Multi-WiFi device adaptive selection method for cooperative tracking in complex environment

Abstract

The invention discloses a multi-WiFi device adaptive selection method for cooperative tracking in a complex environment. The method comprises the steps that when multiple WiFi devices track a target object in a complex environment, CSI data of each antenna of any wifi device is acquired, each wifi device comprises three antennas, and each piece of CSI data is a 3 * 30 complex matrix; based on the CSI data of each antenna of any wifi device, determining a Fisher information matrix of Doppler frequency shift of the CSI data of any wifi device; and based on the Fisher information matrix of the Doppler frequency shift of the CSI data of any wifi device, establishing a relation model between the Doppler frequency shifts of the receivers of all wifi devices and the speed parameters of the target object. The technical problems of low tracking efficiency, large error and poor stability when commercial WiFi equipment tracks a target object in a complex environment in the prior art are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of device-free human tracking for commercial WiFi devices, and more particularly, to an adaptive selection method for multiple WiFi devices for collaborative tracking in complex environments. Background Art

[0002] With the rapid development of intelligent environments, WiFi-based sensing has been widely applied in various scenarios, such as fall detection, gesture recognition, human activity recognition, and vital sign monitoring; since the movement of the target affects the propagation of the signal, it obtains Channel State Information (CSI for short) to capture some motion-related information and perform sensing tasks.

[0003] Among these WiFi-based sensing tasks, WiFi-based tracking has attracted great attention in both academia and industry; WiFi-based tracking has become a promising alternative to traditional tracking systems (such as GPS and wearable sensor-based systems) due to its low cost and ubiquitous coverage; the wide deployment of WiFi networks also provides an ideal platform for seamless integration of various applications, including smart home, security monitoring, and health monitoring.

[0004] The core of WiFi-based tracking is to obtain the wireless signal reflected by the target and extract trajectory-related features from the CSI to track the target; trajectory-related features include Angle of Arrival (AoA), Time of Flight (ToF), and Doppler Frequency Shift (DFS for short); AoA and ToF provide angle and distance information respectively, and DFS provides target speed information; although angle and distance information helps to estimate the spatial environment of the target, accurately obtaining this information is a challenge for fine-grained WiFi-based tracking; specifically, the small number of antennas of WiFi devices and the limited channel bandwidth severely limit the accuracy of obtaining angle and distance information; in contrast to these, the accuracy of estimating DFS depends on the number of signal samples within a time window, rather than hardware or bandwidth; therefore, in WiFi-based tracking, it is crucial to estimate the target's speed through DFS.

[0005] Although it is theoretically possible to have a large number of signal samples within a large time window, in practice, it is difficult to accurately estimate speed based on DFS; there are three challenges to address: First, the estimation of the target speed includes the estimation of the motion speed and the motion direction; the estimation of these two parameters requires at least two independent DFS measurements; it requires a multi-device system in which multiple WiFi devices cooperate to track the target; the multi-device system provides different spatial views of the target's motion and enables more accurate tracking; Second, the DFS estimation of WiFi CSI is inaccurate; the accuracy of DFS estimation on WiFi devices is affected by several time-varying factors, such as the quality of the received signal; in addition, the signals received by WiFi devices placed in different locations may be affected by various multipath and interference signals; this results in different DFS estimation accuracies for different WiFi devices; although it is important to select the optimal set of WiFi devices and estimate the target speed, the accuracy of DFS estimation of WiFi CSI has not been analyzed and quantified in existing work; Third, in terms of speed estimation, there is a lack of a mathematical model for analyzing the complementarity of WiFi devices; different WiFi devices may be good at estimating different parameters; for example, some WiFi devices obtain more accurate speeds, while others may obtain more accurate motion directions; by effectively leveraging the complementary advantages of these devices, the overall speed estimation performance can be improved; however, this requires a deeper understanding of the complementarity between devices, which has not been fully analyzed in the existing technology. Summary of the Invention

[0006] Embodiments of the present invention provide a multi-WiFi device adaptive selection method for collaborative tracking in complex environments, to at least solve the technical problems of low tracking efficiency, large error, and poor stability when commercial WiFi devices track target objects in complex environments in the prior art.

