Indoor Wi-Fi fingerprint positioning AP position selection method based on simulated RSS fingerprint database

Abstract

The invention discloses an indoor Wi-Fi fingerprint positioning AP point position selection method based on a simulated RSS fingerprint database. The method comprises the following steps: establishing a corresponding logarithmic distance path loss model according to an indoor environment; an AP point is placed at the AP position reference point in a simulated mode; rSS simulation sampling is carried out on the sampling points respectively, and a simulation RSS finger library is established; taking the AP position reference point with the maximum information entropy gain as a seed point; calculating the information entropy gain of the seed point and each other position reference point; selecting a position reference point with the maximum difference between the information entropy gain and the seed point information entropy gain as an extended seed point; selecting an extended seed point with the minimum MIC and an extended seed point with the minimum MIC, and adding the selected extended seed point into the seed point; and repeating until the position number in the seed point meets the requirement. According to the method, the problems of low precision of Wi-Fi fingerprint positioning in indoor positioning and difficulty in AP position selection are obviously improved, and the method has a wide research prospect.

Description

Technical Field

[0001] The present invention belongs to the technical field of wireless communication, and particularly relates to a method for selecting the position of an indoor Wi-Fi fingerprint positioning AP point based on an analog RSS fingerprint database. Background Art

[0002] With the wide use of smart phones, tablets, and smart wearable devices, users' demand for location services has extended from traditional outdoor navigation to more complex indoor environments. For example, many activities in people's daily lives, such as shopping, working, medical treatment, learning, and traveling, occur in indoor scenarios. Users hope to quickly find their destinations, track important items, and even obtain accurate location-related information and services. Although the traditional Global Positioning System (GPS) performs excellently in outdoor positioning and can provide relatively high positioning accuracy, it is severely restricted by signal attenuation problems in indoor environments and cannot meet users' requirements for precise positioning. This limitation is mainly due to the blocking of GPS signals by building walls, floors, glass, and other structures, making the signals weak or even completely disappear indoors, resulting in unstable or completely ineffective positioning. Therefore, how to accurately perform indoor positioning has become a major research hotspot in the industry today.

[0003] Current indoor location services mainly rely on wireless communication technologies. With the rapid development of wireless communication technologies, wireless technologies such as Wi-Fi, RFID, ZigBee, and ultra-wideband have become increasingly mature and common, and various technologies for solving indoor positioning using these wireless communication technologies have emerged. Among them, because Wi-Fi networks have been widely deployed in indoor scenarios (such as shopping malls, office buildings, airports, hospitals, etc.), without the need for additional hardware investment or reconstruction of infrastructure, and the coverage range of a single Wi-Fi access point (AP) can usually reach dozens of meters to hundreds of meters, Wi-Fi technology is widely used for positioning in large indoor scenarios. Among them, Wi-Fi fingerprint positioning has become a commonly used indoor positioning method. Wi-Fi fingerprint positioning is mainly achieved through two stages: the offline stage and the online stage. In the offline stage, the indoor area is divided into multiple grid points, the signal strength (RSSI) of the AP points within the range of each grid point and the corresponding location information are collected, and a fingerprint database is established. In the online stage, the user device collects the AP point signals in real time, matches the obtained RSSI values with the fingerprint database, and determines the user's current position through algorithms (such as k-NN or machine learning models or neural network models) and outputs the positioning result. From the above Wi-Fi fingerprint positioning method, it can be seen that when the AP points are placed at different positions, it will directly affect the fingerprint database, and thus affect the positioning accuracy in the online stage. Selecting a reasonable position to place the AP points and establishing a high-resolution fingerprint database have great research significance for Wi-Fi fingerprint positioning. Summary of the Invention

[0004] Objective of the Invention: The objective of the present invention is to provide a method for selecting the position of an AP point in indoor Wi-Fi fingerprint positioning based on an analog RSS fingerprint library, thereby improving the problems of low accuracy and difficult selection of AP point positions in indoor positioning of Wi-Fi fingerprint positioning, increasing the resolution of the fingerprint library established in the offline phase, and the accuracy when matching with the fingerprint database in the online phase.

[0005] Technical Solution: A method for selecting the position of an AP point in indoor Wi-Fi fingerprint positioning based on an analog RSS fingerprint library according to the present invention includes the following steps:

[0006] S1: Establish a logarithmic distance path loss model that matches the different indoor environments.

[0007] S2: Divide the area where APs can be placed into n AP position reference points, and divide the positioning area into m sampling points.

