tojIn a crowdsourcing fingerprint of M (j), construct M (j) × N
(j) the RSS matrix R tieed upM(j)×N(j):
Wherein, rmnReceive the signal strength of n-th of AP for m-th of crowdsourcing fingerprint, m=1 ..., M (j), n=1 ..., N
(j), M (j) and N (j) respectively indicates the number of the number of crowdsourcing fingerprint and the AP received in the grid fingerprint supported collection；
If not receiving the signal of n-th of AP in m-th of crowdsourcing fingerprint, r is enabledmn=0；Since the crowdsourcing in grid refers to
Line includes different AP set and length is different, therefore matrix RM(j)×N(j)For sparse matrix；
(6.2) the grid fingerprint of j-th of grid is constructedWherein, obtained in grid fingerprint
The signal value of n-th of AP
Wherein, pmiIndicate the fractionation probability m-th of crowdsourcing fingerprint being added in the grid fingerprint supported collection of j-th of grid.
The indoor fingerprint map constructing method, which is characterized in that step (6) includes following sub-step:
(6.1) construction fitting fingerprint supported collection C:
Centered on the grid, successively search whether it refers to grid adjacent to grid by way of expanding outwardly
Line, until the grid number with grid fingerprint reaches fit threshold S, S >=6, to obtain the fitting fingerprint support of the grid
Collect C, C is made of the centre coordinate of the grid with grid fingerprint and its corresponding grid fingerprint；
(6.2) objective function θ is constructed:
Wherein,For binary polynomial signal strength fitting functionIn the value of g-th of grid, table
Show in fitting fingerprint supported collection C, the fitted signal intensity value of g-th of grid, s-th of AP；
Wherein, ωscdFor fitting coefficient, Xg、YgCross, ordinate for g-th of grid center, index c=1 ..., p；Index
D=1 ..., q；To avoid over-fitting and reducing computation complexity, p, q=2；
In grid fingerprint for g-th of grid, the signal strength indication of s-th of AP；| C | indicate fitting fingerprint supported collection C
It is middle that there are the numbers of the grid of grid fingerprint；
(6.3) fitting coefficient ω is soughtscd:
Objective function θ is sought about fitting coefficient ωscdPartial derivative, make partial derivative 0, i.e., so that objective function θ has
Minimum value obtains fitting coefficient ωscd:
Wherein, intermediate symbols
Intermediate symbols
Exponent e=1 ..., p, index h=1 ..., q；
(6.4) by the grid centre coordinate (X of the gridj, Yj) and fitting coefficient ωscdSubstitute into binary polynomial signal strength
Fitting functionTo seek the fitted signal intensity value of s-th of AP in the gridThe then grid of the grid
Lattice fingerprint isWherein,S=1 ..., N2 (j).
In sub-step (5.4)In sub-step (6.2)In sub-step (7.4)Indicate in corresponding sequence
The signal value of a-th, n-th or s-th AP received.
During sub-step (7.1) construction fitting fingerprint supported collection, each grid has its 8 neighborhood grid, surrounds 8 neighborhoods
The outer layer of grid is 16 neighborhood grids, and the outer layer for surrounding 16 neighborhood grids is 24 neighborhood grids ..., is so successively looked into from inside to outside
Look for whether it has grid fingerprint adjacent to grid；Construction is not involved in by the grid fingerprint obtained using fingerprint surface fitting technology
The process of other grid fitting fingerprint supported collections.The reason is that the grid fingerprint itself that fitting obtains is by it adjacent to grid fingerprint
It is obtained using fingerprint surface fitting technology.And neighbouring grid fingerprint is obtained by crowdsourcing fingerprint supported collection, itself exists
Error.If being fitted obtained grid fingerprint to participate in constructing other grid fitting fingerprint supported collections, cumulative errors can be increasing.
In addition, the grid that the neighbouring grid fingerprint and needs chosen are fitted should avoid in same functional area due to different function
Energy region causes signal strength indication difference larger, fitting inaccuracy.
Using the present invention construct indoor fingerprint map when, can be used the Probabilistic Localization Methods based on Bayes estimation into
Row positioning (is illustrated, the case where 0.5≤numf < 2 is similar, only AP quantity in grid fingerprint with the case where numf >=2
It is different):
After it experienced each step of the invention, each grid has corresponding grid fingerprint, the grid of j-th of grid
Lattice fingerprint isOrUtilize gaussian kernel function combination grid fingerprint
Signal strengthOrWith variance VaOr VsIt is translated into Gaussian Profile, then obtains test fingerprint in j-th of gridProbability:
Z=1 ..., N3, N3 indicate the AP quantity that test fingerprint receives, when z-th of AP and a-th of AP or s-th of AP are
When the same AP, above formula is just significant.
Wherein,OrIt indicates in j-th of grid, a or s AP signal strength indication areOrIn obtained Gaussian Profile, the signal strength indication received isProbability；
Final choice probability P (Ft|Fj) maximum grid assessment position of the grid centre coordinate as the test fingerprint.
Wherein calculate probability P (Ft|Fj) during, FjWith FtIt is middle to gather comprising different AP.Therefore it provides, it is right
In being present in FjIn may be not present in FtIn a or s AP, directly ignore its conditional probabilityOr
Conversely, for being present in FtIn may be not present in FjIn z-th of AP, set a lesser probability for its conditional probability
ValueOrFor pmin=0.001.
