Indoor fingerprint positioning method and device

Abstract

The embodiment of the invention provides an indoor fingerprint positioning method and device, and relates to the technical field of data processing, and the method comprises the steps: obtaining CSI data of Wi-Fi at an indoor reference point with known coordinates and at a to-be-positioned position, and constructing a fingerprint feature containing amplitude and phase information in the CSI data by using the CSI data collected on a continuous time sequence; calculating a weighted representation coefficient of each feature point in each channel of the CSI fingerprint features, and updating the fingerprint features based on the coefficients to obtain spatial features of the CSI fingerprints; calculating a weighted representation coefficient of each channel of the CSI fingerprint feature, updating the fingerprint feature based on the coefficient, and obtaining a time feature of the CSI fingerprint; fusing the two features to update the fingerprint features representing the reference points; and utilizing known coordinate information of the reference point to train a network to predict the matching probability between the input feature and the position information, and determining the position of the fingerprint feature based on the matching probability. By applying the scheme provided by the embodiment of the invention, the precision and the stability of the indoor fingerprint positioning system can be improved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of data processing, and particularly to an indoor fingerprint positioning method and device. Technical Background

[0002] With the rapid development of the Internet of Things (IoT) industry, location-based services (LBS) have shown great application demands and market potentials in fields such as precision marketing, intelligent logistics, and smart home. As the main technical foundation of LBS, indoor positioning technology has received extensive attention in the academic community. To obtain high-precision and high-stability indoor location services, various sensing technologies have been explored and applied to various indoor scenarios, such as Inertial Measurement Unit (IMU), geomagnetism, radio frequency identification (RFID), camera, Wireless fidelity (Wi-Fi), etc. Among them, Wi-Fi has been favored by researchers due to its unique advantages of low device cost and wide deployment.

[0003] The method based on Wi-Fi fingerprint matching is divided into an offline stage and an online stage. In the offline stage, wireless signals are collected in the space to be located and features are extracted to establish a mapping relationship with the labels of corresponding positions. In the online stage, the wireless signal features of the target to be located are extracted and matched with the position labels to estimate the target position. Such methods usually extract fingerprint features from the received signal strength indicator (RSSI) and channel state information (CSI) of wireless sensing signals. Compared with RSSI, CSI has a fine-grained sensing ability and provides more detailed multipath propagation information by describing amplitude and phase characteristics. Utilizing the location specificity of CSI, related technologies use methods of manually extracting features or extracting features through deep learning algorithms to construct fingerprints for specific positions in the pre-positioning space.

[0004] Although theoretically the extracted features can achieve high-precision matching effects, their actual performance will be greatly limited in actual scenarios. The time sensitivity and environmental dependence of wireless signals will cause the collected signal metrics to fluctuate due to the influence of environmental dynamic changes, such as furniture layout changes, human movement, and the opening and closing of doors and windows. This will result in the collected wireless signals containing abnormal noise, exacerbating the occurrence of fingerprint mis-matching phenomena, causing problems such as decreased positioning accuracy and poor stability, and seriously affecting the performance of indoor location services. Therefore, related technologies face the following problems: There are differences in the fingerprint features extracted from the data collected in the offline stage and the online stage, so the robustness and discriminability are insufficient, and it is difficult to obtain accurate matching results. Summary of the Invention

[0005] The purpose of the embodiments of the present invention is to provide an indoor fingerprint positioning method and device to improve the accuracy and stability of an indoor wireless fingerprint positioning system. The specific technical solutions are as follows:

[0006] In a first aspect, the embodiments of the present invention provide an indoor fingerprint positioning method, and the method includes:

[0007] Obtain CSI data at reference points with known coordinates indoors and at the location to be positioned for Wi-Fi, and apply the CSI data collected in a continuous time series to construct fingerprint features including amplitude and phase information in the CSI data;

[0008] Calculate the weighted representation coefficients of each feature point in each channel of the CSI fingerprint features, and update the fingerprint features based on this coefficient to obtain the spatial features of the CSI fingerprint;

[0009] Calculate the weighted representation coefficients of each channel of the CSI fingerprint features, update the fingerprint features based on this coefficient to obtain the time features of the CSI fingerprint, and fuse this feature with the above-mentioned spatial features of the CSI fingerprint to update the fingerprint features representing the position information of the reference points;

[0010] Use the known coordinate information of the reference points to train the model to predict the matching probability between the input features and the position information, and determine the ability of the fingerprint features to belong to the position based on the matching probability.

