WIFI fingerprint positioning method and system for variable indoor environment

Abstract

The invention relates to a WIFI fingerprint positioning method and system for a changeable indoor environment. The method comprises the following steps: collecting online data of a user; inputting the online data of the user into an indoor positioning model, and outputting a positioning result; wherein the indoor positioning model is constructed through a feature extraction network and a classifier and is obtained through training of a plurality of multi-domain fingerprint data, the feature extraction network extracts intra-domain invariant features and inter-domain invariant features, and the intra-domain invariant features and the inter-domain invariant features are fused and input into the classifier. According to the method, the domain invariant features are formed by splicing the internal invariant features and the inter-domain invariant features, and the domain invariant features are used as the position features for position estimation, so that generalization positioning in a variable environment is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of indoor fingerprint positioning, and in particular to a WIFI fingerprint positioning method and system for changing indoor environments. Background Art

[0002] The advent of the big model era has brought profound changes to areas such as embodied intelligence, smart factories, smart transportation, and emergency rescue. Indoor positioning provides key technical support for location-based services, including autonomous obstacle avoidance and navigation for robots in embodied intelligence, equipment and material tracking in smart factories, vehicle positioning for smart transportation, and locating trapped personnel and optimizing rescue routes in emergency rescue. However, most indoor positioning scenarios are characterized by high personnel turnover, complex environmental structures, and frequent object movement, which severely impact positioning accuracy and stability. To address the challenges posed by complex and unpredictable environmental changes to indoor positioning, research is urgently needed to develop high-precision, low-cost, and continuously updateable positioning technologies in these changing environments.

[0003] In complex and changing indoor environments, satellite signals severely attenuate or are interrupted, making high-precision positioning difficult. Researchers have turned to developing positioning systems based on indoor wireless signals. Existing technologies fall into two main categories: geometric positioning, which relies on distance or angle measurements but is susceptible to non-line-of-sight interference, impacting positioning stability. Fingerprint positioning, which achieves positioning by matching pre-stored signal signatures without directly measuring geometric parameters, is more robust to multipath and non-line-of-sight environments. It also offers flexible deployment and highly scalable algorithms, achieving sub-meter positioning accuracy in dynamic scenarios such as shopping malls and hospitals. Current mainstream fingerprint positioning technologies rely on a variety of signal sources, including WiFi, Bluetooth, RFID, ultra-wideband, and vision. WiFi is the most widely used solution due to its wide infrastructure coverage, low cost, and readily available data.

[0004] Data-driven fingerprint positioning requires a pre-built fingerprint database, and the database samples only correspond to the indoor environment at the time of sampling. Typically, a complete initial fingerprint database is collected in a fixed scene. However, indoor environments are complex and changeable. The wireless signal characteristics that serve as fingerprint data can change with the movement of people and the influence of obstacles such as walls and furniture. This leads to significant distribution differences between offline fingerprint databases and online data in real applications, causing a sharp decline in fingerprint positioning performance.

[0005] Current positioning methods based on domain adversarial training have major limitations: First, they rely on target domain data collection after environmental changes (usually requiring 50-100 sets of samples), making it difficult to meet real-time positioning requirements; second, they only support linear knowledge transfer from a single source domain to a single target domain, and are unable to cope with the distribution drift of fingerprint data caused by continuous dynamic changes indoors, resulting in limited cross-scene generalization capabilities; third, simply learning a multi-domain distribution alignment mechanism makes it difficult to extract sufficient domain-invariant features to achieve generalized positioning in unknown environmental distributions. Therefore, there is an urgent need to study cross-domain generalized positioning methods to fully learn the distribution characteristics of existing fingerprint data to improve positioning generalization capabilities in unknown distribution environments. Summary of the Invention

[0006] The purpose of this invention is to provide a Wi-Fi fingerprint positioning method and system for variable indoor environments. Fourier phase transform and knowledge distillation are used to extract intra-domain invariant features from multi-domain fingerprint data. Distribution-related alignment is then used to learn inter-domain invariant features. These features are then combined to form a position estimator, enabling positioning under unknown environmental variations. By maximizing the distance between intra-domain invariant features and inter-domain invariant features, the overlap of invariant information between them is reduced. By concatenating these two types of features, domain-invariant features are formed and used as position features for position estimation, achieving generalized positioning in variable environments.

