Indoor unknown signal interference source fingerprint positioning method based on signal intensity difference

Abstract

The invention discloses an indoor unknown signal interference source fingerprint positioning method based on a signal intensity difference, and the method comprises the following steps: obtaining off-line feature data of a reference point and on-line feature data of an unknown signal interference source, and constructing an off-line fingerprint database and an on-line fingerprint database; carrying out improved singular value decomposition noise reduction processing based on the offline fingerprint database, and determining an offline noise level; determining contribution degree weights of different receiving ends based on the off-line noise level and the standard deviation of the on-line received signal strength difference; improving a grey relational degree calculation method based on the contribution degree weight, and obtaining an improved grey relational degree of an offline fingerprint database and an online fingerprint database; and sorting the grey correlation degrees, and performing weighting processing on the coordinate with the highest grey correlation degree to obtain an estimated coordinate of the unknown signal interference source. The database obtained by using the positioning method is more accurate, and the positioning precision is also improved.

Description

Technical Field

[0001] The invention belongs to the field of indoor positioning, and in particular is directed to a fingerprint positioning method for indoor unknown signal interference sources based on signal strength difference. Background Art

[0002] With the booming development of the Internet of Things (IoT), smart homes, and wireless sensor network technologies, indoor spaces are being flooded with a variety of radio signals, including Wi-Fi, Bluetooth, Zigbee, RFID, etc. However, unknown or unauthorized wireless signal sources, also known as unknown signal interference sources (USIS), are also mixed in. It may not only leak personal privacy, but also cause network interference and even become a frontier for malicious attacks. These illegal signal sources, with their small size and high concealment, act recklessly in crowded and frequently changing indoor environments, posing a huge challenge to security supervision and crackdown work. Therefore, it is crucial to locate unknown signal interference sources.

[0003] Although satellite positioning systems (such as China's BeiDou system and the United States' GPS system) can provide reliable location information in open areas, their performance is greatly reduced in complex environments such as indoors. Due to the complexity of the indoor environment, such as wall obstruction and multipath effects, the strength of satellite signals is severely weakened, making it difficult for satellite navigation systems to be effective indoors. Therefore, indoor positioning technology has become a hot topic and difficulty in current research. Indoor positioning technology usually obtains signal measurement parameters from the signal receiver, also known as the access point (AP), for positioning. Common signal parameters include time of arrival (TOA), time difference of arrival (TDOA), angle of arrival (AOA), channel state information (CSI), and received signal strength (RSS). However, TOA and TDOA methods are difficult to popularize due to their high requirements for time synchronization; AOA relies on expensive antenna arrays; CSI is robust but has a high acquisition cost. In contrast, RSS-based fingerprint positioning technology has become the most popular choice in fingerprint positioning technology due to its advantages such as low cost, easy deployment, and no need for additional synchronization equipment. RSS-based fingerprint positioning technology mainly achieves fast and accurate estimation of the target location by constructing a fingerprint library containing location information and RSS features. This technology is mainly divided into two stages: offline and online. In the offline stage, researchers arrange reference points (Reference Point, RP) in the positioning area, collect RSS data from different APs, and form location fingerprints together with these RSS data and RP location information and store them in the fingerprint library. In the online stage, the user terminal collects the RSS data of each AP in real time, compares the real-time measured RSS data with the fingerprint data in the fingerprint library, and uses matching algorithms or positioning models to obtain the target location.

[0004] However, current research focuses on locating signal receivers, while there are relatively few studies on locating unknown signal interference sources, but this field is also of great significance. Unknown signal interference sources may interfere with normal communications, steal privacy information, etc., posing a serious threat to communication security. Therefore, in the face of the challenge of locating unknown signal interference sources, this study overcomes the problem that the traditional RSS fingerprint library is difficult to directly apply to the positioning of unknown signal interference sources due to the significant differences in RSS signals generated by different signal sources at the same location. A received signal strength difference (RSSD) fingerprint database based on singular value decomposition denoising was invented, and a weighted grey correlation algorithm was used to locate unknown signal interference sources. Summary of the invention

[0005] The purpose of the present invention is to provide a fingerprint positioning method for indoor unknown signal interference sources based on signal strength difference to solve the problems existing in the above-mentioned prior art.