[0007] According to one aspect of the embodiments of the present invention, a method for adaptively selecting multiple WiFi devices for collaborative tracking in a complex environment is provided. The method may include: when multiple WiFi devices track a target object in a complex environment, obtaining CSI data of each antenna of any one WiFi device, where each WiFi device includes three antennas, and each CSI data is a 3×30 complex matrix; based on the CSI data of each antenna of any one WiFi device, determining the Fisher information matrix of the Doppler frequency shift of the CSI data of any one WiFi device; based on the Fisher information matrix of the Doppler frequency shift of the CSI data of any one WiFi device, establishing a relationship model between the Doppler frequency shift of the receivers of all WiFi devices and the velocity parameters of the target object; based on the relationship model between the Doppler frequency shift of the receivers of all WiFi devices and the velocity parameters of the target object, defining an element model of the Fisher information matrix of all WiFi devices; based on the element model of the Fisher information matrix of all WiFi devices, obtaining the initial Jacobian matrix of all WiFi devices; based on the initial Jacobian matrix of all WiFi devices, defining the initial Fisher information matrix of all WiFi devices; introducing a receiver selection matrix of all WiFi devices to correct the initial Jacobian matrix of all WiFi devices to obtain a Jacobian matrix carrying the selection matrix; based on the Jacobian matrix carrying the selection matrix and the initial Fisher information matrix, obtaining the corrected Fisher information matrix of all WiFi devices; based on the corrected Fisher information matrix of all WiFi devices, determining the target WiFi device when maximizing the determinant of the corrected Fisher information matrix of all WiFi devices; and tracking the target object based on the target WiFi device.

[0008] Optionally, the determining the Fisher information matrix of the Doppler frequency shift of the CSI data of any one WiFi device based on the CSI data of each antenna of any one WiFi device includes: processing the CSI data of any two antennas of any one WiFi device to obtain an initial quotient signal of the CSI data of any one WiFi device; when the initial quotient signal of the CSI data of any one WiFi device is interfered by environmental noise, obtaining a target quotient signal of the CSI data of any one WiFi device, where the target quotient signal carries a noise signal; and obtaining the Fisher information matrix of the Doppler frequency shift of the CSI data of any one WiFi device based on the target quotient signal of the CSI data of any one WiFi device.

[0009] Optionally, the processing of the CSI data of any two antennas of any one Wi-Fi device to obtain the initial quotient signal of the CSI data of any one Wi-Fi device includes: determining the quotient of the CSI data of any two antennas of any one Wi-Fi device as the initial quotient signal of the CSI data of any one Wi-Fi device.

[0010] Optionally, the obtaining of the Fisher information matrix of the Doppler frequency shift of the CSI data of any one Wi-Fi device based on the target quotient signal of the CSI data of any one Wi-Fi device includes: obtaining the log-likelihood function of the target quotient signal of the CSI data of any one Wi-Fi device based on the target quotient signal of the CSI data of any one Wi-Fi device; obtaining the Fisher information matrix of the Doppler frequency shift of the CSI data of any one Wi-Fi device based on the log-likelihood function of the target quotient signal of the CSI data of any one Wi-Fi device.

[0011] Optionally, the obtaining of the log-likelihood function of the target quotient signal of the CSI data of any one Wi-Fi device based on the target quotient signal of the CSI data of any one Wi-Fi device includes: obtaining the probability density function of the target quotient signal of the CSI data of any one Wi-Fi device based on the target quotient signal of the CSI data of any one Wi-Fi device; obtaining the log-likelihood function of the target quotient signal of the CSI data of any one Wi-Fi device based on the probability density function of the target quotient signal of the CSI data of any one Wi-Fi device.

[0012] Optionally, the obtaining of the Fisher information matrix of the Doppler frequency shift of the CSI data of any one Wi-Fi device based on the log-likelihood function of the target quotient signal of the CSI data of any one Wi-Fi device includes: obtaining the Fisher information matrix of the Doppler frequency shift of the CSI data of any one Wi-Fi device based on the first-order derivative, second-order derivative and Fisher information matrix of the log-likelihood function of the target quotient signal of the CSI data of any one Wi-Fi device with respect to the Doppler frequency shift of the CSI data of any one Wi-Fi device.

[0013] Optionally, establishing a relationship model between the Doppler frequency shift of the receivers of all WiFi devices and the velocity parameter of the target object based on the Fisher information matrix of the Doppler frequency shift of the CSI data of any one WiFi device includes: obtaining the Cramer-Rao lower bound of the Doppler frequency shift of the CSI data of any one WiFi device based on the Fisher information matrix of the Doppler frequency shift of the CSI data of any one WiFi device; and establishing a relationship model between the Doppler frequency shift of the receivers of all WiFi devices and the velocity parameter of the target object based on the Cramer-Rao lower bound of the Doppler frequency shift of the CSI data of any one WiFi device.

[0014] Advantages of the present invention: (1) By introducing a multi-device collaborative tracking method based on commercial WiFi devices, the present invention can achieve high-precision human tracking without additional hardware support; this method effectively reduces the hardware cost of the system, and at the same time, taking advantage of the wide deployment of existing WiFi devices, reduces the dependence on traditional sensors or devices, making the system more universal and economical.

[0015] (2) By quantifying the DFS estimation error and combining the complementarity between devices, the proposed device selection strategy significantly improves the accuracy of target tracking; compared with traditional single-device tracking systems, it can work collaboratively through multiple devices to make up for the problem of insufficient estimation accuracy of a single device, thereby enhancing the tracking ability of the system in complex environments.