The present invention is divided into K nonoverlapping functional areas mutually according to the physical structure of target area first, and
By the of substantially equal grid of each functional regional division size；The crowdsourcing fingerprint acquired using ruck, such as utilize walking
The mode of track collects the collectable WiFi signal in target area, and carries out position mark to crowdsourcing fingerprint, such as can use
Path matching algorithm assigns approximate coordinate to crowdsourcing fingerprint each in track.Each crowdsourcing fingerprint is split to certain probability
In one or more grids adjacent with its labeling position.For each grid, grid fingerprint supported collection is formed；And it is right
Fingerprint carries out AP screening in grid fingerprint supported collection, constructs grid fingerprint.For lack grid fingerprint supported collection or supported collection
In contain less fingerprint quantity grid be fitted by fingerprint using it adjacent to the grid fingerprint of grid and construct its grid fingerprint,
It is fitted to obtain grid fingerprint for example, by using fingerprint surface fitting technology.
Using the indoor fingerprint map that the present invention constructs, it is fixed that position assessment is carried out to test fingerprint using the method for probability
Position.
Compared with prior art, the present invention have it is following the utility model has the advantages that
(1) it reduces fingerprint collecting workload: due to using crowdsourcing fingerprint splitting step, fingerprint being split to multiple grids
In, fingerprint quantity is increased, makes it that there is preferable positioning result using less crowdsourcing fingerprint.
(2) be conducive to improve positional accuracy: due to using crowdsourcing fingerprint splitting step, crowdsourcing fingerprint be split to more
In a grid, solves the problems, such as crowdsourcing fingerprint positions mark inaccuracy；Due to using crowdsourcing fingerprint splitting step and referring to
Line surface fitting step solves the problems, such as that fingerprint positions are unevenly distributed using fingerprint fractionation and fingerprint fitting technique；Due to adopting
With AP screening step, AP screening is carried out, the AP that receptance is low in grid is rejected, reduces the influence of noise signal；Above three
The combination of process greatly facilitates raising positional accuracy.
(3) it reduces fingerprint and compares workload: establishing indoor fingerprint map based on grid, positioning stage needs the finger compared
Line quantity is only related to the number of grid, is considerably less than the quantity of crowdsourcing fingerprint.In addition, due to using AP screening step, AP
Screening reduces the computation complexity of fingerprint comparison.
Detailed description of the invention
Fig. 1 is flow diagram of the invention；
Fig. 2 is the positioning scene figure of application example of the present invention；
Fig. 3 is the schematic diagram that a crowdsourcing fingerprint is split to multiple grids in the present invention；
Fig. 4 is AP screening step flow diagram；
Fig. 5 is fingerprint surface fitting steps flow chart schematic diagram；
Fig. 6 is to obtain the schematic diagram of fitting fingerprint supported collection in mean camber fingerprint fitting technique of the present invention；
When Fig. 7 is using different crowdsourcing fingerprint numbers, different constructing plan average localization errors compare；
Fig. 8 is that different constructing plan position error cumulative distribution figures compare；
Fig. 9 is influence of the fingerprint mark error to different constructing plan average localization errors.
Specific embodiment
The present invention is described in more detail with reference to the accompanying drawings and embodiments.
As shown in Figure 1, the present invention includes grid division step, obtains crowdsourcing fingerprint step, crowdsourcing fingerprint splitting step, crowd
Packet fingerprint quantity judgment step, directly constructs grid fingerprint step and fingerprint surface fitting step at AP screening step.
As an embodiment of the present invention:
(1) grid division step:
Establish plane right-angle coordinate, as shown in Fig. 2, by planar target region according to its physical structure be divided into 6 between teach
Room and 1 corridor totally 7 functional areas not overlapped each other, wherein odd number classroom area is respectively 10.5 × 9.56m2, even number
Number classroom area is respectively 10.5 × 7.76m2, corridor area is 32.6 × 3.62m2, between each functional area cement wall with a thickness of
0.3m, the gross area are about 717m2；And the grid that each functional regional division is of substantially equal for size, wherein k-th of function
Region division is a grid of J (k), and the grid centre coordinate of j-th of grid is denoted as (Xj, Yj), j=1,2 ..., J (k), k=1,
2 ..., 7；
For example, being the grid that length and width are respectively about 0.6 meter, odd number by functional regional division in the scene of the present embodiment
Classroom can be divided into 272 grids, and even number classroom can be divided into 221 grids, and corridor can be divided into 324 grids；
(2) crowdsourcing fingerprint step is obtained:
The existing WiFi signal in target area, frequency acquisition 1Hz, run trace limit are acquired by way of run trace
System in the target area walkable region (as shown in Fig. 2, walkable region mainly includes corridor between classroom inside corridor and classroom,
And without crowdsourcing fingerprint near seat in classroom, and run trace is to be randomly dispersed in walkable region, therefore, in target area
