Wireless indoor positioning method for self-supervised learning optimization

Abstract

The invention relates to the technical field of indoor positioning, in particular to a self-supervised learning optimized wireless indoor positioning method, which comprises the following steps of: S1, acquiring RSS (Received Signal Strength) data of fingerprint points in a positioning area in an off-line manner, and establishing a fingerprint database; s2, acquiring positioning point RSS data on line; s3, comparing the selected features with the fingerprint database to obtain a primary positioning result; s4, performing feature subtraction one combination to obtain multiple groups of dimension-reduced positioning point RSS data, and comparing the data with the fingerprint database to obtain multiple secondary positioning results; s5, judging whether the two positioning results are consistent or not; if so, outputting a primary positioning result; and otherwise, taking the multiple groups of dimensionality-reduced positioning point RSS data as newly selected features, and repeatedly executing the steps S3-S5 until the two positioning results are judged to be consistent. According to the method, the accuracy of data positioning is discriminated and verified by using two positioning results, and the validity of positioning data is improved under the condition that the workload and node equipment are not increased through continuous dimension reduction and feedback elimination of invalid data.

Description

Technical Field

[0001] The present invention relates to the field of indoor positioning technology, and more particularly to a wireless indoor positioning method based on self-supervised learning optimization. Background Art

[0002] With the continuous development of computer technology and network technology, the demand for location information services has been increasingly studied and applied. Especially in large shopping malls, exhibition halls and other large areas, a complete indoor positioning system is needed for navigation and positioning services. The current GPS technology can meet the needs of outdoor positioning, but it cannot be well applied to indoor environments because the GPS signal is blocked by indoor walls.

[0003] The current indoor positioning technologies are mainly based on WLAN, Bluetooth, RFID and ZigBee, etc., and the positioning is completed by measuring the distance and matching the fingerprint information. The equipment based on WLAN technology and fingerprint information matching method has low cost and high positioning accuracy and is widely used. The traditional method uses the equipment to set multiple fingerprint points in the specified area to collect feature data and build an offline fingerprint database. The online collected positioning point RSS data will be compared with the offline fingerprint database for positioning, but its positioning is usually one-time positioning, and its results lack verification. Some technologies will filter out unstable signals from the online collected positioning point RSS data through filtering, and use the obtained stable wireless RSS data for positioning. Its positioning also lacks verification and ignores factors such as indoor environment layout and transformation, personnel movement, human body occlusion, equipment collection direction and equipment itself instability. Some stable data that are not easy to filter out will be generated, and these stable data have large errors. For example, human body occlusion may make the difference in the strength of the collected wireless signal more than 10dBm, and the wireless signal is stable and difficult to filter out. Such data is called invalid data in the present invention, and invalid data will cause the positioning result of the wireless feature data to be very different from the true value to a certain extent.

[0004] Therefore, how to improve the effectiveness of positioning data based on wireless technology and fingerprint information matching methods without increasing the workload and node equipment, greatly reduce the problem of high workload caused by body occlusion during positioning, and thus improve the reliability, engineering practicality and accuracy of positioning results. Summary of the invention

[0005] In view of this, the present invention provides a wireless indoor positioning method optimized by self-supervised learning, which solves the problem in the prior art that only attention is paid to solving the problem that signal time-varying property affects positioning accuracy, while ignoring the problem that stable wireless signals also need to extract effective signals to improve positioning accuracy.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] S1, in the offline stage, use the device to collect the RSS data of the fingerprint points in the positioning area and establish a fingerprint database;

[0008] S2, in the online stage, use the device to obtain the RSS data of the positioning point;

[0009] S3, select features from the positioning point RSS data and compare them with the fingerprint database to obtain a positioning result;

[0010] S4, subtract one from the selected features to combine multiple sets of reduced-dimensional positioning point RSS data, and compare them with the fingerprint database respectively to obtain multiple secondary positioning results;

[0011] S5, comparing the multiple secondary positioning results with the primary positioning result to determine whether the results are consistent;

[0012] If the results are consistent, the first positioning result is output; otherwise, the multiple groups of reduced-dimensional positioning point RSS data in S4 are used as newly selected features, and S3 to S5 are repeatedly executed until the second positioning result is consistent with the first positioning result.

