CN108810799B - Multi-floor indoor positioning method and system based on linear discriminant analysis - Google Patents

Multi-floor indoor positioning method and system based on linear discriminant analysis Download PDF

Info

Publication number
CN108810799B
CN108810799B CN201810520501.0A CN201810520501A CN108810799B CN 108810799 B CN108810799 B CN 108810799B CN 201810520501 A CN201810520501 A CN 201810520501A CN 108810799 B CN108810799 B CN 108810799B
Authority
CN
China
Prior art keywords
floor
rss
target
group
pairing
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201810520501.0A
Other languages
Chinese (zh)
Other versions
CN108810799A (en
Inventor
罗娟
张振燕
王纯
郑燕柳
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hunan University
Original Assignee
Hunan University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Hunan University filed Critical Hunan University
Priority to CN201810520501.0A priority Critical patent/CN108810799B/en
Publication of CN108810799A publication Critical patent/CN108810799A/en
Application granted granted Critical
Publication of CN108810799B publication Critical patent/CN108810799B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/02—Services making use of location information
    • H04W4/023—Services making use of location information using mutual or relative location information between multiple location based services [LBS] targets or of distance thresholds
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
    • G06F18/24147—Distances to closest patterns, e.g. nearest neighbour classification
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/30—Services specially adapted for particular environments, situations or purposes
    • H04W4/33—Services specially adapted for particular environments, situations or purposes for indoor environments, e.g. buildings

Abstract

The invention discloses a multi-floor indoor positioning method and system based on linear discriminant analysis, and relates to the technical field of wireless multi-floor positioning.A positioning method is used for carrying out grid division on floors in an off-line stage, an off-line RSS fingerprint database is established by acquiring RSS values of APs sensed by sampling points, then a OvO splitting strategy is adopted to split the multi-floors into a plurality of groups of pairwise paired floors, pairwise pairing and classification are carried out on the APs in each group of pairwise paired floors, a classifier is established for each pairwise paired floor, on the basis of OvO splitting, an AP pairing group corresponding to the maximum Rayleigh quotient is obtained by adopting L DA (data access) to be an optimal AP pairing group, a corresponding classifier floor discrimination model is established by utilizing projection straight lines corresponding to floor attributes of the optimal AP pairing group and projection thresholds, when target positioning to be positioned is carried out, each classifier floor discrimination model is adopted, a voting method is combined to obtain floors where targets to be positioned, and coordinate positions where the targets.