[0007] To achieve the above object, the present invention provides the following solutions:

[0008] A WIFI fingerprint positioning method for a variable indoor environment, comprising:

[0009] Collect users' online data;

[0010] The user's online data is input into an indoor positioning model, and a positioning result is output; wherein, the indoor positioning model is constructed by a feature extraction network and a classifier and obtained by training with a plurality of multi-domain fingerprint data, the feature extraction network extracts intra-domain invariant features and inter-domain mutually invariant features respectively, and the intra-domain invariant features and inter-domain mutually invariant features are fused and input into the classifier.

[0011] Optionally, the feature extraction network extracts intra-domain invariant features and inter-domain mutually invariant features respectively, including:

[0012] Using the multi-domain fingerprint data to train a teacher network to obtain a trained teacher network;

[0013] The student network is made to learn the extracted features of the trained teacher network through knowledge distillation technology, and the student network is trained with a plurality of multi-domain fingerprint data to obtain a domain-invariant feature extractor;

[0014] The intra-domain invariant feature extractor and the inter-domain mutually invariant feature extractor are used to extract the intra-domain invariant features and the inter-domain mutually invariant features respectively, wherein the intra-domain invariant feature extractor and the inter-domain mutually invariant feature extractor are composed of CNN, ReLu, and BatchNorm layers.

[0015] Optionally, using the plurality of multi-domain fingerprint data to train a teacher network, and obtaining the trained teacher network includes:

[0016] Performing Fourier phase transform on the multi-domain fingerprint data to obtain Fourier phase information;

[0017] Inputting the Fourier phase information into the teacher network to extract Fourier phase fingerprint features;

[0018] The Fourier phase fingerprint feature is combined with the reference point label to train the teacher network.

[0019] Optionally, training the teacher network includes:

[0020]

[0021] in, and They represent the feature extractors of the teacher network and classifier The learnable parameters, P tr represents the distribution of training data, E is the expectation, is the classification cross entropy loss function, x is the multi-domain fingerprint data, is the Fourier phase fingerprint feature, and y is the fingerprint data label.

[0022] Optionally, using knowledge distillation technology to enable the student network to learn the features extracted by the trained teacher network includes:

[0023]

[0024] in, They represent the feature extractors of the student network and classifier The learnable parameters, λ1 is a hyperparameter, is the MSE loss, Fingerprint classification loss for student networks.

[0025] Optionally, the objective function of the indoor positioning model is:

[0026]

[0027] Among them, G c and G fRepresent the classifier and feature extraction network respectively, and their corresponding parameters are θ c and θ f , λ1, λ2 and λ3 are hyperparameters, d(.,.) represents the distance function, z1, z2 are invariant features within the domain and invariant features between domains, respectively, C i and C j are the covariance matrix and ||.|| F represents the Frobenius norm form of the matrix, is the mutually invariant feature learning loss, N is the covariance matrix dimension, is the loss of feature diversity.

[0028] Optionally, the positioning result is:

[0029] y=arg min G c (G f (x)).

[0030] The present invention also provides a system for implementing a WIFI fingerprint positioning method for a variable indoor environment, comprising:

[0031] Data collection module, used to collect users' online data;

[0032] The positioning module is used to input the user's online data into an indoor positioning model and output a positioning result; wherein the indoor positioning model is constructed by a feature extraction network and a classifier and is obtained by training with a plurality of multi-domain fingerprint data. The feature extraction network extracts intra-domain invariant features and inter-domain mutually invariant features respectively, and the intra-domain invariant features and inter-domain mutually invariant features are fused and input into the classifier.

[0033] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, a WIFI fingerprint positioning method for a variable indoor environment is implemented.

[0034] The beneficial effect of the present invention is that when the indoor environment changes, in order to maintain the accuracy of positioning, it is usually necessary to re-collect data and train the positioning model. However, the present invention can adapt to certain environmental changes without the need for updating, so it can reduce more data collection and model training costs.