Domain, crowdsourcing fingerprint positions are unevenly distributed, some positions do not have crowdsourcing fingerprint at all), the quantity of run trace should reach target area
Numf is about 10 and (in order to study influence of the crowdsourcing fingerprint quantity to positional accuracy, therefore has collected enough crowds in domain
Packet fingerprint)；(in the present embodiment, walking rate is 1m/s, is received for every run trace near linear and the walking that remains a constant speed substantially
The AP quantity variation range arrived is 100+~400+), and record the starting point coordinate (x of every run traceb, yb) and terminating point
Coordinate (xe, ye), the fingerprint set l of the run tracetraj: ltrai=< (xb, yb), (xe, ye), (f1, f2..., fT)>；
Wherein, t-th of crowdsourcing fingerprint f in run tracet:T is to adopt in run trace
The crowdsourcing fingerprint quantity of sample, rt2For the signal strength for receiving the 2nd AP in t-th of crowdsourcing fingerprint, analogize below；NfFor t
The quantity of AP is received in a crowdsourcing fingerprint；
As Fig. 4 corridor in, the fingerprint set l of the run traceAB:
lAB=< (xA, yA), (xB, yB), (f1, f2..., f14) >,
Wherein, t-th of crowdsourcing fingerprint f in run tracet:Run trace lABIn share
14 crowdsourcing fingerprints；
To each run trace, passing through starting point coordinate (xb, yb) and terminating point coordinate (xe, ye) line segment on, by row
Crowdsourcing fingerprint quantity in track is walked, evenly distributes an approximate coordinate to each crowdsourcing fingerprint；In k-th of functional area, i-th many
Packet fingerprint fiCoordinate be denoted as (xi, yi), i=1,2 ... I (k), I (k) they are the number in k-th of functional area comprising crowdsourcing fingerprint
Mesh；
MEIZU MX5 is used to carry out the acquisition of signal strength as terminal device in testing.The fingerprint of acquisition is divided into two
Part: a part is test fingerprint, acquires 725 test fingerprints altogether, is divided into 1m × 1m therebetween, is evenly distributed in the target
In region, each test point sample within 20 seconds, and sampling per second is primary；Another part is training fingerprint, is referred to for constructing interior
Line map.Training fingerprint includes that (1362, use ginseng for the reference fingerprint positioned at walkable region that is obtained by on-site land survey again
The target of fingerprint is examined merely to increasing the quantity of fingerprint) and the crowdsourcing fingerprint (6830) that is obtained by track-wise.In Fig. 2
The point of black triangle indicates grid fingerprint in classroom 412, and the black dot expression in classroom 408 on curve passes through track-wise
Obtained crowdsourcing fingerprint, the pentagonal point of black indicates the reference fingerprint obtained based on on-site land survey (using all in classroom 409
Reference fingerprint compare experiment).
(3) crowdsourcing fingerprint splitting step:
In same functional area, i-th of crowdsourcing fingerprint f is calculatediTo the Euclidean distance at the grid center of j-th of grid
dij:
Judge whether dij≤ R is, to split Probability pijBy i-th of crowdsourcing fingerprint fiThe grid that j-th of grid is added refers to
Line supported collection GFjIn, otherwise, in grid fingerprint supported collection GFjIn be added without i-th of crowdsourcing fingerprint fi, it is formulated are as follows:dij≤ R in the present embodiment, splits threshold value R and is set as 100cm；Split Probability pij=1；
As shown in figure 3, black dot indicates crowdsourcing fingerprint fi, black side's point indicate grid center, calculate black dot with
Crowdsourcing fingerprint is assigned in the grid by the Euclidean distance between black side's point if the Euclidean distance, which is less than, splits threshold value R；
(4) crowdsourcing fingerprint quantity judgment step:
Judge in target area whether numf >=2, be to carry out step (5)；Otherwise step (6) are carried out；
(5) AP screening step:
Grid each for same functional area judges the crowdsourcing fingerprint whether split in its grid fingerprint supported collection
Number >=3, be that AP screening then is carried out to the grid fingerprint supported collection, and construct grid fingerprint；Otherwise step (7) are carried out；
As shown in figure 4, in the step (5), for j-th of grid, for its grid fingerprint supported collection GFjCarry out AP sieve
Selecting and constructing grid fingerprint includes following sub-step:
(5.1) it will be split to a crowdsourcing fingerprint of M (j) therein, construct a M (j) × N (j) dimension RSS matrix
RM(j)×N(j):
Wherein, rmnReceive the signal strength of n-th of AP for m-th of crowdsourcing fingerprint, m=1 ..., M (j), n=1 ..., N
(j), M (j) and N (j) respectively indicates the number of the number of crowdsourcing fingerprint and the AP received in the grid fingerprint supported collection；
If not receiving the signal of n-th of AP in m-th of crowdsourcing fingerprint, r is enabledmn=0；Since the crowdsourcing in grid refers to
Line includes different AP set and length is different, therefore matrix RM(j)×N(j)For sparse matrix；
(5.2) RSS matrix R is calculatedM(j)×N(j)The variance V of the non-zero values matrix unit of middle each columnn:
Wherein,Indicate the mean value for the signal strength that n-th of AP is received in the grid, VnAs n-th of AP is in phase
Answer the variance of the signal value received in grid；|Wn| representing matrix RM(j)×N(j)In the n-th column nonzero element number；
(5.3) judge whether Vn≤ σ, in the present embodiment, variance threshold values σ is set as 0, is then in RSS matrix RM(j)×N(j)
In, reject VnOtherwise corresponding column retain corresponding column；Obtain the new RSS matrix of M (j) × N1 (j) dimension