[0006] To achieve the above object, the present invention provides a fingerprint positioning method for indoor unknown signal interference source based on signal strength difference, comprising the following steps:

[0007] Obtain offline feature data of reference points and online feature data of unknown signal interference sources, and build offline fingerprint database and online fingerprint database;

[0008] Performing improved singular value decomposition noise reduction processing based on the offline fingerprint database, and determining the offline noise level;

[0009] Determining contribution weights of different receiving ends based on the offline noise level and the standard deviation of the online received signal strength difference;

[0010] Based on the contribution weight, a grey relational degree calculation method is improved, and the improved grey relational degrees of the offline fingerprint database and the online fingerprint database are obtained;

[0011] The grey correlation degrees are sorted, and the coordinates with the highest grey correlation degrees are weighted to obtain the estimated coordinates of the unknown signal interference source;

[0012] Preferably, the process of obtaining offline feature data of reference points and constructing an offline fingerprint database includes:

[0013] The reference points are evenly distributed in the area to be positioned, and different receiving terminals obtain the received signal strength of each reference point;

[0014] The receiving ends are grouped into two groups, and the difference in received signal strength of each group of receiving ends at the same reference point is calculated;

[0015] Building an offline fingerprint database based on the received signal strength difference of the reference point;

[0016] Preferably, the process of obtaining online feature data of unknown signal interference sources and constructing an online fingerprint database includes:

[0017] Obtaining the received signal strength of the unknown signal interference source at different receiving ends in the area to be located;

[0018] The receiving ends are divided into groups of two each, and the difference in the received signal strength of each group of receiving ends at the unknown signal interference source is calculated;

[0019] Based on the received signal strength difference of the unknown signal interference source, construct an online fingerprint database of the unknown signal interference source;

[0020] Preferably, the process of performing improved singular value decomposition denoising processing on the offline fingerprint database includes:

[0021] Convert the received signal strength difference sequences of all reference points obtained by a group of receiving ends into a Hankel matrix form;

[0022] Perform singular value decomposition based on the Hankel matrix constructed above to obtain a singular value sequence;

[0023] Based on the singular value sequence obtained by the decomposition, a singular value threshold is set according to the improved Laida criterion;

[0024] The remaining singular values ​​that do not exceed the singular value threshold are divided into two groups, and the order corresponding to the singular value is used as the independent variable and the singular value is used as the dependent variable for least square fitting to obtain the order corresponding to the minimum fitting error;

[0025] The singular values ​​of the singular value matrix with an order smaller than the above order are set to zero, and the sum of these zeroed singular values ​​is taken as the noise level estimation of the group of receiving ends;

[0026] The denoised Hankel matrix is ​​restored according to the processed singular values ​​and their corresponding singular vectors;

[0027] Restore the above Hankel matrix to sequence form;

[0028] Repeat the above process for the reference point received signal strength difference sequences collected by all receiving end combinations in the large conference room to obtain the noise-reduced offline fingerprint database and the noise level corresponding to each group of receiving ends;

[0029] Preferably, the process of determining the contribution weights of different receiving ends based on the offline noise level and the standard deviation of the online received signal strength difference includes:

[0030] Calculate the standard deviation of the received signal strength difference of the unknown signal interference source received by each group of receiving ends in the online database;

[0031] Determine the contribution of the receiving end to the positioning process based on the standard deviation and the offline noise level, and calculate the contribution weight;

[0032] Preferably, the process of improving the grey relational degree calculation method based on the contribution weight and obtaining the improved grey relational degree of the offline fingerprint database and the online fingerprint database includes:

[0033] Based on the grey correlation coefficient calculation method, the grey correlation coefficient between the offline fingerprint database and the online fingerprint database after noise reduction is obtained;

[0034] Combining the grey correlation coefficient with the contribution weight to obtain an improved grey correlation between each reference point and the unknown signal interference source;

[0035] Preferably, the grey correlation degrees are sorted, and the coordinates with the highest grey correlation degree are weighted to obtain the estimated coordinates of the unknown signal interference source, comprising:

[0036] Sort by grey correlation degree and select the first few reference points with the largest grey correlation degree;

[0037] Based on the received signal strength differences at the first few largest reference points and the received signal strength differences corresponding to the unknown signal interference source, the Euclidean distance of the signal strength differences is calculated, and the distance is used as a weight to assign to the reference point coordinates, and the coordinates of the unknown signal interference source are estimated through the weighted reference point coordinates.

[0038] The technical effects of the present invention are:

[0039] (1) The present invention introduces the received signal strength difference as a positioning feature, which breaks through the limitations of traditional indoor positioning technology when facing heterogeneous radio transmitters and realizes efficient and accurate positioning of unknown signal interference sources.