[0035] The actual operating environment will inevitably have data distribution differences from the fingerprint library. However, based on its generalization mechanism, the present invention can effectively cope with dynamic changes in the environment and improve the accuracy and robustness of the positioning system.

[0036] The present invention uses the knowledge distillation architecture to reduce the amount of computation corresponding to Fourier transform, accelerates the positioning model inference speed while maintaining the same accuracy, reduces the occupation of hardware resources, and is conducive to the application of fingerprint positioning on lightweight devices. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0038] Figure 1 This is a flow chart of a WIFI fingerprint positioning method for a variable indoor environment according to an embodiment of the present invention;

[0039] Figure 2 This is the domain-invariant feature distillation learning architecture of an embodiment of the present invention;

[0040] Figure 3 This is the architecture of the domain generalized positioning method based on Fourier phase information according to an embodiment of the present invention;

[0041] Figure 4 The laboratory structure of an embodiment of the present invention;

[0042] Figure 5 This is a WIFI signal transceiver device according to an embodiment of the present invention, wherein (a) is a receiver and (b) is a transmitter. DETAILED DESCRIPTION

[0043] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0044] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0045] Example 1:

[0046] like Figure 1 As shown, this embodiment provides a WIFI fingerprint positioning method for a variable indoor environment, including:

[0047] Collect users' online data;

[0048] The user's online data is input into the indoor positioning model, and the positioning results are output. The indoor positioning model is constructed through a feature extraction network and a classifier and obtained through training with a number of multi-domain fingerprint data. The feature extraction network extracts intra-domain invariant features and inter-domain mutually invariant features respectively, and the intra-domain invariant features and inter-domain mutually invariant features are fused and input into the classifier.

[0049] Specifically, this embodiment trains on multiple multi-domain fingerprint data sets, using a dual feature extractor to learn both intra-domain invariant features and inter-domain mutually invariant features. These two types of features are then fused to form a unified domain-invariant feature for fingerprint classification. To enhance the information diversity of the localization model, this embodiment introduces an exploration loss mechanism, which minimizes feature overlap by maximizing the difference between the two types of features.

[0050] Furthermore, the feature extraction network extracts intra-domain invariant features and inter-domain invariant features, including:

[0051] Using a number of multi-domain fingerprint data to train the teacher network and obtain the trained teacher network;

[0052] Through knowledge distillation technology, the teacher network extracts features after the student network learns and trains, and the student network is trained with multiple multi-domain fingerprint data to obtain a domain-invariant feature extractor;

[0053] The intra-domain invariant feature extractor and the inter-domain mutually invariant feature extractor are used to extract intra-domain invariant features and inter-domain mutually invariant features respectively. The intra-domain invariant feature extractor and the inter-domain mutually invariant feature extractor are composed of CNN, ReLu, and BatchNorm layers.

[0054] Specifically, such as Figure 3 As shown, the extraction of domain-invariant features is achieved using a distillation framework. In the teacher network, the fingerprint data is Fourier transformed and feature extracted to obtain invariant features that can be used for classification. Knowledge distillation is then used to learn the teacher network's feature extraction capabilities. Knowledge distillation is a simple learning framework that can encourage different networks to extract common feature information.

[0055] The teacher network takes Fourier phase information as input and generates output based on the reference point label, thereby extracting Fourier phase fingerprint features suitable for classification tasks. For single-channel two-dimensional data CSI data \mathbit{x} Fourier transform It can be expressed as:

[0056]

[0057] u and v are indices, H and W are the height and width of the CSI feature map, respectively. The Fourier transform can be efficiently calculated using the FFT algorithm. The phase component is then expressed as:

[0058]

[0059] Where R(x) and I(x) represent For multi-channel CSI feature maps, Fourier transform can be performed on each channel to extract its independent phase information.

