Map construction method and system for automatically deploying WiFi access point information

Abstract

The invention provides an automatic deployment WiFi access point information map construction method and system. The method comprises the following steps: analyzing channel change characteristics, analyzing energy distribution of a signal on a time axis, identifying an energy concentration area, and obtaining a highest point of energy PLCR; establishing a correlation among the energy highest point PLCR, the human body movement speed and the human body position, obtaining and utilizing a model prediction framework, carrying out reverse solution on a track, fusing human body movement constraints and spatial constraints into a simulation data set, and utilizing the speed constraint in the movement constraints to derive a feasible region of the target position, so as to obtain the target position. Calculating a feature data set according to the simulation trajectory sequence, training to obtain a data-driven framework model, and processing to obtain a prediction trajectory; and carrying out similar SAGE iterative solution, and carrying out iterative optimization on the parameter estimation value by utilizing an SAGE algorithm until a preset convergence condition is met. The technical problems that large-scale deployment difficulty is large, dependence on specific antenna layout restricts system adaptability, and modeling and positioning operation precision is low are solved.

Description

Technical Field

[0001] The present invention relates to the field of WiFi access point information map construction in indoor environments, and in particular to a method and system for automatically deploying WiFi access point information map construction. Background Art

[0002] The Internet of Things (IoT) technology has made significant progress in recent years, greatly improving the intelligence and convenience of our lives. Whether it is family life or industrial production, the Internet of Things is changing our daily lives and work patterns in a revolutionary way. The continuous innovation and miniaturization trend of sensor technology have made IoT devices more compact and sophisticated, and can be easily integrated into various devices and environments. This high degree of connectivity allows real-time communication and collaborative operation between devices, which significantly improves work efficiency and intelligence. In the past decade, WiFi device-free positioning technology has experienced rapid development. Since this technology does not require users to carry specific devices, it has shown great potential in multiple application scenarios, such as medical monitoring and home automation. The core idea of ​​device-free positioning technology is to capture and analyze the signals reflected by the human body by analyzing the channel state information (CSI), and then extract motion-related characteristics to achieve target tracking. In order to deeply understand the basic principles of device-free perception and positioning, researchers have introduced a variety of theoretical models, including the well-known Fresnel zone model and hyperbolic zone model.

[0003] For example, the existing invention patent application document "Indoor map establishment system conducive to timely update" with publication number CN107016117A, the existing method includes: a two-dimensional code identification, a handheld terminal and a data processing center; the two-dimensional code identification is a two-dimensional code generated by a two-dimensional code generator using the user-end indoor map update webpage as an information source; the handheld terminal includes a two-dimensional code recognition module, an input module, a WiFi positioning module and an information sending module; the data processing center includes a map construction module, a storage module, an information receiving module, and an analysis and processing module for comparing the information stored in the storage module with the information received by the information receiving module, and for the map construction module to update the indoor map. The indoor map establishment system provided by the present invention facilitates timely updating of indoor maps. And the existing invention patent application document "Contactless motion recognition method, system and laboratory device based on WiFi" with publication number CN116304915A, the existing method includes: obtaining channel state information data through WiFi equipment; preprocessing the received channel state information data, eliminating the noise information caused by machine operation and environmental interference, and selecting subcarriers for the channel state information data of multiple subcarriers between each antenna link using a dynamic subcarrier selection algorithm; making dynamic and static judgments on the acquired channel state information data, and cutting out the action interval; extracting features from the action interval; training the extracted features, and identifying and classifying the action based on the training model. Although the aforementioned WiFi positioning technology has achieved a high level of accuracy, its operating principle is mainly to determine the relative spatial position based on WiFi equipment, which requires a relatively accurate pre-setting of the device position. This shows that in the deployment process of the existing system, it is still necessary to rely on manual calibration of access points, which undoubtedly increases the difficulty of large-scale deployment. With the popularization of smart homes and the advent of the Internet of Things era, the number of Internet of Things devices connected to WiFi in home environments is expected to continue to grow. Considering that smart devices are usually designed to be lightweight and easy to move, coupled with the unpredictability of user behavior and the variability of the home environment, these may lead to frequent changes in the position of the device, thereby causing deviations in the perceived space. If the calibration process requires user participation, this will increase the difficulty of use and limit the adaptability of the perception system. Therefore, it is particularly important to develop an automated calibration technology. Exploring a system that can automatically adapt to environmental changes and achieve higher-precision positioning without user intervention will be a far-reaching challenge. Implementing such a method can not only reduce the deployment and maintenance costs of the system, but also improve the stability of the system and the user's operating experience, laying a solid foundation for the advancement of intelligent perception technology.

