CN106412841A - Indoor location method based on DBSCAN algorithm - Google Patents

Abstract

The present invention is an indoor positioning method based on the DBSCAN algorithm, which belongs to the technical field of indoor positioning; the method first installs a signal transmitter and a signal collector in a building to be positioned, sets the number of sample points, and obtains each sample point Then record and number the coordinates and signal strength sequence of each sample point, use the DBSCAN algorithm to establish an indoor positioning model, and finally obtain the signal strength sequence of the point to be located, and obtain the pending location according to the established indoor positioning model Positioning coordinates of the site; the present invention uses unsupervised learning algorithm to model, classifies and clusters according to the wireless signal strength value (RSSI), instead of artificially performing grid division, so that the grid division is more realistic and improves indoor positioning stability and precision.

Description

An Indoor Positioning Method Based on DBSCAN Algorithm

technical field

The invention belongs to the technical field of indoor positioning, and in particular relates to an indoor positioning method based on a DBSCAN algorithm.

Background technique

With the development of the Internet of Things, the demand for indoor positioning is also increasing day by day. In indoor environments such as shopping malls, schools, office buildings, hospitals, hotels, airports, warehouses, etc., indoor positioning technology is needed to efficiently manage resources and personnel. VR There is also a large demand in the field to locate the player's indoor location information, etc. Therefore, how to better meet the increasing demand for indoor positioning has become a hot issue in the current positioning technology.

In the existing indoor positioning technology, the fingerprint positioning method based on the radio signal strength (RSSI) has been widely used in various indoor positioning systems. Therefore, the measurement value of the wireless signal strength (RSSI) fluctuates greatly; in order to improve the stability and accuracy of the measurement, the method of matching the fingerprint database is commonly used to realize indoor positioning, and the common method of using the matching fingerprint database to realize indoor positioning is Naive Bayesian method, K-nearest neighbor method (KNN), Bayesian estimation, neural network method, etc., these are all methods of supervised learning, usually need to artificially divide the grid, these networks are generally divided by dividing the area equally, due to the actual wireless signal strength ( RSSI) is unevenly distributed, which will cause locations with similar wireless signal strength (RSSI) sequences to be divided into different grids, and multiple adjacent grid signal strength (RSSI) sequences are very similar, but they are finally randomly located in one of the grids. The grid situation occurs, resulting in inaccurate positioning results.

Contents of the invention

In order to solve the deficiencies in the prior art, the present invention proposes a kind of indoor positioning method based on DBSCAN algorithm, comprising the following steps:

Step 1. Install several signal transmitters and several signal collectors in the building to be located;

Step 2. Set the number of sample points, determine the coordinates of each sample point, and obtain the signal strength sequence of each sample point, that is, the signal strength from each sample point to each signal transmitter;

Step 3, record and number the coordinates and signal strength sequences of each sample point;

Step 4, according to the coordinates and signal strength sequences of each sample point after numbering, use the DBSCAN algorithm to establish an indoor positioning model, including the following steps:

Step 4.1, setting density selection parameters, that is, setting the length of the neighborhood and the number of sample points that each core object contains at least;

Step 4.2, according to the signal strength sequence of each sample point after numbering and the set density selection parameter, set the constraint condition of the core object, determine the core object of the sample point according to the constraint condition, and further determine the clustering cluster through the core object; The constraint condition of the core object is: whether the number of other sample points contained in the neighborhood of each sample point is greater than or equal to the set number of sample points contained in each core object at least;

Step 4.3, obtain the average signal strength of each cluster, and obtain the average position coordinates of each cluster according to the coordinates of each sample point after numbering, that is, the establishment of the indoor positioning model is completed;

Step 5. Obtain the signal strength sequence of the point to be located, and obtain the positioning coordinates of the point to be located according to the established indoor positioning model, specifically: obtain the Euclidean value of the average signal strength of the point to be located and each cluster distance, and select the average position coordinates of the cluster with the smallest Euclidean distance as the positioning coordinates of the point to be located.

The constraint condition of the core object described in step 4.2, the formula is as follows:

|N _θ (x _j )|≥Minpts (1)

Among them, N _θ represents the number of samples in the θ neighborhood of a certain sample point; θ represents the length of the neighborhood; x _j represents the jth sample point, and j represents a natural number; Minpts represents the number of sample points that each core object contains at least number;

in,

N _θ (x _j )＝{ _xi ∈ D|d( _xi ,x _j )≤θ} (2)

Among them, x _i represents the i-th sample point; i represents a natural number; D represents the coordinates of each sample point after numbering and the data set of the signal intensity sequence; d( _xi , x _j ) represents two x _i and x _j Euclidean distance between sample points;

in,

Among them, n represents the number of wireless signal strengths recorded at each sample point; k represents a natural number; r _ik represents the kth wireless signal strength value of the i-th sample point; r _jk represents the k-th value of the j-th sample point Wireless signal strength value.

