CN118900397A - A feature compensation and passive positioning method and system - Google Patents

Abstract

The invention provides a feature compensation and passive positioning method and a system, wherein the method comprises the following steps: performing signal characteristic processing operation, and extracting WiFi signal characteristics by analyzing channel state information to obtain PLCR matrix through processing; processing to obtain DFS data according to the change rate PLCR of the reflection path, deriving an observation value-based matrix P ₁ and a reliability-based matrix R according to the PLCR matrix, and distributing weights for various types of predictions for combined prediction operation; calculating predictions of all types, and obtaining a final PLCR prediction by utilizing weight combination, and processing to obtain an applicable PLCR predicted value and a user track; and determining the speed of the user, and predicting the wireless signal characteristics at the next moment to obtain a complete track. The method solves the technical problems that the correlation of signals between WiFi links is not fully utilized, and the wireless signal characteristics are lost in a real application scene, so that the accuracy of WiFi passive positioning operation in a non-continuous communication scene is low.

Description

A feature compensation and passive positioning method and system

Technical Field

The present invention relates to the field of indoor WiFi passive positioning under wireless sensing technology, and in particular to a feature compensation and passive positioning method and system.

Background Art

Indoor tracking and positioning have attracted extensive interest from researchers in recent years. Existing tracking methods can generally be divided into two categories, namely active tracking and passive tracking. Active tracking methods mainly rely on received signal strength indication, which is widely used but difficult to achieve high accuracy, while methods that utilize channel state information can provide more accurate tracking. Passive tracking methods do not require users to carry devices, which has the advantage of requiring continuous link communication, but usually require continuous link communication, which limits their application.

Since common mobile devices (such as mobile phones, tablet computers, and laptops) can be conveniently used as WiFi receivers to provide received signal strength indicators (RSSI), RSSI has become a widely used feature for tracking. For example, the existing invention patent application document "A speed-independent gait recognition method based on WiFi signals" with publication number CN114757237B includes: extracting the CSI amplitude of WiFi that changes over time during a person's walking process; preprocessing the CSI amplitude; determining whether there is a person walking in the environment, and extracting walking activity segments; converting the walking activity segments into time-frequency graphs of the same size; building a speed-independent gait recognition model based on DANN, the model includes a feature extractor, an identity identifier, and a speed identifier, the feature extractor is used to extract potential features from the input time-frequency graph, the identity identifier is used to predict the identity of the target under test using the features extracted by the feature extractor, and the speed identifier is used to predict the speed of the target under test using the features extracted by the feature extractor; training the speed-independent gait recognition model and outputting the identity of the target under test. And the existing invention patent application document "Hot-deployed cross-mode gait recognition system and method based on IMUWiFi" with publication number CN117197888A, the existing method includes: a gait feature extraction unit, which can calculate and solve the posture to compensate for the drift error based on the original IMU data obtained from the inertial measurement unit in combination with the extended Kalman filter, reduce the ankle IMU drift and suppress the waist IMU drift to obtain the footprint and trunk speed curves respectively, extract the IMU-based gait feature vector from the footprint and trunk speed curve and save it in the feature database; a gait recognition unit, which can receive the CSI data of the WiFi signal and eliminate the high-frequency noise of the CSI data through the adaptive PCA method, generate a spectrum diagram with the CSI data and extract the WiFi-based gait feature vector from it, compare the WiFi-based gait feature vector with the IMU-based gait feature vector through a pre-trained classification neural network model, and identify the identity of the walker according to the comparison result. However, multipath effects and asynchrony between devices make high-precision RSSI tracking difficult to achieve. Since CSI provides more information through multiple antennas and multiple carriers, it is now possible to use CSI for relatively accurate active positioning and tracking. Existing studies have used multiple-input multiple-output technology to build antenna arrays to analyze the arrival angle of signals and achieve mobile device tracking. Inspired by RSSI indoor positioning, some studies have shown that CSI fingerprints can eliminate the impact of multipath effects. By developing a solution that automatically updates the CSI fingerprint database, the efficiency of positioning can be improved without the need to collect fingerprints on site. Although active positioning technologies have made progress in indoor tracking, they all require users to carry electronic devices to work, which will undoubtedly reduce users' willingness to use them.