[0011] In an embodiment of the present invention, the step of obtaining CSI data at reference points with known coordinates indoors and at the location to be positioned for Wi-Fi, and applying the CSI data collected in a continuous time series to construct fingerprint features including amplitude and phase information in the CSI data includes:

[0012] Continuously obtain multiple CSI data packets through Wi-Fi devices, retain the amplitude and phase information of all subcarriers and multiple links from each CSI data packet as single-layer features, and splice the single-layer features of the CSI in the continuous time series to construct fingerprint features.

[0013] In an embodiment of the present invention, the step of calculating the weighted representation coefficients of each feature point in each channel of the CSI fingerprint features, and updating the fingerprint features based on this coefficient to obtain the spatial features of the CSI fingerprint includes:

[0014] Build a spatial feature learning sub-network. The sub-network first performs pooling representation on each channel of the CSI fingerprint features, and then calculates the spatial weighting coefficient M through a fully connected layer s , and this coefficient is used to weight each feature point in each channel of the CSI fingerprint features, so as to update the CSI fingerprint features and obtain the spatial features of the CSI fingerprint.

[0015] In one embodiment of the present invention, calculating the weighted representation coefficients of each channel of the CSI fingerprint feature, updating the fingerprint feature based on this coefficient, obtaining the time feature of the CSI fingerprint, and fusing this feature with the spatial feature of the above CSI fingerprint to update the fingerprint feature representing the reference point position information includes:

[0016] Construct a time feature learning sub-network, which performs average pooling on all feature points of each channel of the CSI fingerprint feature, and then calculates the time weighting coefficient M through a fully connected layer t , and this coefficient is used to weight each channel in the CSI fingerprint feature, thereby updating the CSI fingerprint feature and obtaining the time feature of the CSI fingerprint.

[0017] The obtained time feature is used to be fused with the spatial feature obtained above. All information of the above two features is retained here and fused in a splicing manner. The fingerprint feature obtained after fusion updates and replaces the original fingerprint feature to represent the reference point position information.

[0018] In one embodiment of the present invention, the ability to use the known coordinate information of the reference point to train the model to predict the matching probability between the input feature and the position information and determine the position of the fingerprint feature based on the matching probability includes:

[0019] Input the CSI data obtained from the reference point and the reference point coordinates into the indoor fingerprint positioning model together, and the pre-trained model updates and adjusts the model parameters; after obtaining the above pre-trained model, input the CSI data obtained at the unknown coordinate in the same indoor scene, and obtain the position information of the fingerprint feature to be located output by the fingerprint positioning model.

[0020] In a second aspect, an embodiment of the present invention provides an indoor fingerprint positioning device, and the device includes:

[0021] An acquisition module, configured to acquire CSI data at a reference point with known coordinates in the indoor area of Wi-Fi and at the location to be located, and apply the CSI data collected in a continuous time series to construct a fingerprint feature including amplitude and phase information in the CSI data;

[0022] A spatial feature calculation module: calculating the weighted representation coefficients of each feature point in each channel of the CSI fingerprint feature, updating the fingerprint feature based on this coefficient, and obtaining the spatial feature of the CSI fingerprint;

[0023] A time feature calculation module: calculating the weighted representation coefficients of each channel of the CSI fingerprint feature, updating the fingerprint feature based on this coefficient, obtaining the time feature of the CSI fingerprint, and fusing this feature with the spatial feature of the above CSI fingerprint to update the fingerprint feature representing the reference point position information;

[0024] Location prediction module: The ability to train a model using the known coordinate information of reference points to predict the matching probability between input features and location information, and determine the location to which the fingerprint features belong based on the matching probability.