[0040] (2) In view of the problem that the indoor environment is complex and changeable and the signal is easily disturbed by noise, the present invention adopts improved singular value decomposition to effectively solve the problem that the offline fingerprint database is affected by noise and has low accuracy.

[0041] (3) The present invention uses the standard deviation of the online received signal strength difference and the noise level in the offline stage to determine the contribution weight of the receiving end to solve the problem that the RSSD of the unknown signal interference source in the online stage is easily interfered by noise. Due to the complexity of the indoor environment, the signal strength is easily interfered by noise. Therefore, the present invention assigns weights to different received signal strength differences to reduce the impact of noise on the positioning result. At the same time, the contribution weight is used to improve the gray correlation calculation method, and the improved gray correlation is used for positioning.

[0042] (4) In order to improve the precise positioning of the unknown signal interference source, the present invention introduces distance weights on the basis of the grey correlation matching method, selects the first few reference points with the largest grey correlation, and uses the Euclidean distance of the received signal strength difference between the reference point and the unknown signal interference source as the weight to weight the coordinates of the selected reference point to improve the positioning accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0044] Figure 1 is a system structure diagram in an embodiment of the present invention;

[0045] Figure 2 is a flow chart of a positioning method in an embodiment of the present invention;

[0046] Figure 3 The figure is a flow chart of the curve fitting method in the implementation of the present invention. DETAILED DESCRIPTION

[0047] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0048] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0049] Embodiment 1

[0050] Most of the existing indoor positioning technologies are based on RSS. However, due to the unknown signal interference source's unknown properties such as signal strength and frequency, the traditional indoor positioning relies on the RSS fingerprint database established by the fixed signal source and cannot effectively locate the unknown signal interference source. Therefore, we reconstruct the RSS signal into the received signal strength difference (RSSD) signal and use RSSD as a feature to build a fingerprint database. The advantage of RSSD is that it focuses on the difference between signal strengths rather than the absolute signal strength value, which can offset the impact caused by the change of signal source frequency and strength to a certain extent.

[0051] However, both the RSS signal and the reconstructed RSSD signal will be affected by the noise introduced by factors such as multipath effect and personnel flow in a complex indoor environment, which directly affects the accuracy and reliability of the fingerprint database. In order to solve this problem, we introduced the singular value decomposition (SVD) technology for noise reduction. SVD decomposes the signal matrix into a combination of singular values ​​and singular vectors, where larger singular values ​​represent the effective components in the signal, while smaller singular values ​​often represent noise. By reasonably setting the singular value cutoff point, we can effectively filter out the noise component and improve the signal quality. The key to SVD noise reduction is to find the "optimal singular value cutoff point". This method uses local least squares curve fitting to adaptively find the "optimal singular value cutoff point" to ensure that the noise is removed to the maximum extent while retaining the main features of the signal.

[0052] In order to further improve the positioning accuracy, especially in the dynamically changing indoor environment, we introduced the grey correlation algorithm as the core of online positioning. This method is based on the similarity or difference between the development trends of factors to measure, which can capture the trend of signal changes over time and environment, and improves the positioning accuracy compared with the positioning algorithm that only uses the Euclidean distance as the measurement. At a specific location, the RSS signals received by different APs are reconstructed into RSSD signals. Because the change of RSSD mainly depends on the physical distance, rather than the absolute attribute of the signal source, the curve formed by the target to be located and the reference point on the same coordinate axis should show a high grey correlation. The traditional grey correlation method requires the data to be standardized to make it in the same dimension, while RSSD not only makes all data in the same dimension, but also increases the amount of data, increases the complexity of the signal curve, and thus improves the positioning accuracy. At the same time, in view of the noise interference of the RSSD measured online in real time, we designed an AP contribution weight system based on the online RSSD variance and the noise level in the offline stage. This system further improves the online positioning accuracy based on the grey correlation algorithm by reducing the weight of the AP that is more affected by noise in the positioning process.