[0008] S3: Use the corresponding logarithmic distance path loss model to simulate the placement of APs at n AP position reference points, and perform RSS analog sampling at m sampling points respectively to establish an n*m analog RSS fingerprint library.

[0009] S4: Calculate the information entropy gain of each AP position reference point in the analog RSS fingerprint library according to the information entropy gain algorithm, and use the AP position reference point with the largest information entropy gain as the seed point of the region growing algorithm.

[0010] S5: Retrieve the other position reference points except the seed, calculate the information entropy gain between the seed point and each other position reference point, select the position reference point with the largest difference between the information entropy gain and the information entropy gain of the seed point, and use these AP position reference points as extended seed points.

[0011] S6: Calculate the maximum information coefficient (MIC) between each extended seed point and the seed point, select the extended seed point with the smallest MIC, and add it to the seed points.

[0012] S7: Repeat the above two steps until the number of AP point positions in the seed points meets the requirements.

[0013] Specifically, the logarithmic distance path loss model in S1 is:

[0014]

[0015] where n is the environmental factor, which is related to the structure and materials of the building. It represents the proportional exponent between the path length and the path loss, and its value range usually takes 2 to 4; PL(d) is the intensity of the received signal at a distance d from the transmitted signal, d 0Denote the received signal strength at a distance of 1 meter (usually taken as 1 meter), in dB, and ζ is the signal attenuation factor, which is independent of the propagation distance.

[0016] Specifically, the simulated RSS fingerprint library FP(s) in S3 is as follows:

[0017]

[0018] Among them, denotes the RSS value of the j-th sampling point when the AP is simulated and placed at the i-th AP location reference point.

[0019] Specifically, the information entropy gain algorithm in S4 includes:

[0020]

[0021]

[0022]

[0023] IG(AP i ) = H(C) - H(C|AP i )

[0024] C represents the possible location of the mobile device, H(C) is the entropy of C, representing the uncertainty of the location, and H(C|AP i ) is the conditional entropy, that is, the uncertainty of the location under the condition that the value of AP i is known. IG(AP i ) represents the information entropy gain of AP i .

[0025] Specifically, the maximum information coefficient in S6 includes:

[0026] Given two integer values r and c, divide the scatter plot of two random variables into an r×c grid and find the maximum mutual information value:

[0027] I * (D, r, c) = max{I(D|G)}

[0028] Among them, D = {(x i , y i ), i = 1, 2,..., n} is the value set of two feature pairs (X, Y); G is the scatter plot corresponding to D divided into r*c grids;

[0028] Normalize the maximum mutual information value:

[0029]

[0030] The maximum mutual information under different partitioning scales is selected as the maximum information coefficient (MIC):

[0031] MIC(D)=max rc∈B(n) {M(D) r,c}

[0032] Where M(D)=(m r,c ,c) is to form a feature matrix by combining the maximum normalized mutual information obtained under different scale divisions; B(n) is the upper limit of r*c, and the default value is B(n) = n 0.6 .

[0033] Beneficial effects: the present invention has at least the following advantages

[0034] In the prior art, if the signal characteristics of adjacent collection points are too similar, the system is prone to positioning errors during matching. Based on this, the present invention aims at the fact that the quality of the fingerprint library will directly affect the accuracy and stability of positioning in Wi-Fi fingerprint positioning, and increases the fingerprint discrimination and differentiation in the Wi-Fi fingerprint library, thereby helping to more accurately match the current position in the future and avoid errors. The present invention uses an information entropy gain algorithm to screen AP point positions with strong correlation between fingerprints and target locations, which is helpful for later model learning and prediction. The present invention achieves high discrimination between different fingerprints by screening AP point positions with large information entropy gain differences, helps the model capture key patterns of data, reduces dependence on irrelevant information, and thus improves prediction performance. The present invention uses the maximum information coefficient algorithm to avoid high redundancy between fingerprints as much as possible, and the Wi-Fi signal strength distribution of each fingerprint collection point should be as unique as possible, that is, the signal difference between each collection point should be significant enough to effectively distinguish different locations.

[0035] A computer memory for storing a computer program, characterized in that when the computer program is executed by a processor, the method for selecting an indoor Wi-Fi fingerprint positioning AP point based on a simulated RSS fingerprint library according to claims 1 to 5 is implemented.

[0036] A computer device, comprising a processor, a communication interface, a memory and a communication bus, and a computer program stored in the memory, characterized in that when the processor executes the computer program, an indoor Wi-Fi fingerprint positioning AP point location selection method based on a simulated RSS fingerprint library according to claims 1-5 is implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 The present invention provides a flowchart of a method for selecting an indoor Wi-Fi fingerprint positioning AP point based on a simulated RSS fingerprint library.