[0025] In one embodiment of the present invention, the acquisition module is specifically configured to:

[0026] Continuously acquire multiple CSI data packets through a Wi-Fi device, retain the amplitude and phase information of all subcarriers and multi-links in each CSI data packet as single-layer features, and splice the CSI single-layer features of continuous time series to construct fingerprint features.

[0027] In one embodiment of the present invention, the spatial feature calculation module is specifically configured to:

[0028] Build a spatial feature learning sub-network. The sub-network first performs pooling characterization on each channel of the CSI fingerprint features, and then calculates the spatial weighting coefficient M through a fully connected layer. s The coefficient is used to weight each feature point of each channel in the CSI fingerprint features, thereby updating the CSI fingerprint features and obtaining the spatial features of the CSI fingerprint.

[0029] In one embodiment of the present invention, the temporal feature calculation module is specifically configured to:

[0030] Build a temporal feature learning sub-network. The sub-network performs average pooling on all feature points of each channel of the CSI fingerprint features, and then calculates the temporal weighting coefficient M through a fully connected layer. t , The coefficient is used to weight each channel in the CSI fingerprint features, thereby updating the CSI fingerprint features and obtaining the temporal features of the CSI fingerprint.

[0031] The obtained temporal features are used to be fused with the spatial features obtained above. All information of the above two features is retained here and fused in a splicing manner. The fingerprint features obtained after fusion are updated to replace the original fingerprint features to represent the location information of the reference point.

[0032] In one embodiment of the present invention, the location prediction module is specifically configured to:

[0033] Input the CSI data obtained from the reference point and the reference point coordinates into the indoor fingerprint positioning model together, and the pre-trained model updates and adjusts the model parameters; after obtaining the above pre-trained model, input the CSI data obtained at an unknown coordinate in the same indoor scene, and obtain the location information of the fingerprint features to be located output by the fingerprint positioning model.

[0034] Advantageous effects of the embodiments of the present invention:

[0035] The present invention provides an indoor fingerprint positioning method and device, which simultaneously calculates the temporal features and spatial features of CSI and linearly fuses them, and uses the fused features as fingerprints for indoor position estimation, contributing an important technical tool to the development of indoor location services. Through this indoor fingerprint positioning method and device, compared with the related technologies, the present invention can significantly improve the accuracy and stability of fingerprint positioning in complex dynamic indoor scenarios, and increase the feasibility of deploying the fingerprint positioning system in actual indoor scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can also obtain other embodiments based on these drawings.

[0037] Figure 1 It is a schematic flowchart of an indoor fingerprint positioning method provided by an embodiment of the present invention;

[0038] Figure 2 It is a schematic structural diagram of the original fingerprint features provided by an embodiment of the present invention;

[0039] Figure 3 It is a schematic structural diagram of a spatial feature learning sub-network provided by an embodiment of the present invention;

[0040] Figure 4 It is a schematic structural diagram of a temporal feature learning sub-network provided by an embodiment of the present invention;

[0041] Figure 5 It is a schematic structural diagram of an indoor fingerprint positioning device provided by an embodiment of the present invention.

[0042] DETAILED IMPLEMENTATION METHODS

[0043] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art based on the present invention belong to the scope of protection of the present invention.

[0044] The wireless signals obtained by the indoor fingerprint positioning system are affected by the dynamic changes in the environment and fluctuate, resulting in data containing abnormal noise, exacerbating the occurrence of fingerprint mis-matching, causing problems such as decreased positioning accuracy and poor stability, and seriously affecting the performance of indoor location services. Therefore, the indoor positioning method based on Wi-Fi fingerprint matching faces the following problems: the fingerprint features extracted from the data collected in the offline and online stages are different, so the robustness and discriminability are insufficient, and it is difficult to obtain accurate matching results.

[0045] To solve the above problems, the embodiments of the present invention provide an indoor fingerprint positioning method and device, which will be specifically described below.

[0046] First, an indoor fingerprint positioning method provided by the embodiments of the present invention will be described.

[0047] See Figure 1 , which is a schematic flowchart of an indoor fingerprint positioning method provided by the embodiments of the present invention. This method is applied to an electronic device with computing capabilities. Exemplarily, this method is applied to a computer. The above method includes the following steps S101 to step S104.