Wherein, N1 (j) is the AP number rejected after corresponding AP, N1 (j) < N (j)；
(5.4) the grid fingerprint of j-th of grid is constructedWherein, it is obtained in grid fingerprint
A-th of AP signal value
Wherein, pmjIndicate the fractionation probability m-th of crowdsourcing fingerprint being added in the grid fingerprint supported collection of j-th of grid；
Table 1 gives AP quantity received by average each fingerprint in each functional area, executes AP screening process institute
The AP quantity of rejecting, and reject ratio shared by the quantity of AP.As can be seen from Table 2, each functional area rejects the ratio of AP
Rate is more than 25%.This procedure reduces the complexities of comparing calculation in position fixing process；
Table 1
| APs | Corridor | 408 | 409 | 410 | 411 | 412 | 413 |
| Average AP number | 425.7 | 279.5 | 270.1 | 300.4 | 193.3 | 319.1 | 357.7 |
| Reject AP | 162.6 | 83.3 | 96.7 | 79.3 | 48.7 | 85.2 | 161.7 |
| Reject ratio | 0.382 | 0.298 | 0.358 | 0.264 | 0.252 | 0.267 | 0.452 |
(6) grid fingerprint step is directly constructed:
Grid each for same functional area judges the crowdsourcing obtained in its grid fingerprint supported collection with the presence or absence of fractionation
Fingerprint exists and then directly constructs grid fingerprint to the grid；Otherwise step (7) are carried out；
For j-th of grid, directly constructing grid fingerprint includes following sub-step:
(6.1) its grid fingerprint supported collection GF will be split tojIn a crowdsourcing fingerprint of M (j), construct M (j) × N (j)
The RSS matrix R of dimensionM(j)×N(j):
Wherein, rmnReceive the signal strength of n-th of AP for m-th of crowdsourcing fingerprint, m=1 ..., M (j), n=1 ..., N
(j), M (j) and N (j) respectively indicates the number of the number of crowdsourcing fingerprint and the AP received in the grid fingerprint supported collection；
If not receiving the signal of n-th of AP in m-th of crowdsourcing fingerprint, r is enabledmn=0；Since the crowdsourcing in grid refers to
Line includes different AP set and length is different, therefore matrix RM(j)×N(j)For sparse matrix；
(6.2) the grid fingerprint of j-th of grid is constructedWherein, obtained in grid fingerprint
The signal value of n-th of AP
Wherein, pmjIndicate the fractionation probability m-th of crowdsourcing fingerprint being added in the grid fingerprint supported collection of j-th of grid；
(7) fingerprint surface fitting step:
For corresponding j-th of the grid of grid fingerprint supported collection, grid centre coordinate is (Xj, Yj), it is neighbouring based on its
Grid fingerprint constructs the wireless of a local continuous using fingerprint surface fitting for each AP that neighbouring grid fingerprint includes
Electric map, the computation grid centre coordinate (X in the radio mapj, Yj) at each AP fitted signal intensity value, and will
Its grid fingerprint as the grid；It is to be obtained by AP screening step or directly construction grid fingerprint step adjacent to grid fingerprint
The grid fingerprint of the neighbouring grid arrived；
As shown in figure 5, step (7) includes following sub-step:
(7.1) construction fitting fingerprint supported collection C:
Centered on the grid, successively search whether it refers to grid adjacent to grid by way of expanding outwardly
Line, until the grid number with grid fingerprint reaches fit threshold S, in the present embodiment, fit threshold S is set as 6, thus
To the grid fitting fingerprint supported collection C, C by the grid with grid fingerprint centre coordinate and its corresponding grid fingerprint structure
At；
As shown in fig. 6, numerical value 1 represents the grid there are grid fingerprint, numerical value 0 indicates to need in the grid without grid fingerprint
Grid fingerprint is obtained using fingerprint surface fitting technology.For the grid of figure acceptance of the bid black square, refer to obtain fitting
Line supported collection is first looked for the presence or absence of grid fingerprint in the grid that nearest one layer of dotted line frame is included, if grid in these grids
The quantity of lattice fingerprint reaches fit threshold, then stops searching, and the grid fingerprint obtained by lookup constitutes the fitting fingerprint of the grid
Supported collection；Otherwise, seeking scope is expanded to the grid that following dotted line frame is included, continues previous action process；
(7.2) objective function θ is constructed:
Wherein,For binary polynomial signal strength fitting functionIn the value of g-th of grid, table
Show in fitting fingerprint supported collection C, the fitted signal intensity value of g-th of grid, s-th of AP；
Wherein, ωscdFor fitting coefficient, Xg、YgCross, ordinate for g-th of grid center, index c=1 ..., p；Index
D=1 ..., q；To avoid over-fitting and reducing computation complexity, p, q=2；
In grid fingerprint for g-th of grid, the signal strength indication of s-th of AP；| C | indicate fitting fingerprint supported collection C
It is middle that there are the numbers of the grid of grid fingerprint；
(7.3) fitting coefficient ω is soughtscd:
Objective function θ is sought about fitting coefficient ωscdPartial derivative, make partial derivative 0, i.e., so that objective function θ has
Minimum value obtains fitting coefficient ωscd:
Wherein, intermediate symbols
Intermediate symbols
Exponent e=1 ..., p, index h=1 ..., q；
(7.4) by the grid centre coordinate (X of the gridj, Yj) and fitting coefficient ωscdSubstitute into binary polynomial signal strength
Fitting functionTo seek the fitted signal intensity value of s-th of AP in the gridThe then grid of the grid
Lattice fingerprint isWherein,S=1 ..., N2 (j).