[0038] Figure 2It is the division of the installable AP position area and the positioning area provided by the embodiments of the present invention, and the establishment of the simulated RSS fingerprint database thereafter.

[0039] Figure 3 It is the error cumulative distribution function graph provided by the embodiments of the present invention for comparison with the random placement of AP points. Specific embodiments Next, the technical solutions in the embodiments of the present invention will be completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative work shall fall within the protection scope of the present invention.

[0040] Next, the technical solutions of the present invention will be further described in conjunction with the accompanying drawings.

[0041] As Figure 1 shown, a method for selecting the position of an indoor Wi-Fi fingerprint positioning AP point based on a simulated RSS fingerprint database includes the following steps:

[0042] S1: According to different indoor environments, establish a corresponding log-distance path loss model;

[0043] S2: As Figure 2 shown, divide the installable AP position area into n AP position reference points, and divide the positioning area into m sampling points. The finer the division of the AP position reference points and the sampling points, the more complex the calculation, the more time-consuming the online positioning stage, but the more accurate the later positioning.

[0044] S3: As Figure 2 shown, use the corresponding log-distance path loss model to simulate the placement of AP points at n AP position reference points, and perform RSS simulation sampling at m sampling points respectively to establish an n*m simulated RSS fingerprint database;

[0045] S4: According to the information entropy gain algorithm, calculate the information entropy gain of each AP position reference point in the simulated RSS fingerprint database, and use the AP position reference point with the largest information entropy gain as the seed point of the region growing algorithm;

[0046] S5: Retrieve the other position reference points except the seed, calculate the information entropy gain between the seed point and each other position reference point, select the position reference point with the largest difference between the information entropy gain and the information entropy gain of the seed point, and use these AP position reference points as the extended seed points;

[0047] S6: Calculate the maximum information coefficient (MIC) between each extended seed point and the seed point, select the extended seed point with the smallest MIC, and add it to the seed points;

[0048] S7: Repeat the above two steps until the number of AP points within the seed points meets the requirements.

[0049] Specifically, the logarithmic distance path loss model in S1 is:

[0050]

[0051] where n is the environmental factor, related to the structure and materials of the building, which represents the proportional exponent between the path length and the path loss, and its value range usually takes 2 - 4; PL(d) is the intensity of the received signal at a distance d from the transmitted signal, and d 0 represents the received signal intensity when the distance is (usually taken as 1 meter), in dB; ζ is the signal attenuation factor, independent of the propagation distance.

[0052] Specifically, the simulated RSS fingerprint library FP(s) in S3 is:

[0053]

[0054] where represents the RSS value of the jth sampling point when the AP is simulated and placed at the reference point of the ith AP position.

[0055] Specifically, the information entropy gain algorithm in S4 includes:

[0056]

[0057]

[0058]

[0059] IG(AP i ) = H(C) - H(C|AP i )

[0060] C represents the possible positions of the mobile device, H(C) is the entropy of C, representing the uncertainty of the position, and H(C|AP i ) is the conditional entropy, that is, the uncertainty of the position under the condition that the value of AP i is known, and IG(AP i ) represents the information entropy gain of AP i .

[0061] Specifically, the maximum information coefficient in S6 includes:

[0062] Given two integer values r and c, divide the scatter plot of two random variables into an r - row and c - column grid, and find the maximum mutual information value:

[0063] I * (D, r, c) = max{I(D|G)}

[0064] where D = {(x i , y i ), i = 1, 2, …, n} is the value set of two feature pairs (X, Y); G is to divide the scatter plot corresponding to D into r * c grids;

[0065] Normalize the maximum mutual information value:

[0066]

[0067] Select the maximum value of mutual information under different partitioning scales as the maximum information coefficient (MIC):

[0068] MIC(D) = max rc∈B(n) {M(D) r,c}

[0069] where M(D) = (m r,c , c) is to form a feature matrix with the maximum normalized mutual information obtained under different scale partitions; B(n) is the upper limit value of r * c, and by default B(n) = n 0.6 .

[0070] As Figure 3 shown, Base Station 1 is the AP location provided by an indoor Wi-Fi fingerprint positioning AP point location selection method based on an analog RSS fingerprint library according to the present invention, and Base Station 2 is randomly placed AP points. The error cumulative distribution function graph of the comparison between the two after simulation positioning is obtained; it can be seen that the average error of Base Station 1 is much smaller than that of Base Station 2. In summary, an AP point location selection method based on an analog RSS fingerprint library proposed by the present invention can effectively improve the accuracy of Wi-Fi fingerprint positioning.