Passive tracking has significant advantages over active tracking because it extracts and analyzes the features of signals without requiring the user to carry any equipment. The Widar system proposed in 2017 first inferred the user's speed by the rate of change of the path length of multiple WiFi links. Widar2.0 and md-Track proposed in 2018 and 2019, respectively, further combined the angle of arrival, flight time, and Doppler shift to achieve single-link tracking. In addition, there is work to implement methods to eliminate tracking restrictions when users cross WiFi links, and track and recognize gait through smart microphones. The newly proposed NNE-Tracking in 2024 invented a wireless perception architecture that can effectively combat environmental noise by generating large-scale data sets to train the system while using mathematical models to supervise the training process. Although wireless positioning under line of sight has been significantly improved, the above-mentioned ideal scenario-based models often perform poorly when dealing with a large number of occlusions in the environment. In order to break through the limitations of obstacles, researchers have also begun to look for ways to track positioning under non-line-of-sight paths. NLoc proposed in 2022 models blocked reflections and virtual direct signals to achieve positioning under non-line-of-sight. HyperTracking, proposed in 2024, develops a non-line-of-sight tracking method that combines spatial model features and neural networks to eliminate the interference of obstacles on positioning.

With the development of Internet of Things technology, there are usually many smart devices in daily environment. Multiple transceiver links allow to obtain rich user movement information, but the communication in each link is not continuous. The above passive tracking systems cannot track users who are missing link information, and they all require continuous link information. Communication Duty Cycles (CDC) is used to measure the degree of missing wireless signal features. It refers to the proportion of valid communication data packets that can be used for perception. Existing work is based on the premise that CDC is 100%.

In summary, the existing technology has technical problems such as not fully utilizing the correlation of signals between WiFi links and missing wireless signal features in real application scenarios, resulting in low accuracy of WiFi passive positioning operations in non-continuous communication scenarios.

Summary of the invention

The technical problem to be solved by the present invention is: how to solve the technical problem that the correlation between signals between WiFi links is not fully utilized in the prior art, wireless signal characteristics are missing in real application scenarios, and the accuracy of WiFi passive positioning operations in non-continuous communication scenarios is low.

The present invention adopts the following technical solution to solve the above technical problem: a feature compensation and passive positioning method comprises:

S1. Perform signal feature processing operations, extract WiFi signal features by analyzing channel state information, and process to obtain a PLCR matrix; collect and extract the reflection path change rate PLCR in the CSI data, and process to obtain DFS data, and derive the observation-based matrix _P1 and the reliability-based matrix R according to the PLCR matrix, and assign weights to the first type prediction, the second type prediction, and the third type prediction for combined prediction operations;

S2, performing preparation operations for PLCR prediction in the tracking phase, obtaining a first type prediction, a second type prediction, and a third type prediction, obtaining a final PLCR prediction by combining weights, and obtaining a suitable PLCR prediction value and a user trajectory by processing the final PLCR prediction;

S3. Determine the user's speed and predict the wireless signal characteristics at the next moment. Design and use a neural network to map the user's trajectory according to the reflection path change rate PLCR. Obtain the complete trajectory by performing tracking operations, prediction operations, algorithm adjustment operations, and local trajectory prediction and tuning operations.

The present invention utilizes WiFi multi-link communication to achieve accurate passive tracking of the human body in indoor environments when wireless signal features are severely missing. The goal of the present invention is to compensate for the missing wireless signal features in real application scenarios by mining and utilizing the correlation between signals of multiple WiFi links, thereby achieving accurate WiFi passive positioning function in non-continuous communication scenarios.

In a more specific technical solution, S1 includes:

S11, obtaining CSI data, and extracting the reflection path change rate PLCR from the original CSI reading of the CSI data by performing a short-time Fourier transform STFT, wherein the function of CSI with respect to frequency f and time t can be expressed as follows:

Where _Hs (f,t) represents the static CSI component, L(t) is the path length corresponding to the dynamic CSI component _Hd (f,t), λ is the wavelength, A(f,t) is the signal amplitude, and e ^-j2πL(t)/λ is the phase;

S12. According to the Doppler effect, the following formula is derived:

Where _fD represents DFS data, r represents the reflection path change rate PLCR, and L(t) is the dynamic path length at time t;

S13, check each element in the PLCR matrix, and when a missing value is found, scan upward from the position of the missing value until the most recent observed true value in the same link is encountered, and fill the most recent observed true value into the preset observation matrix to obtain a matrix P ₁ based on the observation value;

S14. In the reliability-based matrix R, when the actual reflection path change rate PLCR can be observed at a specific location, the weight assigned to the first type of prediction is 1; when the signal feature is continuously lost, the weight of the first type of prediction is reduced; when a link experiences continuous feature loss, a quadratic function is used to reduce the weight of the first type of prediction, and weights are assigned to the second and third types of predictions.