[0016] (3) By dynamically selecting devices and real-time optimizing the tracking strategy, the system can adapt to environmental changes and provide efficient and stable tracking performance in different scenarios; this dynamic adjustment mechanism effectively avoids the negative impact brought by environmental changes, improves the robustness and flexibility of the system, and ensures efficient operation in a changing environment.

[0017] (4) The system of the present invention has a wide range of application scenarios, not only applicable to common fields such as smart homes and security monitoring, but also can be extended to special needs scenarios such as medical health monitoring and elderly care; through device-free human tracking based on commercial WiFi devices, the system can meet the requirements of different fields for high-precision tracking and low-cost deployment, and has significant market potential and commercial value. Description of the Drawings

[0018] The drawings described herein are used to provide a further understanding of the present invention, form a part of this application, and the illustrative embodiments and descriptions of the present invention are used to explain the present invention, and do not constitute an improper limitation to the present invention. In the drawings: Figure 1 is a flowchart of a multi-WiFi device adaptive selection method for collaborative tracking in complex environments according to an embodiment of the present invention; Figure 2 It is a schematic structural diagram of the AdaptTrack system constructed according to the multi-WiFi device adaptive selection method of the embodiment of the present invention. Detailed implementation manners

[0019] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of the present invention.

[0020] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are used to describe a specific order or sequence. It should be understood that such terms can be interchanged under appropriate circumstances so that the embodiments of the present invention described here can be implemented in an order different from those illustrated or described here. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0021] Embodiment 1 According to the embodiment of the present invention, a multi-WiFi device adaptive selection method for collaborative tracking in a complex environment is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system including at least one set of computer-executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that here.

[0022] Figure 1 It is a flowchart of the multi-WiFi device adaptive selection method for collaborative tracking in a complex environment according to the embodiment of the present invention. As Figure 1 shown, the method may include the following steps: Step S101, when multiple WiFi devices track a target object in a complex environment, obtain the CSI data of each antenna of any one WiFi device. Among them, each WiFi device includes three antennas, and each CSI data is a 3×30 complex matrix.

[0023] In the technical solution provided in step S101 of the present invention, for commercial WiFi devices, there is a lack of time synchronization between the transmitter (Tx) and the receiver (Rx) on each WiFi device, resulting in a time-varying random phase offset in each CSI sample. , which makes it challenging to directly analyze human motion using CSI data; however, for commercial WiFi cards (such as Intel 5300), the time-varying phase offsets of different receiving antennas on the same network card are consistent because they share the same radio frequency oscillator; by calculating the quotient of the CSI data from two receiving antennas on the same network card of the receiver, the phase noise can be effectively eliminated.

[0024] Step S102: Based on the CSI data of each antenna of any one WiFi device, determine the Fisher information matrix of the Doppler frequency shift of the CSI data of any one WiFi device.

[0025] In the technical solution provided in step S102 of the present invention, the CSI data of any two antennas of any one WiFi device are calculated to obtain the Fisher information matrix of the Doppler frequency shift of the CSI data of any one WiFi device.

[0026] Step S103: Based on the Fisher information matrix of the Doppler frequency shift of the CSI data of any one WiFi device, establish a relationship model between the Doppler frequency shift of the receivers of all WiFi devices and the velocity parameters of the target object.

[0027] In the technical solution provided in step S103 of the present invention, in order to select a suitable WiFi device, first establish a relationship expression between the DFS of the receiver of each WiFi device and the velocity parameters of the target object ( and ); The relationship expression between the DFS of the receiver of each WiFi device and the velocity parameters of the target object ( and ) is:

[0028] Wherein, is the relationship between the DFS of the th receiver and the velocity parameters of the target object ( and ), is a known constant, which depends on the position and angle of the th receiver, is the frequency of the wave source, is the speed of light, is the distance between the target object and the The included angle formed by Tx and Rx of a receiver For the target object moving at a speed approaching the th receiver, the propagation direction forms an angle with the direction of the angle bisector.

[0029] Step S104: Based on the relationship model between the Doppler frequency shift of the receivers of all WiFi devices and the speed parameters of the target object, define the element model of the Fisher information matrix of all WiFi devices.

[0030] In the technical solution provided in step S104 of the present invention above, by calculating the Fisher information matrix to evaluate the contribution of each receiver to estimating the speed parameters of the target object. For example, when 5 WiFi devices track a target object in a complex environment, define the elements of the Fisher information matrix of the 5 WiFi devices, that is, the moving speed ( ) and moving direction ( ) of each WiFi device for each target object; The expression of the element model of the Fisher information matrix of all WiFi devices is:

[0031] Wherein, is the element model of the Fisher information matrix of all WiFi devices, is the relationship between the DFS of each receiver and the speed parameters of the target object, is the number of receivers, is the th receiver's DFS estimation error.

[0032] Step S105: Based on the element model of the Fisher information matrix of all WiFi devices, obtain the initial Jacobian matrix of all WiFi devices.