Advantages of the present invention:

The present invention proposes an indoor positioning method based on the DBSCAN algorithm, uses a non-supervised learning algorithm for modeling, and classifies and clusters according to the wireless signal strength value (RSSI), instead of artificially performing grid division, so that the grid division is more realistic , improving the stability and accuracy of indoor positioning.

Description of drawings

Fig. 1 is the flow chart of the indoor positioning method based on DBSCAN algorithm of an embodiment of the present invention;

Fig. 2 is a schematic diagram of indoor positioning according to an embodiment of the present invention.

detailed description

An embodiment of the present invention will be further described below in conjunction with the accompanying drawings.

In the embodiment of the present invention, an indoor positioning method based on the DBSCAN algorithm, the method flow chart is shown in Figure 1, including the following steps:

Step 1. Install several signal transmitters and several signal collectors in the building to be located;

Step 2. Set the number of sample points, determine the coordinates (Z ₁ , Z ₂ ) of each sample point, and obtain the signal intensity sequence {r ₁ ,r ₂ ,···,r _n } of each sample point , that is, the signal strength from each sample point to each signal transmitter;

Step 3. Record and number the coordinates and signal strength sequence of each sample point as a sample data to obtain a data set D={x ₁ ,x ₂ ,···,x _m }; where m represents a natural number, that is, the sample data serial number;

Step 4, according to the coordinates and signal strength sequences of each sample point after numbering, use the DBSCAN algorithm to establish an indoor positioning model, including the following steps:

Step 4.1, set the density selection parameters, that is, set the length θ of the neighborhood and the number Minpts of the sample points that each core object contains at least;

Step 4.2, according to the numbered signal strength sequence of each sample point and the set density selection parameters, set the constraint conditions of the core objects, determine all the core objects of the sample points according to the constraints, and further determine the clustering through the core objects Cluster; the constraint condition of the core object is: whether the number of other sample points contained in the neighborhood of each sample point is greater than or equal to the number of sample points that each core object contains at least, the specific steps as follows:

Step 4.2.1, set the constraint condition of the core object according to the signal strength sequence of each sample point after numbering and the length of the set neighborhood, and determine all the core objects in the sample point according to the constraint condition, the constraint condition formula as follows:

|N _θ (x _j )|≥Minpts (1)

Among them, N _θ represents the number of samples in the θ neighborhood of a certain sample point; θ represents the length of the neighborhood; x _j represents the jth sample point, j represents a natural number, that is, the serial number of the sample point; Minpts represents each core object At least the number of sample points included;

In the embodiment of the present invention, for x _j ∈ D, its θ neighborhood includes samples whose distance from x _j in the sample set D is less than or equal to θ, and the formula is:

N _θ (x _j )＝{ _xi ∈ D|d( _xi ,x _j )≤θ} (2)

Among them, x _i represents the i-th sample point; i represents a natural number, that is, the serial number of the sample point; D represents the coordinates of each sample point after numbering and the data set of the signal intensity sequence;

In the embodiment of the present invention, d( _xi , x _j ) represents the Euclidean distance between two sample points x _i and x _j , that is, the signal strength sequence {r ₁ ,r ₂ ,...,r _n } consists of The distance between two points x _i and x _j in the n-dimensional Euclidean space of , the formula is as follows:

Among them, n represents the number of wireless signal strengths recorded at each sample point; k represents a natural number, that is, the serial number of the wireless signal strength value; r _ik represents the kth wireless signal strength value of the i-th sample point; r _jk represents the The kth wireless signal strength value of the j sample point;

Step 4.2.2, further determine all the clusters through the core object;

In the embodiment of the present invention, if x _j is located in the field of θ of x _i , and x _i is the core object, it is said that x _j is directly reached by the density of x _i ; for x _i and x _j , if there is a sample sequence {p ₁ , p ₂ ,···,p _n }, where p ₁ = x _i , p _n = x _j , and each sample in the sample sequence can be directly reached by the previous sample density, then x _j is said to be density-reachable by _xi ; A core object in the sample set D is arbitrarily selected as the "seed", and all sample points with reachable density are found from this point, that is, the corresponding cluster is determined; all core objects are accessed to obtain all clusters. clusters;