[0048] Step S101: Obtain the CSI data of Wi-Fi at the reference points with known coordinates indoors and the location to be determined, and apply the CSI data collected in the continuous time series to construct fingerprint features including the amplitude and phase information in the CSI data.

[0049] During the process of indoor fingerprint positioning, multiple CSI data packets are continuously obtained through Wi-Fi devices, and the amplitude and phase information of all subcarriers and multiple links in each CSI data packet are retained as single-layer features, and the CSI single-layer features of the continuous time series are spliced to construct fingerprint features.

[0050] Exemplarily, each collected CSI data packet contains fine-grained physical information, which records the amplitude and phase information of the wireless signal propagation between the transmitter and the receiver, and the complex propagation link characteristics of its multiple outputs and multiple receptions are determined by the number of antennas of the receiver and the transmitter. The CSI received from each propagation link contains multiple subcarrier information at the same time.

[0051] Therefore, the CSI amplitude data received once can be represented as the matrix CSI A :

[0052]

[0053] Each row vector in the matrix represents the amplitudes of the CSI corresponding to multiple propagation links collected in a data packet, where n is the number of propagation links; each column vector represents the amplitudes of each subcarrier of the CSI of a single link, where m is the number of subcarriers.

[0054] Similarly, the phase matrix CSI can be obtained from the collected CSI data P , and its structure is as follows:

[0055]

[0056] In order to retain the multi-link and multi-carrier information of the wireless signal propagating in space, as Figure 2 shown, in the embodiments of the present invention, the CSI amplitude matrix CSI A and the phase matrix CSI P are spliced together as one layer of the fingerprint feature. And the above-mentioned CSI amplitude matrix CSI A and phase matrix CSI P are obtained from multiple continuously collected data packets, and the single-layer features of T data packets are spliced together to obtain the original fingerprint feature CSI in .

[0057] Step S102: Calculate the weighted representation coefficients of each feature point in each channel of the CSI fingerprint feature, and update the fingerprint feature based on this coefficient to obtain the spatial feature of the CSI fingerprint.

[0058] Specifically, a spatial feature learning sub-network is built. The sub-network first performs pooling representation on each channel of the CSI fingerprint feature, and then calculates the spatial weighting coefficient M s through a fully connected layer. The coefficient is used to weight each feature point in each channel of the CSI fingerprint feature, so as to update the CSI fingerprint feature and obtain the spatial feature of the CSI fingerprint.

[0059] Exemplarily, the above original fingerprint feature is passed through a basic convolutional layer to obtain the output feature CSI basic , whose size is C×H×W. As Figure 3 shown, first perform average pooling and max pooling on the channel dimension of CSI basic to obtain two feature matrices of size 1×H×W, and then splice the two feature matrices to obtain an intermediate layer feature matrix m of size 2×H×W. Apply a convolutional layer to perform convolution on the intermediate layer feature matrix m, and then use the sigmoid function to activate it to obtain a spatial feature weight map M s , whose size is 1×H×W. Finally, multiply the above spatial feature weight map and the feature CSI basic layer by layer to obtain a robust spatial feature weighted by spatial attention. The calculation formula of the spatial attention module is as follows:

[0060]

[0061] where σ is the sigmoid operation, and f CNN is the size of the convolution kernel.

[0062] Step S103: Calculate the weighted representation coefficients for each channel of the CSI fingerprint features, update the fingerprint features based on these coefficients to obtain the temporal features of the CSI fingerprint, and fuse this feature with the spatial features of the above CSI fingerprint to update the fingerprint features representing the reference point position information.

[0063] Specifically, build a temporal feature learning sub-network. The sub-network performs average pooling on all feature points of each channel of the CSI fingerprint features, and then calculates the temporal weighting coefficient M t through a fully connected layer. The coefficient is used to weight each channel in the CSI fingerprint features, thereby updating the CSI fingerprint features to obtain the temporal features of the CSI fingerprint.