[0033] In the technical solution provided in step S105 of the present invention above, for each pair of Fisher information matrices of WiFi devices ( and ) take the partial derivative, and organize these derivatives into a matrix to obtain the Jacobian matrix ; To evaluate the contribution of each receiver to the speed parameter estimation, first, it is necessary to calculate the sensitivity of each measurement (i.e., DFS) to the parameter to be estimated (speed and direction ); specifically, by calculating the sensitivity of DFS to and The partial derivatives are taken and these derivatives are organized into a matrix to obtain the initial Jacobian matrix :

[0034] where is the initial Jacobian matrix, is a known constant, that is , is the target object approaching the th receiver at speed , and the propagation direction forms an angle with the angular bisector of . The value range of is from 1 to n .

[0035] Step S106: Define the initial Fisher information matrix of all Wi-Fi devices based on the initial Jacobian matrix of all Wi-Fi devices.

[0036] In the technical solution provided in step S106 of the present invention, according to the initial Jacobian matrix of all Wi-Fi devices, the initial Fisher information matrix of all Wi-Fi devices is calculated to evaluate the confidence of parameter estimation ( and ). Because the receiver has an estimated position at different locations, the expression of the initial Fisher information matrix of all Wi-Fi devices is:

[0037] where is the covariance matrix of the estimation error (since different receivers have different measurement errors and thus different estimation errors), represents the estimation error of the th receiver for DFS, is the transpose of , J vv describes the estimation accuracy of parameter v . Specifically, Jvv the larger it is, the higher the sensitivity of the observed data to parameter v , and the higher the accuracy of estimating v . J vθ and J θv describe the information coupling relationship between parameters v and θ . If J vθ or J θv is not zero, it means that the parameterv and θ are related, and estimating the precision of one parameter will be affected by the other parameter. J θθ describes the estimated precision of the parameter θ which is similar to J vv , J θθ the larger it is, the higher the sensitivity of the observed data to the parameter θ is, and the higher the precision of the estimation θ will be.

[0038] Step S107: Introduce the receiver selection matrix of all Wi-Fi devices to correct the initial Jacobian matrix of all Wi-Fi devices, and obtain the Jacobian matrix with the selection matrix.

[0039] In the technical solution provided in step S107 of the present invention, in order to achieve optimal device selection in a multi-receiver system, a receiver selection matrix is introduced, which can flexibly control which receivers are included in the speed estimation process. According to the receiver selection matrix, the initial Jacobian matrix of all Wi-Fi devices is corrected to obtain the Jacobian matrix with the selection matrix. The receiver selection matrix is expressed as: The receiver selection matrix S is defined as:

[0040] indicates whether to select the th receiver for calculation; by introducing the selection matrix , the Jacobian matrix can be modified to generate the Jacobian matrix with the selection matrix :

[0041] where is the Jacobian matrix with the selection matrix.

[0042] Step S108: Based on the Jacobian matrix with the selection matrix and the initial Fisher information matrix, obtain the corrected Fisher information matrix of all Wi-Fi devices.

[0043] In the technical solution provided in step S108 of the present invention, the initial Fisher information matrix is corrected by the Jacobian matrix with the selection matrix to obtain the corrected Fisher information matrix of all Wi-Fi devices; The expression of the corrected Fisher information matrix of all Wi-Fi devices is:

[0044] Among them, is the Fisher information matrix of all the corrected Wi-Fi devices, is the transpose of.

[0045] Step S109: Based on the Fisher information matrix of all the corrected Wi-Fi devices, determine the target Wi-Fi device when maximizing the determinant of the Fisher information matrix of all the corrected Wi-Fi devices.

[0046] In the technical solution provided in step S109 of the present invention, the determinant of the Fisher information matrix of all the corrected Wi-Fi devices is maximized to obtain the target Wi-Fi device; When defining the optimization objective, the goal is to optimize the estimation speed magnitude and direction of the performance; this goal can be achieved by adopting the D-optimality criterion, which involves maximizing the determinant of the FIM; the optimization objective can be expressed as:

[0047] Among them, is to take the determinant, is to take the maximum value of the determinant. During the optimization process, the following constraints need to be considered: , represents the total number of receivers to be selected; the core idea of D-optimality is to improve the accuracy of parameter estimation by minimizing the volume of the confidence ellipsoid of the regression parameters; this is mathematically equivalent to maximizing the determinant of the FIM because the determinant is inversely proportional to the volume of the confidence ellipsoid in the parameter space; by doing so, the uncertainty of the estimated parameters is minimized, resulting in more accurate and robust estimation results.

[0048] Step S110: Track the target object based on the target Wi-Fi device.

[0049] In the technical solution provided in step S110 of the present invention, track the target object according to the target Wi-Fi device.

[0050] The above method of this embodiment will be further introduced below.