[0053] The specific process includes:

[0054] 1. Construction of positioning feature parameters

[0055] When describing the propagation characteristics of indoor radio signals, the log-normal shadow model is usually used, which can effectively express the RSS of any point in the indoor environment. Specifically, the relationship between the received power P(d) and the distance d from the AP is shown in formula (1):

[0056]

[0057] Where P(d) represents the received power at a distance d from the AP; d 0 is the reference distance, generally set to 1 meter, used for the calculation of standardized path loss; P s is the transmission power of the radio transmitter; G s is the transmitting antenna gain; G r is the antenna gain at the receiving end; λ is the wavelength of the transmitter carrier; L r is the system loss factor; α is the path loss index, which varies depending on whether there is a line-of-sight path in the indoor environment and the number of obstructions, and is usually between 1.6 and 1.8 (line-of-sight condition) or between 4 and 6 (obstruction condition); χ is a Gaussian random variable, which represents the shadow fading caused by multipath effects, obstacle scattering, etc., with a mean of 0 and a standard deviation of σ.

[0058] In the actual positioning process, since the transmitting power and antenna gain of the transmitter are often unknown or difficult to measure accurately, it is difficult to locate the unknown signal interference source by directly using the original RSS value as the positioning feature. Therefore, in order to deal with the problem of unknown signal interference sources, the RSS signal is reconstructed into RSSD. This reconstruction aims to extract features that are more directly related to the distance change through mathematical or signal processing methods, while minimizing or compensating for the influence of unknown factors and environmental randomness at the transmitter, thereby providing a more reliable and stable basis for indoor positioning.

[0059] From equation (1), we can get the RSS of the p-th AP and the q-th AP receiving the same transmitter as equation (2) and equation (3), and the RSSD of the p-th AP and the q-th AP can be shown as (4).

[0060]

[0061] From formula (4), we can see that RSSD still obeys Gaussian distribution.

[0062] 2. Offline Phase

[0063] Assume that M APs and N RPs are deployed in the indoor positioning area. The RSS fingerprint database collected by each AP is shown in Table 1:

[0064] Table 1 Offline RSS fingerprint database

[0065]

[0066] Subtract the RSS collected by any two APs to get RSSD, and we get Set AP combination, The RSSD is matched with the position coordinates to form a fingerprint, and all fingerprint information is stored in the fingerprint library. The RSSD-based offline fingerprint database is obtained as shown in Table 2:

[0067] Table 2 Offline RSSD fingerprint database

[0068]

[0069] SVD is used to perform noise reduction on the RSSD fingerprint database.

[0070] Assume that the RSS received by any AP is expressed as formula (5):

[0071]

[0072] in Expressed as the pure error-free measurement RSS value in watts (W), n i Indicates the noise content.

[0073] According to formula (5), the RSSD obtained by subtracting any two APs can be expressed as:

[0074]

[0075] The noise in the database is introduced by each AP collection. In order to remove the noise in the fingerprint database, i,j , SVD is used to reduce the noise of the RSSD vectors of all RPs collected by each group of APs. The RSSD vectors of all reference points from the same group of APs can be expressed as follows:

[0076] RSSD α =[RSSD 1，α , RSSD 2,α ,...,RSSD N，α ] T (7)

[0077] RSSD k,α It indicates that the kth RSSD data is collected for the αth group of APs.

[0078] However, since SVD denoising works on a rectangular form, the RSSD vector is converted into a matrix form using the special structure of the Hankel matrix as shown in formula (8):

[0079]

[0080] Where N = m + n - 1.

[0081] SVD is an orthogonal transformation that can transform the original matrix into a diagonal matrix whose eigenvalues ​​reflect some of the main features of the original matrix. According to SVD theory, there exists an orthogonal matrix So that:

[0082] H ɑ=UΣV T (9)

[0083] in: is an n×m singular value matrix; ∑ r =diag(δ 1 , δ 2 , ..., δ r )The elements on the diagonal are H ɑ Singular values ​​of the matrix; H α The singular values ​​are arranged in descending order, i.e.

[0084] δ 1 ≥δ 2 ≥...≥δ r ≥δ r+1 =δ r+2 =... =δ n =0,r=rank(H ɑ ) (10)

[0085] The SVD formula of the Hankel matrix can be written in vector form as follows:

[0086]

[0087] According to the low-rank approximation principle of matrices, a noisy matrix can be approximated by a low-rank matrix. The larger the singular value, the more information it contains. The smaller the singular value, the less information. Since the singular value representing noise is generally small, the signal can be denoised by selecting appropriate singular values, and the main information of the signal can be retained. By selecting appropriate non-zero singular values ​​for reconstruction, a new matrix can be obtained

[0088]