[0064] Exemplarily, the above original fingerprint features are also passed through a basic convolutional layer to obtain the output feature CSI basic , with a size of C×H×W. As Figure 4 shown, first perform global average pooling on the obtained CSI basic to obtain a feature vector CSI z with a global receptive field, with a size of 1×1×C. The calculation formula for the global average pooling is as follows:

[0065]

[0066] where CSI basic-C is the c-th layer channel feature in the feature CSI basic .

[0067] Then use a fully connected network to perform a non-linear transformation on the result obtained after the above global average pooling to obtain a channel weight vector M t , with the same size of 1×1×C. Finally, use the obtained channel weight vector M t as the weight to weight the input features of the corresponding channels to obtain stable temporal features. The calculation formula for the above temporal attention module is as follows:

[0068] M t (CSI basic ) = σ(g(CSI z , W))

[0069] = σ(W2δ(W1CSI z ))

[0070] where σ is the sigmoid operation and g is the fully connected network. After that, the CSI spatial features and temporal feature vectors obtained above are linearly concatenated to obtain CSI spatio-temporal features of size 2C×H×W, and the original fingerprint features are replaced by the updated spatio-temporal features as the fingerprint features representing the reference point position information.

[0071] Step S104: Use the known coordinate information of the reference points to train the model to predict the matching probability between the input features and the position information and determine the ability of the fingerprint features to belong to the position based on the matching probability.

[0072] Specifically, the CSI data obtained from the reference points and the reference point coordinates are jointly input into the indoor fingerprint positioning model, and the pre-training model updates and adjusts the model parameters; after obtaining the above pre-training model, the CSI data obtained at the unknown coordinates in the same indoor scene is input to obtain the position information of the fingerprint features to be located output by the fingerprint positioning model.

[0073] Exemplarily, the updated fingerprint features are subjected to a convolution operation to obtain a one-dimensional feature vector h with the same dimension as the number of reference points at the known positions, and the vector dimension is d h = num_RPs. The feature vector is input into the SoftMax classifier for position classification prediction. The probability that the positioning prediction result is the i-th position point is:

[0074]

[0075] where h i represents the i-th eigenvalue in the feature vector h.

[0076] To train the model corresponding to the method end-to-end in the offline stage, cross-entropy is used as the overall loss function, which is expressed as:

[0077]

[0078] where y i and are the true label and the prediction result of the feature to be located, respectively.

[0079] In the application stage, the collected CSI data is input into the above model, and the model will output the matching probability values between the fingerprint features to be located and each reference point. To further obtain the accurate position estimation of the fingerprint features, the reference point positions ranked top three in terms of prediction probability are selected, and the final position prediction is obtained through the weighted centroid method:

[0080]

[0081] where L is the final predicted coordinate, L i 、Pi The reference point coordinates corresponding to the three predicted maximum probabilities and the corresponding probability values.

[0082] Corresponding to the indoor fingerprint positioning method described above, an embodiment of the present invention further provides an indoor fingerprint positioning device.

[0083] See Figure 5 , which is a schematic structural diagram of an indoor fingerprint positioning device provided by an embodiment of the present invention. The device is applied to an electronic device with computing capabilities. The above device includes:

[0084] An acquisition module 501, configured to acquire CSI data at reference points with known coordinates indoors and at the location to be located by Wi-Fi, and apply the CSI data collected in a continuous time series to construct fingerprint features including amplitude and phase information in the CSI data.

[0085] A spatial feature calculation module 502: Calculate the weighted representation coefficients of each feature point in each channel of the CSI fingerprint feature, and update the fingerprint feature based on this coefficient to obtain the spatial feature of the CSI fingerprint.

[0086] A time feature calculation module 503: Calculate the weighted representation coefficients of each channel of the CSI fingerprint feature, update the fingerprint feature based on this coefficient to obtain the time feature of the CSI fingerprint, and fuse this feature with the spatial feature of the above CSI fingerprint to update the fingerprint feature representing the reference point position information.

[0087] A position prediction module 504: Use the known coordinate information of the reference point to train a model to predict the matching probability between the input feature and the position information, and determine the ability of the fingerprint feature to belong to a position based on the matching probability.