In a more specific technical solution, in S14, the weight of the first type of prediction is reduced using the following logic:

In a more specific technical solution, S2 includes:

S21. Calculate the prediction based on the observed data as the first type of prediction, denote the t-th element of the n-th link in the matrix P ₁ based on the observed value as P ₁ (t,n), and obtain the prediction based on the observed data:

Where t ^′ is the minimum time index such that P(t ^′ ,n)≠0 and 1≤t′<t.

S22, calculating a prediction based on a proportional relationship as a second type of prediction, obtaining and using non-missing values, calculating a prediction based on a proportional relationship, and if all reflection path change rates PLCR at t= _t2 are missing, jumping to a prediction using a mathematical model;

S23, calculating the prediction based on the mathematical model as the third type of prediction, and obtaining the PLCR prediction matrix P ₃ based on the mathematical model through mathematical modeling, and reversely deducing the reflection path change rate PLCR of the first t time slots;

S24, performing a weighted operation on the first type prediction, the second type prediction, and the third type prediction to obtain a final PLCR prediction, and updating elements of the PLCR prediction matrix;

S25. Use the final PLCR prediction to fill in the missing features and obtain the user trajectory.

The present invention expands the application scenarios of traditional passive positioning systems, relaxes the unrealistic requirement of traditional perception technology for long-term continuous communication, and further realizes the integration of communication and perception of WiFi-based wireless perception systems in practical applications.

In a more specific technical solution, in S22, the following logic is used to obtain a prediction based on the proportional relationship:

In a more specific technical solution, S23 includes:

S231. For the preset link, assume the transmitter position is: l _t =(x _t ,y _t ); assume the receiver position is: l _r =(x _r ,y _r ), the current person position is: l _h =(x _h ,y _h ), and the person's speed is: v =(v _x , _vy ) ^T ;

S232, using the following logic, obtain the PLCR prediction matrix based on the mathematical model:

P ₃ (t,n)=A×v=a _x v _x +a _y v _y .

S233: When the PLCR data of all links are lost during the preset time slot, PLCR prediction is performed based on the known position sequence.

In a more specific technical solution, the PLCR prediction matrix based on the mathematical model satisfies:

In a more specific technical solution, in S24, each element of the PLCR prediction matrix P is updated using the following logic:

Where w represents the weight of the prediction based on the observed value.

The present invention proposes three mechanisms to effectively compensate for the missing WiFi features in passive tracking scenarios. By combining these three signal feature prediction methods, the present invention proposes an algorithm called simultaneous tracking and prediction. The algorithm achieves accurate passive tracking in the presence of severe WiFi feature loss.

In a more specific technical solution, S3 includes:

S31, during the tracking process, reversely predict the reflection path change rate PLCR through the local user trajectory;

S32, performing tracking operations and prediction operations through continuous iterations to compensate for missing PLCR features;

S33, continuously executing the algorithm adjustment operation and the local trajectory prediction and optimization operation to obtain the complete trajectory.

In a more specific technical solution, a feature compensation and passive positioning system includes:

The signal feature processing module is used to perform signal feature processing operations, extract WiFi signal features by analyzing channel state information, and process to obtain a PLCR matrix; collect and extract the reflection path change rate PLCR in the CSI data, and process it to obtain DFS data, and derive the observation-based matrix _P1 and the reliability-based matrix R according to the PLCR matrix, and assign weights to the first type prediction, the second type prediction, and the third type prediction for combined prediction operations;

A prediction combination module is used to prepare for the PLCR prediction in the tracking phase, obtain the first type prediction, the second type prediction and the third type prediction, obtain the final PLCR prediction by weight combination, and obtain the applicable PLCR prediction value and user trajectory by processing the final PLCR prediction. The prediction combination module is connected to the signal feature processing module;

The wireless signal feature prediction module is used to determine the user's speed and predict the wireless signal features at the next moment. A neural network is designed and used to map the user trajectory according to the reflection path change rate PLCR. The complete trajectory is obtained by performing tracking operations, prediction operations, algorithm adjustment operations, and local trajectory prediction and tuning operations. The wireless signal feature prediction module is connected to the prediction combination module.