[0070] A computer memory for storing a computer program, characterized in that when the computer program is executed by a processor, it implements an indoor Wi-Fi fingerprint positioning AP point location selection method according to claims 1 - 5.

[0071] A computer device, including a processor, a communication interface, a memory, a communication bus, and a computer program stored on the memory, characterized in that when the processor executes the computer program, it implements an indoor Wi-Fi fingerprint positioning AP point location selection method according to claims 1 - 5.

Claims

1. A method for selecting indoor Wi-Fi fingerprint positioning AP points based on simulated RSS fingerprint library, characterized in that: include: S1: According to different indoor environments, a matching logarithmic distance path loss model is established; S2: Divide the AP location area into n AP location reference points and divide the positioning area into m sampling points; S3: Using the corresponding logarithmic distance path loss model, simulate placing AP points at n AP position reference points, and perform RSS simulation sampling at m sampling points respectively to establish an n*m simulated RSS fingerprint library; S4: According to the information entropy gain algorithm, the information entropy gain of each AP position reference point in the simulated RSS fingerprint library is calculated, and the AP position reference point with the largest information entropy gain is used as the seed point of the region growing algorithm; S5: Retrieve other location reference points except the seed, calculate the information entropy gain between the seed point and each other location reference point, select the location reference point with the largest difference between the information entropy gain and the information entropy gain of the seed point, and use these AP location reference points as extended seed points; S6: Calculate the maximum information coefficient (MIC) between each extended seed point and the seed point, select the extended seed point with the smallest MIC, and add it to the seed point; S7: Repeat the above two steps until the number of AP point positions within the seed point meets the requirement.

2. According to claim 1, a method for selecting an indoor Wi-Fi fingerprint positioning AP point based on a simulated RSS fingerprint library, characterized in that: The logarithmic distance path loss model in S1 is: Among them, n refers to the environmental factor, which is related to the structure and material of the building. It represents the proportional exponent between the path length and the path loss, and its value range is usually 2 to 4; PL(d) is the strength of the received signal at a distance d from the transmitted signal, d0 represents the received signal strength when the distance is (usually 1 meter), in dB; ζ is the signal attenuation factor, which is independent of the propagation distance.

3. According to claim 1, a method for selecting indoor Wi-Fi fingerprint positioning AP point location based on simulated RSS fingerprint library, characterized in that: The simulated RSS fingerprint library FP(s) in S3 is: in, Indicates the RSS value of the j-th sampling point simulating the placement of the AP at the i-th AP location reference point.

4. According to claim 1, a method for selecting indoor Wi-Fi fingerprint positioning AP point location based on simulated RSS fingerprint library, characterized in that: The information entropy gain algorithm in S4 includes: IG(AP i )=H(C)-H(C∣AP i , C represents the possible location of the mobile device, H(C) is the entropy of C, representing the uncertainty of the location, and H(C|AP i ) is the conditional entropy, that is, in AP i The uncertainty of the position under the condition of known value, IG(AP i ) indicates AP i The information entropy gain.

5. According to claim 1, a method for selecting indoor Wi-Fi fingerprint positioning AP point location based on simulated RSS fingerprint library, characterized in that , the maximum information coefficient in S6 includes: S51: Given two integer values ​​r and c, the scatter plot of the two random variables is divided into r rows and c columns, and the maximum mutual information value is calculated: I * (D,r,c)=max{I(D|G)} Where D = {(x i ,y i ),i=1,2,…,n} is the value set of two feature pairs (X,Y); G is the scatter plot corresponding to D divided into r*c grids; S52: Normalize the maximum mutual information value: S53: Select the maximum value of mutual information under different partitioning scales as the maximum information coefficient (MIC): MIC(D)=max rc∈B(n) {M(D) r,c } Where M(D)=(m r,c ,c) is to form a feature matrix by combining the maximum normalized mutual information obtained under different scale divisions; B(n) is the upper limit of r*c, and the default value is B(n) = n 0.6 .

6. A computer memory for storing a computer program, characterized in that: When the computer program is executed by a processor, the method for selecting an indoor Wi-Fi fingerprint positioning AP point based on a simulated RSS fingerprint library described in claims 1-5 is implemented.

7. A computer device comprising a processor, a communication interface, a memory and a communication bus and a computer program stored in the memory, characterized in that: When the processor executes the computer program, the method for selecting an indoor Wi-Fi fingerprint positioning AP point based on a simulated RSS fingerprint library described in claims 1-5 is implemented.

Images