Description

Multi-floor indoor positioning method and system based on linear discriminant analysis
Technical Field
The invention belongs to the technical field of wireless multi-floor positioning, and particularly relates to a multi-floor indoor positioning method and system based on linear discriminant analysis.
Background
The application demand based on the location service (L BS) is generally concerned with the popularization and wide application of mobile devices, such as the location of firefighters in fire scenes, the location of patients in medical centers, the location of personnel in businesses, the location of underground parking spaces, and the like.
To date, most indoor positioning studies have been based on two-dimensional space, i.e., single-floor studies. For multi-floor measurements, two-dimensional space cannot meet this requirement. At present, the indoor environment is mostly a multi-floor scene, such as a market, an airport, an office building and the like. Indoor positioning of multiple floors is increasingly receiving extensive attention and research.
Currently, the main research direction of the multi-floor indoor positioning technology is based on a WiFi fingerprint database, and there are two main positioning methods for distinguishing multiple floors by the WiFi fingerprint database: firstly, training reference position fingerprint sample data with floor numbers in an off-line stage (the common training method comprises artificial neural network, Bayesian classification and K nearest neighbor) to obtain a floor discrimination model, and judging a to-be-positioned point by using the obtained floor discrimination model in an on-line stage. However, as the number of ap (access point) and floors increases, the computational complexity of the method also increases. Secondly, the workload of manually collecting samples in an off-line stage is reduced, and a floor discrimination model is established by combining other methods (such as clustering), but the positioning precision is reduced to a certain extent by the method.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a multi-floor indoor positioning method and a system based on linear discriminant analysis, wherein the positioning method is used for carrying out grid division in an off-line stage and then collecting data in each grid so as to establish a complete off-line fingerprint database; then establishing a classifier floor discrimination model by using a linear discrimination analysis method; in the online stage, a floor where a target to be positioned is located is judged according to a classifier floor judgment model in the offline stage and by combining a voting method (the number of floors is more than 2); and finally, positioning the specific position of the target to be positioned by utilizing an improved KNN algorithm.
The invention solves the technical problems through the following technical scheme: a multi-floor indoor positioning method based on linear discriminant analysis comprises an off-line stage and an on-line stage, wherein the off-line stage comprises the following steps:
step (1), establishing an RSS fingerprint database: carrying out grid division on multi-floor indoor areas, collecting RSS values of all APs in each grid, and generating an RSS fingerprint database;
splitting the floor and the AP: adopting an OvO splitting strategy to split the multiple floors into multiple groups of pairwise paired floors, and then carrying out pairwise pairing classification on all APs contained in each group of paired floors to form multiple groups of AP pairing groups, wherein each group of AP pairing group only contains two APs;
step (3) building a classifier floor discrimination model: constructing a classifier for each group of matched floors, and calculating the floor attribute generalized Rayleigh quotient of each group of AP matched groups in each classifier by adopting a linear discriminant analysis method; selecting an AP pairing group corresponding to the maximum generalized Rayleigh quotient as an optimal AP pairing group corresponding to the classifier, and constructing a corresponding classifier floor discrimination model by using a projection straight line and a projection threshold value corresponding to the floor attribute of the optimal AP pairing group;
the floor attribute of each group of AP pairing group refers to the average value of RSS values of all grids of each floor in the corresponding pairing floor of each AP in the AP pairing group;
the number of classifiers is consistent with the number of matched floors; training the classifier by adopting a linear discriminant analysis method, and selecting an AP pairing group corresponding to the maximum generalized Rayleigh quotient as an optimal AP pairing group corresponding to the classifier; the input of the classifier is the average value of RSS values, and all APs contained in each group of matched floors are output as an AP matched group; training the classifiers by a linear discriminant analysis method and establishing classifier floor discriminant models, so that the floor where each classifier floor discriminant model is judged by a plurality of groups of AP pairing groups is changed into the floor judged by only one group of AP pairing groups, thereby greatly reducing the calculation amount and the calculation complexity;
all APs contained in the multiple floors have unique labels, after linear discriminant analysis processing, each group of paired floors corresponds to one classifier, each classifier corresponds to one group of optimal AP paired groups, the specific labels, the corresponding projection straight lines and projection thresholds of the APs in the optimal AP paired groups and the corresponding classifier floor discriminant models are uniquely determined, and the paired floors, the classifiers, the optimal AP paired groups, the specific labels, the projection straight lines and the projection thresholds of the APs are in one-to-one correspondence;
the online phase includes the following steps:
step (4) acquiring RSS values of all APs acquired by a target to be positioned;
step (5), judging the floor where the target to be positioned is located: combining RSS values of all APs collected by a target to be positioned according to an optimal AP pairing group, substituting the combined RSS values of the APs collected by the target to be positioned into corresponding classifier floor discrimination models, obtaining a floor discrimination result by each classifier floor discrimination model, and selecting the floor discrimination result with the highest voting number by combining a voting method to serve as a final result of the floor where the target to be positioned is located;
and (6) calculating the specific position of the target to be positioned.
Further, the specific operation steps of establishing the RSS fingerprint database in the step (1) are as follows:
step (1.1) setting N APs in a multi-floor indoor area, carrying out homogenization grid division on the multi-floor indoor area, dividing each floor into M grids, and enabling plane projection coordinates of the M grids of each floor to be the same; the uniform grid division is convenient for sample collection in an off-line stage;
uniformly selecting x sampling points in each grid, collecting RSS values of all APs sensed by each sampling point, and if y times, carrying out x × y times of sampling on each grid, then calculating the average value of the RSS values collected by each sampling point in each grid, the average value of the RSS values of each grid and the average value of the RSS values of all grids of each AP on each floor;
step (1.3) forming an RSS fingerprint library by using a fingerprint sequence of a plurality of sampling points, wherein the fingerprint sequence of each sampling point is represented as [ F ]i,Gij,RSS1,RSS2,...,RSSk,...,RSSn]，FiIndicating the ith floor, GijJ is less than or equal to M, RSSkAnd representing the RSS value of the kth AP acquired by the sampling point, wherein N represents the number of the APs perceived by the sampling point, N is less than N, and i, j and k all represent independent variables.
Further, the generalized rayleigh quotient J in the step (3) is calculated as:
Figure BDA0001674679250000031
wherein w represents the projection vector of the line onto which the sample points in the RSS fingerprint library are projected, u1Mean vector, u, representing the RSS values collected by each AP in each group of AP pairing groups at one of the paired floors2Mean vector representing the RSS values collected by each AP in each AP pairing group on the other of the paired floors, ∑1A covariance matrix representing the RSS values collected by each AP in each AP pairing group on one of the paired floors, ∑2Representing a covariance matrix of RSS values collected by each AP in each group of AP pairing groups on another floor in the pairing floors, | | | | sweet2Representing a norm.