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

The present invention utilizes WiFi multi-link communication to achieve accurate passive tracking of the human body in indoor environments when wireless signal features are severely missing. The goal of the present invention is to compensate for the missing wireless signal features in real application scenarios by mining and utilizing the correlation between signals of multiple WiFi links, thereby achieving accurate WiFi passive positioning function in non-continuous communication scenarios.

The present invention expands the application scenarios of traditional passive positioning systems, relaxes the unrealistic requirement of traditional perception technology for long-term continuous communication, and further realizes the integration of communication and perception of WiFi-based wireless perception systems in practical applications.

The present invention proposes three mechanisms to effectively compensate for the missing WiFi features in passive tracking scenarios. By combining these three signal feature prediction methods, the present invention proposes an algorithm called simultaneous tracking and prediction. The algorithm achieves accurate passive tracking in the presence of severe WiFi feature loss.

The present invention solves the technical problems in the prior art that the correlation between signals of WiFi links is not fully utilized, wireless signal characteristics are missing in real application scenarios, and the accuracy of WiFi passive positioning operations in non-continuous communication scenarios is low.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a schematic diagram of basic steps of a feature compensation and passive positioning method according to Embodiment 1 of the present invention;

FIG2 is a schematic diagram of data stream processing of a feature compensation and passive positioning method according to Embodiment 1 of the present invention;

FIG3 is a schematic diagram of a Fresnel zone cut by human body motion according to Embodiment 1 of the present invention;

FIG4 is a schematic diagram of a calculated PLCR matrix, a matrix based on observation values, and a reliability matrix according to Embodiment 1 of the present invention;

FIG5 is a schematic diagram of establishing a mapping between wireless signal features and trajectories using an LSTM neural network according to Embodiment 1 of the present invention;

FIG. 6 is a diagram showing the passive positioning effect of Embodiment 1 of the present invention.

DETAILED DESCRIPTION

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.

Example 1

As shown in FIG. 1 and FIG. 2 , a feature compensation and passive positioning method provided by the present invention includes the following basic steps:

S1, perform signal feature processing operation;

As shown in FIG3 , in this embodiment, the WiFi signal characteristics are extracted by analyzing the channel state information to obtain the PLCR matrix; specifically, since the user cuts the Fresnel zone ellipse when moving, the receiver can calculate the PLCR through the received signal; the matrix based on the observation value and the reliability matrix are calculated to assist the next stage of PLCR prediction; the system alternates tracking and prediction operations and repeats the cycle.

In this embodiment, the system first acquires CSI; PLCR is extracted from the raw CSI reading by performing a short-time Fourier transform (STFT). In this embodiment, the wavelength is represented as λ, the signal amplitude is represented as A(f, t), and the phase is represented as e ^-j2πL(t)/λ , then the function of CSI with respect to frequency f and time t can be expressed as follows:

Where H _s (f, t) represents the static CSI component, and L(t) is the path length corresponding to the dynamic CSI component H _d (f, t). PLCR is mathematically defined as:

Where L(t) is the dynamic path length at time t. Due to the Doppler effect, PLCR is a constant multiple of DFS, so the following equation can be derived:

Where _fD and r represent DFS and PLCR, respectively.

In this embodiment, the PLCR matrix can be represented by, for example, a two-dimensional matrix P of size T×N, where T is the total number of time slots and N is the number of links. The row index t represents the time index, and the column index n represents the signal characteristics of different WiFi links, i.e., the signal characteristics received by different receivers. When there is no WiFi communication at the location corresponding to time t and link n, the system sets the corresponding value P(t,n) to 0. The matrix is continuously updated in each iteration of the prediction phase.

In this embodiment, the matrix _P1 based on the observed values is the same size as the PLCR matrix and can be directly derived from the PLCR matrix. Specifically, the present invention compensates for the missing WiFi features based on the most recent true PLCR features that can be observed in the same WiFi link. In order to calculate the observation matrix, it is necessary to check each element in the original PLCR matrix. When one of the missing values is checked, the system scans upward from that position until a non-missing PLCR value, i.e., the most recently observed true value in the same link, is encountered. This value is filled into the observation matrix as a prediction based on the observed values. The elements in the observation matrix are called predictions based on the observed values.