[0088] In an embodiment of the present invention, the above acquisition module 501 is specifically configured to:

[0089] Continuously acquire multiple CSI data packets through a Wi-Fi device, retain the amplitude and phase information of all subcarriers and multi-links in each CSI data packet as a single-layer feature, and splice the CSI single-layer features in a continuous time series to construct fingerprint features.

[0090] In an embodiment of the present invention, the above spatial feature calculation module 502 is specifically configured to:

[0091] Build a spatial feature learning sub-network. The sub-network first performs pooling representation on each channel of the CSI fingerprint feature, and then calculates the spatial weighting coefficient M through a fully connected layer s , and the coefficient is used to weight each feature point in each channel of the CSI fingerprint feature, so as to update the CSI fingerprint feature and obtain the spatial feature of the CSI fingerprint.

[0092] In one embodiment of the present invention, the above-mentioned time feature calculation module 503 is specifically configured to:

[0093] Build a time feature learning sub-network, which performs average pooling on all feature points of each channel of the CSI fingerprint feature, and then calculates the time weighting coefficient M through a fully connected layer t , and the coefficient is used to weight each channel in the CSI fingerprint feature, so as to update the CSI fingerprint feature and obtain the time feature of the CSI fingerprint. The obtained time feature is used to fuse with the spatial feature obtained above. All information of the above two features is retained and fused in a splicing manner. The fused fingerprint feature is updated to replace the original fingerprint feature to represent the reference point position information.

[0094] In one embodiment of the present invention, the above-mentioned position prediction module 504 is specifically configured to:

[0095] Input the CSI data obtained from the reference point and the reference point coordinates into the indoor fingerprint positioning model together, and the pre-trained model updates and adjusts the model parameters; after obtaining the above pre-trained model, input the CSI data obtained at the unknown coordinates in the same indoor scene, and obtain the position information of the fingerprint feature to be located output by the fingerprint positioning model.

Claims

1. An indoor fingerprint positioning method, characterized in that, The method includes: Obtaining CSI data at reference points with known indoor coordinates of Wi-Fi and at the location to be located, and applying the CSI data collected in a continuous time series to construct fingerprint features including amplitude and phase information in the CSI data; Calculating the weighted representation coefficients of each feature point in each channel of the CSI fingerprint features, and updating the fingerprint features based on this coefficient to obtain the spatial features of the CSI fingerprint; Calculating the weighted representation coefficients of each channel of the CSI fingerprint features, updating the fingerprint features based on this coefficient to obtain the temporal features of the CSI fingerprint, and fusing this feature with the above-mentioned spatial features of the CSI fingerprint to update the fingerprint features representing the location information of the reference points; Using the known coordinate information of the reference points to train a model to predict the matching probability between the input features and the location information and determining the ability of the fingerprint features to belong to a location based on the matching probability.

2. The method according to claim 1, characterized in that, The obtaining of CSI data at reference points with known indoor coordinates of Wi-Fi and at the location to be located, and applying the CSI data collected in a continuous time series to construct fingerprint features including amplitude and phase information in the CSI data includes: Continuously obtaining multiple CSI data packets through a Wi-Fi device, retaining the amplitude and phase information of all subcarriers and multiple links from each CSI data packet as single-layer features, and splicing the single-layer features of the CSI in the continuous time series to construct fingerprint features.

3. The method according to claim 1, wherein The calculating of the weighted representation coefficients of each feature point in each channel of the CSI fingerprint features, and updating the fingerprint features based on this coefficient to obtain the spatial features of the CSI fingerprint includes: Build a spatial feature learning sub-network. The sub-network first performs pooling representation on each channel of the CSI fingerprint feature, and then calculates the spatial weighting coefficient M through a fully connected layer. s The coefficient is used to weight each feature point of each channel in the CSI fingerprint feature, so as to update the CSI fingerprint feature and obtain the spatial feature of the CSI fingerprint.