Wherein, in sub-step (5.4)In sub-step (6.2)In sub-step (7.4)Indicate corresponding
The signal value of a-th, n-th or s-th AP received in sequence.
Crowdsourcing fingerprint to propose in more of the invention splits, the property of AP screening and fingerprint surface fitting these three steps
Can, 6 kinds of schemes for constructing indoor fingerprint map are tested altogether, are respectively:
(a)+without screening+without fitting, it is benchmark scheme, i.e., only includes the steps that (1) of the invention and step (2) without fractionation；
(b) without split+without screening+fitting, i.e., only include the steps that (1) of the invention, step (2) and step (7)；
(c) split+without screening+and without fitting, i.e., only include the steps that (1) of the invention, step (2) and step (3)；
(d) split+without screening+fitting, that is, include the steps that (1) of the invention, step (2), step (3), step (6) and step
Suddenly (7), suitable for target area, the case where 0.5≤numf < 2；
(e) fractionation+screening+fitting includes the steps that (1) of the invention, step (2), step (3), step (5) and step
(7), suitable for target area, the case where numf >=2；；
0.6 × 0.6m of on-site land survey interval2
When Fig. 7 is indicated using different crowdsourcing fingerprint numbers, the comparison of scheme (a), (b), (c), (d) average localization error.
It can be seen from figure 7 that scheme (c), (d) use crowdsourcing fingerprint and split technology, it is average fixed for relative plan (a), (b)
Level exactness tool improves a lot.It is also seen that crowdsourcing fingerprint splits technology with respect to fingerprint surface fitting technology performance from Fig. 7
More preferably, this is because fitting fingerprint supported collection is using grid fingerprint, there are certain errors for itself, quasi- by fingerprint curved surface
After conjunction, error can be bigger.In this Fig. 7, the reason of not describing the performance of AP screening scheme is the grid for being able to carry out AP screening
Premise be need to guarantee each grid have a certain number of crowdsourcing fingerprints.
Fig. 8 shows scheme (a), (d), (e) and position error cumulative distribution based on on-site land survey, table 2 gives this
The average localization error of four kinds of schemes, 50% position error and 90% position error.
Table 2
| Average localization error (m) | It is average | 50% | 90% |
| 0.6 × 0.6m of on-site land survey interval2 | 3.06 | 1.92 | 7.50 |
| (a) without split+without screening+without fitting | 1.96 | 1.20 | 4.63 |
| (b) split+without screening+fitting | 1.65 | 1.02 | 3.65 |
| (c) fractionation+screening+fitting | 1.48 | 1.01 | 3.16 |
It can be seen that in conjunction with Fig. 8 and table 2, scheme (e) relative plan (a), (d) and the positioning performance based on on-site land survey have
Biggish raising.
Fig. 9 gives influence of the fingerprint mark error to locating scheme (a), (d), (e) average localization error.From Fig. 9
As can be seen that the average localization error of three kinds of schemes all increases as the standard deviation of mark error increases, but scheme (d),
(e) variation is relatively gentle, and the average localization error of scheme (e) is still the smallest.Therefore, crowdsourcing fingerprint is split to multiple grid
It goes to can reduce the influence due to caused by fingerprint mark inaccuracy in lattice.
Claims (5)
1. a kind of indoor fingerprint map constructing method based on crowdsourcing fingerprint, including grid division step, acquisition crowdsourcing fingerprint step
Suddenly, crowdsourcing fingerprint splitting step, crowdsourcing fingerprint quantity judgment step, AP screening step, directly construction and refer to grid fingerprint step
Line surface fitting step, it is characterised in that:
(1) grid division step:
Plane right-angle coordinate is established, planar target region is divided into the K functions of not overlapping each other according to its physical structure
Region, and be the of substantially equal grid of size by each functional regional division, wherein k-th of functional regional division is that J (k) is a
The grid centre coordinate of grid, j-th of grid is denoted as (Xj, Yj), j=1,2 ..., J (k), k=1,2 ..., K；K is positive integer；
(2) crowdsourcing fingerprint step is obtained:
The existing WiFi signal in target area is acquired by way of run trace, frequency acquisition is 0.5~2Hz, run trace
Walkable region, the quantity of run trace should reach numf >=0.5 in target area in the target area for limitation, and numf is every
Square metre include crowdsourcing fingerprint quantity；Every run trace near linear and the walking that remains a constant speed substantially, and record every walking
Starting point coordinate (the x of trackb, yb) and terminating point coordinate (xe, ye), the fingerprint set l of the run tracetraj: ltraj=<
(xb, yb), (xe, ye), (f1, f2..., fr) >；
Wherein, t-th of crowdsourcing fingerprint f in run tracet:T is the crowd sampled in run trace
Packet fingerprint quantity, rt2For the signal strength for receiving the 2nd AP in t-th of crowdsourcing fingerprint, analogize below；NfFor t-th of crowdsourcing
The quantity of AP is received in fingerprint；
To each run trace, passing through starting point coordinate (xb, yb) and terminating point coordinate (xe, ye) line segment on, by walking rail
Crowdsourcing fingerprint quantity in mark evenly distributes an approximate coordinate to each crowdsourcing fingerprint；In k-th of functional area, i-th of crowdsourcing refers to