[0060] Furthermore, the teacher network is trained using multiple multi-domain fingerprint data. The trained teacher network includes:

[0061] Performing Fourier phase transform on a number of multi-domain fingerprint data to obtain Fourier phase information;

[0062] Input the Fourier phase information into the teacher network to extract the Fourier phase fingerprint features;

[0063] The Fourier phase fingerprint features are combined with the reference point labels to train the teacher network.

[0064] Furthermore, after obtaining the Fourier phase fingerprint feature x of the original data x, it is combined with the label (x, y) to train the teacher network including:

[0065]

[0066] in, and They represent the feature extractors of the teacher network and classifier The learnable parameters, P tr represents the distribution of training data, E is the expectation, is the classification cross entropy loss function, x is the multi-domain fingerprint data, is the Fourier phase fingerprint feature, and y is the fingerprint classification label.

[0067] Furthermore, if Figure 2 As shown in Figure 2, the features extracted by the teacher network after the student network is trained through knowledge distillation technology include:

[0068] After the teacher network is trained, the feature knowledge distillation technique can be used to guide the student network to learn Fourier phase information. This process can be expressed as:

[0069]

[0070] in, They represent the feature extractors of the student network and classifier The learnable parameters, λ1 is a hyperparameter, is the MSE loss, Fingerprint classification loss for student networks.

[0071] Furthermore, the objective function of the indoor positioning model is:

[0072]

[0073] Among them, G c and G f Represent the classifier and feature extraction network respectively, and their corresponding parameters are θ c and θ f , λ1, λ2 and λ3 are hyperparameters.

[0074] Specifically, using only Fourier phase transform is not enough to obtain sufficient domain-invariant position information in fingerprint data for discrimination and classification. Therefore, this embodiment explores mutually invariant features by utilizing cross-domain knowledge contained in multiple fingerprint training domains. Given two domains This embodiment uses the correlation alignment method to align their second-order statistics (correlation):

[0075]

[0076] in, and are the covariance matrix and ||.|| F represents the Frobenius norm form of the matrix, is the mutual invariant feature learning loss, N is the dimension of the covariance matrix, n i For S i The number of domain samples, X i For S i Domain sample, i is S i Field label, n i For S j The number of domain samples, X i For S j Domain sample, j is S j Field label.

[0077] To avoid duplication and redundancy between internally invariant features and mutually invariant features, this method designs a dual fingerprint feature extractor to extract as many different invariant features as possible. Diverse invariant features are beneficial for achieving better domain generalization and positioning. To achieve this goal, this embodiment uses an additional regularization term to ensure that a certain distance is maintained between the internally invariant features (z1) and the mutually invariant features (z2):

[0078]

[0079] Among them, d(.,.) represents the distance function, z1 and z2 are the invariant features within the domain and the invariant features between domains respectively.

[0080] In general, this embodiment has two steps. First, the teacher network is trained, and then the following objective function is optimized:

[0081]

[0082] The above formula integrates four types of loss functions: classification loss Internal invariant feature learning loss Mutually Invariant Feature Learning Loss and feature diversity loss The first and third losses are two general objective losses for learning invariant features for domain generalization. Existing work has shown that these two losses alone are insufficient for accurate localization in changing environments. By leveraging internally invariant features and feature diversity, this embodiment can mitigate the impact of these issues and achieve better performance. Currently, the hyperparameters in the above formula are primarily adjusted empirically, but we plan to design more heuristic methods to automatically determine hyperparameters in the future.

[0083] Furthermore, for the determination of the user's location, the reference point can be predicted based on the online data. The positioning result is:

[0084] y=arg minG c (G f (x)).

[0085] In summary, the positioning method of this embodiment uses Fourier phase transform and knowledge distillation to extract intra-domain invariant features from each multi-domain fingerprint data set. It then uses distribution-related alignment to learn inter-domain invariant features. These features are then combined to form domain-invariant features, which are then used by a position estimator to achieve positioning under unknown environmental changes. Simultaneously, by maximizing the distance between intra-domain and inter-domain invariant features, the overlap of invariant information between the two is reduced. By concatenating these two types of features to form domain-invariant features, these features are used as position features for position estimation, achieving generalized positioning in changing environments.