[0051] As an alternative embodiment, in step S102, based on the CSI data of each antenna of any one Wi-Fi device, the Fisher information matrix of the Doppler frequency shift of the CSI data of any one Wi-Fi device is determined, including: processing the CSI data of any two antennas of any one Wi-Fi device to obtain an initial quotient signal of the CSI data of any one Wi-Fi device; when the initial quotient signal of the CSI data of any one Wi-Fi device is interfered by environmental noise, obtaining a target quotient signal of the CSI data of any one Wi-Fi device, where the target quotient signal carries a noise signal; based on the target quotient signal of the CSI data of any one Wi-Fi device, obtaining the Fisher information matrix of the Doppler frequency shift of the CSI data of any one Wi-Fi device.

[0052] In this embodiment, since calculating the quotient of the CSI data of two receiving antennas on the same network card of the receiver can effectively eliminate phase noise, therefore, perform a quotient operation on the CSI data of any two antennas of any one Wi-Fi device to obtain an initial quotient signal of the CSI data of any one Wi-Fi device. The expression of the initial quotient signal is:

[0053] where is the initial quotient signal, represents frequency, represents time, and respectively represent the amplitude and phase of the static path in the CSI data received by the first receiving antenna and the second receiving antenna, and respectively represent the attenuation of the dynamic components of the first antenna and the second antenna, is the phase shift introduced on the first antenna due to the Doppler frequency shift, is the Doppler frequency shift, represents the phase difference between the first antenna and the second antenna, is the time-varying random phase shift existing in each CSI sample due to the lack of time synchronization between Tx and Rx in commercial Wi-Fi devices, ( ) is the attenuation of the dynamic component of the first antenna, ( ) is the attenuation of the dynamic component of the second antenna, where the first antenna and the second antenna are any two antennas of the receiver of any one Wi-Fi device; For simplicity of representation, rewrite the expression of the initial quotient signal as:

[0054] Among them, , , and .

[0055] Since the parameter mainly affects the position of the center of the circle in the complex plane, while the parameters and play a crucial role in determining the radius of the circle; however, in practical applications, the received initial CSI quotient signal will inevitably be affected by environmental noise, thus bringing uncertainty to the characteristics of the signal. To capture these effects and model the system more precisely, the received initial CSI quotient signal is represented as follows:

[0056] Among them, is the radius of the circle formed by the CSI signal in the complex plane, represents the random perturbation introduced by environmental noise, and these perturbations come from two main sources: channel noise and dynamic multipath interference; the latter includes reflections caused by moving objects (such as other people or mobile devices) in the environment that are irrelevant to the target object; according to existing research, can be effectively modeled as a random variable conforming to the complex Gaussian distribution, denoted as , represents a complex Gaussian distribution with a mean of 0 and a variance of , that is, the real and imaginary parts of the complex random variable both follow independent normal distributions; this distribution accurately captures the statistical characteristics of the noise; to calculate the error of calculating from , the Cramér-Rao Lower Bound (CRLB) can be used to quantify the error of the DFS estimation; Perform multiple calculations on the target quotient signal of the CSI data of any one wifi device to obtain the Fisher information matrix of the Doppler frequency shift of the CSI data of any one wifi device.

[0057] As an alternative embodiment, process the CSI data of any two antennas of any one wifi device to obtain the initial quotient signal of the CSI data of any one wifi device, including: determining the quotient of the CSI data of any two antennas of any one wifi device as the initial quotient signal of the CSI data of any one wifi device.

[0058] In this embodiment, the quotient of the CSI data of any two antennas of any one wifi device is determined as the initial quotient signal of the CSI data of any one wifi device.

[0059] As an alternative embodiment, based on the target quotient signal of the CSI data of any one Wi-Fi device, obtaining the Fisher information matrix of the Doppler frequency shift of the CSI data of any one Wi-Fi device includes: based on the target quotient signal of the CSI data of any one Wi-Fi device, obtaining the log-likelihood function of the target quotient signal of the CSI data of any one Wi-Fi device; based on the log-likelihood function of the target quotient signal of the CSI data of any one Wi-Fi device, obtaining the Fisher information matrix of the Doppler frequency shift of the CSI data of any one Wi-Fi device.

[0060] In this embodiment, calculate the target quotient signal of the CSI data of any one Wi-Fi device to obtain the log-likelihood function of the target quotient signal of the CSI data of any one Wi-Fi device; calculate the log-likelihood function of the target quotient signal of the CSI data of any one Wi-Fi device to obtain the Fisher information matrix of the Doppler frequency shift of the CSI data of any one Wi-Fi device.

[0061] As an alternative embodiment, based on the target quotient signal of the CSI data of any one Wi-Fi device, obtaining the log-likelihood function of the target quotient signal of the CSI data of any one Wi-Fi device includes: based on the target quotient signal of the CSI data of any one Wi-Fi device, obtaining the probability density function of the target quotient signal of the CSI data of any one Wi-Fi device; based on the probability density function of the target quotient signal of the CSI data of any one Wi-Fi device, obtaining the log-likelihood function of the target quotient signal of the CSI data of any one Wi-Fi device.