[0004] At present, the research on WiFi sensing technology is mostly focused on enhancing its functionality and improving the accuracy of positioning, while the research on practical application problems is relatively lacking. First, the core task of WiFi sensing technology is to analyze the signal changes caused by user activities and effectively model them. However, since the change of device position is often short-lived and instantaneous, the captured signal change characteristics are relatively weak, so it is difficult to accurately model such changes. Secondly, the appearance and structure of home smart devices are different, resulting in the diversity of their antenna layouts. The perception system that relies on a specific antenna layout is difficult to meet the perception needs of different devices, so there are certain application limitations. In view of these challenges, a method of using mobile robots to assist perception is proposed to enhance the perception ability of the scene. Although this method has made some progress in improving positioning accuracy, considering the cost and the wide applicability of indoor environments, this solution still faces many difficulties in commercial promotion and actual deployment.

[0005] In summary, the existing technology has technical problems such as difficulty in large-scale deployment, dependence on specific antenna layout that restricts system adaptability, and low accuracy in modeling and positioning operations. Summary of the invention

[0006] The technical problem to be solved by the present invention is: how to solve the technical problems in the prior art that large-scale deployment is difficult, the dependence on a specific antenna layout restricts the adaptability of the system, and the modeling and positioning operation accuracy is low.

[0007] The present invention adopts the following technical solutions to solve the above technical problems: the method for constructing an automatic deployment WiFi access point information map comprises:

[0008] S1. Perform high-pass and low-pass filtering on the signal to extract information related to human motion; calculate the ratio of signals received by different antennas to perform phase offset correction on the original CSI signal; obtain human motion signal change data through time-frequency analysis, use the STFT method to analyze the signal, analyze the energy distribution of the signal on the time axis, identify the energy concentration area, and obtain the highest energy point PLCR;

[0009] S2. Establish the correlation between the highest energy point PLCR, human body movement speed and human body position, obtain and use the model prediction framework, reversely solve the trajectory, integrate human body movement constraints and space constraints into the simulation data set, use the speed constraints in the movement constraints to derive the feasible domain of the target position, generate a preset number of simulation trajectory sequences to cover the feasible domain, calculate the simulated PLCR sequence according to the simulation trajectory sequence to obtain the feature data set, train the data-driven framework model based on it, and process it to obtain the predicted trajectory;

[0010] S3. Fix the current parameter estimate θ(t) , assuming it to be the device position; calculate the conditional expectation of the latent variable Z relative to the observed data X, perform the expectation solving operation, obtain the latent variable, update the parameter estimation value according to the expected value of the latent variable, alternately perform the expectation solving operation and the maximization operation, and use the SAGE algorithm to iteratively optimize the parameter estimation value until the preset convergence condition is met.

[0011] The present invention automatically constructs a WiFi device map, based on the characteristic changes of CSI signals in indoor environments, relies on the reverse reasoning effect of collaborative sensing activities on the spatial deployment of devices, constructs a reverse solution model, and uses the excellent feature matching performance of spatial alternating generalized expectation maximization theory and the excellent batch processing performance of neural networks to achieve rapid construction and automatic deployment of WiFi access point information maps in indoor environments. The present invention utilizes collaborative sensing activities in indoor environments and reversely reasoning about device spatial attributes to achieve accurate positioning of device locations.

[0012] In a more specific technical solution, in S1, in order to quantify the change information in the aforementioned signal, the highest energy point PLCR is associated with the data to quantify the change of the human motion signal:

[0013]

[0014] In the formula, f D is the Doppler shift of the signal, corresponding to the part with the highest energy in the spectrum, represents the PLCR of the nth link, and λ represents the wavelength of the transmitted signal.

[0015] In a more specific technical solution, S2 includes:

[0016] S21. According to Fresnel theory, referring to the change of reflection path length, the correlation between the highest energy point PLCR, human body movement speed and human body position is established:

[0017]

[0018] In the formula, and Indicates the user speed (v x ,v y ) on the reflection path.

[0019] S22, obtaining a model prediction framework, using the model prediction framework to perform batch processing operations, and performing reverse solution of preset batch trajectories;

[0020] S23, performing a method for self-generating a simulation data set, integrating human motion constraints and space constraints into the simulation data set, and using speed constraints in the motion constraints to derive a feasible domain of the target position;

[0021] S24. Randomly set the initial position of the track in the preset area:

[0022]

[0023] S25. Generate a preset number of simulation trajectory sequences according to formula (2) to cover the feasible domain, and calculate the simulated PLCR sequence according to the simulation trajectory sequence and the equipment position to serve as the feature data set D e ;

[0024] S26. Design a neural network architecture and use the neural network structure to perform fitting operations, wherein the neural network architecture includes: an LSTM module; using the LSTM module as the core layer of the neural network architecture, and using the LSTM module to learn the mapping relationship between the user position and the highest energy point PLCR.

[0025] S27, the highest energy point PLCR and the currently retrieved initial position Input into the data-driven tracking model and predict the predicted trajectory.

[0026] The present invention utilizes the excellent data batch processing performance of neural networks and can rely on neural networks to achieve fine-grained spatial retrieval. As the spatial grid is refined, the computing power resources consumed by the device for solving increases exponentially. While ensuring the accuracy of the solution, it solves the problem that the granularity of spatial traversal is large and the computing power resources of the device cannot meet the demand.