In this embodiment, the reliability matrix is represented by R and has the same size as the above two matrices. However, the elements are not PLCRs, but weights between 0 and 1. Although the system has obtained basic predictions based on observations, accurate tracking cannot be achieved by relying on it alone. In order to improve the accuracy of the final prediction, the present invention introduces two other predictions obtained by other methods, which are respectively recorded as the second type prediction and the third type prediction.

In this embodiment, in order to effectively and accurately combine the three predictions, the system assigns weights to them and then calculates the final PLCR prediction value. Since the PLCR in a given link remains stable over short time intervals, as long as the signal in a certain link is not missing for a long time, the system is more inclined to trust the prediction based on the observation value. In order to measure the degree of trust in the prediction based on the observation value, each element in the reliability matrix is a weight assigned to the prediction 1, representing the degree of trust the system has in the prediction based on the observation value at the corresponding time slot. When the actual PLCR can be observed at a specific location, the weight assigned to the prediction based on the observation value is 1 because the observation is true and accurate. However, if the signal features are continuously lost, the reliability of past PLCR observations will decrease. Despite this, the most recent observable value of the same link is still meaningful. Based on the above analysis, the weight assigned to the prediction based on the observation value will gradually decrease over time until it is reduced to 0 after T _w time indexes. When a link experiences continuous feature loss, a quadratic function is used in this embodiment to reduce the weight:

Where w represents the above weight. This weight is reduced to zero at time slot t = _Tw . Depending on the specific situation of missing signal features, another part of the weight (1-w) will be allocated to the prediction based on the proportional relationship or the prediction based on the model. In the above process, the calculation process of the PLCR matrix, the matrix based on the observation value and the reliability matrix is shown in Figure 4.

S2, preparing for the PLCR prediction in the tracking phase, obtaining and combining the first type prediction, the second type prediction and the third type prediction to obtain the final PLCR prediction;

In this embodiment, the first type of prediction, the second type of prediction and the third type of prediction are combined. Specifically, the first type of prediction can be: a prediction based on an observed value; the second type of prediction can be: a prediction based on a proportional relationship; the third type of prediction can be: a prediction based on a model;

In this embodiment, the prediction based on observations is calculated directly from a small number of actual PLCR values that can be directly observed; the prediction based on proportional relationships is determined by the correlation between different WiFi links; the system considers the degree of feature missingness and combines the above predictions with the model-based prediction weights to obtain the final PLCR prediction.

In this embodiment, the prediction based on the observed data is calculated. Specifically, the actual PLCR values that can be observed are stored in the original PLCR matrix P, and the positions of missing data are marked with 0. For each position of missing data, the system scans upward until a non-missing value is encountered. This value is the most recently observed PLCR value of the same link, which is regarded as the prediction based on the observed data at that position. The t-th element of the n-th link in the observation matrix is represented as P ₁ (t,n), and the corresponding prediction based on the observed data can be calculated as:

Where t ^′ is the minimum time index such that P(t ^′ ,n)≠0 and 1≤t′<t.

In this embodiment, a prediction based on a proportional relationship is calculated. Specifically, in two given links, the ratio of the PLCRs at two adjacent times t=t ₁ (previous time) and t=t ₂ (next time) is approximately constant. Assuming that the system is processing data corresponding to t=t ₂ in the current iteration, the system can confirm that the PLCR of each link at its previous time t=t ₁ has been obtained in the previous iteration and is therefore not missing. Therefore, in link n ₁ , as long as there is at least one non-missing PLCR at time t=t ₂ , the system can use the proportional relationship to fill in the missing data in link n _1. Denoting the missing PLCR at t=t ₂ as P ₂ (t ₂ ,n ₁ ), the system can use three non-missing values to calculate a prediction based on a proportional relationship:

Otherwise, if all PLCR data at t = _t2 are missing, the system must resort to predictions based on mathematical models.