Line fiCoordinate be denoted as (xi, yi), i=1,2 ... I (k), I (k) they are the number in k-th of functional area comprising crowdsourcing fingerprint；
(3) crowdsourcing fingerprint splitting step:
In same functional area, i-th of crowdsourcing fingerprint f is calculatediTo the Euclidean distance d at the grid center of j-th of gridij:
Judge whether dij≤ R is, to split Probability pijBy i-th of crowdsourcing fingerprint fiThe grid fingerprint branch of j-th of grid is added
Support collection GFjIn, otherwise, in grid fingerprint supported collection GFjIn be added without i-th of crowdsourcing fingerprint fi, it is formulated are as follows:dij≤ R, 90cm≤fractionation threshold value R≤150cm；
(4) crowdsourcing fingerprint quantity judgment step:
Judge in target area whether numf >=2, be to carry out step (5)；Otherwise step (6) are carried out；
(5) AP screening step:
Grid each for same functional area judges the number of the crowdsourcing fingerprint whether split in its grid fingerprint supported collection
Mesh >=Q screens threshold value Q >=3, is then to carry out AP screening to the grid fingerprint supported collection, and construct grid fingerprint；Otherwise it is walked
Suddenly (7)；
(6) grid fingerprint step is directly constructed:
Grid each for same functional area judges that the crowdsourcing obtained in its grid fingerprint supported collection with the presence or absence of fractionation refers to
Line is then directly to construct grid fingerprint to the grid；Otherwise step (7) are carried out；
(7) fingerprint surface fitting step:
For corresponding j-th of the grid of grid fingerprint supported collection, grid centre coordinate is (Xj, Yj), based on it adjacent to grid
Fingerprint, for each AP that neighbouring grid fingerprint includes, with constructing the radio of a local continuous using fingerprint surface fitting
Figure, the computation grid centre coordinate (X in the radio mapj, Yj) at each AP fitted signal intensity value, and made
For the grid fingerprint of the grid；It is obtained adjacent to grid fingerprint by AP screening step or directly construction grid fingerprint step
The grid fingerprint of neighbouring grid.
2. interior fingerprint map constructing method as described in claim 1, it is characterised in that:
In the step (3), Probability p is splitij=1, or
Wherein, dij≤ R, L are i-th of crowdsourcing fingerprint according to threshold value R is split, and are split to the quantity of neighbouring grid.
3. interior fingerprint map constructing method as claimed in claim 1 or 2, it is characterised in that:
In the step (5), for j-th of grid, for its grid fingerprint supported collection GFjIt carries out AP screening and constructs grid and refer to
Line includes following sub-step:
(5.1) it will be split to a crowdsourcing fingerprint of M (j) therein, construct a M (j) × N (j) dimension RSS matrix RM(j)×N(j):
Wherein, rmnReceive the signal strength of n-th of AP for m-th of crowdsourcing fingerprint, m=1 ..., M (j), n=1 ..., N (j), M
(j) number of the number of crowdsourcing fingerprint and the AP received in the grid fingerprint supported collection is respectively indicated with N (j)；
If not receiving the signal of n-th of AP in m-th of crowdsourcing fingerprint, r is enabledmn=0；Due to the crowdsourcing fingerprint packet in grid
It is different containing different AP set and length, therefore matrix RM(j)×N(j)For sparse matrix；
(5.2) RSS matrix R is calculatedM(j)×N(j)The variance V of the non-zero values matrix unit of middle each columnn:
Wherein,Indicate the mean value for the signal strength that n-th of AP is received in the grid, VnAs n-th of AP is in corresponding grid
The variance of the signal value received in lattice；|Wn| representing matrix RM(j)×N(j)In the n-th column nonzero element number；
(5.3) judge whether Vn≤ σ, 0≤variance threshold values σ≤0.5 are then in RSS matrix RM(j)×N(j)In, reject VnCorresponding
Otherwise column retain corresponding column；Obtain the new RSS matrix of M (j) × N1 (j) dimension
Wherein, N1 (j) is the AP number rejected after corresponding AP, N1 (j) < N (j)；
(5.4) the grid fingerprint of j-th of grid is constructedWherein, a obtained in grid fingerprint
The signal value of a AP
Wherein, pmjIndicate the fractionation probability m-th of crowdsourcing fingerprint being added in the grid fingerprint supported collection of j-th of grid.
4. interior fingerprint map constructing method as claimed in claim 1 or 2, which is characterized in that in the step (6), for
J-th of grid, directly constructing grid fingerprint includes following sub-step:
(6.1) its grid fingerprint supported collection GF will be split tojIn a crowdsourcing fingerprint of M (j), construct M (j) × N (j) dimension
RSS matrix RM(j)×N(j):
Wherein, rmnReceive the signal strength of n-th of AP for m-th of crowdsourcing fingerprint, m=1 ..., M (j), n=1 ..., N (j), M
(j) number of the number of crowdsourcing fingerprint and the AP received in the grid fingerprint supported collection is respectively indicated with N (j)；
If not receiving the signal of n-th of AP in m-th of crowdsourcing fingerprint, r is enabledmn=0；Due to the crowdsourcing fingerprint packet in grid
It is different containing different AP set and length, therefore matrix RM(j)×N(j)For sparse matrix；
(6.2) the grid fingerprint of j-th of grid is constructedWherein, n-th obtained in grid fingerprint
The signal value of AP
Wherein, pmjIndicate the fractionation probability m-th of crowdsourcing fingerprint being added in the grid fingerprint supported collection of j-th of grid.