Further, a projection vector w of a straight line projected by a sampling point in the RSS fingerprint library is obtained by the following formula:
Figure BDA0001674679250000032
wherein S isw＝∑1+∑2，SwAnd the two samples are respectively an RSS value set acquired by each AP in each group of AP pairing groups at one floor in the pairing floors and an RSS value set acquired by each AP in each group of AP pairing groups at the other floor in the pairing floors.
Further, the calculation expression of the projection threshold of the corresponding classifier floor discrimination model is as follows:
Figure BDA0001674679250000033
wherein theta represents a projection threshold of the floor discrimination model of the corresponding classifier, wTAnd the transposed matrix represents the projection vector of the projection straight line in the floor discrimination model of the corresponding classifier.
Further, in the step (5), the RSS values of the AP pairing group collected by the combined target to be positioned are projected onto the corresponding projection straight lines, and then the projection points of the RSS values are compared with the corresponding projection thresholds, if the projection points are greater than the projection thresholds, the RSS values belonging to the corresponding optimal AP pairing group on one floor are projected on the floor greater than the projection thresholds, and if the RSS values are less than the projection thresholds, the RSS values belonging to the other floor in the paired floors are projected on the floor.
Further, the step (6) adopts LL _ KNN (improved K-nearest neighbor algorithm) to calculate the specific position of the target to be located.
Further, the specific operation of calculating the specific position of the target to be positioned by adopting LL _ KNN is as follows:
step (6.1) setting RSS values of all APs collected by the target to be positioned
Figure BDA0001674679250000041
RSSkRepresenting the RSS value of the kth AP acquired by the target to be positioned, wherein k is an independent variable;
step (6.2) calculating the distance from the RSS value of the target to be positioned to the RSS value collected by all sampling points in the RSS fingerprint database of the floor where the target to be positioned is located, wherein the distance calculation formula is as follows:
Figure BDA0001674679250000042
wherein the RSSpkRepresenting the RSS value of the kth AP acquired by an acquisition point p in an RSS fingerprint library, wherein k and p are independent variables; n represents the number of APs perceived by the sampling point p; lpRepresenting the RSS value distance between the target to be positioned and the sampling point p; the number of the APs collected by the target to be positioned is equal to the number of the APs perceived by each sampling point;
step (6.3) selecting four sampling points with the nearest distance, and endowing the four sampling points with different weights lambda according to the distanceiI is 1,2,3,4, namely four sampling points with the nearest distance, and the weight calculation formula is
Figure BDA0001674679250000043
liDenotes the ith closest distance, l1+l2+l3+l4Represents the sum of the four closest distances;
step (6.4) calculating the coordinates (p) of the target to be positionedx,py)＝λ1(x1,y1)+λ2(x2,y2)+λ3(x3,y3)+λ4(x4,y4) Wherein (x)1,y1)、(x2,y2)、(x3,y3)、(x4,y4) Respectively representing the position coordinates of the four nearest sampling points in the RSS fingerprint database.
Further, a multi-floor indoor positioning system based on linear discriminant analysis is based on the multi-floor indoor positioning method based on linear discriminant analysis, and comprises an RSS signal acquisition unit, an RSS fingerprint library unit, an OvO floor splitting unit, a classifier floor discriminant model establishing unit, a target floor determining unit to be positioned and a target position determining unit to be positioned;
the RSS signal acquisition unit is used for acquiring RSS values of all APs sensed by sampling points in grids according to grid division of multi-floor indoor areas and transmitting floor information, grid information, sampling point information and corresponding RSS values to the RSS fingerprint database; the RSS signal acquisition unit is also used for acquiring RSS values of the APs sensed by the target to be positioned and transmitting the information to the classifier floor discrimination model establishing unit;
the RSS fingerprint database unit is used for establishing an RSS fingerprint database according to the floor information, the grid information, the sampling point information and the corresponding RSS value acquired by the RSS signal acquisition unit;
the OvO floor splitting unit is used for splitting multiple floors into multiple groups of pairwise matching floors, and pairwise matching and classifying all APs contained in each group of matching floors to form multiple groups of AP matching groups;
the classifier floor discrimination model establishing unit is used for respectively calculating generalized Rayleigh quotient of each group of AP pairing group in each group of pairing floors by adopting a linear discrimination analysis method, and establishing a floor discrimination model of the pairing floor classifier according to the AP pairing group corresponding to the solved maximum generalized Rayleigh quotient;
the system comprises a positioning target floor determining unit, a classifier floor judging model and a positioning target locating unit, wherein the positioning target floor determining unit is used for combining RSS values of all APs sensed by the RSS signal acquisition unit according to an optimal AP pairing group and substituting the RSS values of the APs acquired by the combined positioning target into the corresponding classifier floor judging model to obtain the floor where the positioning target is located;
and the target position determining unit is used for calculating the coordinates of the target to be positioned according to the RSS value of the target to be positioned and the RSS value in the RSS fingerprint database.
Has the advantages that:
the invention provides a multi-floor indoor positioning method and a system based on linear discriminant analysis, wherein the positioning method is used for carrying out grid division on floors in an off-line stage, selecting sampling points to obtain RSS values of all APs sensed by the sampling points, and establishing an off-line RSS fingerprint database; then, adopting an OvO splitting strategy to split the multiple floors into multiple groups of pairwise matching floors, and carrying out pairwise matching classification on the APs in each group of matching floors to form multiple groups of AP matching groups; each group of matched floors corresponds to one classifier, on the basis of OvO splitting, an AP matched group corresponding to the maximum Rayleigh quotient is obtained by adopting a linear discrimination analysis method, namely the optimal AP matched group, and a corresponding classifier floor discrimination model is constructed by utilizing a projection straight line corresponding to the floor attribute of the optimal AP matched group and a projection threshold value; when positioning a target to be positioned, combining RSS values of the target to be positioned according to an optimal AP pairing group, substituting the RSS values of the APs acquired by the combined target to be positioned into the corresponding classifier floor discrimination model, acquiring the floor where the target to be positioned is located by combining a voting method, and finally calculating the coordinate position where the target to be positioned is located;
according to the method, on the basis of OvO splitting, linear discriminant analysis methods are adopted to construct each classifier floor discriminant model, each group of AP pairing groups only comprises two APs, when each classifier floor discriminant model is adopted to position a target to be positioned, all the APs are not required to be considered, only two APs in the optimal AP pairing group are required to be considered, the complexity and the calculation amount of calculation are reduced, the floor discriminant precision is improved, and after the floor is determined, LL _ KNN is adopted to calculate the coordinates of the target to be positioned finally, and the positioning precision is improved.
Drawings
In order to more clearly illustrate the technical solution of the present invention, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only one embodiment of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on the drawings without creative efforts.