[0089] Where γ is the "optimal singular value cutoff point", that is, the number of non-zero singular values ​​selected. How to determine whether γ is the reconstructed matrix If γ is too large, the noise reduction effect is not obvious, and there will still be a lot of noise in the signal. If γ is too small, some useful information of the original signal will be lost, and it may even cause distortion of the signal after noise reduction. The essence of SVD noise reduction is to distinguish between singular values ​​representing useful information of the signal and singular values ​​representing noise by analyzing the size distribution of singular values. In situations with high signal-to-noise ratio, the singular value distribution curve often shows an obvious "elbow" feature, which marks a significant boundary between the signal and noise singular values. However, when the signal-to-noise ratio is low, this distinction becomes blurred, the entire curve tends to be flat, and it is difficult to directly identify the "elbow", so it is difficult to determine the "optimal singular value cutoff point".

[0090] The present invention uses two least squares curves to fit the singular value curve, and divides the singular values ​​of different orders into useful information feature groups and noise feature groups. Because the first few largest singular values ​​have far exceeded the remaining values, these high values ​​usually strongly reflect the main characteristics of the signal, and the influence of noise on these high values ​​is minimal. However, directly fitting these high values ​​to the curve will cause the "optimal singular value cutoff point" to be too far forward, and then mistakenly classify some useful singular values ​​into the noise group, resulting in excessive noise reduction.

[0091] In order to avoid this problem, the present invention introduces a threshold t based on the improved Laida criterion. Only when the singular value does not exceed this threshold, it is included in the fitting range. In this way, it can not only eliminate the unreasonable influence of a very small number of extremely large singular values ​​on the fitting results, but also ensure that the fitting process more accurately reflects the actual boundary between the signal and the noise, thereby optimizing the noise reduction effect, reducing signal distortion, and improving the accuracy of the reconstructed signal. The setting of the threshold t is as shown in formula (13):

[0092]

[0093] in Collectively referred to as discrete coefficient of variation; σ RSSD is the standard deviation to remove the singular extreme values; μ RSSD is the mean after removing the outliers.

[0094] The singular values ​​below the threshold are fitted with the least square method. If the first a singular values ​​exceed the threshold, the singular values ​​S′ below the threshold and their corresponding order O′ are divided into two groups, as shown in equations (14)-(15)

[0095] S 1 =[δ a+1 , δ a+2 , ..., δ γ ],O 1 =[a+1,a+2,...,γ] (14)

[0096] S 2 =[δ γ+1 , δ γ+2 , ..., δ r ],O 1 =[γ+1,γ+2,...,r] (15)

[0097] With the order as the independent variable and the singular value as the dependent variable, the least squares method is used to fit the linear function f 1 and f 2 , calculate the fitting error FE γ :

[0098]

[0099] When FE γ When it reaches the minimum, γ is the "optimal singular value cutoff point".

[0100] After finding the optimal singular value inflection point, the sum of the singular values ​​after γ+1 represents the noise level NL to which the group of APs are subjected in the offline stage. α :

[0101]

[0102] According to the SVD denoising principle, the singular values ​​after γ+1 are set to zero, and the first γ singular values ​​are reconstructed. Then the Hankel matrix of the RSSD after denoising is reconstructed as shown in formula (12).

[0103] After denoising the RSSD signal space, the ideal estimated RSSD denoising space needs to be restored to the RSSD vector form. According to the characteristics of the Hankel matrix, the estimated ideal signal space The same Hankel matrix is The RSSD vector of the corresponding AP combination and even the fingerprint database can be restored by taking the average of the diagonal matrix. The formula for the diagonal average is shown in formula (18):

[0104]

[0105] where x i,j Represents the Hankel matrix The element in the i-th row and j-th column of , where n is the Hankel matrix The number of rows, m is the Hankel matrix The number of columns, and n<m.

[0106] Then the RSSD vector of the AP combination after noise reduction is expressed as formula (19):

[0107]

[0108] The RSSD vector denoising process corresponding to the remaining AP combinations is as above. All denoised RSSD vectors and the corresponding RP coordinates are used as features to form a denoised offline fingerprint database as shown in Table 3:

[0109] Table 3 Offline fingerprint database after denoising

[0110]

[0111] 3. Online Stage

[0112] When an unknown signal interference source (USIS) enters the area to be positioned, all APs in the positioning area will collect relevant RSS measurement results from the USIS in real time, and then reconstruct RSSD online data.

[0113] Assume that the RSSD obtained online is:

[0114] RSSD USIS =[RSSD USIS，1 , RSSD USIS，2 , ..., RSSD USIS，A ] (20)

[0115] RSSD k,USIS The RSSD signal obtained by subtracting the kth group of APs.

[0116] The traditional grey relational algorithm needs to standardize the data to make it in the same dimension, while RSSD has effectively eliminated the dimensional difference of the data, making the offline database data and the online measurement data in the same dimension, and the grey relational analysis can be directly used.

[0117] The RSSD sequence obtained online is regarded as the comparison sequence, and the RSSD corresponding to all RPs from the same group of APs in the database is regarded as the reference sequence. Then all reference sequences form a reference matrix, and the absolute difference between all reference sequences and the comparison sequence is taken to form an absolute difference matrix AD:

[0118] AD=[AD 1 , A.D. 2 , ..., A.D. A ] (twenty one)

[0119] in:

[0120]

[0121] Correlation coefficient between each comparison sequence and the reference sequence:

[0122]

[0123] where d max is the maximum value of all elements in D, d min is the minimum value of all elements in D, ρ represents the resolution coefficient, which is used to adjust the size of the comparison environment and is generally taken as 0.5.

[0124] When using the traditional grey correlation analysis model for indoor positioning, the traditional approach is to average the correlation coefficients between the RSSD measured by each AP and the reference sequence, and use this as the overall grey correlation. Considering that when collecting RSSD features in both the offline and online stages, they will be affected by various indoor noises, resulting in inaccurate RSSD measured by each AP. In order to more accurately reflect the reliability of each AP in the current environment and its contribution to positioning, we propose the concept of introducing offline noise level and online RSSD variance as weights. Specifically, the AP is affected by noise differently in the offline and online stages. In order to further improve the positioning accuracy, the singular value noise level representing the noise is introduced, and in the online stage, the variance of the RSSD data measured by each group of APs online can be regarded as an indicator of the degree to which the AP is disturbed by environmental noise. The larger the RSSD variance, the greater the fluctuation of the RSSD data of the AP and the greater the impact of noise. Therefore, its importance should be reduced accordingly in the positioning process, that is, a smaller weight should be given.

[0125] Assume that the variance of RSSD obtained by the kth group of APs is σ k , then the AP contribution weight is set to:

[0126]

[0127] Then the grey correlation degree between each RP and USIS is:

[0128]

[0129] The larger the GRD, the higher the correlation, and the smaller the GRD, the smaller the correlation. The coordinates of the K RPs with the largest grey correlation are selected for the subsequent USIS position estimation.

[0130] Due to the interference of indoor signal propagation, in some cases, the RP coordinates with a larger grey correlation are actually quite different from the USIS coordinates. To solve this problem and further improve the positioning accuracy, the distance D between the RSSD corresponding to the selected K RP fingerprint coordinates and the RSSD corresponding to the USIS is calculated and used as the weight Assign weights to the RP coordinates and obtain the weighted coordinates as the final positioning result, as shown in the following formula:

[0131]

[0132] Existing RSS-based indoor positioning technology cannot be directly applied to the positioning of unknown signal interference sources because the pre-built RSS fingerprint database lacks flexibility and cannot adapt to the changes of different signal sources, thereby limiting the accuracy of positioning. To solve this problem, the present invention innovatively reconstructs the RSS fingerprint database into an RSSD fingerprint database to achieve effective positioning of different signal sources.

[0133] In view of the fact that indoor environments are affected by factors such as multipath effects and personnel flow, the database contains a large amount of noise. The present invention uses improved SVD noise reduction processing and local least squares curve fitting to adaptively find the optimal singular value inflection point to ensure that the noise is removed to the maximum extent while retaining the main features of the signal, making the offline database more accurate.

[0134] In order to more accurately locate the unknown signal interference source in a dynamic environment, the gray correlation algorithm is used to replace the traditional positioning algorithm based on Euclidean distance. Considering that the contribution of each AP to online positioning is different due to the influence of noise during online positioning, the present invention designs a set of AP contribution weights, assigns different weights to each AP, so that the final gray correlation is more in line with the actual situation, thereby achieving a significant improvement in positioning accuracy.

[0135] The above is only a preferred specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A fingerprint positioning method for indoor unknown signal interference source based on signal strength difference, characterized in that: The following steps are involved: Obtain offline feature data of reference points and online feature data of unknown signal interference sources, and build offline fingerprint database and online fingerprint database; Performing improved singular value decomposition noise reduction processing based on the offline fingerprint database, and determining the offline noise level; Determining contribution weights of different receiving ends based on the offline noise level and the standard deviation of the online received signal strength difference; Based on the contribution weight, a grey relational degree calculation method is improved, and the improved grey relational degrees of the offline fingerprint database and the online fingerprint database are obtained; The grey correlation degrees are sorted, and the coordinates with the highest grey correlation degree are weighted to obtain the estimated coordinates of the unknown signal interference source.

2. The method for locating illegal unknown signal interference sources in a large conference room according to claim 1 is characterized in that: The process of obtaining offline feature data of reference points and building an offline fingerprint database includes: The reference points are evenly distributed in the area to be positioned, and different receiving terminals obtain the received signal strength of each reference point; The receiving ends are grouped into two groups, and the difference in received signal strength of each group of receiving ends at the same reference point is calculated; Based on the received signal strength differences of the reference points, an offline fingerprint database is constructed.

3. The method for locating illegal unknown signal interference sources in a large conference room according to claim 1 is characterized in that: The process of obtaining online feature data of unknown signal interference sources and building an online fingerprint database includes: Obtaining the received signal strength of the unknown signal interference source at different receiving ends in the area to be located; The receiving ends are divided into groups of two each, and the difference in the received signal strength of each group of receiving ends at the unknown signal interference source is calculated; Based on the received signal strength difference of the unknown signal interference source, an online fingerprint database of the unknown signal interference source is constructed.

4. The method for locating illegal unknown signal interference sources in a large conference room according to claim 1 is characterized in that: The process of performing improved singular value decomposition denoising on the offline fingerprint database and determining the offline noise level includes: Convert the received signal strength difference sequences of all reference points obtained by a group of receiving ends into a Hankel matrix form; Perform singular value decomposition based on the Hankel matrix constructed above to obtain a singular value sequence; Based on the singular value sequence obtained by the decomposition, a singular value threshold is set according to the improved Laida criterion; The remaining singular values ​​that do not exceed the singular value threshold are divided into two groups, and the order corresponding to the singular value is used as the independent variable and the singular value is used as the dependent variable for least square fitting to obtain the order corresponding to the minimum fitting error; The singular values ​​of the singular value matrix with an order smaller than the above order are set to zero, and the sum of these zeroed singular values ​​is taken as the noise level estimation of the group of receiving ends; The processed singular values ​​and their corresponding singular vectors are used to restore the denoised Hankel matrix; Restore the above Hankel matrix to sequence form; The above process is repeated for the reference point received signal strength difference sequences collected by all receiving end combinations in the large conference room to obtain the noise-reduced offline fingerprint database and the noise level corresponding to each group of receiving ends.

5. The method for locating illegal unknown signal interference sources in a large conference room according to claim 1 is characterized in that: The process of determining contribution weights of different receiving ends based on the offline noise level and the standard deviation of the online received signal strength difference includes: Calculate the standard deviation of the received signal strength difference of the unknown signal interference source received by each group of receiving ends in the online database; Based on the standard deviation and the offline noise level described in Requirement 4, the contribution of the receiving end to the positioning process is determined, and the contribution weight is calculated.

6. The method for locating illegal unknown signal interference sources in a large conference room according to claim 1 is characterized in that: The process of improving the grey correlation calculation method based on the contribution weight and obtaining the improved grey correlation between the offline fingerprint database and the online fingerprint database includes: Based on the grey correlation coefficient calculation method, the grey correlation coefficient between the offline fingerprint database and the online fingerprint database after noise reduction is obtained; The grey relational coefficient is combined with the contribution weight to obtain an improved grey relational degree between each reference point and the unknown signal interference source.

7. The method for locating illegal unknown signal interference sources in a large conference room according to claim 1 is characterized in that: The process of sorting the grey correlation degrees and performing weighted processing on the coordinates with the highest grey correlation degree to obtain the estimated coordinates of the unknown signal interference source includes: Sort by grey correlation degree and select the first few reference points with the largest grey correlation degree; Based on the received signal strength differences at the first few largest reference points and the received signal strength differences corresponding to the unknown signal interference source, the Euclidean distance of the signal strength differences is calculated, and the distance is used as a weight to assign to the reference point coordinates, and the coordinates of the unknown signal interference source are estimated through the weighted reference point coordinates.

Images