[0013] Furthermore, S1 includes:

[0014] S11, selecting M suitable fingerprint points indoors according to the layout of the indoor environment;

[0015] S12, using the device to collect the wireless signal strength received on N AP nodes around each fingerprint point;

[0016] S13, constructing a fingerprint database of the offline phase of the positioning area by collecting all the fingerprint points and the wireless signal strengths received on all AP nodes;

[0017] The fingerprint database is represented by D, and its expression is:

[0018]

[0019] Where M is the total number of suitable fingerprint points, N is the total number of wireless signal AP nodes, {r i1 ,r i2 ,...,r iN} is the set of N RSS data collected at the i-th fingerprint point, r ij Represents the RSS data of the i-th fingerprint point on the j-th AP node.

[0020] Furthermore, there are N RSS data of the positioning points obtained in S2, which are represented by {s 1 ,s 2 ,...,s N}.

[0021] Furthermore, in S3, all the data obtained in S2 are selected as feature data, compared with the fingerprint database and a positioning result is obtained, which is represented by L 1 =(l 1x ,l 1y ).

[0022] Further, in S4, the N positioning point RSS data selected in S3 are subjected to feature reduction by one, and N groups of positioning point RSS data with one dimension reduced are combined, which are represented as {s 1 ,s 2 ,...,s N-1},…,{s 2 ,s 3 ,…,s N}; they are all compared with the fingerprint database and N secondary positioning results are obtained, denoted as P 1 ,P 2 ,…,P N .

[0023] Further, in S5, the steps of comparing the N secondary positioning results obtained in S4 with the primary positioning result and determining whether the two positioning results are consistent are:

[0024] S51, preset a distance threshold, the distance threshold is the distance between two most adjacent fingerprint points in the grid method. times;

[0025] S52, calculating the position difference between the N secondary positioning results and the primary positioning result respectively;

[0026] S53, selecting a maximum value from the N position differences and comparing it with a set distance threshold;

[0027] S54. If the following formula is satisfied, it means that the two positioning results of the data before and after the dimensionality reduction are consistent;

[0028] The formula is:

[0029] d≥max{d 1 ,d 2 ,...,d N}

[0030] Among them, d is the pre-set distance threshold, d i is the position difference between the two i-th positionings, and the value range of i is [1, N].

[0031] Furthermore, S5 further includes the following steps:

[0032] S55, if the above formula is not satisfied, it means that the two positioning results of the data before and after the dimension reduction are consistent, and it is necessary to return to S3, and select one group from the N groups of one-dimensional positioning point RSS data as feature data, and repeat S3 to S5;

[0033] If the two positioning results of N groups of one-dimensional positioning point RSS data are inconsistent after repeating S3 to S5, S3 continues to select one group from (N-1) groups of two-dimensional positioning point RSS data as feature data, and repeats S3 to S5;

[0034] After multiple dimensionality reduction and self-supervised learning optimization, S3 selects a set of feature data from the (N-k+1) groups of k-dimensional reduced positioning point RSS data. If the two positioning results are consistent after repeatedly executing S3 to S5, the first positioning result of this group of data is output.

[0035] Furthermore, S5 further includes the following steps:

[0036] S55, if the above formula is not satisfied, it means that the two positioning results of the data before and after the dimension reduction are consistent, and it is necessary to return to S3, and select one group from the N groups of one-dimensional positioning point RSS data as feature data, and repeat S3 to S5;

[0037] If the two positioning results of N groups of one-dimensional positioning point RSS data are inconsistent after repeating S3 to S5, S3 continues to select one group from (N-1) groups of two-dimensional positioning point RSS data as feature data, and repeats S3 to S5;

[0038] After multiple dimensionality reduction and self-supervised learning optimization, S3 selects a set of feature data from the (N-k+1) groups of k-dimensional reduced positioning point RSS data. If the two positioning results are consistent after repeatedly executing S3 to S5, the first positioning result of this group of data is output.

[0039] Furthermore, the devices used in S1 and S2 are smart phones or tablets, and the collected wireless signals are WiFi or Bluetooth signals.

[0040] Furthermore, the first positioning result and the second positioning result in S3 and S4 are obtained by using the wknn algorithm.

[0041] It can be seen from the above technical solutions that, compared with the prior art, the present invention has the following beneficial effects:

[0042] (1) The present invention obtains multiple secondary positioning results through dimensionality reduction, combination and secondary positioning, which are used to compare with the primary positioning results. The positioning effect of multiple wireless signal node signal strength characteristic data collected at a certain positioning point in the online stage can be verified. This method does not increase the workload and node equipment, and is very simple, practical and feasible.

[0043] (2) The present invention utilizes dimension reduction, combination, secondary positioning, feature discrimination, feedback and other components to form a self-supervised learning optimization process for indoor wireless positioning data, which can gradually eliminate invalid data from the selected data and ultimately output a primary positioning result that has passed the consistency verification of the positioning result.