FIG. 1 is a block diagram of a multi-floor indoor positioning method and system based on linear discriminant analysis according to the present invention;
FIG. 2 is a schematic diagram of OvO floor determination in conjunction with voting in the present invention;
FIG. 3 is a classification result obtained by the maximum generalized Rayleigh quotient according to an embodiment of the present invention;
FIG. 4 is a classification result obtained from the least generalized Rayleigh quotient according to an embodiment of the present invention;
FIG. 5 is a comparison of the positioning accuracy between the floor discrimination model and the case of no floor discrimination according to the present invention.
Detailed Description
The technical solutions in the present invention are clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As shown in fig. 1, the multi-floor indoor positioning method based on linear discriminant analysis provided by the present invention includes an offline stage and an online stage, wherein the offline stage includes the following steps:
step (1), establishing an RSS fingerprint database: the method comprises the following steps of performing grid division on multi-floor indoor areas, collecting RSS values of APs in each grid, and generating an RSS fingerprint database, wherein the method comprises the following specific operation steps:
the method comprises the following steps that (1.1) N APs are arranged in a multi-floor indoor area, the multi-floor indoor area is subjected to uniform grid division, each floor is divided into M grids, and the plane projection coordinates of the M grids of each floor are the same, in the embodiment, 3 × 3 meters are used as one grid;
uniformly selecting x sampling points (averagely selecting one point as a sampling point every 1 meter) in each grid, acquiring the RSS value of each AP sensed by each sampling point, and carrying out x × y times of sampling on each grid if y times, then calculating the average value of the RSS values acquired by each sampling point in each grid, the average value of the RSS values of each grid and the average value of the RSS values of all grids of each AP on each floor;
step (1.3) forming an RSS fingerprint library by using a fingerprint sequence of a plurality of sampling points, wherein the fingerprint sequence of each sampling point is represented as [ F ]i,Gij,RSS1,RSS2,...,RSSk,...,RSSn]，FiIndicating the ith floor, GijJ is less than or equal to M, RSSkAnd representing the RSS value of the kth AP acquired by the sampling point, wherein N represents the number of the APs perceived by the sampling point, N is less than N, and i, j and k all represent independent variables.
Splitting the floor and the AP: adopt OvO split strategy to disassemble the multi-floor into two liang of pairs of floors of multiunit (if the floor number is two, then need not to disassemble the floor, directly for a set of pairs of floors), then pair two liang of all APs that contain in the floor to every group and pair the classification, form multiunit AP and match the group, only contain two APs in every group AP matches the group.
In the indoor positioning of multiple floors (the number of floors is more than or equal to 3), because the number of floors is more, the complexity and the difficulty of positioning are increased, the splitting strategy is adopted to split the multiple floors into the multiple floors which are pairwise paired, namely, the split multiple floors are split into the multiple two-classification tasks for solving, so that the problem of the multiple floors is solved, and the complexity of positioning is reduced.
The splitting strategy is to split a multi-classification task (multi-floor) into a plurality of two-classification tasks (a plurality of groups of paired floors), namely, splitting a problem firstly, and then constructing a classifier for each split two-classification task; and when in on-line positioning, voting is carried out on the prediction results of the classifier floor discrimination models constructed by the classifiers, and the prediction result with the largest number of votes is taken as the final classification result.
The most classical resolution strategies have three types: One-to-One (One vs. One, OvO for short), One-to-the-rest (One vs. rest, OvR for short), and Many-to-Many (man y vs. man y, MvM for short). For K classes, OvR and OvO training K (K-1)/2 classifiers are required, so the storage overhead and test time overhead of OvO is usually greater than OvR, but when training, OvR each classifier uses all training samples, and OvO each classifier uses only two class samples, so when there are more classes, OvO training time overhead is usually less than OvR. For the prediction performance, depending on the specific data distribution, the two are mostly not very different. Therefore, the invention adopts OvO splitting strategy to split the floor and AP attributes.
As shown in fig. 2, taking 3 floors (f1, f2, f3 represent 3 floors) as an example, first, 3 floors are disassembled into 3 groups of paired floors ((f1, f2), (f2, f3), (f1, f3)), and then all APs included in the 3 groups of paired floors are pairwise classified to form N (N-1)/2 AP pairing groups.
Step (3) building a classifier floor discrimination model: constructing a classifier for each group of matched floors, and calculating the floor attribute generalized Rayleigh quotient of each group of AP matched groups in each classifier by adopting a linear discriminant analysis method; selecting an AP pairing group corresponding to the maximum generalized Rayleigh quotient as an optimal AP pairing group corresponding to the classifier, and constructing a corresponding classifier floor discrimination model by using a projection straight line and a projection threshold value corresponding to the floor attribute of the optimal AP pairing group;
the floor attribute of each group of AP pairing group refers to an average of RSS values of all grids of each floor in the corresponding pairing floor of each AP in the AP pairing group.
Taking 3 floors, 2 APs per floor as an example, 3 floors are respectively represented by f1, f2 and f3, two APs of f1 are represented by AP1-1 and AP1-2, two APs of f2 are represented by AP2-1 and AP2-2, and two APs of f3 are represented by AP3-1 and AP 3-2. The 3 floors are pairwise matched, and then the 3 floors are divided into 3 groups of matched floors (f1, f2), (f2, f3), (f1 and f 3); taking the paired floors (f1, f2) as an example, the APs in the paired floors of the group are pairwise classified, so that 6 groups of AP paired groups are respectively (AP1-1, AP1-2), (AP1-1, AP2-1), (AP1-1, AP2-2), (AP1-2, AP2-1), (AP1-2, AP2-2), (AP2-1, AP 2-2). And (4) sequentially analogizing other groups to floors to obtain AP (access point) matching groups of each group to floors.
Taking the AP pairing group (AP1-1, AP2-1) in the pairing floor (f1, f2) as an example, the floor attributes of each AP in the pairing floor (f1, f2) of the AP pairing group (AP1-1, AP2-1) include the average of the RSS values of the AP1-1 in all grids of f1, the average of the RSS values of the AP1-1 in all grids of f2, the average of the RSS values of the AP2-1 in all grids of f2, the average of the RSS values of the AP2-1 in all grids of f1, and the attributes of the other AP pairing groups in the pairing floor (f1, f2) are analogized in turn.
By ui＝[uij]The mean vector of RSS values collected at the ith floor by each group of AP pairing groups in each group of pairing floors is represented, wherein the vector element uijRepresents the average of the RSS values of all grids collected by the jth AP on the ith floor; since each paired floor group only contains 2 floors, i is 1,2, and since each paired AP group only contains 2 APs, j is 1,2, resulting in u1＝[u11,u12]And u2＝[u21,u22]，u1For each group of APs, pairing each AP in the group with a mean vector, u, of RSS values collected at one of the paired floors11Pairing groups for APsAverage of all grid RSS values, u, collected by an AP on a floor12The average value of the RSS values of all grids acquired by the other AP in the AP pairing group on one floor is obtained; u. of2For each group of APs, the mean vector u of RSS values collected by each AP in the paired group on the other floor in the paired floors21Average, u, of all grid RSS values collected on one floor of another in a group of AP pairs22The average of the RSS values of all the grids collected on another floor by another AP in the AP pairing group.