5. interior fingerprint map constructing method as claimed in claim 1 or 2, which is characterized in that step (7) includes following sub-step
It is rapid:
(7.1) construction fitting fingerprint supported collection C:
Centered on the grid, whether adjacent to grid have grid fingerprint, directly if successively searching it by way of expanding outwardly
Reach fit threshold S, S >=6 to the grid number with grid fingerprint, to obtain fitting fingerprint the supported collection C, C of the grid
It is made of the centre coordinate of the grid with grid fingerprint and its corresponding grid fingerprint；
(7.2) objective function θ is constructed:
Wherein,For binary polynomial signal strength fitting functionIn the value of g-th of grid, indicate
It is fitted in fingerprint supported collection C, the fitted signal intensity value of g-th of grid, s-th of AP；
Wherein, ωscdFor fitting coefficient, Xg、YgCross, ordinate for g-th of grid center, index c=1 ..., p；Index d=
1 ..., q；To avoid over-fitting and reducing computation complexity, p, q=2；
In grid fingerprint for g-th of grid, the signal strength indication of s-th of AP；| C | it indicates to deposit in fitting fingerprint supported collection C
In the number of the grid of grid fingerprint；
(7.3) fitting coefficient ω is soughtscd:
Objective function θ is sought about fitting coefficient ωscdPartial derivative, make partial derivative 0, i.e., so that objective function θ has minimum
Value, obtains fitting coefficient ωscd:
Wherein, intermediate symbols
Intermediate symbols
Exponent e=1 ..., p, index h=1 ..., q；
(7.4) by the grid centre coordinate (X of the gridj, Yj) and fitting coefficient ωscdSubstitute into the fitting of binary polynomial signal strength
FunctionTo seek the fitted signal intensity value of s-th of AP in the gridThen the grid of the grid refers to
Line isWherein,S=1 ..., N2 (j).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611219485.9A CN106792465B (en) | 2016-12-26 | 2016-12-26 | A kind of indoor fingerprint map constructing method based on crowdsourcing fingerprint |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611219485.9A CN106792465B (en) | 2016-12-26 | 2016-12-26 | A kind of indoor fingerprint map constructing method based on crowdsourcing fingerprint |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106792465A CN106792465A (en) | 2017-05-31 |
| CN106792465B true CN106792465B (en) | 2019-08-30 |
Family
ID=58926757
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611219485.9A Active CN106792465B (en) | 2016-12-26 | 2016-12-26 | A kind of indoor fingerprint map constructing method based on crowdsourcing fingerprint |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106792465B (en) |
Families Citing this family (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107290715B (en) * | 2017-07-18 | 2019-05-14 | 华中科技大学 | A kind of indoor orientation method based on fingerprint calibration and fingerprint conversion |
| CN108534779B (en) * | 2018-03-09 | 2020-09-08 | 华中科技大学 | Indoor positioning map construction method based on track correction and fingerprint improvement |
| CN108413975B (en) * | 2018-03-15 | 2021-11-09 | 斑马网络技术有限公司 | Map acquisition method and system, cloud processor and vehicle |
| CN108449709B (en) * | 2018-03-16 | 2019-06-18 | 华中科技大学 | A kind of indoor radio map constructing method based on crowdsourcing sample |
| CN108810817B (en) * | 2018-05-18 | 2020-10-27 | 东南大学 | Indoor map generation and track encryption method based on mobile crowdsourcing mode |
| CN108709557B (en) * | 2018-05-18 | 2021-09-24 | 东南大学 | Indoor map generation method based on multi-user track fitting |
| CN110530363B (en) * | 2018-05-24 | 2022-08-19 | 北京智慧图科技有限责任公司 | Building-level AP crowdsourcing generation method |
| CN109059919B (en) * | 2018-06-08 | 2020-05-19 | 华中科技大学 | Indoor positioning method based on crowdsourcing sample weighted surface fitting |
| CN108919177B (en) * | 2018-07-16 | 2020-12-18 | 华中科技大学 | Positioning map construction method based on virtual information source estimation and track correction |
| CN110264154B (en) * | 2019-05-28 | 2023-06-09 | 南京航空航天大学 | Crowd-sourced signal map construction method based on self-encoder |
| CN110716512A (en) * | 2019-09-02 | 2020-01-21 | 华电电力科学研究院有限公司 | Environmental protection equipment performance prediction method based on coal-fired power plant operation data |
| CN111474516B (en) * | 2020-04-17 | 2022-01-11 | 华中科技大学 | Multi-level indoor positioning method and system based on crowdsourcing sample surface fitting |
| CN112068073B (en) * | 2020-09-09 | 2023-09-01 | 中山大学 | UWB positioning error correction method, UWB positioning system, device and storage medium |
| CN112543938B (en) * | 2020-09-29 | 2022-02-08 | 华为技术有限公司 | Generation method and device of grid occupation map |
| CN115685060A (en) * | 2021-07-28 | 2023-02-03 | 华为技术有限公司 | Indoor fingerprint map construction method and related device |