[0044] (3) The present invention can solve the problem that the wireless signal strength collected at different AP nodes in the online stage of the collected positioning points is stable but has large differences due to indoor environmental changes and human obstruction. That is, invalid data is eliminated from the collected stable wireless signal feature data to avoid its influence on the indoor wireless positioning accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0046] Figure 1 A flowchart of a self-supervised learning optimized wireless indoor positioning method provided by the present invention;

[0047] Figure 2 The diagram is a secondary positioning result diagram including and excluding invalid data when 6 data are selected for primary and secondary dimensionality reduction in the embodiment. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0049] The embodiment of the present invention discloses a wireless indoor positioning method for self-supervised learning optimization, comprising the following steps:

[0050] S1, in the offline stage, use the device to collect the RSS data of the fingerprint points in the positioning area and establish a fingerprint database;

[0051] S2, in the online stage, use the device to obtain the RSS data of the positioning point;

[0052] S3, select features from the positioning point RSS data and compare them with the fingerprint database to obtain a positioning result;

[0053] S4, subtract one from the selected features to combine multiple sets of reduced-dimensional positioning point RSS data, and compare them with the fingerprint database respectively to obtain multiple secondary positioning results;

[0054] S5, comparing the multiple secondary positioning results with the primary positioning result to determine whether the results are consistent;

[0055] If the results are consistent, the first positioning result is output; otherwise, the multiple groups of reduced-dimensional positioning point RSS data in S4 are used as newly selected features, and S3 to S5 are repeatedly executed until the second positioning result is consistent with the first positioning result.

[0056] In this embodiment, S1 includes:

[0057] S11, selecting M suitable fingerprint points indoors according to the layout of the indoor environment;

[0058] S12, using the device to collect the wireless signal strength received on N AP nodes around each fingerprint point;

[0059] S13, constructing a fingerprint database of the offline phase of the positioning area by collecting all the fingerprint points and the wireless signal strengths received on all AP nodes;

[0060] The fingerprint database is represented by D, and its expression is:

[0061]

[0062] Where M is the total number of suitable fingerprint points, N is the total number of wireless signal AP nodes, {r i1 ,r i2 ,...,r iN} is the set of N RSS data collected at the i-th fingerprint point, r ij Represents the RSS data of the i-th fingerprint point on the j-th AP node.

[0063] In this embodiment, there are N positioning point RSS data obtained in S2, represented by {s 1 ,s 2 ,...,s N}.

[0064] In this embodiment, in S3, all the data obtained in S2 are selected as feature data, compared with the fingerprint database and a positioning result is obtained, which is represented by L 1 =(l 1x ,l 1y ).

[0065] In this embodiment, in S4, the N positioning point RSS data selected in S3 are subjected to feature reduction by one, and N groups of positioning point RSS data with one dimension reduced are combined, which are respectively expressed as

[0066] {s 1 ,s 2 ,...,s N-1},…,{s 2 ,s 3 ,…,s N}; they are all compared with the fingerprint database and N secondary positioning results are obtained, denoted as P 1 ,P 2 ,…,P N .

[0067] In this embodiment, in S5, the steps of comparing the N secondary positioning results obtained in S4 with the primary positioning result and determining whether the two positioning results are consistent are:

[0068] S51, preset a distance threshold, the distance threshold is the distance between two most adjacent fingerprint points in the grid method. times;

[0069] S52, calculating the position difference between the N secondary positioning results and the primary positioning result respectively;

[0070] S53, selecting a maximum value from the N position differences and comparing it with a set distance threshold;

[0071] S54. If the following formula is satisfied, it means that the two positioning results of the data before and after the dimensionality reduction are consistent;

[0072] The formula is:

[0073] d≥max{d 1 ,d 2 ,…,d N}

[0074] Among them, d is the pre-set distance threshold, d i is the position difference between the two i-th positionings, and the value range of i is [1, N].

[0075] In this embodiment, S5 further includes the following steps:

[0076] S55, if the above formula is not satisfied, it means that the two positioning results of the data before and after the dimension reduction are consistent, and it is necessary to return to S3, and select one group from the N groups of one-dimensional positioning point RSS data as feature data, and repeat S3 to S5;

[0077] If the two positioning results of N groups of one-dimensional positioning point RSS data are inconsistent after repeating S3 to S5, S3 continues to select one group from (N-1) groups of two-dimensional positioning point RSS data as feature data, and repeats S3 to S5;

[0078] After multiple dimensionality reduction and self-supervised learning optimization, S3 selects a set of feature data from the (N-k+1) groups of k-dimensional reduced positioning point RSS data. If the two positioning results are consistent after repeatedly executing S3 to S5, the first positioning result of this group of data is output.