Taking the paired floors (f1, f2) and the AP paired group (AP1-1, AP2-1) as an example, u11、u12Respectively, the average of the RSS values of all grids collected by the AP1-1 and AP2-1 at the floor f1, and if u is 3 grids per floor, u is11(70.28+68.67+ 70.39)/3: 69.78, 70.28, 68.67, 70.39 are RSS values of AP1-1 in 3 cells of floor f1, respectively, u12(72.72+74.11+71.50)/3 ═ 72.78, 72.72, 74.11, 71.50 are RSS values of AP2-1 in 3 cells of floor f1, respectively, and u is the RSS value of AP2-1 in 3 cells of floor f11＝[69.78,72.78](ii) a In the same way, u21、u22Represents the average, u, of the RSS values of all grids collected by AP1-1 and AP2-1, respectively, at floor f22＝[62.54,59.96]. The actual RSS values are negative values, which are convenient to calculate here, and the negative signs are omitted, as shown in fig. 3 and 4.
The generalized Rayleigh quotient J is calculated by the formula:
Figure BDA0001674679250000091
wherein w represents the line projected by the sampling point in the RSS fingerprint database (the projection vector projected by the sampling point to the line is used for representing in calculation), and u represents the projection vector projected by the sampling point to the line1Mean vector, u, representing the RSS values collected by each AP in each group of AP pairing groups at one of the paired floors2Mean vector representing the RSS values collected by each AP in each AP pairing group on the other of the paired floors, ∑1Representing the covariance of the RSS values collected by each AP in each AP pairing group at one of the paired floorsMatrix, ∑1Each element in (1) is u1Covariance between the elements of medium vector ∑2A covariance matrix representing the RSS values collected by each AP in each AP pairing group on the other of the paired floors, ∑2Each element in (1) is u2The covariance between medium vector elements, | | | | non-woven phosphor2Representing a norm.
When calculating the generalized Rayleigh quotient, firstly, u is calculated according to the floor attribute of each group of AP pairing groups1、u2、∑1、∑2And w, and then J is obtained according to the formula (1).
The invention relates to a linear discriminant analysis method (L DA) which is a supervised dimension reduction technology and can only process two classifications, wherein the invention adopts a OvO splitting strategy to solve the problem of multiple floors, and then adopts L DA to process, the principle of L DA is that a sample set with labels is projected on a straight line by a projection method, so that the projection points of the same type of samples are as close as possible, the projection points of the different types of samples are as far as possible, when classifying new samples, the new samples are projected on the straight line of the same type of samples, and then the classification of the new samples is determined according to the positions of the projection points, in indoor multi-floor positioning, the AP with the best classification effect (the number of the AP is between 1 and N-1, and N is the total number of the APs in the area) is selected as the attribute of floor classification by the dimension reduction technology.
Let the projected straight line be w, and to make the projection points of the same type of samples (samples collected on the same floor) as close as possible, the covariance of the projection points of the same type of samples can be made as small as possible, that is, wT×(∑1+∑2) × w are as small as possible, to make the projection points of the heterogeneous samples (samples taken on different floors) as far as possible, the distance between the class centers can be made as large as possible, i.e.:
Figure BDA0001674679250000092
as large as possible. The quotient of the two is considered at the same time, and the target to be maximized is the formula (1).
Definition 1: the intra-class divergence matrix (here the sum of the covariance matrices of the two classes of samples in each set of paired floors) is denoted by the symbol SwIs shown to be
Figure BDA0001674679250000093
Wherein X represents X1、X2One sample point of (2), X1Representing the set of RSS values, X, collected by each AP in each group of AP pairing groups at one of the paired floors2Representing a set of RSS values collected by each AP in each set of AP pairing groups at another floor in the pairing floor.
Definition 2: the inter-class divergence matrix (here the product of the mean vectors of the two classes of samples in each set of paired floors) is given by SbThe symbol indicates that
Sb＝(u1-u2)×(u1-u2)T(3)
Rewriting formula (1) as
Figure BDA0001674679250000101
The formula (4) is the target to be maximized, i.e. SbAnd SwGeneralized Rayleigh quotient of (1).
As can be seen from the expression (4), the numerator and the denominator are both quadratic terms related to w, so that the solution of the expression (4) has no relation with the length of w, and is only related to the direction thereof, let wTSww is 1, then formula (4) is equivalent to
Figure BDA0001674679250000102
Introducing Lagrange multiplier lambda by Lagrange multiplier method, wherein the formula (5) is equivalent to Sbw＝λSww (6) due to Sbw is constantly (u)1-u2) Let us order
Sbw＝λ(u1-u2) Substituting into equation (6) yields: w ═ Sw -1(u1-u2) (7), the projection line w can be obtained.
Taking OvO two-storey buildings as an example, 6 APs are uniformly distributed in the first and second-storey scenes, the generalized rayleigh quotient of the floor attribute of each group of AP pairing groups is calculated according to formula (1), and the floor classification results obtained by calculating the maximum value and the minimum value are shown in fig. 3 and 4 below. In fig. 3, the maximum value is shown as the classification result, and fig. 4 is the minimum value is shown as the classification result, where in APX _ Y, X represents the floor number and Y represents the reference number of the AP. For example, AP1_3 represents an AP numbered 3 deployed on floor 1, and AP2_1 represents an AP numbered 1 deployed on floor 2. The results shown in fig. 3 and 4 indicate that an optimal AP pairing group can be selected according to the calculated maximum generalized rayleigh quotient, and then a corresponding classifier floor discrimination model is constructed by using a projection straight line and a projection threshold corresponding to the floor attribute of the optimal AP pairing group, wherein the corresponding classifier floor discrimination model is used for determining the number of floors where the target to be positioned is located, and the determination result is accurate.
The online phase includes the following steps:
and (4) acquiring the RSS value of each AP acquired by the target to be positioned.
Step (5), judging the floor where the target to be positioned is located: and combining the RSS values of the APs acquired by the targets to be positioned according to the optimal AP pairing group, substituting the combined RSS values of the APs acquired by the targets to be positioned into the corresponding classifier floor judgment models, obtaining a floor judgment result by each classifier floor judgment model, and selecting the floor judgment result with the highest voting number by combining a voting method to serve as the final result of the floor where the targets to be positioned are located.
And projecting the RSS value of the combined AP pairing group acquired by the target to be positioned onto the corresponding projection straight line, then comparing the projection point with the corresponding projection threshold, if the RSS value is greater than the projection threshold, determining that the projection point of the RSS value of one floor of the corresponding optimal AP pairing group is greater than the projection threshold, and if the RSS value is less than the projection threshold, determining that the RSS value belongs to the other floor of the paired floors.
Taking 3 groups of paired floors (f1, f2), (f2, f3), (f1, f3) as an example, setting the optimal AP paired groups of (f1, f2) as (AP1-1, AP2-1), (f2, f3) as (AP2-2, AP3-1), (f1, f3) as (AP1-2, AP3-2), determining the optimal AP paired group of each group of paired floors, determining specific labels of APs in the optimal AP paired groups, and constructing a corresponding classifier floor discrimination model by using projection lines and projection thresholds corresponding to floor attributes of the optimal AP paired groups.