In this embodiment, a prediction based on a mathematical model is calculated. Specifically, the prediction matrix based on the mathematical model obtained by mathematical modeling is represented as P ₃ . In previous work, mathematical models have been used to obtain trajectories from PLCR features. However, the present invention utilizes mathematical models from another perspective, that is, to infer the PLCR features corresponding to the trajectory through a model method. For a specific link, assuming that the positions of the transmitter and the receiver are l _t = (x _t , y _t ) and l _r = (x _r , y _r ) respectively, the current position of the person is l _h = (x _h , y _h ), and the speed of the person is v = (v _x , _vy ) ^T , then the model-based PLCR prediction can be calculated as follows:

P ₃ (t,n)=A×v=a _x v _x +a _y v _y .

in,

Specifically, if the trajectory prediction of the first t time slots can be obtained through iteration, then the system can derive the speed prediction of the first t time slots by approximate difference of the position sequence. With this speed prediction, the system can reversely derive the PLCR of the first t time slots. In fact, the continuous nature of PLCR allows the PLCR at the (t+1)th time to be approximately fitted with the value at time t. Therefore, even if the PLCR data of all links at a specific time slot are lost, the system can still make PLCR predictions based on the previous position sequence.

In this embodiment, after the calculation of the three predictions is completed, the three predictions are combined by the following weighted method. Assume that in the current iteration, the first t rows have been processed and the (t+1)th row is being processed. Then, for each value in the (t+1)th row. If the value is not missing, the system can simply use the first type of prediction, that is, the observed value, without performing additional operations; if the value is missing and there is at least one observable non-missing PLCR value in the (t+1)th row, the first type of prediction is combined with the second type of prediction; if the value is missing and all values in the (t+1)th row are missing, the first type of prediction is combined with the third type of prediction.

After obtaining the three predictions, each element of the final PLCR prediction matrix P can be updated as follows:

Where w represents the weight of the prediction based on the observation. The system then fills in the missing features with the weighted final PLCR prediction. After integrating the three predictions, the system obtains the complete PLCR matrix for the first (t+1) time slots. The system then sets t to (t+1) and proceeds to the next row of the PLCR matrix. The system continues to loop until it finishes processing the last row of the PLCR matrix and obtains the final trajectory.

S3, determine the user's speed and predict the wireless signal characteristics at the next moment;

In this embodiment, the neural network model provides local tracking results in the initial period of time; the system obtains speed estimates by calculating the differences between adjacent positions of the trajectory; through the Fresnel zone model, the model-based PLCR prediction can be derived for the moment corresponding to each time index; due to the continuity of human movement, this prediction can be reused in the next time slot, and the system continues to operate in the prediction phase of the next iteration, processing the data of the next time slot until the final trajectory is obtained at the end of the iteration.

As shown in FIG5 , in this embodiment, a neural network is designed to map the PLCR (input) to the user's trajectory (output). The basic idea of the simultaneous tracking and prediction algorithm is that during the tracking process, the PLCR is reversely predicted through the local trajectory. By continuously iterating the simultaneous tracking and prediction, the system can gradually compensate for the missing PLCR features. In order to realize this idea, the system must obtain a relatively accurate local trajectory prediction at the beginning. Otherwise, if there is no initial local trajectory prediction, it is impossible to reversely compensate for the missing features. Subsequently, the system gradually adjusts and optimizes the local trajectory prediction by continuing to execute the algorithm, and gradually obtains the complete trajectory.

As shown in FIG6 , in this embodiment, to obtain the initial local trajectory prediction, the system can only rely on the first type of prediction based on the observation value. Because only this prediction can be obtained in the case of an unknown user position sequence. Therefore, the system takes the first N _f rows from the observation matrix and uses them as the input of the neural network to obtain a preliminary local trajectory prediction. According to experience, N _f is set to the number of time slots corresponding to 1s. It is worth noting that since the system only obtains the position sequence corresponding to the first N _f time slots at this time, the trajectory is not complete. Then, the system starts to iteratively perform simultaneous tracking and prediction. Since this algorithm adopts the method of gradually filling missing data, the system can fill a row of missing data in the original PLCR matrix after each iteration. By using the prediction integration method introduced above, the system can obtain the local trajectory through tracking in each iteration and obtain the PLCR prediction for the next moment. With this prediction, the system can compensate for the missing signal features and then re-predict a more complete trajectory until the iteration ends. The system continues to perform simultaneous tracking and prediction through iterations, and finally can obtain the final complete tracking result. The passive positioning effect of this method is shown in FIG6 . Experiments show that when the communication duty cycle is as low as 20%, the tracking error of the present invention is only 0.47m. Compared with the existing state-of-the-art work, the present invention reduces the tracking error by 79.19%.