[0086] Let’s take the above overall technical solution and put it into a specific application environment to illustrate:

[0087] Hardware and environment configuration of the embodiment:

[0088] This embodiment collects a fingerprint database of a small laboratory environment and uses online fingerprints before and after the environment changes for positioning. There are many obstacles in the laboratory, including desks, sofas, display cabinets, chairs and other facilities. Figure 4As shown in the figure, the laboratory area is about 4.0m×8.0m. In the whole laboratory space, there are 24 training fingerprint reference points and 15 test fingerprint reference points (the distance between adjacent reference points of the same type is 0.8 meters). Figure 5 As shown in Figures (a) and (b), the WiFi positioning signal data transceiver equipment consists of a transmitter with one omnidirectional antenna and a receiver with three omnidirectional antennas. Both the receiver and transmitter are industrial computers with built-in Intel 5300 wireless network cards. The transmitter, acting as the mobile terminal, continuously sends WiFi signal data packets at a 20MHz bandwidth. The receiver is placed in a fixed location indoors and stores the received data packets.

[0089] This embodiment sets up the five spatial conditions in Table 1 containing different environmental changes in the experimental environment. In spatial condition 1, each reference point is sampled continuously 1200 times. Then, using the method proposed in 3.2, every 30 consecutive CSI data are grouped together to form a sample image. Therefore, each reference point has 40 samples, thereby establishing a complete initial fingerprint database. At the same time, in order to simulate the lack of data in the scene spatial condition to be expanded, only 20 fingerprint samples are collected for each training reference point in scene spatial conditions 2-4, and the corresponding real fingerprint database is established. Scene spatial condition 1 corresponds to the initial fingerprint database, and scene spatial conditions 2-4 are the real fingerprint databases of the scene spatial conditions to be expanded. The CSI data is then sent to DLEI-DAM to expand the data set in scene spatial conditions 2-4. The DLEI-DAM network is based on the PyTorch framework and uses a GPU (GTX 2080Ti) for accelerated training.

[0090] Table 1

[0091]

[0092] The generalized positioning system training process is as follows:

[0093] 1. After collecting a CSI fingerprint dataset encompassing various environmental variations, a CSI amplitude matrix containing 30 subcarriers was extracted for each of the three independent antenna links. This matrix was then normalized using min-max normalization to form a single-channel feature map. Setting the time-domain sampling count to 30, the time-frequency feature map dimension for each antenna link was 30×30 (time-subcarrier dimension). Finally, the normalized amplitude matrices of the three links were mapped to the RGB channels, creating a 3×30×30 three-dimensional feature tensor map, thus constructing a fingerprint dataset encompassing various environmental variations.

[0094] 2. Use the datasets of environments 1-4 to train the teacher network using the formula The teacher network is trained, and the structure of the feature extractor in the teacher network is a multi-layer CNN network. After the teacher network training is completed, the teacher network parameters are fixed.

[0095] 3. Use the datasets from Environments 1-4 to train a domain generalized localization model based on Fourier phase information. The entire model is trained by optimizing the objective function of DGLM-FPI. After training, the model parameters are fixed. The two feature extractors consist of CNN, ReLu, and BatchNorm layers, and the classifier consists of a fully connected layer.

[0096] 4. After training the generalized positioning model, the parameters are fixed and used as the location inference model. When positioning is required, the user's Wi-Fi CSI signal is collected, converted into a fingerprint feature map, and fed into the location inference model to obtain the user's predicted location.

[0097] Example 2:

[0098] A system for implementing a WIFI fingerprint positioning method for a variable indoor environment, comprising:

[0099] Data collection module, used to collect users' online data;

[0100] The positioning module is used to input the user's online data into an indoor positioning model and output a positioning result; wherein the indoor positioning model is constructed by a feature extraction network and a classifier and is obtained by training with a plurality of multi-domain fingerprint data. The feature extraction network extracts intra-domain invariant features and inter-domain mutually invariant features respectively, and the intra-domain invariant features and inter-domain mutually invariant features are fused and input into the classifier.