Collecting RSS values of all APs at a target to be positioned, setting RSS values of APs 1-1, AP1-2, AP2-1, AP2-2, AP3-1 and AP3-2 collected at the target to be positioned, substituting the RSS values of the APs collected by the targets to be positioned into corresponding classifier floor discrimination models according to the labels (AP1-1, AP1-2, AP2-1, AP2-2, AP3-1 and AP3-2) of the target to be positioned and the specific labels of the APs in the optimal AP pairing group of each classifier floor discrimination model, projecting the RSS values of the APs 1-1 and AP2-1 collected by the targets to be positioned to (f1 and f2) classifiers (the combination of the AP pairing groups is AP1-1 and AP2-1) on the projection straight line corresponding to the floor discrimination models, and collecting the RSS values of the targets to be positioned by the AP 2-2-1, The RSS value of the AP3-1 is projected to a projection straight line corresponding to a floor discrimination model of a classifier (optimal AP pairing combination is AP2-2 and AP3-1) of the (f2 and f3), the RSS values of the AP1-2 and AP3-2 collected by the target to be positioned are projected to a projection straight line corresponding to a floor discrimination model of the (f1 and f3) classifier (optimal AP pairing combination is AP1-2 and AP3-2), each projection point is compared with a corresponding projection threshold value respectively to obtain a floor discrimination result in each classifier floor model, and a floor discrimination result with the highest voting number is selected as a final floor of the target to be positioned by combining a voting method, as shown in FIG. 2, the highest voting number of f1 is 2, and the floor of the target to be positioned is f 1.
The optimal AP pairing combination, obtained from the calculated largest relegator, is not necessarily distributed on two different floors, but may be the same floor.
On a projection straight line corresponding to the classifier floor discrimination model, taking the mean value of the central values of the two types of samples as a projection threshold, wherein the calculation expression of the projection threshold corresponding to the classifier floor discrimination model is as follows:
Figure BDA0001674679250000111
wherein θ represents the correspondenceProjection threshold, w, of classifier floor discrimination modelTAnd the transposed matrix represents the projection vector of the projection straight line in the floor discrimination model of the corresponding classifier.
And (6) calculating the specific position of the target to be positioned.
Step (6.1) setting RSS values of all APs collected by the target to be positioned
Figure BDA0001674679250000121
RSSkRepresenting the RSS value of the kth AP acquired by the target to be positioned, wherein k is an independent variable;
step (6.2) calculating the distance from the RSS value of the target to be positioned to the RSS value collected by all sampling points in the RSS fingerprint database of the floor where the target to be positioned is located, wherein the distance calculation formula is as follows:
Figure BDA0001674679250000122
wherein the RSSpkRepresenting the RSS value of the kth AP acquired by an acquisition point p in an RSS fingerprint library, wherein k and p are independent variables; n represents the number of APs perceived by the sampling point p; lpRepresenting the RSS value distance between the target to be positioned and the sampling point p, wherein the distance is calculated by the square sum of the difference of the corresponding RSS values of the APs between the target to be positioned and the sampling point; the number of the APs collected by the target to be positioned is equal to the number of the APs perceived by each sampling point;
step (6.3) selecting four sampling points with the nearest distance, and endowing the four sampling points with different weights lambda according to the distanceiI is 1,2,3,4, namely four sampling points with the nearest distance, and the weight calculation formula is
Figure BDA0001674679250000123
liDenotes the ith closest distance, l1+l2+l3+l4Representing the sum of the four closest distances. In the embodiment, 4 sampling points are selected according to experience, 3 or less sampling points are selected relatively less, the error is large, more than 4 sampling points are selected far away, and the error is large, so that 4 sampling points are selected most appropriately;
step (6.4) calculating the coordinates (p) of the target to be positionedx,py)＝λ1(x1,y1)+λ2(x2,y2)+λ3(x3,y3)+λ4(x4,y4) Wherein (x)1,y1)、(x2,y2)、(x3,y3)、(x4,y4) Respectively representing the position coordinates of the four nearest sampling points in the RSS fingerprint database.
After the floors are determined by adopting an improved K-nearest neighbor algorithm, the APs with the closer distances of other floors are filtered, the APs are only selected in the same floor, and finally, positioning is carried out. Compared with the traditional K-neighbor algorithm, the method only finds the nearest AP for positioning and does not distinguish floors, so that the positioning accuracy of the improved K-neighbor algorithm is higher for the multi-floor condition.
The method comprises the steps of selecting 30 targets to be positioned for testing, wherein one is floor judgment by using the floor judgment model of the invention, and the other is floor judgment by testing without floor judgment, specifically, as shown in figure 5, the average positioning accuracy of positioning by using the floor judgment model of the invention is 1.38m, the average positioning accuracy of positioning without floor judgment is 3.41m, and the positioning accuracy is obviously improved, and meanwhile, the positioning by using the floor judgment model of the invention has small relative difference of the positioning accuracy, and floats around 2 meters, and the positioning is performed without floor judgment, so that the positioning accuracy greatly fluctuates, the best positioning accuracy reaches within 1 meter, and the worst positioning accuracy reaches over 7 meters.
As shown in fig. 1, a multi-floor indoor positioning system based on linear discriminant analysis, based on the multi-floor indoor positioning method based on linear discriminant analysis of any one of claims 1 to 8, includes an RSS signal acquisition unit, an RSS fingerprint library unit, an OvO floor splitting unit, a classifier floor discriminant model establishing unit, a target floor to be positioned determining unit, and a target position to be positioned determining unit;
the RSS signal acquisition unit is used for acquiring RSS values of all APs sensed by sampling points in grids according to grid division of multi-floor indoor areas and transmitting floor information, grid information, sampling point information and corresponding RSS values to the RSS fingerprint database; the RSS signal acquisition unit is also used for acquiring RSS values of the APs sensed by the target to be positioned and transmitting the information to the classifier floor discrimination model establishing unit;
the RSS fingerprint database unit is used for establishing an RSS fingerprint database according to the floor information, the grid information, the sampling point information and the corresponding RSS value acquired by the RSS signal acquisition unit;
the OvO floor splitting unit is used for splitting multiple floors into multiple groups of pairwise matching floors, and pairwise matching and classifying all APs contained in each group of matching floors to form multiple groups of AP matching groups;
the classifier floor discrimination model establishing unit is used for respectively calculating generalized Rayleigh quotient of each group of AP pairing group in each group of pairing floors by adopting a linear discrimination analysis method, and establishing a floor discrimination model of the pairing floor classifier according to the AP pairing group corresponding to the solved maximum generalized Rayleigh quotient;
the system comprises a positioning target floor determining unit, a classifier floor judging model and a positioning target locating unit, wherein the positioning target floor determining unit is used for combining RSS values of all APs sensed by the RSS signal acquisition unit according to an optimal AP pairing group and substituting the RSS values of the APs acquired by the combined positioning target into the corresponding classifier floor judging model to obtain the floor where the positioning target is located;
and the target position determining unit is used for calculating the coordinates of the target to be positioned according to the RSS value of the target to be positioned and the RSS value in the RSS fingerprint database.
The above-mentioned embodiments are further described in detail in the technical field, background, objects, schemes and advantages of the present invention, and it should be understood that the embodiments are only preferred embodiments of the present invention and are not intended to limit the present invention, and any modifications, equivalents, improvements and the like made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (9)