[0062] In this embodiment, in order to further characterize the statistical behavior of the observed CSI quotient signal its probability density function (PDF) is derived, and its expression is:

[0063] where is the probability density function of the target quotient signal of the CSI data of any one Wi-Fi device, represents the observed CSI quotient signal, is the ideal CSI quotient signal in the case of no noise. Based on the PDF, the corresponding log-likelihood function can be expressed as:

[0064] where is the log-likelihood function of the target quotient signal of the CSI data of any one Wi-Fi device.

[0065] As an alternative embodiment, the Fisher information matrix of the Doppler frequency shift of the CSI data of any one Wi-Fi device is obtained based on the log-likelihood function of the target commercial signal of the CSI data of any one Wi-Fi device, including: obtaining the Fisher information matrix of the Doppler frequency shift of the CSI data of any one Wi-Fi device based on the first-order derivative, second-order derivative and Fisher information matrix of the log-likelihood function of the target commercial signal of the CSI data of any one Wi-Fi device with respect to the Doppler frequency shift of the CSI data of any one Wi-Fi device.

[0066] In this embodiment, the log-likelihood function provides a basis for evaluating the information content of the received signal with respect to DFS; to quantify the accuracy of the DFS estimate, the Fisher information matrix (FIM) is calculated; the FIM is derived from the first-order and second-order derivatives of the log-likelihood function with respect to as follows: First-order derivative: Captures the sensitivity of the likelihood function to the change; Second-order derivative: Quantifies the curvature of the likelihood function, which is directly related to the information content of: The first-order derivative of the log-likelihood function with respect to :

[0067]

[0068]

[0069] where is the conjugate of . Due to complex conjugate symmetry, the above formula can be simplified to:

[0070] where is the real number extraction operation; The second-order derivative of the log-likelihood function with respect to :

[0071]

[0072] Fisher information matrix:

[0073] When taking the expectation, the mean of the noise term is 0; thus it can be simplified to:

[0074]

[0075] wherein, is time, is the expected value; Therefore, the final Fisher information matrix is:

[0076] This expression emphasizes that the radius of the circle formed by the quotient of the FIM and the CSI signal in the complex plane is proportional to, and inversely proportional to the noise variance

[0077] As an alternative embodiment, in step S103, based on the Fisher information matrix of the Doppler shift of the CSI data of any one Wi-Fi device, a relationship model between the Doppler shift of the receivers of all Wi-Fi devices and the velocity parameter of the target object is established, including: obtaining the Cramér-Rao lower bound of the Doppler shift of the CSI data of any one Wi-Fi device based on the Fisher information matrix of the Doppler shift of the CSI data of any one Wi-Fi device; establishing a relationship model between the Doppler shift of the receivers of all Wi-Fi devices and the velocity parameter of the target object based on the Cramér-Rao lower bound of the Doppler shift of the CSI data of any one Wi-Fi device.

[0078] In this embodiment, the reciprocal of the Fisher information matrix of the Doppler shift of the CSI data of any one Wi-Fi device is used as the Cramér-Rao lower bound of the Doppler shift of the CSI data of any one Wi-Fi device, wherein the expression of the Cramér-Rao lower bound of the Doppler shift of the CSI data of any one Wi-Fi device is:

[0079] wherein, is the Cramér-Rao lower bound of the Doppler shift of the CSI data of any one Wi-Fi device.

[0080] Figure 2 FIG. is a schematic structural diagram of the AdaptTrack system constructed according to the multi-WiFi device adaptive selection method of the embodiments of the present invention. This system uses commercial Wi-Fi devices to achieve multi-device collaborative tracking and mainly consists of four modules: a data acquisition and preprocessing module, a multi-WiFi device selection module, a speed estimation module, and a trajectory reconstruction module; 1. Data acquisition and preprocessing module (as shown in the blue part of Figure 2 ) ​Collect CSI data from five receivers, each receiver equipped with three antennas; taking the Intel 5300 wireless network card as an example, the CSI data contains information from 30 subcarriers; combined with the three antennas supported by the 5300 network card, each CSI data sample is represented as a 3 × 30 complex matrix; by element-wise splitting the CSI data from two antennas on the same receiver, the channel quotient signal is calculated to obtain a 1×30 complex vector; subsequently, all the collected data is transmitted from the receiver to a dedicated computer using the SCP protocol for further processing; although the CSI quotient effectively reduces the noise in the CSI data, its representation in the complex plane may still exhibit an irregular shape; to better visualize the circular features and quantify the estimation error, smoothing is applied to the CSI quotient; in the present invention, the Savitzky-Golay filter is used, which is a smoothing method based on polynomial fitting that can effectively remove noise while maintaining the trend characteristics of the signal; by fitting a polynomial to the CSI quotient signal within a sliding window and calculating the fitted values, the Savitzky-Golay filter reduces signal fluctuations while avoiding over-smoothing or distortion that may occur in traditional filters.