Example 2

Example: The system was deployed on five commercial WiFi devices equipped with Intel 5300 wireless network cards. One device was designated as a transmitter and equipped with a single antenna. The remaining four devices were receivers, each equipped with three antennas and installed in a linear array (antenna spacing was 2.5 cm). Linux802.11n CSITool was installed on the devices to collect CSI readings. Data packets were transmitted at a frequency of 1000 Hz. The transmitter was configured in injection mode, and the receiver operated in monitoring mode in channel 64 and 5.32 GHz band. The transmitter was located at (2.4,-2.4), and the four receivers were located at (2.4,2.4), (-2.4,-2.4), (2.4,0), and (0,-2.4).

The present invention uses commercial WiFi equipment to implement a system prototype. Experimental results show that when the communication duty cycle is 20.00%, the system tracking error is 0.47m. Compared with the existing state-of-the-art work, the present invention reduces the tracking error by 79.19%.

In summary, the present invention uses WiFi multi-link communication to achieve accurate passive tracking of the human body in an indoor environment when wireless signal features are severely missing. The goal of the present invention is to compensate for the missing wireless signal features in real application scenarios by mining and utilizing the correlation of signals between multiple WiFi links, thereby achieving accurate WiFi passive positioning function in non-continuous communication scenarios.

The present invention expands the application scenarios of traditional passive positioning systems, relaxes the unrealistic requirement of traditional perception technology for long-term continuous communication, and further realizes the integration of communication and perception of WiFi-based wireless perception systems in practical applications.

The present invention proposes three mechanisms to effectively compensate for the missing WiFi features in passive tracking scenarios. By combining these three signal feature prediction methods, the present invention proposes an algorithm called simultaneous tracking and prediction. The algorithm achieves accurate passive tracking in the presence of severe WiFi feature loss.

The present invention solves the technical problems in the prior art that the correlation between signals of WiFi links is not fully utilized, wireless signal characteristics are missing in real application scenarios, and the accuracy of WiFi passive positioning operations in non-continuous communication scenarios is low.

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 (10)

1. A method of feature compensation and passive positioning, the method comprising:

s1, performing signal characteristic processing operation, and extracting WiFi signal characteristics by analyzing channel state information to obtain PLCR matrixes through processing; collecting and extracting a reflection path change rate PLCR in the CSI data, processing to obtain DFS data, deriving an observation value-based matrix P ₁ and a reliability-based matrix R according to the PLCR matrix, and distributing weights for the first type of prediction, the second type of prediction and the third type of prediction for combined prediction operation;

S2, performing preparation operation for PLCR prediction in a tracking stage, solving the first type prediction, the second type prediction and the third type prediction, acquiring a final PLCR prediction by using the weight combination, and acquiring an applicable PLCR prediction value and a user track by using final PLCR prediction processing;

S3, determining the speed of a user, predicting the wireless signal characteristics at the next moment, designing and utilizing a neural network, mapping the user track according to the reflection path change rate PLCR, and obtaining a complete track by executing the tracking operation, the predicting operation, the algorithm adjusting operation and the local track predicting and optimizing operation.

2. The method for feature compensation and passive positioning according to claim 1, wherein S1 comprises:

S11, obtaining CSI data, and extracting a reflection path change rate PLCR from original CSI readings of the CSI data by performing short-time Fourier transform (STFT), wherein the function of the CSI on frequency f and time t can be expressed as the following formula:

Wherein H _s (f, t) represents a static CSI component, L (t) is a path length corresponding to a dynamic CSI component H _d (f, t), lambda is a wavelength, A (f, t) is a signal amplitude, and e ^-j2πL(t)/λ is a phase;

s12, according to Doppler effect, the following formula is deduced:

Where f _D denotes the DFS data, r denotes the reflected path change rate PLCR, and L (t) is the dynamic path length at time t;

S13, checking each element in the PLCR matrix, and when a missing value is checked, scanning upwards from the position of the missing value until a latest observed true value in the same link is encountered, and filling the latest observed true value into a preset observed matrix to obtain an observed value-based matrix P ₁;

s14, in the reliability-based matrix R, when the actual reflection path change rate PLCR can be observed at a specific position, the weight allocated to the first type of prediction is 1; reducing the weight of the first type of prediction when signal features continue to be lost; and when one link experiences continuous characteristic missing, adopting a quadratic function to reduce the weight of the first type of prediction, and distributing weights for the second type of prediction and the third type of prediction.