[0027] In a more specific technical solution, in S231, the motion constraint is expressed using the following logic:

[0028]

[0029] In the formula, It represents the coordinate point of the simulation trajectory at time t, and Δt represents the simulation sampling time interval, which is consistent with the STFT sliding window time interval. and Represent the simulation speed and simulation acceleration respectively;

[0030] Use the following logic to express spatial constraints:

[0031]

[0032] In the formula, M allow Indicates the feasible area indoors, which can be flexibly adjusted when deploying a new environment.

[0033] In a more specific technical solution, in S25, the simulation trajectory sequence is used as the label data set T e , where the following logic is used, based on the label data set T e, processed to obtain the feature data set:

[0034]

[0035] Wherein, t represents time.

[0036] In a more specific technical solution, in S27, the following logic is used to process the predicted trajectory:

[0037]

[0038] In the formula, F LSTM (·) represents the trained data-driven tracking framework model, Indicates the actual collected PLCR.

[0039] The present invention trains the twin neural network by self-generating data sets, which not only improves the training efficiency but also significantly improves the performance of the network. In addition, the framework has good scalability and can be extended to assist in large-scale spatial retrieval.

[0040] In a more specific technical solution, S3 includes:

[0041] S31, perform the expected solution operation and fix the current parameter estimate θ (t) , the parameter estimate θ (t) Set as the device position; calculate the conditional expectation of the latent variable Z relative to the observed data X; according to the conditional expectation, under the conditions of the current observed data and the current parameter estimation, the corresponding latent variable values ​​are weighted averaged to obtain the expected log-likelihood function Q θ (Z):

[0042]

[0043] Where L(θ) is the complete log-likelihood function and θ is the parameter we wish to estimate.

[0044] S32, perform maximization operation; update parameter estimation using conditional expectation;

[0045] S33, by alternately executing S31 and step S32, using the SAGE algorithm to tune the parameter estimation value θ (t) , until the preset convergence condition is met.

[0046] The present invention makes up for the deficiency of automatic deployment of access point information map in existing smart home solutions, and realizes high-precision and fast access point information map construction through convergent matching of channel characteristics and data-driven spatial fine-grained spatial retrieval.

[0047] In a more specific technical solution, in S32, the expected log-likelihood function Q θ (Z) solves the maximum value and finds the expected log-likelihood function Q θ (Z) Obtain the parameter θ under the maximum value:

[0048] θ t+1 = arg max θ Q θ (Z), (8).

[0049] In a more specific technical solution, in S33, the parameter estimation value θ is tuned using the following logic: (t) :

[0050]

[0051] In the formula, θ (t) is the parameter estimate for the tth iteration.

[0052] The present invention uses a new parameter extraction method to obtain accurate PLCR solution. PLCR is a relatively accurate channel change feature that can be obtained under existing technical conditions. The present invention solves the access point information based on this feature. Therefore, the accurate solution of PLCR feature is the key to the effectiveness of the algorithm.

[0053] In a more specific technical solution, the automated deployment of the WiFi access point information map construction system includes:

[0054] The channel change feature analysis module is used to perform high-pass and low-pass filtering on the signal to extract information related to human motion; calculate the ratio of received signals from different antennas to perform phase offset correction on the original CSI signal; obtain human motion signal change data through time-frequency analysis, use the STFT method to analyze the signal, analyze the energy distribution of the signal on the time axis, identify the energy concentration area, and obtain the highest energy point PLCR;

[0055] The data-driven framework training module is used to establish the correlation between the highest energy point PLCR, human body movement speed and human body position, obtain and use the model prediction framework, reversely solve the trajectory, integrate human body movement constraints and space constraints into the simulation data set, use the speed constraints in the movement constraints to derive the feasible domain of the target position, generate a preset number of simulation trajectory sequences to cover the feasible domain, calculate the simulated PLCR sequence based on the simulation trajectory sequence to obtain the feature data set, and train the data-driven framework model based on it, and process it to obtain the predicted trajectory. The data-driven framework training module is connected to the channel change feature analysis module;

[0056] SAGE-like iterative solution module to fix the current parameter estimate θ (t), assuming it to be the device position; calculate the conditional expectation of the latent variable Z relative to the observed data X, perform the expectation solving operation, obtain the latent variable, update the parameter estimation value according to the expected value of the latent variable, alternately perform the expectation solving operation and the maximization operation, and use the SAGE algorithm to iteratively optimize the parameter estimation value until the preset convergence condition is met. The SAGE-like iterative solution module is connected to the data-driven framework training module.

[0057] Compared with the prior art, the present invention has the following advantages:

[0058] The present invention automatically constructs a WiFi device map, based on the characteristic changes of CSI signals in indoor environments, relies on the reverse reasoning effect of collaborative sensing activities on the spatial deployment of devices, constructs a reverse solution model, and uses the excellent feature matching performance of spatial alternating generalized expectation maximization theory and the excellent batch processing performance of neural networks to achieve rapid construction and automatic deployment of WiFi access point information maps in indoor environments. The present invention utilizes collaborative sensing activities in indoor environments and reversely reasoning about device spatial attributes to achieve accurate positioning of device locations.