[0081] 2. Multi-WiFi device selection module In the multi-device collaborative tracking system constructed based on the present invention, the multi-WiFi device selection module is not an independent module, but an important part of the speed estimation module. Its main task is to dynamically select the optimal device combination according to the CSI signal quality of the devices and the geometric relationship between the devices, so as to improve the speed estimation accuracy of the target.

[0082] First, calculate the standard deviation of the CSI quotient signal to detect the motion state of the target; after detecting the motion, calculate the DFS estimation error using the smoothed CSI quotient data generated in the preprocessing stage; specifically, the CSI quotient data is divided into multiple time periods, each time period corresponding to a fixed number of data packets; for each period, the system calculates the DFS estimation error based on the smoothed CSI quotient signal to evaluate the quality of the CSI data within that time period; the calculated estimation error provides an important basis for the selection of subsequent devices, optimizing the accuracy of the target speed estimation; since the target speed estimation accuracy varies among different devices, factors such as the position of the target relative to the device, the motion direction, the received CSI signal quality, and the geometric relationship between the devices will all affect the estimation accuracy; therefore, by comprehensively considering the recently calculated estimation error and the geometric relationship between the devices, the device selection process can be optimized to determine the optimal device combination, thereby improving the target speed estimation accuracy.

[0083] 3. Speed estimation module (as shown in the orange part in Figure 2 Figure After the device selection is completed, continuous wavelet transform (CWT) is performed on the CSI commercial signal of the selected device to generate a time-frequency spectrogram; through the analysis of the spectrogram, the frequency value with the highest energy is extracted as the DFS; subsequently, the current speed of the target is calculated using the extracted DFS.

[0084] 4. Trajectory reconstruction module (as shown in the green part in Figure 2 ) Given the initial position of the target, the position of the target can be calculated based on the speed estimated by the speed estimation module, thereby reconstructing the motion trajectory of the target.

[0085] In the embodiments of the present invention, when multiple WiFi devices track a target object in a complex environment, CSI data of each antenna of any one WiFi device is obtained, where each WiFi device includes three antennas, and each CSI data is a 3×30 complex matrix; based on the CSI data of each antenna of any one WiFi device, a Fisher information matrix of the Doppler frequency shift of the CSI data of any one WiFi device is determined; based on the Fisher information matrix of the Doppler frequency shift of the CSI data of any one WiFi device, a relationship model between the Doppler frequency shift of the receivers of all WiFi devices and the speed parameters of the target object is established; based on the relationship model between the Doppler frequency shift of the receivers of all WiFi devices and the speed parameters of the target object, an element model of the Fisher information matrix of all WiFi devices is defined; based on the element model of the Fisher information matrix of all WiFi devices, an initial Jacobian matrix of all WiFi devices is obtained; based on the initial Jacobian matrix of all WiFi devices, an initial Fisher information matrix of all WiFi devices is defined; a receiver selection matrix of all WiFi devices is introduced to correct the initial Jacobian matrix of all WiFi devices to obtain a Jacobian matrix with the selection matrix; based on the Jacobian matrix with the selection matrix and the initial Fisher information matrix, a corrected Fisher information matrix of all WiFi devices is obtained; based on the corrected Fisher information matrix of all WiFi devices, a target WiFi device is determined when the determinant of the corrected Fisher information matrix of all WiFi devices is maximized; the target object is tracked based on the target WiFi device, solving the technical problems of low tracking efficiency, large error, and poor stability when commercial WiFi devices track a target object in a complex environment in the prior art, achieving the technical effect that the device selection strategy proposed by quantifying the DFS estimation error through the Cramer-Rao lower bound of the Doppler frequency shift and combining the complementarity between devices significantly improves the accuracy of target tracking, and designing a real-time optimized tracking strategy to enable the system to adapt to environmental changes and operate efficiently in multiple environments.

[0086] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0087] In the above embodiments of the present invention, the descriptions of the various embodiments have their own emphases. For the parts not detailed in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.

[0088] In the several embodiments provided by the present application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only illustrative. For example, the division of units can be a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of units or modules can be in electrical or other forms.