[0079] In this embodiment, S5 further includes the following steps:

[0080] S55, if the above formula is not satisfied, it means that the two positioning results of the data before and after the dimension reduction are consistent, and it is necessary to return to S3, and select one group from the N groups of one-dimensional positioning point RSS data as feature data, and repeat S3 to S5;

[0081] If the two positioning results of N groups of one-dimensional positioning point RSS data are inconsistent after repeating S3 to S5, S3 continues to select one group from (N-1) groups of two-dimensional positioning point RSS data as feature data, and repeats S3 to S5;

[0082] After multiple dimensionality reduction and self-supervised learning optimization, S3 selects a set of feature data from the (N-k+1) groups of k-dimensional reduced positioning point RSS data. If the two positioning results are consistent after repeatedly executing S3 to S5, the first positioning result of this group of data is output.

[0083] In this embodiment, the devices used in S1 and S2 are smart phones or tablets, and the collected wireless signals are WiFi or Bluetooth signals.

[0084] Furthermore, in this embodiment, the first positioning result and the second positioning result in S3 and S4 are both obtained by using the wknn algorithm.

[0085] Ideally, the RSS data of the initial positioning points collected in the online stage are directly consistent with the judgment results after one dimensionality reduction combination and two positioning result judgments, and a positioning result is output; however, in practice, the RSS data of the initial positioning points collected in the online stage may have one or more invalid data in the collected RSS data of the positioning points due to changes in the indoor environment, signal collection direction, and personnel obstruction. The positioning result obtained by only one positioning lacks reliable verification and cannot avoid the influence of the collected stable data with large errors on the indoor wireless positioning accuracy. Therefore, the present invention proposes the above-mentioned specific solution to this problem to realize the self-supervision process of feature data during indoor positioning.

[0086] The secondary positioning result of the present invention can verify the accuracy of the primary positioning result, which is better than the traditional method of positioning only by filtering because its positioning accuracy can be verified.

[0087] In order to facilitate the understanding of the wireless positioning method of the present invention, it is assumed that 6 feature data are selected in S3 to compare with the fingerprint database to obtain a positioning result; 5 features are randomly extracted from them, that is, a combination of 6 out of 5, which will be combined into 6 groups of one-dimensional positioning point RSS data containing 5 feature data for secondary positioning, and 6 secondary positioning results are obtained;

[0088] If all six feature data are valid data, the error between the six secondary positioning results and the primary positioning result will not be too large. In other words, the six secondary positioning positions obtained with five features are close and close to the primary positioning position obtained with six features.

[0089] If there is an invalid data that is stable but has a large error among the 6 feature data, then the 6 feature data will be combined into 6 groups of positioning point RSS data containing 5 feature data after the feature is subtracted by one, that is, the one-dimensional positioning point feature data will be reduced. Comparing them with the fingerprint database will obtain 6 secondary positioning results.

[0090] Since 5 of the 6 sets of one-dimensional positioning point feature data contain stable but invalid data with large errors, the positioning results of the 5 sets of one-dimensional positioning point feature data have large errors, which will exceed the set position threshold and are inconsistent with the first positioning results before dimensionality reduction;

[0091] Similarly, by continuing to reduce the dimension in the above manner, 5 sets of positioning point RSS data containing 4 feature data will be obtained, that is, the reduced 2D positioning point feature data, which are compared with the fingerprint database respectively to obtain 5 secondary positioning results;

[0092] Since one of the six groups of one-dimensional positioning point feature data has eliminated invalid feature data, when the dimension is further reduced, combined and the two positioning results are judged, the two positioning results will be consistent.

[0093] Figure 2 In the figure, Figure 1 is the secondary positioning result of selecting 5 feature data when the RSS data of the positioning point contains feature data with large errors, and Figure 2 is the secondary positioning result of selecting 4 feature data when the RSS data of the positioning point does not contain feature data with large errors;

[0094] From the two figures, it is not difficult to see that the secondary positioning results that do not contain feature data with large errors are close in position, otherwise the positions are far apart.

[0095] Through multiple experiments similar to the above, it is found that when the RSS data of the positioning point contains feature data with large errors, the average positioning accuracy of the positioning result can be improved by about 20% after using the wireless positioning method of the present invention.