[0059] The present invention utilizes the excellent data batch processing performance of neural networks and can rely on neural networks to achieve fine-grained spatial retrieval. As the spatial grid is refined, the computing power resources consumed by the device for solving increases exponentially. While ensuring the accuracy of the solution, it solves the problem that the granularity of spatial traversal is large and the computing power resources of the device cannot meet the demand.

[0060] The present invention trains the twin neural network by self-generating data sets, which not only improves the training efficiency but also significantly improves the performance of the network. In addition, the framework has good scalability and can be extended to assist in large-scale spatial retrieval.

[0061] The present invention makes up for the deficiency of automatic deployment of access point information map in existing smart home solutions, and realizes high-precision and fast access point information map construction through convergent matching of channel characteristics and data-driven spatial fine-grained spatial retrieval.

[0062] The present invention uses a new parameter extraction method to obtain accurate PLCR solution. PLCR is a relatively accurate channel change feature that can be obtained under existing technical conditions. The present invention solves the access point information based on this feature. Therefore, the accurate solution of PLCR feature is the key to the effectiveness of the algorithm.

[0063] The present invention solves the technical problems existing in the prior art, such as the difficulty of large-scale deployment, the restriction of system adaptability due to the dependence on a specific antenna layout, and the low accuracy of modeling and positioning operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1This is a schematic diagram of basic steps of a method for constructing an automatic deployment WiFi access point information map according to Embodiment 1 of the present invention;

[0065] Figure 2 This is a data flow processing diagram of a method for constructing an automatic deployment WiFi access point information map according to Embodiment 1 of the present invention;

[0066] Figure 3 A schematic diagram of feasibility analysis of automatic solution of access point information map according to Embodiment 1 of the present invention;

[0067] Figure 4 This is a schematic diagram of a signal feature analysis data flow in Example 1 of the present invention;

[0068] Figure 5 Schematic diagram of the Fresnel zone model and velocity mapping of Example 1 of the present invention;

[0069] Figure 6 Schematic diagram of the specific steps of the SAGE-like iterative solution of Example 1 of the present invention. DETAILED DESCRIPTION

[0070] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in combination with the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0071] Example 1

[0072] In this embodiment, in order to ensure the wide applicability of the experimental environment, the line-of-sight environment is selected as the test environment, and the Intel 5300 wireless network card commonly used in the market is used to implement the system construction. In order to meet the basic requirements of the system operation, a configuration scheme consisting of a transmitter and two receivers is designed. All relevant equipment is placed at a height of 0.8 meters from the ground to simulate the typical height of home automation equipment in actual applications. At the transmitter, a single antenna is used to perform the task of transmitting data packets; and at the receiver, a triple antenna array is used to capture and decode CSI. The transmission and reception programs are implemented based on the open source CSITool software library. Considering the reduction of interference from surrounding WiFi signals, the 5GHz frequency band is selected for signal tracking. In this prototype system, the transmitter sends data packets at a frequency of 1000 times per second. After the data is collected, a portable computer is used for data processing to calculate the user's movement trajectory. Matlab was selected as the programming language during the entire system development process.

[0073] like Figure 1 and Figure 2 As shown, the method for constructing an automatic deployment WiFi access point information map provided by the present invention includes the following basic steps:

[0074] S1, channel change characteristics analysis;

[0075] like Figure 3 As shown, in this embodiment, the accurate location solution for the existing perception theory depends on the accurate analysis of the channel change characteristics. Due to the influence of various factors such as environmental interference and equipment disturbance, it is difficult to directly obtain the available location-related features from the original CSI signal. The present invention analyzes the wireless sensing part and focuses on the dynamic part caused by human body movement. Specifically, the CSI signal includes: a dynamic part and a static part. In order to achieve the above goals, we have developed a set of feature extraction methods.

[0076] like Figure 4 As shown, in this embodiment, in order to extract the key information related to human body movement in the signal, the signal is first subjected to high-pass and low-pass filtering, and the analysis focus is placed on the signal changes caused by human activities; since the original CSI signal is easily affected by phase offset noise, an elimination technique is used, that is, the ratio of signals received by different antennas is calculated to correct the error caused by the phase offset; when the signal passes through the human body, significant power changes will occur due to the scattering characteristics of the human body; time-frequency analysis technology is used to track these signal changes caused by human body movement. Specifically, the STFT method is used to analyze the signal, and by analyzing the energy distribution of the signal on the time axis, the area with the most concentrated energy is identified. In order to quantify the change information in the aforementioned signal, the highest energy point PLCR obtained in the STFT analysis is correlated, and the formula is as follows:

[0077]

[0078] Among them, f D is the Doppler shift of the signal, corresponding to the part with the highest energy in the spectrum, represents the PLCR of the nth link, and λ represents the wavelength of the transmitted signal.