4. The method according to claim 1, wherein The calculating of the weighted representation coefficients of each channel of the CSI fingerprint features, updating the fingerprint features based on this coefficient to obtain the temporal features of the CSI fingerprint, and fusing this feature with the above-mentioned spatial features of the CSI fingerprint to update the fingerprint features representing the location information of the reference points includes: Build a time feature learning sub-network, which performs average pooling on all feature points of each channel of the CSI fingerprint feature, and then calculates the time weighting coefficient M through a fully connected layer t , and the coefficient is used to weight each channel in the CSI fingerprint feature, so as to update the CSI fingerprint feature and obtain the time feature of the CSI fingerprint. The obtained temporal features are used to be fused with the spatial features obtained above, and all the information of the above two features is retained here and fused in a splicing manner. The fingerprint features obtained after the fusion are updated to replace the original fingerprint features to represent the location information of the reference points.

5. The method according to claim 1, wherein The using of the known coordinate information of the reference points to train a model to predict the matching probability between the input features and the location information and determining the ability of the fingerprint features to belong to a location based on the matching probability includes: Jointly inputting the CSI data obtained at the reference points and the reference point coordinates into an indoor fingerprint positioning model, and the pre-trained model updates and adjusts the model parameters; after obtaining the above-mentioned pre-trained model, inputting the CSI data obtained at an unknown coordinate in the same indoor scene, and obtaining the location information of the fingerprint features to be located output by the fingerprint positioning model.

6. An indoor fingerprint positioning device, characterized in that, The device includes: An obtaining module, configured to obtain CSI data at reference points with known indoor coordinates of Wi-Fi and at the location to be located, and apply the CSI data collected in a continuous time series to construct fingerprint features including amplitude and phase information in the CSI data; A spatial feature calculation module: calculating the weighted representation coefficients of each feature point in each channel of the CSI fingerprint features, and updating the fingerprint features based on this coefficient to obtain the spatial features of the CSI fingerprint; Time Feature Calculation Module: Calculate the weighted representation coefficients of each channel of the CSI fingerprint features, update the fingerprint features based on these coefficients, obtain the time features of the CSI fingerprint, and fuse these features with the spatial features of the above CSI fingerprint to update the fingerprint features representing the reference point location information; Location Prediction Module: Use the known coordinate information of the reference point to train the model to predict the matching probability between the input features and the location information and determine the ability of the fingerprint features to belong to a location based on the matching probability.

7. The device according to claim 6, characterized in that, The obtaining module is specifically used for: Continuously obtain multiple CSI data packets through a Wi-Fi device, retain the amplitude and phase information of all subcarriers and multi-links in each CSI data packet as single-layer features, and splice the CSI single-layer features of the continuous time series to construct fingerprint features.

8. The device according to claim 6, characterized in that, The spatial feature calculation module is specifically used for: Build a spatial feature learning sub-network. The sub-network first performs pooling representation on each channel of the CSI fingerprint feature, and then calculates the spatial weighting coefficient M through a fully connected layer. s The coefficient is used to weight each feature point of each channel in the CSI fingerprint feature, so as to update the CSI fingerprint feature and obtain the spatial feature of the CSI fingerprint.

9. The device according to claim 6, characterized in that, The time feature calculation module is specifically used for: Build a time feature learning sub-network, which performs average pooling on all feature points of each channel of the CSI fingerprint feature, and then calculates the time weighting coefficient M through a fully connected layer t , and this coefficient is used to weight each channel in the CSI fingerprint feature, so as to update the CSI fingerprint feature and obtain the time feature of the CSI fingerprint. The obtained time features are used to be fused with the spatial features obtained above. All information of the above two features is retained here and fused in a splicing manner. The fingerprint features obtained after fusion are updated to replace the original fingerprint features to represent the reference point location information.

10. The device according to claim 6, characterized in that, The obtaining module, spatial feature calculation module, time feature calculation module, and location prediction module are specifically used for: Input the CSI data obtained at the reference point and the reference point coordinates into the indoor fingerprint positioning model together, and the pre-trained model updates and adjusts the model parameters; after obtaining the above pre-trained model, input the CSI data obtained at an unknown coordinate in the same indoor scene, and obtain the location information of the to-be-located fingerprint features output by the fingerprint positioning model.

Images