1. A multi-floor indoor positioning method based on linear discriminant analysis comprises an off-line stage and an on-line stage, and is characterized in that the off-line stage comprises the following steps:
step (1), establishing an RSS fingerprint database: carrying out grid division on multi-floor indoor areas, collecting RSS values of all APs in each grid, and generating an RSS fingerprint database;
splitting the floor and the AP: adopting an OvO splitting strategy to split multiple floors into multiple groups of pairwise paired floors, and then carrying out pairwise pairing classification on all APs contained in each group of paired floors to form multiple groups of AP paired groups; OvO denotes one-to-one;
step (3) building a classifier floor discrimination model: constructing a classifier for each group of matched floors, and calculating the floor attribute generalized Rayleigh quotient of each group of AP matched groups in each classifier by adopting a linear discriminant analysis method; selecting an AP pairing group corresponding to the maximum generalized Rayleigh quotient as an optimal AP pairing group corresponding to the classifier, and constructing a corresponding classifier floor discrimination model by using a projection straight line and a projection threshold value corresponding to the floor attribute of the optimal AP pairing group;
the floor attribute of each group of AP pairing group refers to the average value of RSS values of all grids of each floor in the corresponding pairing floor of each AP in the AP pairing group;
the online phase includes the following steps:
step (4) acquiring RSS values of all APs acquired by a target to be positioned;
step (5), judging the floor where the target to be positioned is located: combining RSS values of all APs collected by a target to be positioned according to an optimal AP pairing group, substituting the combined RSS values of the APs collected by the target to be positioned into corresponding classifier floor discrimination models, obtaining a floor discrimination result by each classifier floor discrimination model, and selecting the floor discrimination result with the highest voting number by combining a voting method to serve as a final result of the floor where the target to be positioned is located;
and (6) calculating the specific position of the target to be positioned.
2. The linear discriminant analysis-based multi-floor indoor positioning method as claimed in claim 1, wherein the step (1) of establishing the RSS fingerprint database comprises the following steps:
step (1.1) setting N APs in a multi-floor indoor area, carrying out homogenization grid division on the multi-floor indoor area, dividing each floor into M grids, and enabling plane projection coordinates of the M grids of each floor to be the same;
uniformly selecting x sampling points in each grid, collecting the RSS value of each AP sensed by each sampling point, and if y times, carrying out x × y times of sampling on each grid, and then calculating the average value of the RSS values collected by each sampling point in each grid, the average value of the RSS values of each grid and the average value of the RSS values of all grids of each AP on each floor;
step (1.3) forming an RSS fingerprint library by using a fingerprint sequence of a plurality of sampling points, wherein the fingerprint sequence of each sampling point is represented as [ F ]i,Gij,RSS1,RSS2,...,RSSk,...,RSSn]，FiIndicating the ith floor, GijJ is less than or equal to M, RSSkAnd representing the RSS value of the kth AP acquired by the sampling point, wherein N represents the number of the APs perceived by the sampling point, N is less than N, and i, j and k all represent independent variables.
3. The linear discriminant analysis-based multi-floor indoor positioning method as claimed in claim 1, wherein the generalized rayleigh quotient J in step (3) is calculated by:
Figure FDA0002526321260000021
wherein w represents the projection vector of the line onto which the sampling points in the RSS fingerprint library are projected, and u represents the distance between the sampling points1Mean vector, u, representing the RSS values collected by each AP in each group of AP pairing groups at one of the paired floors2Mean vector representing the RSS values collected by each AP in each AP pairing group on the other of the paired floors, ∑1Each AP in each group of AP pairing group is represented on one floor in pairing floorsCovariance matrix of collected RSS values, ∑2Representing a covariance matrix of RSS values collected by each AP in each group of AP pairing groups on another floor in the pairing floors, | | | | sweet2Representing a norm.
4. The linear discriminant analysis-based multi-floor indoor positioning method as claimed in claim 3, wherein a projection vector w of a straight line projected by the sampling points in the RSS fingerprint database is obtained by the following formula:
Figure FDA0002526321260000022
wherein S isw＝∑1+∑2。
5. The linear discriminant analysis-based multi-floor indoor positioning method of claim 4, wherein the computational expression of the projection threshold of the corresponding classifier floor discriminant model is as follows:
Figure FDA0002526321260000023
wherein theta represents a projection threshold of the floor discrimination model of the corresponding classifier, wTAnd the transposed matrix represents the projection vector of the projection straight line in the floor discrimination model of the corresponding classifier.
6. The linear discriminant analysis-based multi-floor indoor positioning method according to claim 1, wherein in the step (5), the RSS values of the combined AP pairing group collected by the target to be positioned are projected onto the corresponding projection straight line, and then the projection points are compared with the corresponding projection threshold, if the RSS values of the combined AP pairing group collected by the target to be positioned are greater than the projection threshold, the RSS value of the projection point belonging to the corresponding optimal AP pairing group on one floor is greater than the projection threshold, and if the RSS value of the combined AP pairing group on the other floor is less than the projection threshold, the combined AP pairing group belongs to the other floor.
7. The linear discriminant analysis-based multi-floor indoor positioning method as claimed in claim 1, wherein the step (6) employs LL _ KNN to calculate the specific location of the target to be positioned, LL _ KNN representing an improved K-nearest neighbor algorithm.
8. The linear discriminant analysis-based multi-floor indoor positioning method as claimed in claim 7, wherein the specific operation of calculating the specific position of the target to be positioned using LL _ KNN is as follows:
step (6.1) setting RSS values of all APs collected by the target to be positioned
Figure FDA0002526321260000031
RSSkRepresenting the RSS value of the kth AP acquired by the target to be positioned, wherein k is an independent variable;
step (6.2) calculating the distance from the RSS value of the target to be positioned to the RSS value collected by all sampling points in the RSS fingerprint database of the floor where the target to be positioned is located, wherein the distance calculation formula is as follows:
Figure FDA0002526321260000032
wherein the RSSpkRepresenting the RSS value of the kth AP acquired by an acquisition point p in an RSS fingerprint library, wherein k and p are independent variables; n represents the number of APs perceived by the sampling point p; lpRepresenting the RSS value distance between the target to be positioned and the sampling point p;
step (6.3) selecting four sampling points with the nearest distance, and endowing the four sampling points with different weights lambda according to the distanceiI is 1,2,3,4, namely four sampling points with the nearest distance, and the weight calculation formula is
Figure FDA0002526321260000033
liDenotes the ith closest distance, l1+l2+l3+l4Represents the sum of the four closest distances;
step (6.4) calculating the coordinates (p) of the target to be positionedx,py)＝λ1(x1,y1)+λ2(x2,y2)+λ3(x3,y3)+λ4(x4,y4) Wherein (x)1,y1)、(x2,y2)、(x3,y3)、(x4,y4) Respectively representing the position coordinates of the four nearest sampling points in the RSS fingerprint database.
9. A multi-floor indoor positioning system based on linear discriminant analysis, which is characterized in that the multi-floor indoor positioning method based on linear discriminant analysis is based on any one of claims 1 to 8, and comprises an RSS signal acquisition unit, an RSS fingerprint library unit, an OvO floor splitting unit, a classifier floor discriminant model establishing unit, a target floor to be positioned determining unit and a target position to be positioned determining unit;
the RSS signal acquisition unit is used for acquiring RSS values of all APs sensed by sampling points in grids according to grid division of multi-floor indoor areas and transmitting floor information, grid information, sampling point information and corresponding RSS values to the RSS fingerprint database; the RSS signal acquisition unit is also used for acquiring RSS values of the APs sensed by the target to be positioned and transmitting the information to the classifier floor discrimination model establishing unit;
the RSS fingerprint database unit is used for establishing an RSS fingerprint database according to the floor information, the grid information, the sampling point information and the corresponding RSS value acquired by the RSS signal acquisition unit;
the OvO floor splitting unit is used for splitting multiple floors into multiple groups of pairwise matching floors, and pairwise matching and classifying all APs contained in each group of matching floors to form multiple groups of AP matching groups;
the classifier floor discrimination model establishing unit is used for calculating the floor attribute generalized Rayleigh quotient of each group of AP pairing group in the classifier corresponding to each group of pairing floors by adopting a linear discrimination analysis method and establishing a corresponding classifier floor discrimination model according to the AP pairing group corresponding to the maximum generalized Rayleigh quotient;
the to-be-positioned target floor determining unit is used for combining RSS values of all APs sensed by the RSS signal collecting unit according to the optimal AP pairing group, substituting the RSS values of the APs collected by the combined to-be-positioned targets into the corresponding classifier floor distinguishing model, and finally combining a voting method to obtain the floor where the to-be-positioned target is located;
and the target position determining unit is used for calculating the coordinates of the target to be positioned according to the RSS value of the target to be positioned and the RSS value in the RSS fingerprint database.
CN201810520501.0A 2018-05-28 2018-05-28 Multi-floor indoor positioning method and system based on linear discriminant analysis Active CN108810799B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810520501.0A CN108810799B (en) 2018-05-28 2018-05-28 Multi-floor indoor positioning method and system based on linear discriminant analysis