[0079] In the process of wireless signal transmission and reception, the human body will have a significant impact on the propagation path of the signal, mainly manifested in the absorption and scattering of the signal. This impact is manifested as a change in power distribution in the time-frequency spectrum, especially when the signal undergoes Doppler shift due to human activity. In order to accurately estimate the Doppler shift and effectively compensate for it, in this embodiment, the area with the highest power peak is identified and selected in the time-frequency spectrum as the focus of analysis.

[0080] S2, data-driven framework training;

[0081] The existing tracking perception includes: user behavior related information, device location related information; in this embodiment, taking the Fresnel zone as an example, see Figure 5 , PLCR is affected by the projection of the user's speed on the two reflection paths. As the length of the reflection path changes, the moving user can be positioned on a series of ellipses with the same focus. By analyzing this process, the correlation between the three physical parameters of PLCR, user speed and user position can be quantitatively established, as shown below:

[0082]

[0083] in, and Indicates the user speed (v x ,v y ) on the reflection path.

[0084] Although the spatial retrieval can be achieved theoretically by relying on the channel features extracted from formula (2) and the aforementioned step S1, under fine-grained conditions, the computational resources required for trajectory reverse solution are too large to be deployed in a real environment. Therefore, the main purpose of this embodiment is to obtain a model prediction framework and complete large-scale trajectory reverse solution with the help of batch processing performance.

[0085] Wireless sensing is based on the collection and analysis of wireless communication signals. By capturing, processing and interpreting wireless signals, it enables the system to understand and interpret various events, entities or behaviors in the surrounding environment. These signals may come from wireless transmission tools such as WiFi routers, Bluetooth devices, and radio frequency identification tags. The research and application of wireless sensing technology are extensive, and researchers often use it to develop a variety of high-tech solutions, such as accurate wireless navigation, monitoring of breathing rhythm, and the construction of smart home systems. Through in-depth analysis and effective use of wireless signals, the relevant system can extract key information to facilitate effective monitoring and in-depth analysis of individual behavior patterns.

[0086] WiFi wireless sensing is an emerging field of wireless environmental monitoring. In indoor environments, WiFi routers not only provide network access, but also become multifunctional due to the environmental information contained in their signals. The properties of these signals, including signal strength, dynamic changes in path loss, and signal phase, provide a rich data source for monitoring environmental changes. With this data, people and objects can be identified and tracked, and even accurately located without the need for additional monitoring hardware. A significant advantage of WiFi sensing technology is its reuse of existing network facilities, which reduces the need for additional hardware, reduces costs, and simplifies the deployment process. It improves the flexibility and economy of the system by reducing the reliance on dedicated sensors. In addition, WiFi sensing technology has demonstrated its wide applicability in application scenarios such as indoor navigation and smart homes, providing an innovative solution for the design and implementation of intelligent systems.

[0087] The performance and effectiveness of neural network models depend largely on the quality of the training data set. A high-quality data set should have the following characteristics: (1) The data set contains a large enough amount of data to ensure that the neural network can fully learn the distribution pattern of the data. (2) The data set is rich and diverse, which can ensure that the neural network can adapt to different scenarios and needs. (3) The annotation format of the data set is consistent to ensure that the neural network can understand the meaning of the data. However, the data set in the field of WiFi wireless sensing does not show the above three characteristics. The existing open source data set samples are relatively small and only target specific scenarios. When the device position changes, the system will be completely unavailable. Therefore, the present invention designs a method for self-generation of simulation data sets, which incorporates human motion constraints and spatial constraints into the data set.

[0088] In this embodiment, motion constraints are introduced into the data set; specifically, for human motion in indoor scenes, the acceleration of the real human motion will be within a certain range, and the probability of turning when the target moves will tend to stabilize over time. Based on the mathematical and physical relationship between speed and acceleration, it can be inferred that the simulated speed of human motion will change within a feasible domain, and there will be no sudden and abnormal changes. Based on the relationship between speed and position, we can use the speed constraint to derive the feasible domain of the target position. The above motion constraints can be expressed as follows:

[0089]

[0090] in, It represents the coordinate point of the simulation trajectory at time t, and Δt represents the simulation sampling time interval, which is consistent with the STFT sliding window time interval. and They represent the simulated speed and simulated acceleration respectively. Since there is usually no intense exercise in indoor environments, the acceleration is usually less than 2m / s2 , the speed is usually less than 3m / s. Under simulation conditions, we assign a value of 0.05 to the probability of a turning event. α(t) and β(t) represent the angles of the speed direction and the acceleration direction relative to the reference coordinate axis, respectively;

[0091] Access point information map construction refers to the creation of a map that can display the geographical location and related information of network access points. These access points may include WiFi hotspots, mobile communication base stations, access points of Internet service providers, etc. In the field of geographic information systems, such maps can help network operators, urban planners and researchers understand network coverage, optimize network layout, improve service quality, and conduct network-related planning and analysis. In the field of wireless sensing, access point information is an important prior parameter of the sensing algorithm.