Step 4.3, obtain the average signal strength of each cluster, and calculate and obtain the average position coordinates of each cluster according to the coordinates of each sample point after numbering, that is, the establishment of the indoor positioning model is completed;

In the embodiment of the present invention, to obtain the average signal strength of each cluster, the following formula is used:

Among them, h represents the number of sample points in each cluster; r _i1 represents the first wireless signal strength of the i-th sample point in each cluster; r _i2 represents the i-th sample point in each cluster The second wireless signal strength of the sample point; r _in represents the nth wireless signal strength of the i-th sample point in each cluster;

In the embodiment of the present invention, the following formula is used to obtain the average position coordinates of each cluster:

Among them, Z _i1 represents the abscissa value of the i-th sample point in each cluster; Z _i2 represents the ordinate value of the i-th sample point in each cluster;

Step 5. Obtain the signal strength sequence of the point to be located by online measurement, etc., and obtain the positioning coordinates of the point to be located according to the established indoor positioning model, specifically: obtain the average value of the point to be located and each cluster The Euclidean distance of the signal strength, and select the average position coordinates of the cluster with the smallest Euclidean distance as the positioning coordinates of the point to be located;

In the embodiment of the present invention, as shown in Figure 2, the method is simulated on the computer, 14 sample points are set, and a point to be located is randomly set, and the wireless signal strength sequence received by each sample point is {r ₁ , r ₂ }, use the DBSCAN algorithm to divide the sample points into two clusters, C1 and C2, and obtain the center point of each cluster, that is, the average signal strength point; obtain the points to be located and each cluster The Euclidean distance of the center point, where the Euclidean distance to the cluster C1 is the smallest, the coordinates of the location are the average position coordinates of the cluster C1.

Claims (2)

1. a kind of indoor positioning method based on DBSCAN algorithm, it is characterized in that: comprise the following steps: Step 1. Install several signal transmitters and several signal collectors in the building to be located; Step 2. Set the number of sample points, determine the coordinates of each sample point, and obtain the signal strength sequence of each sample point, that is, the signal strength from each sample point to each signal transmitter; Step 3, record and number the coordinates and signal strength sequences of each sample point; Step 4, according to the coordinates and signal strength sequences of each sample point after numbering, use the DBSCAN algorithm to establish an indoor positioning model, including the following steps: Step 4.1, setting density selection parameters, that is, setting the length of the neighborhood and the number of sample points that each core object contains at least; Step 4.2, according to the signal strength sequence of each sample point after numbering and the set density selection parameter, set the constraint condition of the core object, determine the core object of the sample point according to the constraint condition, and further determine the clustering cluster through the core object; The constraint condition of the core object is: whether the number of other sample points contained in the neighborhood of each sample point is greater than or equal to the set number of sample points contained in each core object at least; Step 4.3, obtain the average signal strength of each cluster, and obtain the average position coordinates of each cluster according to the coordinates of each sample point after numbering, that is, the establishment of the indoor positioning model is completed; Step 5. Obtain the signal strength sequence of the point to be located, and obtain the positioning coordinates of the point to be located according to the established indoor positioning model, specifically: obtain the Euclidean value of the average signal strength of the point to be located and each cluster distance, and select the average position coordinates of the cluster with the smallest Euclidean distance as the positioning coordinates of the point to be located. 2. the indoor positioning method based on DBSCAN algorithm according to claim 1, is characterized in that: the constraint condition of the core object described in step 4.2, formula is as follows: |N _θ (x _j )|≥Minpts (1) Among them, N _θ represents the number of samples in the θ neighborhood of a certain sample point; θ represents the length of the neighborhood; x _j represents the jth sample point, and j represents a natural number; Minpts represents the number of sample points that each core object contains at least number; in, N _θ (x _j )＝{ _xi ∈ D|d( _xi ,x _j )≤θ} (2) Among them, x _i represents the i-th sample point; i represents a natural number; D represents the coordinates of each sample point after numbering and the data set of the signal intensity sequence; d( _xi , x _j ) represents two x _i and x _j Euclidean distance between sample points; in, $\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\sqrt{{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}_{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}^{\mathrm{<span\; class="notranslate">\; no</span>}}{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; r</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$ Among them, n represents the number of wireless signal strengths recorded at each sample point; k represents a natural number; r _ik represents the kth wireless signal strength value of the i-th sample point; r _jk represents the k-th value of the j-th sample point Wireless signal strength value.