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810520501.0A CN108810799B (en) 2018-05-28 2018-05-28 Multi-floor indoor positioning method and system based on linear discriminant analysis

Publications (2)

Publication Number Publication Date
CN108810799A CN108810799A (en) 2018-11-13
CN108810799B true CN108810799B (en) 2020-07-24

Family

ID=64090553

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810520501.0A Active CN108810799B (en) 2018-05-28 2018-05-28 Multi-floor indoor positioning method and system based on linear discriminant analysis

Country Status (1)

Country Link
CN (1) CN108810799B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110231593A (en) * 2019-04-11 2019-09-13 深圳市城市交通规划设计研究中心有限公司 Indoor orientation method, device, computer readable storage medium and terminal device
CN110401977B (en) * 2019-06-21 2021-03-02 湖南大学 Multi-floor indoor positioning method based on Softmax regression multi-classification recognizer
CN111372212A (en) * 2020-03-17 2020-07-03 杭州十域科技有限公司 Fingerprint matching method with low algorithm complexity
CN115685060A (en) * 2021-07-28 2023-02-03 华为技术有限公司 Indoor fingerprint map construction method and related device

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102791025A (en) * 2011-05-20 2012-11-21 盛乐信息技术（上海）有限公司 Wireless fidelity (WIFI) based layered positioning system and implementing method
CN103874200A (en) * 2014-03-19 2014-06-18 北京邮电大学 Floor recognition method and system
CN103916821A (en) * 2014-04-11 2014-07-09 北京工业大学 Floor distinguishing method based on RSSI difference between floors
CN106814345A (en) * 2017-03-24 2017-06-09 西北工业大学 It is easy to data acquisition and high-precision indoor orientation method

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9374678B2 (en) * 2014-02-28 2016-06-21 Texas Instruments Incorporated Kalman filter iteratively performing time/measurement update of user/relative floor locations

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102791025A (en) * 2011-05-20 2012-11-21 盛乐信息技术（上海）有限公司 Wireless fidelity (WIFI) based layered positioning system and implementing method
CN103874200A (en) * 2014-03-19 2014-06-18 北京邮电大学 Floor recognition method and system
CN103916821A (en) * 2014-04-11 2014-07-09 北京工业大学 Floor distinguishing method based on RSSI difference between floors
CN106814345A (en) * 2017-03-24 2017-06-09 西北工业大学 It is easy to data acquisition and high-precision indoor orientation method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
A light WLAN radio map for floor detection in multifloor environment localization;Iyad H Alshami;《2015 9TH Malaysian software engineering conference》;20151217;全文 *

Also Published As

Publication number Publication date
CN108810799A (en) 2018-11-13

Similar Documents

Publication Publication Date Title
CN108810799B (en) Multi-floor indoor positioning method and system based on linear discriminant analysis
Zheng et al. Exploiting fingerprint correlation for fingerprint-based indoor localization: A deep learning-based approach
CN112699913A (en) Transformer area household variable relation abnormity diagnosis method and device
CN111199016A (en) DTW-based improved K-means daily load curve clustering method
CN107727095B (en) 3D indoor positioning method based on spectral clustering and weighted back propagation neural network
CN103954940B (en) Radar fence interferes discrimination method with distributed compacting based on the centralization that cross bearing point clusters
Wang et al. Graph-based multiprototype competitive learning and its applications
CN108627798B (en) WLAN indoor positioning algorithm based on linear discriminant analysis and gradient lifting tree
CN109063757A (en) It is diagonally indicated based on block and the multifarious multiple view Subspace clustering method of view
CN110097088A (en) A kind of dynamic multi-objective evolvement method based on transfer learning Yu particular point strategy
CN108520310B (en) Wind speed forecasting method of G-L mixed noise characteristic v-support vector regression machine
CN109121083B (en) Indoor positioning method based on fingerprint similarity of AP (Access Point) sequence
CN110188825B (en) Image clustering method, system, device and medium based on discrete multi-view clustering
CN111726765A (en) WIFI indoor positioning method and system for large-scale complex scene
CN110401977B (en) Multi-floor indoor positioning method based on Softmax regression multi-classification recognizer
CN112966773A (en) Unmanned aerial vehicle flight condition mode identification method and system
CN110796159A (en) Power data classification method and system based on k-means algorithm
CN108495263A (en) A kind of indoor orientation method based on broad sense multi-kernel function
Beltran et al. Predicting protein-protein interactions based on biological information using extreme gradient boosting
Basiouny et al. Enhancing Wi-Fi fingerprinting for indoor positioning system using single multiplicative neuron and PCA algorithm
CN111223528A (en) Multi-group data clustering method and device
CN109583727A (en) A kind of method and system determining cable status using the mobile inspection data of cable run
Xu et al. A novel Minkowski-distance-based consensus clustering algorithm
CN109800854A (en) A kind of Hydrophobicity of Composite Insulator grade determination method based on probabilistic neural network
CN114679683A (en) Indoor intelligent positioning method based on derivative fingerprint migration

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