[0092] In this embodiment, spatial constraints are added to the simulation data set. These constraints mainly come from the actual size of the indoor space and the position of the object. The formula is as follows:

[0093]

[0094] Among them, allow Indicates the feasible area indoors, which can be flexibly adjusted when deploying a new environment;

[0095] In this embodiment, the initial positions of a series of trajectories are randomly set in the area Then, according to the above formula (2), a sufficient number of simulation trajectory sequences are generated to ensure coverage of the entire feasible area. The coordinates of these trajectory devices are used to calculate the simulated PLCR sequence as the feature data set D e ;

[0096] The Path Length Change Rate (PLCR) is a key parameter in the field of wireless signal propagation. It measures how fast the length of the signal propagation path changes over time during the reflection process. In complex propagation environments, such as indoors or in urban areas, the reflection of signals causes the path length to change continuously, which poses a challenge to the positioning and tracking of signals. This rate of change not only reflects the dynamics of the signal propagation environment, but is also important for accurately estimating the target position. In the positioning system, by monitoring and analyzing the rate of change of the reflection path length, the dynamic behavior of the target can be captured more accurately, thereby improving the positioning accuracy. In addition, the rate of change of the reflection path length is also closely related to the multipath effect and signal attenuation of the signal, which together affect the performance of the wireless system.

[0097] In addition, this embodiment takes into account that the above simulation process adopts a supervised learning method and uses the corresponding trajectory sequence as the label data set Te Specifically, the training data set can be expressed as follows:

[0098]

[0099] Wherein, t represents time. After the above processing, a training data set can be obtained. After testing, the present invention can generate 150,000 simulation data in 7 seconds. Using these data as inputs for the training process, we can obtain the corresponding solution function represented by the neural network model. For a new scenario, model training can be completed in a few minutes without any additional manual labor.

[0100] After solving the problem of data set, we need to design the core part of neural network, that is, the design architecture of neural network. Our method is to use theoretical model to guide neural network. The overall design idea is to focus on theoretical model and less on neural network, so that neural network only plays a fitting role.

[0101] In this embodiment, considering the good prediction properties of LSTM for time series, a layer of LSTM is used as the core layer of the network, and the LSTM neural network is used to learn the mapping relationship between the user location and the DPLCR. Input into the designed data-driven tracking framework to obtain the final predicted trajectory. It can be expressed as follows:

[0102]

[0103] Among them, F LSTM (·) represents the trained data-driven tracking framework model, Indicates the actual collected PLCR.

[0104] S3, SAGE-like iterative solution;

[0105] like Figure 6 As shown, in this embodiment, the mathematical expression of the SAGE-like algorithm includes the following specific steps:

[0106] S31, performing an expected solution operation;

[0107] In this embodiment, the current parameter estimate θ is fixed (t) , assuming it is the device position; calculate the conditional expectation of the latent variable Z relative to the observed data X. Specifically, the aforementioned conditional expectation reflects the weighted average of the possible values ​​of the latent variable under the conditions of given observed data and current parameter estimates.

[0108] The data formula is as follows:

[0109] Among them, L(θ) is the complete log-likelihood function. Here, the log-likelihood function is only used as a reference and can be replaced according to the needs of the algorithm. It depends on the observed data X and the latent variable Z, and θ is the parameter you want to estimate. The goal of this step is to find Q θ (Z) is the calculation expression.

[0110] The spatial alternating generalized expectation-maximization algorithm is an iterative estimation method for processing spatial data and spatial statistical models. The algorithm combines the generalized expectation-maximization framework and the optimization strategy of the alternating direction multiplier method, and is particularly suitable for solving high-dimensional data problems with complex spatial correlation. A key advantage of the spatial alternating generalized expectation-maximization algorithm is its flexibility and scalability. The spatial alternating generalized expectation-maximization algorithm can easily adapt to a variety of complex spatial statistical models, including but not limited to: spatial autocorrelation models, spatial error models, and spatial panel data models.

[0111] S32, performing a maximization operation;

[0112] In this embodiment, the expected value of the latent variable obtained in the expectation solution step is used to update the parameter estimate. Specifically, the expected log-likelihood function Q obtained in the expectation solution step is updated. θ (Z) is solved for the maximum value, which can be done analytically or by numerical optimization techniques. The goal of the maximization step is to find Q θ (Z) Obtain the parameter θ under the maximum value. The specific mathematical expression is as follows: θ t+1 = arg max θ Q θ (Z), (8)

[0113] In the actual operation of this embodiment, this step S32 involves a complex optimization problem, especially when the number of model parameters and hidden variables is large. Therefore, it may be necessary to use an approximate method or a heuristic algorithm to find the optimal solution. By alternating between the expectation solution step and the maximization step, the SAGE algorithm gradually improves the parameter estimate until convergence is satisfied. Each iteration aims to optimize the value of the target estimation function, thereby obtaining a better parameter estimate. This process can be summarized as

[0114] for:

[0115] Among them, θ (t) is the parameter estimate for the tth iteration.