[0096] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0097] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A wireless indoor positioning method based on self-supervised learning optimization, characterized in that: The main steps include: S1, in the offline stage, use the device to collect the RSS data of the fingerprint points in the positioning area and establish a fingerprint database; S2, in the online stage, use the device to obtain the RSS data of the positioning point; S3, select features from the positioning point RSS data and compare them with the fingerprint database to obtain a positioning result; S4, subtract one from the selected features to combine multiple sets of reduced-dimensional positioning point RSS data, and compare them with the fingerprint database respectively to obtain multiple secondary positioning results; S5, comparing the multiple secondary positioning results with the primary positioning result to determine whether the results are consistent; If the results are consistent, the first positioning result is output; otherwise, the multiple groups of reduced-dimensional positioning point RSS data in S4 are used as newly selected features, and S3 to S5 are repeatedly executed until the second positioning result is consistent with the first positioning result.

2. The wireless indoor positioning method based on self-supervised learning optimization according to claim 1, characterized in that: S1 includes: S11, selecting M suitable fingerprint points indoors according to the layout of the indoor environment; S12, using the device to collect the wireless signal strength received on N AP nodes around each fingerprint point; S13, constructing a fingerprint database of the offline phase of the positioning area by collecting all the fingerprint points and the wireless signal strengths received on all AP nodes; The fingerprint database is represented by D, and its expression is: Where M is the number of suitable fingerprint points, N is the number of wireless signal AP nodes, {r i1 ,r i2 ,...,r iN } is the set of N RSS data collected at the i-th fingerprint point, r ij Represents the RSS data of the i-th fingerprint point on the j-th AP node.

3. The wireless indoor positioning method based on self-supervised learning optimization according to claim 2, characterized in that: There are N positioning point RSS data obtained in S2, represented as {s1,s2,…,s N }.

4. The wireless indoor positioning method based on self-supervised learning optimization according to claim 3, characterized in that: In S3, all the data obtained in S2 are selected as feature data, compared with the fingerprint database and a positioning result is obtained, which is expressed as L1=(l 1x ,l 1y ).

5. The wireless indoor positioning method based on self-supervised learning optimization according to claim 4, characterized in that: In S4, the N positioning point RSS data selected in S3 are subjected to feature reduction by one, and N groups of positioning point RSS data with one dimension reduced are combined, which are represented as {s1, s2, ..., s N-1 },...,{s2,s3,...,s N }; they are all compared with the fingerprint database and obtain N secondary positioning results, denoted as P1, P2, ..., P N .

6. The wireless indoor positioning method based on self-supervised learning optimization according to claim 5, characterized in that: In S5, the steps of comparing the N secondary positioning results obtained in S4 with the primary positioning result and determining whether the two positioning results are consistent are: S51, preset a distance threshold, the distance threshold is the distance between two most adjacent fingerprint points in the grid method. times; S52, calculating the position difference between the N secondary positioning results and the primary positioning result respectively; S53, selecting a maximum value from the N position differences and comparing it with a set distance threshold; S54. If the following formula is satisfied, it means that the two positioning results of the data before and after the dimensionality reduction are consistent; The formula is: d≥max{d1,d2,...,d N } Among them, d is the pre-set distance threshold, d i is the position difference between the two i-th positionings, and the value range of i is [1, N].

7. The wireless indoor positioning method based on self-supervised learning optimization according to claim 6, characterized in that: S5 also includes the following steps: S55, if the above formula is not satisfied, it means that the two positioning results of the data before and after the dimension reduction are consistent, and it is necessary to return to S3, and select one group from the N groups of one-dimensional positioning point RSS data as feature data, and repeat S3 to S5; If the two positioning results of N groups of one-dimensional positioning point RSS data are inconsistent after repeating S3 to S5, S3 continues to select one group from (N-1) groups of two-dimensional positioning point RSS data as feature data, and repeats S3 to S5; After multiple dimensionality reduction and self-supervised learning optimization, S3 selects one group of feature data from the (N-k+1) groups of k-dimensional reduced positioning point RSS data. After repeatedly executing S3 to S5, if the two positioning results are consistent, the first positioning result of this group of data is output.

8. The wireless indoor positioning method based on self-supervised learning optimization according to claim 8, characterized in that: Also includes: S6. If all the data obtained by S2 cannot achieve the consistency of the two positioning results after multiple dimensionality reduction and self-supervised learning optimization, return to S2 to re-acquire a new set of positioning point RSS data and continue to execute S3 to S5.

9. The wireless indoor positioning method based on self-supervised learning optimization according to claim 1, characterized in that: The devices used in S1 and S2 are smartphones or tablets, and the wireless signals collected are WiFi or Bluetooth signals.

10. The wireless indoor positioning method based on self-supervised learning optimization according to claim 1, characterized in that: The first positioning result and the second positioning result in S3 and S4 are obtained by using the wknn algorithm.

Images