[0089] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0090] In addition, the functional units in the various embodiments of the present invention can be integrated into a first processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0091] The above is only the preferred embodiment of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. An adaptive selection method for multi-WiFi devices for collaborative tracking in complex environments, characterized in that Including: When multiple WiFi devices track a target object in a complex environment, obtain the CSI data of each antenna of any one WiFi device. Each WiFi device includes three antennas, and each CSI data is a 3×30 complex matrix; Based on the CSI data of each antenna of any one WiFi device, determine the Fisher information matrix of the Doppler frequency shift of the CSI data of any one WiFi device; Based on the Fisher information matrix of the Doppler frequency shift of the CSI data of any one WiFi device, establish a relationship model between the Doppler frequency shift of the receivers of all WiFi devices and the velocity parameters of the target object; Based on the relationship model between the Doppler frequency shift of the receivers of all WiFi devices and the velocity parameters of the target object, define the element model of the Fisher information matrix of all WiFi devices; Based on the element model of the Fisher information matrix of all WiFi devices, obtain the initial Jacobian matrix of all WiFi devices; Based on the initial Jacobian matrix of all WiFi devices, define the initial Fisher information matrix of all WiFi devices; Introduce the receiver selection matrix of all WiFi devices, and correct the initial Jacobian matrix of all WiFi devices to obtain the Jacobian matrix carrying the selection matrix; Based on the Jacobian matrix carrying the selection matrix and the initial Fisher information matrix, obtain the corrected Fisher information matrix of all WiFi devices; Based on the corrected Fisher information matrix of all WiFi devices, determine the target WiFi device when maximizing the determinant of the corrected Fisher information matrix of all WiFi devices; Track the target object based on the target WiFi device.

2. The method according to claim 1, characterized in that The determining the Fisher information matrix of the Doppler frequency shift of the CSI data of any one WiFi device based on the CSI data of each antenna of any one WiFi device includes: Process the CSI data of any two antennas of any one WiFi device to obtain the initial quotient signal of the CSI data of any one WiFi device; When the initial quotient signal of the CSI data of any one WiFi device is interfered by environmental noise, obtain the target quotient signal of the CSI data of any one WiFi device, where the target quotient signal carries a noise signal; Based on the target quotient signal of the CSI data of any one WiFi device, obtain the Fisher information matrix of the Doppler frequency shift of the CSI data of any one WiFi device.

3. The method according to claim 2, wherein The processing the CSI data of any two antennas of any one WiFi device to obtain the initial quotient signal of the CSI data of any one WiFi device includes: Determine the quotient of the CSI data of any two antennas of any one WiFi device as the initial quotient signal of the CSI data of any one WiFi device.

4. The method according to claim 2, characterized in that, The obtaining of the Fisher information matrix of the Doppler frequency shift of the CSI data of any one Wi-Fi device based on the target quotient signal of the CSI data of any one Wi-Fi device includes: Obtaining the log-likelihood function of the target quotient signal of the CSI data of any one Wi-Fi device based on the target quotient signal of the CSI data of any one Wi-Fi device; Obtaining the Fisher information matrix of the Doppler frequency shift of the CSI data of any one Wi-Fi device based on the log-likelihood function of the target quotient signal of the CSI data of any one Wi-Fi device.

5. The method according to claim 4, wherein The obtaining of the log-likelihood function of the target quotient signal of the CSI data of any one Wi-Fi device based on the target quotient signal of the CSI data of any one Wi-Fi device includes: Obtaining the probability density function of the target quotient signal of the CSI data of any one Wi-Fi device based on the target quotient signal of the CSI data of any one Wi-Fi device; Obtaining the log-likelihood function of the target quotient signal of the CSI data of any one Wi-Fi device based on the probability density function of the target quotient signal of the CSI data of any one Wi-Fi device.

6. The method according to claim 5, characterized in that, The obtaining of the Fisher information matrix of the Doppler frequency shift of the CSI data of any one Wi-Fi device based on the log-likelihood function of the target quotient signal of the CSI data of any one Wi-Fi device includes: Obtaining the Fisher information matrix of the Doppler frequency shift of the CSI data of any one Wi-Fi device based on the first derivative and the second derivative of the log-likelihood function of the target quotient signal of the CSI data of any one Wi-Fi device with respect to the Doppler frequency shift of the CSI data of any one Wi-Fi device and the Fisher information matrix.

7. The method according to claim 1, wherein The establishing of the relationship model between the Doppler frequency shift of the receivers of all Wi-Fi devices and the velocity parameter of the target object based on the Fisher information matrix of the Doppler frequency shift of the CSI data of any one Wi-Fi device includes: Obtaining the Cramér-Rao lower bound of the Doppler frequency shift of the CSI data of any one Wi-Fi device based on the Fisher information matrix of the Doppler frequency shift of the CSI data of any one Wi-Fi device; Establishing the relationship model between the Doppler frequency shift of the receivers of all Wi-Fi devices and the velocity parameter of the target object based on the Cramér-Rao lower bound of the Doppler frequency shift of the CSI data of any one Wi-Fi device.

8. A computer system, characterized in that Including: One or more processors, a computer-readable storage medium for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method according to claim 1.

9. A computer-readable storage medium, characterized in that Stored with computer-executable instructions, the instructions being used to implement the method according to claim 1 when executed.

10. A computer program product, characterized in that Including computer-executable instructions, the instructions being used to implement the method according to claim 1 when executed.

Images