[0116] According to the above description, the specific implementation process of the class space alternating generalized expectation maximization algorithm proposed in this paper can be expressed as follows: set the upper limit of iteration rounds θ, the convergence threshold of iteration error Error ζ, and generate a dedicated data-driven prediction model F according to the simulation data set LSTM (·), taking the scenario of one transmitter and two receivers as an example, the DFS sequence parsed by the two receivers The device location for which location information is requested is set to The pseudo code of the algorithm can be expressed as follows:

[0117]

[0118] The core of the present invention is to use commercial WiFi equipment available on the market, without additional hardware investment, to achieve system deployment, greatly reducing the cost of manual operation and maintenance. The advantages of the present invention are its excellent flexibility and adaptability to different home environments, and the ability to build customized information maps according to the specific scenarios of smart homes, meeting users' needs for personalization and high efficiency.

[0119] In summary, the present invention automatically constructs a WiFi device map, based on the characteristic changes of CSI signals in indoor environments, relies on the reverse reasoning effect of collaborative sensing activities on the spatial deployment of devices, constructs a reverse solution model, and uses the excellent feature matching performance of spatial alternating generalized expectation maximization theory and the excellent batch processing performance of neural networks to achieve rapid construction and automatic deployment of WiFi access point information maps in indoor environments. The present invention utilizes collaborative sensing activities in indoor environments and reversely infers the spatial properties of devices to achieve accurate positioning of device locations.

[0120] The present invention utilizes the excellent data batch processing performance of neural networks and can rely on neural networks to achieve fine-grained spatial retrieval. As the spatial grid is refined, the computing power resources consumed by the device for solving increases exponentially. While ensuring the accuracy of the solution, it solves the problem that the granularity of spatial traversal is large and the computing power resources of the device cannot meet the demand.

[0121] The present invention trains the twin neural network by self-generating data sets, which not only improves the training efficiency but also significantly improves the performance of the network. In addition, the framework has good scalability and can be extended to assist in large-scale spatial retrieval.

[0122] The present invention makes up for the deficiency of automatic deployment of access point information map in existing smart home solutions, and realizes high-precision and fast access point information map construction through convergent matching of channel characteristics and data-driven spatial fine-grained spatial retrieval.

[0123] The present invention uses a new parameter extraction method to obtain accurate PLCR solution. PLCR is a relatively accurate channel change feature that can be obtained under existing technical conditions. The present invention solves the access point information based on this feature. Therefore, the accurate solution of PLCR feature is the key to the effectiveness of the algorithm.

[0124] The present invention solves the technical problems existing in the prior art, such as the difficulty of large-scale deployment, the restriction of system adaptability due to the dependence on a specific antenna layout, and the low accuracy of modeling and positioning operations.

[0125] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for constructing a WiFi access point information map by automatically deploying the WiFi access point information map, characterized in that: The method comprises: S1. Perform high-pass and low-pass filtering on the signal to extract information related to human motion; calculate the ratio of signals received by different antennas to perform phase offset correction on the original CSI signal; obtain human motion signal change data through time-frequency analysis, use the STFT method to analyze the signal, analyze the energy distribution of the signal on the time axis, identify the energy concentration area, and obtain the highest energy point PLCR; S2, establish the correlation between the highest energy point PLCR, human body movement speed and human body position, obtain and use the model prediction framework, reversely solve the trajectory, integrate human body movement constraints and space constraints into the simulation data set, use the speed constraints in the movement constraints to derive the feasible domain of the target position, generate a preset number of simulation trajectory sequences to cover the feasible domain, calculate the simulated PLCR sequence according to the simulation trajectory sequence to obtain the feature data set, train the data-driven framework model based on it, and process it to obtain the predicted trajectory; S3. Fix the current parameter estimate θ (t) , assuming it to be the device position; calculate the conditional expectation of the latent variable Z relative to the observed data X, perform the expectation solving operation, obtain the latent variable, update the parameter estimation value according to the expected value of the latent variable, alternately perform the expectation solving operation and the maximization operation, and use the SAGE algorithm to iteratively optimize the parameter estimation value until the preset convergence condition is met.

2. The method for constructing an automated deployment WiFi access point information map according to claim 1, characterized in that: In S1, in order to quantify the change information in the aforementioned signal, the energy highest point PLCR is associated with the energy highest point PLCR to quantify the change data of the human motion signal: In the formula, f D is the Doppler shift of the signal, corresponding to the part with the highest energy in the spectrum, represents the PLCR of the nth link, and λ represents the wavelength of the transmitted signal.