| CN113727273B (en) * | 2021-08-19 | 2022-04-26 | 武汉大学 | Personnel indoor semantic track reconstruction method based on wireless crowdsourcing data |
| CN115683121B (en) * | 2022-12-30 | 2023-03-17 | 武汉大学 | Method for generating multi-source ubiquitous positioning fingerprint based on measured track |
| CN117135573A (en) * | 2023-04-14 | 2023-11-28 | 荣耀终端有限公司 | Cell location updating method, server and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104394588A (en) * | 2014-11-15 | 2015-03-04 | 北京航空航天大学 | Indoor positioning method based on Wi-Fi fingerprints and multi-dimensional scaling analysis |
| CN104462281A (en) * | 2014-11-26 | 2015-03-25 | 杭州优体科技有限公司 | Distributed type parallel surveying system for fingerprint positioning |
| CN104869639A (en) * | 2015-05-29 | 2015-08-26 | 福建星网锐捷网络有限公司 | Indoor positioning method and device |
| CN105242239A (en) * | 2015-10-19 | 2016-01-13 | 华中科技大学 | Indoor subarea positioning method based on crowdsourcing fingerprint clustering and matching |
| CN105263113A (en) * | 2015-10-26 | 2016-01-20 | 深圳大学 | Wi-Fi location fingerprint map building method and system based on crowd-sourcing |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9537889B2 (en) * | 2011-12-05 | 2017-01-03 | Microsoft Technology Licensing, Llc | Trusting crowdsourced data with issued tags |
| US9107178B2 (en) * | 2012-12-24 | 2015-08-11 | Intel Corporation | Geo-location signal fingerprinting |
-
2016
- 2016-12-26 CN CN201611219485.9A patent/CN106792465B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104394588A (en) * | 2014-11-15 | 2015-03-04 | 北京航空航天大学 | Indoor positioning method based on Wi-Fi fingerprints and multi-dimensional scaling analysis |
| CN104462281A (en) * | 2014-11-26 | 2015-03-25 | 杭州优体科技有限公司 | Distributed type parallel surveying system for fingerprint positioning |
| CN104869639A (en) * | 2015-05-29 | 2015-08-26 | 福建星网锐捷网络有限公司 | Indoor positioning method and device |
| CN105242239A (en) * | 2015-10-19 | 2016-01-13 | 华中科技大学 | Indoor subarea positioning method based on crowdsourcing fingerprint clustering and matching |
| CN105263113A (en) * | 2015-10-26 | 2016-01-20 | 深圳大学 | Wi-Fi location fingerprint map building method and system based on crowd-sourcing |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106792465A (en) | 2017-05-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106792465B (en) | A kind of indoor fingerprint map constructing method based on crowdsourcing fingerprint | |
| US9313617B2 (en) | Identification of points of interest and positioning based on points of interest | |
| US8983490B2 (en) | Locating a mobile device | |
| Cengiz | Comprehensive analysis on least-squares lateration for indoor positioning systems | |
| CN106233798A (en) | The location of wireless user equipment (ue) device in the target area | |
| CN103197280A (en) | Access point (AP) location estimation method based on radio-frequency signal strength | |
| CN106255059B (en) | It is a kind of based on geometric ways without device target localization method | |
| CN104038901A (en) | Indoor positioning method for reducing fingerprint data acquisition workload | |
| CN105974361A (en) | Indoor positioning method based on fingerprint section indexes and WiFi-FM fusion fingerprints | |
| KR20110116565A (en) | Method determining indoor location using bayesian algorithm | |
| CN104581945B (en) | The WLAN indoor orientation methods of semi-supervised APC clustering algorithms based on distance restraint | |
| Alitaleshi et al. | EA-CNN: A smart indoor 3D positioning scheme based on Wi-Fi fingerprinting and deep learning | |
| CN110213710A (en) | A kind of high-performance indoor orientation method, indoor locating system based on random forest | |
| CN104144499A (en) | Wireless sensor network positioning method based on RSSI vector similarity degree and generalized inverse | |
| CN103249144A (en) | C-type-based wireless sensor network node location method | |
| CN103167605B (en) | A kind of WiFi outdoor positioning method that satellite auxiliary signal coverage diagram is set up/upgraded | |
| CN104039008B (en) | A kind of hybrid locating method | |
| CN113596989A (en) | Indoor positioning method and system for intelligent workshop | |
| CN103476113B (en) | System is set up based on MNL probability identification indoor locating system and method, location model | |
| Ma et al. | Subway station real-time indoor positioning system for cell phones | |
| Vy et al. | iBeacon indoor localization using trusted-ranges model | |
| Yang et al. | Unmanned aerial vehicle–assisted node localization for wireless sensor networks | |
| CN105960011A (en) | Indoor target positioning method based on sensor network and Bayesian method | |
| Lau et al. | Novel indoor localisation using an unsupervised Wi-Fi signal clustering method | |
| Zhang et al. | Indoor localization method based on AP and local linear regression algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |