Indoor positioning method, device, computer equipment, readable storage medium and program product

By acquiring and processing WiFi channel status information, combining signal processing and machine learning models, the problem of low accuracy of WiFi indoor positioning in high-density environments is solved, and high-precision and real-time indoor positioning are achieved.

CN118509990BActive Publication Date: 2024-11-05CHINA TELECOM CORP LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410971129.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-19
Publication Date
2024-11-05
Estimated Expiration
2044-07-19

AI Technical Summary

Technical Problem

The existing indoor positioning method based on WiFi is low in positioning accuracy in high-density signal environments and is susceptible to noise interference.

Method used

By obtaining the channel state information between the indoor to be located source and multiple signal receiving ends within a preset time interval, a signal processing model is used to process the channel state information to obtain the candidate positioning information set, and processing it using the pre-trained machine learning model. Candidate positioning information is collected to obtain indoor positioning information, and combined with signal processing models to reduce noise interference and improve positioning accuracy.

Benefits of technology

It significantly improves indoor positioning accuracy in high-density signal environments, realizes real-time, non-invasive indoor positioning, and can adapt to environmental changes and noise interference without additional hardware equipment using existing WiFi infrastructure.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118509990B_ABST
    Figure CN118509990B_ABST
Patent Text Reader

Abstract

The present application relates to an indoor positioning method, device, computer equipment, computer-readable storage medium and computer program product. The method comprises: obtaining channel state information between a signal source to be located indoors and multiple signal receiving terminals indoors within a preset time interval; using a signal processing model to process the channel state information to obtain a candidate positioning information set of the signal source to be located; using a pre-trained machine learning model to process the candidate positioning information set to obtain indoor positioning information of the signal source to be located. The method can improve the accuracy of indoor positioning.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to an indoor positioning method, apparatus, computer equipment, computer-readable storage medium, and computer program product. Background Art

[0002] In recent years, with the rapid development of wireless technology, the role of WiFi signals has expanded from a sole communication medium to a non-invasive human sensing tool. The popularity of WiFi devices and the ubiquitous coverage of WiFi networks provide people with the opportunity to extend WiFi capabilities beyond communication, especially in sensing the physical environment. Indoor positioning / tracking can be achieved through contactless wireless sensing.

[0003] However, the current indoor positioning method based on WIFI is still susceptible to noise interference, resulting in low positioning accuracy in high-density signal environments. Summary of the invention

[0004] Based on this, it is necessary to provide an indoor positioning method, device, computer equipment, computer-readable storage medium and computer program product that can improve the indoor positioning accuracy in response to the above technical problems.

[0005] In a first aspect, the present application provides an indoor positioning method, comprising:

[0006] Acquiring, within a preset time interval, channel state information between a signal source to be located indoors and a plurality of signal receiving terminals indoors;

[0007] Using a signal processing model to process the channel state information, and obtaining a set of candidate positioning information of the information source to be located;

[0008] A pre-trained machine learning model is used to process the candidate positioning information set to obtain indoor positioning information of the information source to be located.

[0009] In one of the embodiments, the signal processing model includes a first signal processing model and a second signal processing model;

[0010] The adopting of the signal processing model to process the channel state information to obtain a set of candidate positioning information of the information source to be located includes:

[0011] Using the first signal processing model to process the channel state information, obtain information transmission time difference information of any two signal receiving ends among the multiple signal receiving ends for the information source to be located;

[0012] Using the second signal processing model to process the channel state information, and obtaining the signal transmission direction of the to-be-located signal source relative to the multiple signal receiving ends;

[0013] A candidate positioning information set of the information source to be located is determined according to the information transmission time difference information and the signal transmission direction.

[0014] In one embodiment, the adopting the second signal processing model to process the channel state information to obtain the signal transmission direction of the information source to be located relative to the multiple signal receiving ends includes:

[0015] Determine, according to the channel state information, a corresponding eigenvector of a signal subspace and a eigenvector of a noise subspace;

[0016] constructing a spatial spectrum function according to the eigenvector of the signal subspace and the eigenvector of the noise subspace;

[0017] The signal transmission direction of the information source to be located relative to the multiple signal receiving ends is determined according to the spatial spectrum function.

[0018] In one embodiment, determining the eigenvector of the corresponding signal subspace and the eigenvector of the noise subspace according to the channel state information includes:

[0019] Preprocessing the channel state information to obtain processed channel state information;

[0020] Determining a covariance matrix corresponding to the processed channel state information;

[0021] Performing eigenvalue decomposition processing on the covariance matrix to obtain the eigenvector of the signal subspace.

[0022] In one embodiment, determining the eigenvector of the corresponding signal subspace and the eigenvector of the noise subspace according to the channel state information includes:

[0023] Obtaining a conjugate reconstruction matrix corresponding to the channel state information according to a preset inverse unit matrix and a complex conjugate matrix corresponding to the channel state information;

[0024] Performing eigenvalue decomposition processing on the conjugate reconstruction matrix to obtain the eigenvector of the noise subspace.

[0025] In one embodiment, before using the pre-trained machine learning model to process the candidate positioning information set to obtain the indoor positioning information of the source to be positioned, the method further includes:

[0026] Using a preset feature extraction model, performing feature extraction processing on the candidate positioning information set to obtain feature information corresponding to the candidate positioning information set;

[0027] The adopting of the pre-trained machine learning model to process the candidate positioning information set to obtain the indoor positioning information of the information source to be positioned includes:

[0028] The pre-trained machine learning model is used to process the feature information to obtain indoor positioning information of the information source to be located.

[0029] In one of the embodiments, before using a preset feature extraction model to perform feature extraction processing on the candidate positioning information set to obtain feature information corresponding to the candidate positioning information set, the method further includes:

[0030] Standardize the sample data of indoor positioning to obtain processed sample data;

[0031] Determine the correlation matrix corresponding to the processed sample data, and perform eigenvalue decomposition processing on the correlation matrix to obtain sample eigenvalues ​​and sample eigenvectors;

[0032] Determining a characteristic factor for the processed sample data according to the sample characteristic value and the sample characteristic vector;

[0033] The preset feature extraction model is determined according to the feature factor.

[0034] In a second aspect, the present application also provides an indoor positioning device, comprising:

[0035] A channel state acquisition module, used to acquire channel state information between a signal source to be located indoors and a plurality of signal receiving terminals indoors within a preset time interval;

[0036] A candidate positioning acquisition module, used to process the channel state information using a signal processing model to obtain a set of candidate positioning information of the information source to be located;

[0037] The indoor positioning acquisition module is used to use a pre-trained machine learning model to process the candidate positioning information set to obtain the indoor positioning information of the information source to be located.

[0038] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0039] Acquiring, within a preset time interval, channel state information between a signal source to be located indoors and a plurality of signal receiving terminals indoors;

[0040] Using a signal processing model to process the channel state information, and obtaining a set of candidate positioning information of the information source to be located;

[0041] A pre-trained machine learning model is used to process the candidate positioning information set to obtain indoor positioning information of the information source to be located.

[0042] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0043] Acquiring, within a preset time interval, channel state information between a signal source to be located indoors and a plurality of signal receiving terminals indoors;

[0044] Using a signal processing model to process the channel state information, and obtaining a set of candidate positioning information of the information source to be located;

[0045] A pre-trained machine learning model is used to process the candidate positioning information set to obtain indoor positioning information of the information source to be located.

[0046] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:

[0047] Acquiring, within a preset time interval, channel state information between a signal source to be located indoors and a plurality of signal receiving terminals indoors;

[0048] Using a signal processing model to process the channel state information, and obtaining a set of candidate positioning information of the information source to be located;

[0049] A pre-trained machine learning model is used to process the candidate positioning information set to obtain indoor positioning information of the information source to be located.

[0050] The above-mentioned indoor positioning method, device, computer equipment, computer-readable storage medium and computer program product first obtain channel state information between the indoor signal source to be located and multiple signal receiving ends in the room within a preset time interval. The channel state information provides detailed channel information, including signal strength, phase and multipath effect, etc., which can more accurately describe the state of the wireless channel; then, a signal processing model is used to process the channel state information to obtain a candidate positioning information set of the signal source to be located, and the signal processing model is used to process the channel state information obtained by multiple samplings, so as to obtain a candidate positioning information set containing multiple candidate positionings, which provides high-quality initial data for subsequent machine learning processing; finally, a pre-trained machine learning model is used to process the candidate positioning information set to obtain the indoor positioning information of the signal source to be located. The machine learning model has a certain generalization ability by training a large amount of historical data, and can adapt to different environmental changes and noise interference, so that more accurate indoor positioning information can be obtained from the candidate positioning information set. In the above method, high-precision indoor positioning is achieved by comprehensively utilizing the signal processing model to process the channel state information and combining the machine learning model to optimize the candidate positioning information. The signal processing model is used to reduce noise interference and further denoise is performed through the machine learning model, which significantly improves the positioning accuracy in high-density signal environments. In addition, the existing WiFi infrastructure is used for positioning without the need for additional hardware equipment. Channel state information is continuously acquired and processed within a preset time interval, achieving real-time, non-invasive indoor positioning. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0052] Figure 1 is a schematic diagram of a flow chart of an indoor positioning method in an embodiment;

[0053] Figure 2 It is a flowchart of the steps of obtaining a candidate positioning information set in one embodiment;

[0054] Figure 3 A schematic diagram of a flow chart of a step of determining a signal transmission direction in an embodiment;

[0055] Figure 4 is a flow chart of an indoor positioning method according to another embodiment;

[0056] Figure 5is a structural block diagram of an indoor positioning device in an embodiment;

[0057] Figure 6 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0058] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0059] In one embodiment, Figure 1 As shown, an indoor positioning method is provided. This embodiment uses the method applied to a terminal as an example. It can be understood that the method can also be applied to a server, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. Among them, the terminal can be but is not limited to various personal computers, laptops, smart phones, and tablet computers. The server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides cloud computing services. In this embodiment, the method includes the following steps:

[0060] Step S101: obtaining channel state information between a signal source to be located indoors and a plurality of signal receiving terminals indoors within a preset time interval.

[0061] Exemplarily, first, the user needs to determine a preset time interval, such as 50 milliseconds, 200 milliseconds, etc. Then, the terminal collects channel state information multiple times based on a preset frequency within the preset time interval, and the channel state information generally includes information such as signal strength and phase. Among them, the multiple signal receiving ends include at least three signal receiving ends.

[0062] It should be noted that the terminal executing the above method may be the information source to be located itself (such as a mobile phone, a laptop computer, a tablet computer, etc.), or may be another terminal device.

[0063] Step S102: using a signal processing model to process channel state information, and obtaining a set of candidate positioning information of a source to be positioned.

[0064] Exemplarily, first, the terminal cleans the collected channel state information to remove obvious noise and outliers to ensure the quality of the data and improve the accuracy of subsequent processing. For example, statistical methods or filtering techniques are used to remove outliers and noise interference. The terminal can extract the amplitude and phase information of the signal from the channel state information. The amplitude information reflects the change in signal strength, while the phase information provides the change in the signal propagation path. By extracting these features, a detailed description of the channel state can be obtained, providing rich information for subsequent positioning. Analyze the multipath effect in the CSI data, that is, the phenomenon that the signal reaches the receiving end through multiple paths. The multipath effect will cause delay and interference of the signal in time and space. Identifying and separating multipath signals can improve the accuracy of positioning. For example, delay spread and spatial correlation analysis are used to distinguish between direct paths and reflected paths. Then, the terminal performs delay estimation based on the phase and frequency information in the channel state information. The arrival time of the signal can be estimated by phase unwrapping and frequency domain to time domain conversion. Using the delay estimation result, the distance information of the signal propagation can be inferred. According to the delay estimation result, the propagation distance of the signal from the source to each receiving end is calculated. Combined with the propagation speed of the WiFi signal (usually the speed of light), the distance estimation value can be obtained. The distance information provides basic data for subsequent positioning calculations. The terminal combines the distance estimation results of multiple receiving ends to calculate the candidate location points of the source. Common methods include geometric method, least squares method, etc. For example, the intersection between multiple receiving ends and the source is solved by geometric method to form a set of candidate locations. The terminal can obtain a candidate location based on the channel state information obtained by each sampling, and the channel state information obtained by multiple samplings can obtain a set of candidate locations.

[0065] Step S103: Use a pre-trained machine learning model to process the candidate positioning information set to obtain indoor positioning information of the source to be located.

[0066] Exemplarily, the terminal performs standardization on the candidate positioning information set so that the data distribution of each feature has the same scale. Common methods include standardization (mean is 0, standard deviation is 1) and normalization (scaling the data to between 0 and 1). Standardization helps to eliminate the order of magnitude differences between different features and improve the training effect and convergence speed of the machine learning model. Then, the terminal selects the features that have a greater impact on the positioning results for modeling based on the feature importance of the candidate positioning information set. Feature selection can be completed by calculating the importance index of the feature (such as information gain, mutual information, variance, etc.). Through feature selection, the feature dimension can be reduced, the complexity of the model can be reduced, and the training efficiency and generalization ability of the model can be improved. Then, the terminal uses the pre-trained machine learning model to predict the indoor position of the source. By inputting new candidate positioning information, the model will output the specific location coordinates of the source indoors. The machine learning model can be a neural network model, a random forest, a support vector machine, etc.

[0067] In the above-mentioned indoor positioning method, first, within a preset time interval, the channel state information between the indoor signal source to be located and multiple signal receiving ends in the room is obtained. The channel state information provides detailed channel information, including signal strength, phase and multipath effect, etc., which can more accurately describe the state of the wireless channel; then, a signal processing model is used to process the channel state information to obtain a candidate positioning information set of the signal source to be located, and the signal processing model is used to process the channel state information obtained by multiple samplings, so as to obtain a candidate positioning information set containing multiple candidate positionings, which provides high-quality initial data for subsequent machine learning processing; finally, a pre-trained machine learning model is used to process the candidate positioning information set to obtain the indoor positioning information of the signal source to be located. The machine learning model has a certain generalization ability by training a large amount of historical data, and can adapt to different environmental changes and noise interference, and thus can obtain more accurate indoor positioning information from the candidate positioning information set. In the above method, high-precision indoor positioning is achieved by comprehensively utilizing the signal processing model to process the channel state information and combining the machine learning model to optimize the candidate positioning information. The signal processing model is used to reduce noise interference and further denoise is performed through the machine learning model, which significantly improves the positioning accuracy in high-density signal environments. In addition, the existing WiFi infrastructure is used for positioning without the need for additional hardware equipment. Channel state information is continuously acquired and processed within a preset time interval, achieving real-time, non-invasive indoor positioning.

[0068] In an exemplary embodiment, Figure 2 As shown, the signal processing model includes a first signal processing model and a second signal processing model; the above step S102 uses the signal processing model to process the channel state information to obtain a candidate positioning information set of the source to be located, which can also be implemented by the following steps:

[0069] Step S201, using a first signal processing model to process channel state information, and obtaining information transmission time difference information of any two signal receiving ends among a plurality of signal receiving ends for a signal source to be located;

[0070] Step S202, using a second signal processing model to process channel state information to obtain a signal transmission direction of a signal source to be located relative to a plurality of signal receiving ends;

[0071] Step S203: determining a candidate positioning information set of the information source to be located according to the information transmission time difference information and the signal transmission direction.

[0072] Exemplarily, the first signal processing model can be used to obtain information transmission time difference information of information sent by the source to be located received by any two signal receiving ends. The terminal collects channel state information from multiple signal receiving ends (i.e., WiFi access points) and preprocesses the data, including removing noise and outliers to ensure data quality. Then, the terminal extracts the phase information of the subcarrier from the channel state information. The phase information reflects the phase change experienced by the signal during propagation. The extracted phase information is subjected to phase unwrapping processing to eliminate the phase jump phenomenon and make the phase information continuous. Phase unwrapping helps to accurately calculate the phase difference. Calculate the phase difference between any two receiving ends. The phase difference can be used to infer the relative time difference of arrival (TDoA) of the signal between the two receiving ends. The second signal processing model can be used to obtain the signal transmission direction of the source to be located relative to the signal receiving end. The second signal processing model can be a Music algorithm model (an algorithm based on matrix feature space decomposition), and the signal transmission direction can be obtained from the channel state information based on the Music model. The Music model processes matrix data, and the channel state information in this application is also recorded in the form of a matrix. One dimension in the matrix is ​​the channel state information of each signal receiving end in a certain sampling, and the other dimension is the number of samplings. Finally, the terminal constructs a set of hyperbolic equations based on the obtained TDoA information. The TDoA value of each pair of receiving ends corresponds to a hyperbola, and the location of the source should be near the intersection of these hyperbolas. Integrate the obtained signal transmission direction with the hyperbolic equation group. The signal transmission direction provides an angle constraint on the source, and combined with the intersection point information of the TDoA hyperbola, the source position can be determined more accurately. By solving the hyperbolic equation group and combining the signal transmission direction, a set of candidate positioning information of the source is generated, which contains multiple possible location points, reflecting the multiple possible locations of the source indoors. The channel state information of a group of three signal receiving ends obtained by one sampling can determine the intersection point of a hyperbola; if there are more than three signal receiving ends, then one sampling can obtain the intersection points corresponding to multiple groups of signal receiving ends. Then, in a specific example, these intersection points can be merged into one intersection point, and a fixed intersection point is determined as a candidate position each time sampling; in another specific example, these intersection points can all be used as candidate positions as long as they correspond to the subsequent data processing dimensions.

[0073] In this embodiment, by using the phase information in the channel state information to calculate TDoA, there is no need to calculate ToA first, which simplifies the calculation process and provides high-precision time difference information, which helps to accurately locate the position of the source. Through the second signal processing model, the direction information of the source is provided, which improves the positioning resolution. Combining TDoA and the transmission direction of the source to generate a candidate positioning information set effectively integrates the information of the time and space dimensions and improves the positioning accuracy. Through the application of data preprocessing and multi-signal processing models, the system has strong noise resistance and robustness, and can achieve reliable positioning in complex indoor environments. The generated candidate positioning information set contains multiple possible location points, takes into account various uncertainty factors, provides comprehensive positioning possibilities, and improves the reliability of the final positioning result.

[0074] In an exemplary embodiment, Figure 3 As shown, the above step S202 adopts the second signal processing model to process the channel state information to obtain the signal transmission direction of the information source to be located relative to multiple signal receiving ends, which can also be achieved by the following steps:

[0075] Step S301, determining a corresponding eigenvector of a signal subspace and a eigenvector of a noise subspace according to channel state information;

[0076] Step S302, constructing a spatial spectrum function according to the eigenvector of the signal subspace and the eigenvector of the noise subspace;

[0077] Step S303: determining the signal transmission direction of the information source to be located relative to the plurality of signal receiving ends according to the spatial spectrum function.

[0078] Exemplarily, the second signal processing model is a Music algorithm model. First, the terminal can obtain an array observation signal matrix in matrix form for the channel state information, and then construct a covariance matrix corresponding to the array observation signal matrix, and perform eigenvalue decomposition based on the covariance matrix to obtain eigenvalues ​​and eigenvectors. Eigenvectors with larger eigenvalues ​​can constitute a signal subspace, and eigenvectors with smaller eigenvalues ​​can constitute a noise subspace. Then, the terminal uses the eigenvectors of the signal subspace and the noise subspace obtained by eigenvalue decomposition, and uses their orthogonality to construct a spatial spectrum function. The orthogonality of the signal subspace and the noise subspace means that the eigenvectors of the signal subspace and the eigenvectors of the noise subspace are orthogonal in space, and the direction of the signal can be determined by calculating the orthogonality of the two. The terminal constructs a Music spatial spectrum function based on the eigenvectors of the signal subspace and the eigenvectors of the noise subspace. The spatial spectrum function is a function about the spatial direction, and its value reflects the strength of the signal in different directions. The peak of the spatial spectrum function corresponds to the direction of the incoming wave of the signal. Finally, the terminal performs a full-domain search on the spatial spectrum function to find the spectrum peak. The direction corresponding to the spectrum peak is the direction of the incoming wave of the source.

[0079] Spatial Spectrum Function :

[0080]

[0081] where θ is the spatial direction, is the conjugate transpose of the direction vector, is the eigenvector of the noise subspace.

[0082] In this embodiment, the Music algorithm uses the eigenvalue decomposition of the covariance matrix and the orthogonality of the signal subspace and the noise subspace to achieve high-resolution direction estimation and accurately determine the direction of the incoming wave of the source. The Music algorithm uses the eigenvectors of the noise subspace to construct a spatial spectrum function, effectively separates the signal and noise, and improves the noise resistance and robustness of the direction estimation. The Music algorithm can handle complex multipath environments, separate the direct path and the reflected path through eigenvalue decomposition, and ensure the accuracy of the direction estimation.

[0083] In an exemplary embodiment, the above-mentioned step S301 determines the eigenvector of the corresponding signal subspace and the eigenvector of the noise subspace according to the channel state information, and also includes: preprocessing the channel state information to obtain processed channel state information; determining the covariance matrix corresponding to the processed channel state information; performing eigenvalue decomposition processing on the covariance matrix to obtain the eigenvector of the signal subspace.

[0084] Exemplarily, the terminal can construct an array observation signal of the signal receiving array according to the channel state information of each signal receiving end. For example, in an equidistant linear array (i.e., an array composed of any number of signal receiving ends), the array element spacing is also d, the signal wavelength is λ, there is one signal source in space, but due to reflection, interference, scattering and other factors, N homologous but incoherent signals are generated, then the received array observation signal X is ;in, is the ith signal, is the incident angle of the ith signal, is the array manifold vector (Steering Vector), which indicates the angle of the source The following is the response to the array.

[0085] The terminal constructs the corresponding covariance matrix for the array observation signal in matrix form. The covariance matrix reflects the correlation between the signals at different receiving ends, and its elements are the covariance of the receiving end signals. Based on the covariance matrix, eigenvalue decomposition is performed to obtain eigenvalues ​​and eigenvectors. The eigenvectors corresponding to the larger eigenvalues ​​constitute the signal subspace.

[0086] In this embodiment, the Music algorithm uses the eigenvalue decomposition of the covariance matrix and the orthogonality of the signal subspace and the noise subspace to achieve high-precision direction estimation and accurately determine the direction of the incoming wave of the source. The Music algorithm uses the eigenvectors of the noise subspace to construct a spatial spectrum function, effectively separates the signal and noise, and improves the noise resistance and robustness of the direction estimation.

[0087] In an exemplary embodiment, the above-mentioned step S301 determines the eigenvector of the corresponding signal subspace and the eigenvector of the noise subspace according to the channel state information, and also includes: obtaining a conjugate reconstruction matrix corresponding to the channel state information according to a preset inverse unit matrix and a complex conjugate matrix corresponding to the channel state information; performing eigenvalue decomposition processing on the conjugate reconstruction matrix to obtain the eigenvector of the noise subspace.

[0088] Exemplarily, the preset inverse unit matrix is ​​a block matrix with 1 on the sub-diagonal (upper right to lower left) and 0 on the other elements, and the dimension of the matrix is ​​equal to the number of receiving ends. The terminal first obtains the complex conjugate matrix of the array observation signal corresponding to the channel state information, and then multiplies the complex conjugate matrix with the preset inverse unit matrix to obtain the matrix Y. The covariance matrix of matrix Y is calculated to obtain Ry. Then the covariance matrices Rx and Ry of the array observation signal are added to obtain the conjugate reconstruction matrix R. According to matrix theory, matrices Rx, Ry and R have the same noise subspace. The conjugate reconstruction matrix R is eigendecomposed to obtain its eigenvalues ​​and corresponding eigenvectors, and then the noise subspace is separated. The new noise subspace is subsequently used to construct the spatial spectrum, and the estimated value of the position direction is obtained by seeking the peak value.

[0089] In this embodiment, through the improved Music algorithm, high-resolution direction estimation can be achieved by utilizing the eigenvalue decomposition of the conjugate reconstruction matrix and the orthogonality of the noise subspace and the signal subspace. The conjugate reconstruction matrix can better separate the signal and noise, improve the accuracy of the noise subspace, and improve the noise resistance and robustness of the direction estimation. The improved Music algorithm can process related signal sources or closely spaced signals with small signal-to-noise ratios, and improve the performance of the algorithm in complex environments. By combining the channel state information and the information of the conjugate reconstruction matrix, the multi-dimensional information is integrated to improve the accuracy and stability of positioning.

[0090] In an exemplary embodiment, before the above step S103 uses a pre-trained machine learning model to process the candidate positioning information set to obtain the indoor positioning information of the source to be located, it also includes: using a preset feature extraction model to perform feature extraction processing on the candidate positioning information set to obtain feature information corresponding to the candidate positioning information set.

[0091] Furthermore, in an exemplary embodiment, the above-mentioned step S103 uses a pre-trained machine learning model to process a candidate positioning information set to obtain indoor positioning information of the signal source to be located, and also includes: using a pre-trained machine learning model to process feature information to obtain indoor positioning information of the signal source to be located.

[0092] Exemplarily, first, the terminal uses a preset feature extraction model to extract important features from a set of candidate positioning information. The feature extraction model can be various data dimension reduction or feature selection methods, such as PCA (principal component analysis), LDA (linear discriminant analysis), etc. The purpose of feature extraction is to reduce the data dimension, extract the most useful information for positioning, and improve the processing efficiency and accuracy of subsequent models. Then, the terminal uses a pre-trained machine learning model suitable for positioning tasks, such as random forests, support vector machines, neural networks, etc. The extracted feature information is input into the pre-trained machine learning model for processing, and the model predicts the specific location of the source to be located based on the input feature information. The pre-trained machine learning model can use a support vector machine model (SVM). The SVM model is suitable for high-dimensional data, can effectively process nonlinear relationships, and has good generalization capabilities.

[0093] In this embodiment, the most useful features for positioning are extracted through the feature extraction model, which reduces the data dimension, removes redundant information, and improves the processing efficiency and accuracy of the machine learning model. The feature extraction model can filter out noise and abnormal data while extracting key features, thereby enhancing the robustness of the overall system. Feature extraction reduces the data dimension, significantly reduces the computational complexity of the machine learning model when processing data, and improves the real-time processing capability. By pre-training the machine learning model on a large amount of data, the model can better generalize to unseen data and provide stable and reliable positioning results. The combination of feature extraction and machine learning models comprehensively processes multi-dimensional information and enhances the system's positioning capability in complex environments.

[0094] In an exemplary embodiment, before the preset feature extraction model is used to perform feature extraction processing on the candidate positioning information set to obtain feature information corresponding to the candidate positioning information set, it also includes: standardizing the sample data of indoor positioning to obtain processed sample data; determining the correlation matrix corresponding to the processed sample data, and performing eigenvalue decomposition processing on the correlation matrix to obtain sample eigenvalues ​​and sample eigenvectors; determining the characteristic factors for the processed sample data based on the sample eigenvalues ​​and the sample eigenvectors; and determining the preset feature extraction model based on the characteristic factors.

[0095] Exemplarily, the preset feature extraction model is a factor extraction model determined based on factor analysis. The sample data used to determine the preset feature extraction model does not need sample labels with accurate positions. However, the sample data used to train the machine learning model needs sample labels with accurate positions. Based on the consistency of application and training data, the sample data is sample data of the candidate position information set.

[0096] First, the terminal standardizes the sample data so that the data distribution of each feature has the same scale, which is usually achieved by removing the mean and scaling to unit variance. The correlation matrix of the processed sample data is calculated, which reflects the linear correlation between different features. The correlation matrix is ​​decomposed by eigenvalue to obtain eigenvalues ​​and eigenvectors. Then, the terminal selects the main eigenfactors according to the size of the eigenvalues, usually selecting the first few eigenfactors with higher cumulative variance contribution rates. The standard for selecting eigenfactors is that the cumulative variance contribution rate reaches more than 80%. The factor loading matrix of each eigenfactor can be calculated by principal component analysis, that is, the square root of the eigenvalue of the eigenfactor multiplied by the eigenvector is determined as the factor loading of the eigenfactor, and the factor loadings of all eigenfactors constitute the factor loading matrix. Based on the factor loading matrix, the preset feature extraction model can be obtained. It should be noted that after obtaining the factor loading matrix, in order to make the eigenfactors more interpretable, the user can judge the interpretability of the eigenfactors corresponding to the current factor loading matrix. If the meaning of the eigenfactor is unclear, the user can send the demand for rotation factors to the terminal, and the terminal will perform factor rotation processing to obtain the processed factor loading matrix.

[0097] In this embodiment, the sample data of indoor positioning is standardized by adopting the factor analysis method, the correlation matrix is ​​calculated and the eigenvalue decomposition is performed, the characteristic factors are determined and the feature extraction model is constructed accordingly, the dimension of the data is reduced, thereby reducing the complexity of subsequent calculations and enhancing the system's positioning capability in complex environments.

[0098] In another exemplary embodiment, Figure 4 As shown, the present application provides an indoor positioning method, which includes the following steps:

[0099] Step S401: obtaining channel state information between an indoor signal source to be located and multiple indoor signal receiving terminals within a preset time interval.

[0100] Step S402: using a first signal processing model to process channel state information, and obtaining information transmission time difference information between any two signal receiving ends among a plurality of signal receiving ends for a signal source to be located.

[0101] Step S403, preprocessing the channel state information to obtain processed channel state information, determining a covariance matrix corresponding to the processed channel state information, performing eigenvalue decomposition processing on the covariance matrix, and obtaining an eigenvector of a signal subspace.

[0102] Step S404: obtain a conjugate reconstruction matrix corresponding to the channel state information according to the preset inverse unit matrix and the complex conjugate matrix corresponding to the channel state information, perform eigenvalue decomposition on the conjugate reconstruction matrix, and obtain an eigenvector of the noise subspace.

[0103] Step S405: construct a spatial spectrum function according to the eigenvector of the signal subspace and the eigenvector of the noise subspace.

[0104] Step S406: determining the signal transmission direction of the signal source to be located relative to the plurality of signal receiving ends according to the spatial spectrum function.

[0105] Step S407: determining a candidate positioning information set of the information source to be located according to the information transmission time difference information and the signal transmission direction.

[0106] Step S408, standardize the sample data of indoor positioning to obtain processed sample data, determine the correlation matrix corresponding to the processed sample data, and perform eigenvalue decomposition on the correlation matrix to obtain sample eigenvalues ​​and sample eigenvectors.

[0107] Step S409: determining a characteristic factor for the processed sample data according to the sample characteristic value and the sample characteristic vector, and determining a preset feature extraction model according to the characteristic factor.

[0108] Step S410: Using a preset feature extraction model, perform feature extraction processing on the candidate positioning information set to obtain feature information corresponding to the candidate positioning information set.

[0109] Step S411: Use a pre-trained machine learning model to process feature information to obtain indoor positioning information of the signal source to be located.

[0110] In this embodiment, the channel state information (CSI) in the WiFi signal is used as a positioning parameter, and the TDoA positioning method, the Music algorithm, factor analysis, support vector machine (SVM) and other machine learning methods are used to assist in position estimation and accuracy improvement. Based on the channel state information as a positioning parameter, the Music algorithm is used for position estimation. The Music algorithm is characterized by high direction finding resolution, and can perform asymptotic unbiased estimation of the number of signals, TDOA, polarization, noise interference intensity, the intensity of the incoming wave and the coherence relationship. In particular, the conjugate reconstruction matrix is ​​used to obtain the noise subspace to construct the spatial spectrum function, which can solve the estimation problem of multipath signals and can be used for wireless direction finding in high-density signal environments. Based on the position estimation of the Music algorithm, the dimensionality reduction results obtained based on factor analysis are used to extract the main information of the data, and then the support vector machine (SVM) is used for position-based classification learning to obtain a position classification model. This algorithm combination has good positioning performance and accuracy.

[0111] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0112] Based on the same inventive concept, the embodiment of the present application also provides an indoor positioning device for implementing the indoor positioning method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more indoor positioning device embodiments provided below can refer to the limitations of the indoor positioning method above, and will not be repeated here.

[0113] In an exemplary embodiment, Figure 5 As shown, an indoor positioning device is provided, including: a channel state acquisition module 501, a candidate positioning acquisition module 502 and an indoor positioning acquisition module 503, wherein:

[0114] The channel state acquisition module 501 is used to acquire the channel state information between the indoor signal source to be located and multiple indoor signal receiving terminals within a preset time interval;

[0115] The candidate positioning acquisition module 502 is used to process the channel state information using a signal processing model to obtain a set of candidate positioning information of the source to be located;

[0116] The indoor positioning acquisition module 503 is used to use a pre-trained machine learning model to process a candidate positioning information set to obtain indoor positioning information of a source to be positioned.

[0117] In one embodiment, the signal processing model includes a first signal processing model and a second signal processing model; the above-mentioned candidate positioning acquisition module 502 is also used to use the first signal processing model to process channel state information to obtain information transmission time difference information of any two signal receiving ends among multiple signal receiving ends for the information source to be located; use the second signal processing model to process the channel state information to obtain the signal transmission direction of the information source to be located relative to the multiple signal receiving ends; determine the candidate positioning information set of the information source to be located based on the information transmission time difference information and the signal transmission direction.

[0118] In one of the embodiments, the candidate positioning acquisition module 502 is further used to determine the eigenvector of the corresponding signal subspace and the eigenvector of the noise subspace according to the channel state information; construct a spatial spectrum function according to the eigenvector of the signal subspace and the eigenvector of the noise subspace; and determine the signal transmission direction of the signal source to be located relative to multiple signal receiving ends according to the spatial spectrum function.

[0119] In one embodiment, the candidate positioning acquisition module 502 is further used to pre-process the channel state information to obtain processed channel state information; determine the covariance matrix corresponding to the processed channel state information; perform eigenvalue decomposition on the covariance matrix to obtain eigenvectors of the signal subspace.

[0120] In one of the embodiments, the candidate positioning acquisition module 502 is further used to obtain a conjugate reconstruction matrix corresponding to the channel state information based on a preset inverse unit matrix and a complex conjugate matrix corresponding to the channel state information; and perform eigenvalue decomposition processing on the conjugate reconstruction matrix to obtain an eigenvector of the noise subspace.

[0121] In one of the embodiments, the indoor positioning device further includes a feature extraction module, which is used to perform feature extraction processing on the candidate positioning information set using a preset feature extraction model to obtain feature information corresponding to the candidate positioning information set.

[0122] In one of the embodiments, the indoor positioning acquisition module 503 is further used to use a pre-trained machine learning model to process feature information to obtain indoor positioning information of the signal source to be located.

[0123] In one of the embodiments, the indoor positioning device is also used to standardize the sample data of indoor positioning to obtain processed sample data; determine the correlation matrix corresponding to the processed sample data, and perform eigenvalue decomposition on the correlation matrix to obtain sample eigenvalues ​​and sample eigenvectors; determine the characteristic factors for the processed sample data based on the sample eigenvalues ​​and sample eigenvectors; and determine a preset feature extraction model based on the characteristic factors.

[0124] Each module in the indoor positioning device can be implemented in whole or in part by software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module.

[0125] In an exemplary embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 6As shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, near field communication (NFC) or other technologies. When the computer program is executed by the processor, an indoor positioning method is implemented. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0126] Those skilled in the art will understand that Figure 6 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0127] In one embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiments when executing the computer program.

[0128] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0129] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0130] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0131] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0132] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0133] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. An indoor positioning method, characterized in that: The method comprises: Acquiring, within a preset time interval, channel state information between a signal source to be located indoors and a plurality of signal receiving ends indoors; the signal receiving ends are WiFi access points; Using a signal processing model to process the channel state information, a candidate positioning information set of the information source to be located is obtained; the candidate positioning information set is obtained according to the phase information in the channel state information and the array observation signal matrix corresponding to the channel state information; A pre-trained machine learning model is used to process the important features corresponding to the candidate positioning information set to obtain the indoor positioning information of the source to be located; the important features are obtained by performing feature extraction processing on the candidate positioning information set through a preset feature extraction model; the preset feature extraction model is a factor extraction model determined based on sample data of indoor positioning.

2. The method according to claim 1, characterized in that The signal processing model includes a first signal processing model and a second signal processing model; The adopting of the signal processing model to process the channel state information to obtain a set of candidate positioning information of the information source to be located includes: Using the first signal processing model to process the channel state information, obtain information transmission time difference information of any two signal receiving ends among the multiple signal receiving ends for the information source to be located; Using the second signal processing model to process the channel state information, and obtaining the signal transmission direction of the to-be-located signal source relative to the multiple signal receiving ends; A candidate positioning information set of the information source to be located is determined according to the information transmission time difference information and the signal transmission direction.

3. The method according to claim 2, characterized in that The adopting the second signal processing model to process the channel state information to obtain the signal transmission direction of the to-be-located information source relative to the multiple signal receiving ends includes: Determine, according to the channel state information, a corresponding eigenvector of a signal subspace and a eigenvector of a noise subspace; constructing a spatial spectrum function according to the eigenvector of the signal subspace and the eigenvector of the noise subspace; The signal transmission direction of the information source to be located relative to the multiple signal receiving ends is determined according to the spatial spectrum function.

4. The method according to claim 3, characterized in that The step of determining a corresponding eigenvector of a signal subspace and a eigenvector of a noise subspace according to the channel state information includes: Preprocessing the channel state information to obtain processed channel state information; Determining a covariance matrix corresponding to the processed channel state information; Performing eigenvalue decomposition processing on the covariance matrix to obtain the eigenvector of the signal subspace.

5. The method according to claim 3, characterized in that: The step of determining a corresponding eigenvector of a signal subspace and a eigenvector of a noise subspace according to the channel state information includes: Obtaining a conjugate reconstruction matrix corresponding to the channel state information according to a preset inverse unit matrix and a complex conjugate matrix corresponding to the channel state information; Performing eigenvalue decomposition processing on the conjugate reconstruction matrix to obtain the eigenvector of the noise subspace.

6. The method according to claim 1, characterized in that Before using the pre-trained machine learning model to process the candidate positioning information set to obtain the indoor positioning information of the information source to be positioned, the method further includes: Using a preset feature extraction model, performing feature extraction processing on the candidate positioning information set to obtain feature information corresponding to the candidate positioning information set; The adopting of the pre-trained machine learning model to process the candidate positioning information set to obtain the indoor positioning information of the information source to be positioned includes: The pre-trained machine learning model is used to process the feature information to obtain indoor positioning information of the information source to be located.

7. The method according to claim 6, characterized in that Before performing feature extraction processing on the candidate positioning information set by using a preset feature extraction model to obtain feature information corresponding to the candidate positioning information set, the method further includes: Standardize the sample data of indoor positioning to obtain processed sample data; Determine the correlation matrix corresponding to the processed sample data, and perform eigenvalue decomposition processing on the correlation matrix to obtain sample eigenvalues ​​and sample eigenvectors; Determining a characteristic factor for the processed sample data according to the sample characteristic value and the sample characteristic vector; The preset feature extraction model is determined according to the feature factor.

8. An indoor positioning device, characterized in that: The device comprises: A channel state acquisition module, used to acquire channel state information between a signal source to be located indoors and a plurality of signal receiving ends indoors within a preset time interval; the signal receiving ends are WiFi access points; A candidate positioning acquisition module is used to process the channel state information using a signal processing model to obtain a set of candidate positioning information of the source to be located; the set of candidate positioning information is obtained based on the phase information in the channel state information and the array observation signal matrix corresponding to the channel state information; The indoor positioning acquisition module is used to use a pre-trained machine learning model to process the important features corresponding to the candidate positioning information set to obtain the indoor positioning information of the source to be located; the important features are obtained by performing feature extraction processing on the candidate positioning information set through a preset feature extraction model; the preset feature extraction model is a factor extraction model determined based on sample data of indoor positioning.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

11. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.