3. The method for constructing an automated deployment WiFi access point information map according to claim 1, characterized in that: The S2 includes: S21. According to the Fresnel theory, with reference to the change of the reflection path length, a correlation between the energy highest point PLCR, the human body movement speed and the human body position is established: In the formula, and Indicates the user speed (v x ,v y ) on the reflection path. S22, obtaining a model prediction framework, using the model prediction framework to perform batch processing operations, and performing reverse solution of preset batch trajectories; S23, performing a method for self-generating a simulation data set, integrating human motion constraints and space constraints into the simulation data set, and using speed constraints in the motion constraints to derive a feasible domain of a target position; S24. Randomly set the initial position of the track in the preset area: S25, generating a preset number of simulation trajectory sequences according to the formula (2) to cover the feasible domain, and calculating a simulated PLCR sequence according to the simulation trajectory sequence and the equipment position to serve as the feature data set D e ; S26. Design a neural network architecture, and use the neural network structure to perform fitting operations, wherein the neural network architecture includes: an LSTM module; using the LSTM module as the core layer of the neural network architecture, and using the LSTM module to learn the mapping relationship between the user position and the highest energy point PLCR. S27, the energy highest point PLCR and the currently retrieved initial position The data is input into the data driven tracking model to predict the predicted trajectory.

4. The method for constructing an automated deployment WiFi access point information map according to claim 3, characterized in that: In S231, the motion constraint is expressed using the following logic: In the formula, It represents the coordinate point of the simulation trajectory at time t, and Δt represents the simulation sampling time interval, which is consistent with the STFT sliding window time interval. and Represent the simulation speed and simulation acceleration respectively; The spatial constraints are expressed using the following logic: In the formula, M allow Indicates the feasible area indoors, which can be flexibly adjusted when deploying a new environment.

5. The method for constructing an automated deployment WiFi access point information map according to claim 3, characterized in that: In S25, the simulation trajectory sequence is used as a label data set T e , wherein, using the following logic, according to the label data set T e , and process to obtain the feature data set: Wherein, t represents time.

6. The method for constructing an automated deployment WiFi access point information map according to claim 3, characterized in that: In S27, the following logic is used to process and obtain the predicted trajectory. In the formula, F LSTM (·) represents the trained data-driven tracking framework model, Indicates the actual collected PLCR.

7. The method for constructing an automated deployment WiFi access point information map according to claim 1, characterized in that: The S3 includes: S31, perform the expected solution operation, fix the current parameter estimation value θ (t) , the parameter estimate θ (t) Set as the device position; calculate the conditional expectation of the latent variable Z relative to the observed data X; according to the conditional expectation, under the conditions of the current observed data and the current parameter estimation, the corresponding value of the latent variable is weighted averaged to obtain the expected log-likelihood function Q θ (Z): Where L(θ) is the complete log-likelihood function and θ is the parameter we wish to estimate. S32, performing the maximization operation; updating the parameter estimation using the conditional expectation; S33, by alternately executing the step S31 and the step S32, using the SAGE algorithm to tune the parameter estimation value θ (t) , until the preset convergence condition is met.

8. The method for constructing an automated deployment WiFi access point information map according to claim 7, characterized in that: In S32, the expected log-likelihood function Q θ (Z) Perform the maximum value solution to find the expected log-likelihood function Q θ (Z) Obtain the parameter θ under the maximum value: θ t+1 =argmax θ Q θ (Z), (8)。 9. The method for constructing an automated deployment WiFi access point information map according to claim 7, characterized in that: In S33, the parameter estimation value θ is tuned using the following logic: (t) : In the formula, θ (t) is the parameter estimate for the tth iteration.

10. Automated deployment of WiFi access point information map construction system, characterized in that: The system comprises: The channel change feature analysis module is used to perform high-pass and low-pass filtering on the signal to extract information related to human motion; calculate the ratio of signals received by different antennas to perform phase offset correction on the original CSI signal; obtain human motion signal change data through time-frequency analysis, use the STFT method to analyze the signal, analyze the energy distribution of the signal on the time axis, identify the energy concentration area, and obtain the highest energy point PLCR; A data-driven framework training module is used to establish the correlation between the energy highest point PLCR, human body movement speed and human body position, obtain and use the model prediction framework, reversely solve the trajectory, integrate human body movement constraints and space constraints into the simulation data set, use the speed constraints in the movement constraints to derive the feasible domain of the target position, generate a preset number of simulation trajectory sequences to cover the feasible domain, calculate the simulated PLCR sequence according to the simulation trajectory sequence to obtain a feature data set, train the data-driven framework model based on it, and process to obtain the predicted trajectory. The data-driven framework training module is connected to the channel change feature analysis module; SAGE-like iterative solution module to fix the current parameter estimate θ (t) , assuming it to be the device position; calculating the conditional expectation of the latent variable Z relative to the observed data X, performing the expectation solving operation to obtain the latent variable, updating the parameter estimation value according to the expected value of the latent variable, alternately performing the expectation solving operation and the maximization operation, using the SAGE algorithm to iteratively optimize the parameter estimation value until the preset convergence condition is met, and the SAGE-like iterative solving module is connected to the data-driven framework training module.

Images