[0101] Example 3:

[0102] This embodiment also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, a WIFI fingerprint positioning method for a variable indoor environment is implemented.

[0103] The embodiments described above are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by persons skilled in the art should fall within the scope of protection defined by the claims of the present invention.

Claims

1. A WIFI fingerprint positioning method for a variable indoor environment, characterized in that: include: Collect users' online data; The user's online data is input into an indoor positioning model, and a positioning result is output; wherein, the indoor positioning model is constructed by a feature extraction network and a classifier and obtained by training with a plurality of multi-domain fingerprint data, the feature extraction network extracts intra-domain invariant features and inter-domain mutually invariant features respectively, and the intra-domain invariant features and inter-domain mutually invariant features are fused and input into the classifier.

2. The WIFI fingerprint positioning method for a variable indoor environment according to claim 1, characterized in that: The feature extraction network extracts intra-domain invariant features and inter-domain invariant features respectively, including: Using the multi-domain fingerprint data to train a teacher network to obtain a trained teacher network; The student network is made to learn the extracted features of the trained teacher network through knowledge distillation technology, and the student network is trained with a plurality of multi-domain fingerprint data to obtain a domain-invariant feature extractor; The intra-domain invariant feature extractor and the inter-domain mutually invariant feature extractor are used to extract the intra-domain invariant features and the inter-domain mutually invariant features respectively, wherein the intra-domain invariant feature extractor and the inter-domain mutually invariant feature extractor are composed of CNN, ReLu, and BatchNorm layers.

3. The WIFI fingerprint positioning method for a variable indoor environment according to claim 2, characterized in that: The teacher network is trained using the multi-domain fingerprint data to obtain the trained teacher network, including: Performing Fourier phase transform on the multi-domain fingerprint data to obtain Fourier phase information; Inputting the Fourier phase information into the teacher network to extract Fourier phase fingerprint features; The Fourier phase fingerprint feature is combined with the reference point label to train the teacher network.

4. The WIFI fingerprint positioning method for a variable indoor environment according to claim 3, characterized in that: Training the teacher network includes: in, and They represent the feature extractors of the teacher network and classifier The learnable parameters, P tr represents the distribution of training data, E is the expectation, is the classification cross entropy loss function, x is the multi-domain fingerprint data, is the Fourier phase fingerprint feature, and y is the fingerprint classification label.

5. The WIFI fingerprint positioning method for a variable indoor environment according to claim 4, characterized in that: The features extracted by the teacher network after training are learned by the student network through knowledge distillation technology, including: in, They represent the feature extractors of the student network and classifier The learnable parameters, λ1 is a hyperparameter, is the MSE loss, Fingerprint classification loss for student networks.

6. The WIFI fingerprint positioning method for a variable indoor environment according to claim 5, characterized in that: The objective function of the indoor positioning model is: Among them, G c and G f Represent the classifier and feature extraction network respectively, and their corresponding parameters are θ c and θ f , λ1, λ2 and λ3 are hyperparameters, d(.,.) represents the distance function, z1, z2 are invariant features within the domain and invariant features between domains, respectively, C i and C j are the covariance matrix and ||·|| F represents the Frobenius norm form of the matrix, is the mutually invariant feature learning loss, N is the covariance matrix dimension, is the loss of feature diversity.

7. The WIFI fingerprint positioning method for a variable indoor environment according to claim 6, characterized in that: The positioning result is: y=argminG c (G f (x))。 8. A system for implementing the WIFI fingerprint positioning method for a variable indoor environment according to any one of claims 1 to 7, characterized in that: include: Data collection module, used to collect users' online data; The positioning module is used to input the user's online data into an indoor positioning model and output a positioning result; wherein the indoor positioning model is constructed by a feature extraction network and a classifier and is obtained by training with a plurality of multi-domain fingerprint data. The feature extraction network extracts intra-domain invariant features and inter-domain mutually invariant features respectively, and the intra-domain invariant features and inter-domain mutually invariant features are fused and input into the classifier.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the WIFI fingerprint positioning method for a variable indoor environment according to any one of claims 1 to 7 is implemented.

Images