3. The method of claim 2, wherein in S14, the weight of the first type of prediction is reduced by using the following logic:

4. A method of feature compensation and passive positioning as claimed in claim 1, characterized in that S2 comprises:

S21, calculating prediction based on the observed data, wherein the prediction based on the observed data is obtained by taking the t element of the nth link in the matrix P ₁ based on the observed value as P ₁ (t, n) as the first type of prediction:

Where t ' is the minimum time index such that P (t ', n) noteq0 and 1+.t ' < t.

S22, calculating a prediction based on a proportional relation, which is used as the second type of prediction, solving and utilizing a non-missing value, calculating the prediction based on the proportional relation, wherein all the reflection path change rates PLCR at t=t ₂ are missing, and skipping the prediction using a mathematical model;

s23, calculating a mathematical model-based prediction to serve as a third type of prediction, wherein a PLCR mathematical model-based prediction matrix P ₃ is obtained through mathematical modeling, and the reflection path change rates PLCR of the first t time slots are obtained through reverse deduction;

S24, weighting the first type of prediction, the second type of prediction and the third type of prediction to obtain the final PLCR prediction, and updating PLCR elements of a prediction matrix;

S25, predicting and filling the missing features by using the final PLCR to obtain a user track.

5. The method of claim 4, wherein in S22, the prediction based on the proportional relationship is obtained by using the following logic:

6. the method of feature compensation and passive positioning according to claim 4, wherein S23 comprises:

S231, setting the positions of transmitters of preset links as follows: l _t＝(x_t,y_t); let the receiver position be: l _r＝(x_r,y_r), the current person's location is: l _h＝(x_h,y_h), the speed of the person is: v= (v _x,v_y)^T;

S232, calculating PLCR prediction matrix based on the mathematical model by using the following logic:

P₃(t,n)＝A×v＝a_xv_x+a_yv_y.

s233, when PLCR data of all links are lost in a preset time slot, PLCR prediction is performed based on a known position sequence.

7. The method of claim 6, wherein the PLCR prediction matrix based on mathematical model satisfies:

8. The method of claim 1, wherein in S24, each element of the PLCR prediction matrix P is updated using the following logic:

Where w represents the weight of the prediction based on the observed value.

9. A method of feature compensation and passive positioning as claimed in claim 1, characterized in that the step S3 comprises:

S31, reversely predicting the change rate PLCR of the reflection path through the local user track in the tracking process;

s32, performing the tracking operation and the prediction operation through continuous iteration, and compensating missing PLCR features;

and S33, continuously executing the algorithm adjustment operation and the local track prediction tuning operation, and processing to obtain the complete track.

10. A feature compensation and passive positioning system, the system comprising:

The signal characteristic processing module is used for performing signal characteristic processing operation, extracting WiFi signal characteristics by analyzing channel state information, and processing to obtain PLCR matrixes; collecting and extracting a reflection path change rate PLCR in the CSI data, processing to obtain DFS data, deriving an observation value-based matrix P ₁ and a reliability-based matrix R according to the PLCR matrix, and distributing weights for the first type of prediction, the second type of prediction and the third type of prediction for combined prediction operation;

The prediction combination module is used for preparing operations for PLCR predictions in a tracking stage, solving the first type of predictions, the second type of predictions and the third type of predictions, acquiring a final PLCR prediction by using the weight combination, acquiring an applicable PLCR predicted value and a user track by using final PLCR prediction processing, and the prediction combination module is connected with the signal characteristic processing module;

The wireless signal characteristic prediction module is used for determining the speed of a user and predicting the wireless signal characteristic at the next moment, the neural network is designed and utilized, the user track is mapped according to the reflection path change rate PLCR, the complete track is obtained through executing the tracking operation, the prediction operation, the algorithm adjustment operation and the local track prediction optimization operation, and the wireless signal characteristic prediction module is connected with the prediction combination module.