A method, system, and storage medium for respiratory detection based on WiFi signals.

Abstract

This invention discloses a method, system, and storage medium for respiratory detection based on WiFi signals, applicable to the field of respiratory detection. The WiFi-based respiratory detection method includes the following steps: acquiring the CSI signal corresponding to the target object; performing discrete value removal and wavelet denoising on the CSI signal to obtain a preprocessed signal waveform; performing waveform adjustment and data fusion on the preprocessed signal waveform to obtain a principal component analysis-variable mode decomposition (PCA-VMD) signal; and determining the target respiratory frequency based on the PCA-VMD signal. This invention improves the accuracy of obtaining the target respiratory frequency by performing discrete value removal, wavelet denoising, waveform adjustment, and data fusion on the acquired CSI signal to obtain the PCA-VMD signal, and then using this signal to determine the target respiratory frequency.

Description

Technical Field

[0001] This invention relates to the field of respiratory detection, and in particular to a respiratory detection method, system, and storage medium based on WiFi signals. Background Technology

[0002] Respiratory rate is a crucial physiological indicator of an individual's health, and respiratory-related diseases include chronic obstructive pulmonary disease (COPD), sleep apnea, and nocturnal hypopnea syndrome. Therefore, respiratory monitoring during sleep is significant for early disease prevention and health assessment. In recent years, non-contact respiratory sensing based on WiFi signals has attracted widespread attention from academia and industry. However, because WiFi-based sensing relies on weak reflected signals, the different positions of the person relative to the transceiver and the diversity of subcarrier frequencies can affect the subcarrier's ability to detect respiratory movements. Existing WiFi CSI-based respiratory detection methods mostly detect human respiratory movements by combining multiple subcarriers with different sensing capabilities. Traditional subcarrier signal combining methods are mostly based on signal weighting, which cannot avoid the influence of some interference signals, resulting in low accuracy in respiratory detection. Summary of the Invention

[0003] This invention aims to at least solve one of the technical problems existing in the prior art. To this end, this invention proposes a respiratory detection method, system, and storage medium based on WiFi signals, which can effectively improve the accuracy of respiratory detection.

[0004] A first aspect of the present invention provides a respiratory detection method based on WiFi signals, comprising the following steps:

[0005] Obtain the CSI signal corresponding to the target object;

[0006] The CSI signal is subjected to discrete value removal and wavelet denoising to obtain a preprocessed signal waveform;

[0007] The preprocessed signal waveform is subjected to waveform adjustment and data fusion to obtain the principal component analysis-variational mode decomposition signal;

[0008] The target respiratory rate is determined based on the principal component analysis-variable mode decomposition signal.

[0009] According to an embodiment of the present invention, a respiratory detection method based on WiFi signals has at least the following beneficial effects: The present invention improves the accuracy of respiratory detection by performing discrete value removal and wavelet denoising on the CSI signal corresponding to the acquired object, followed by waveform adjustment and data fusion, principal component analysis and variational mode decomposition on the signal, and then determining the target respiratory frequency based on the principal component analysis-variational mode decomposition signal.

[0010] According to some embodiments of the present invention, the method for obtaining the CSI signal corresponding to the target object includes:

[0011] Acquire several real-time data points collected by the acquisition device, wherein the real-time data includes the complex values ​​of the 128 subcarriers after orthogonal frequency division multiplexing;

[0012] The complex value is moduloed to obtain the corresponding CSI signal.

[0013] According to some embodiments of the present invention, the discrete value removal of the CSI signal includes:

[0014] Get the first sliding window with the first preset duration;

[0015] Obtain discrete signals whose CSI signals fall within the first sliding window and are greater than the average value;

[0016] Remove the discrete signal from the CSI signal.

[0017] According to some embodiments of the present invention, wavelet denoising of the CSI signal includes:

[0018] Perform wavelet transform on each subcarrier sequence corresponding to the CSI signal after removing discrete values ​​according to a preset decibel level;

[0019] The wavelet-transformed signal is reconstructed based on the approximation coefficients.

[0020] According to some embodiments of the present invention, the waveform adjustment of the preprocessed signal waveform includes:

[0021] Adjust the direction of all subcarriers corresponding to the preprocessed signal waveform;

[0022] Get the second sliding window with the second preset duration;

[0023] The low-frequency components on the direction-adjusted signal are filtered out by the second sliding window.

[0024] According to some embodiments of the present invention, the data fusion of the preprocessed signal waveform includes:

[0025] The covariance matrix is ​​calculated after centering all signals after filtering out low-frequency components.

[0026] The eigenvalues ​​and eigenvectors of the signal after filtering out low-frequency components are calculated based on the truncated singular values ​​and the covariance matrix.

[0027] Calculate the principal component analysis signal based on the largest eigenvalue and the eigenvector;

[0028] The principal component analysis signal is decomposed into multiple modal components;

[0029] The modal component with the largest variance is taken as the optimal solution.

[0030] According to some embodiments of the present invention, determining the target respiratory rate based on the principal component analysis-variable mode decomposition signal includes:

[0031] Get the third sliding window;

[0032] The local maximum point of the principal component analysis-variational mode decomposition signal is found according to the third sliding window;

[0033] Multiple peak points are determined based on the local maximum point;

[0034] Calculate the average of the intervals between all adjacent peak points;

[0035] The reciprocal of the average value is taken as the target respiratory rate.

[0036] A second aspect of the present invention provides a respiratory detection system based on WiFi signals, comprising:

[0037] The acquisition module is used to acquire the CSI signal corresponding to the target object;

[0038] The data preprocessing module performs discrete value removal and wavelet denoising on the CSI signal to obtain the preprocessed signal waveform;

[0039] The adaptive data fusion module performs waveform adjustment and data fusion on the preprocessed signal waveform to obtain the principal component analysis-variational mode decomposition signal.

[0040] The calculation module determines the target respiratory rate based on the principal component analysis-variational mode decomposition signal.

[0041] A third aspect of the present invention provides a respiratory detection system based on WiFi signals, comprising:

[0042] At least one memory for storing programs;

[0043] At least one processor is configured to load the program to execute a WiFi signal-based respiratory detection method according to a first aspect embodiment.

[0044] A fourth aspect of the present invention provides a storage medium storing a processor-executable program, which, when executed by a processor, performs a respiratory detection method based on a WiFi signal according to a first aspect embodiment.

[0045] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0046] The present invention will be further described below with reference to the accompanying drawings and embodiments, wherein:

[0047] Figure 1 This is a flowchart of a respiratory detection method based on WiFi signals according to an embodiment of the present invention. Detailed Implementation

[0048] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0049] In the description of this invention, "several" means one or more, "multiple" means two or more, "greater than," "less than," and "exceeding" are understood to exclude the stated number, while "above," "below," and "within" are understood to include the stated number. The use of "first" and "second" in the description is merely for distinguishing technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or implicitly indicating the order of the indicated technical features.

[0050] In the description of this invention, unless otherwise explicitly defined, terms such as "setup" should be interpreted broadly, and those skilled in the art can reasonably determine the specific meaning of the above terms in this invention in conjunction with the specific content of the technical solution.

[0051] In the description of this invention, the terms "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., refer to specific features or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0052] In one example scenario, indoors where a WiFi device is placed, the radio frequency signal propagates from the transmitter to the receiver along multiple paths. When someone is present within the sensing range, the chest movement caused by breathing continuously alters the length of the propagation path. Reflection and diffraction of the signal by the human body introduce additional paths. These additional paths cause variations in the subcarrier amplitude, which are described in this embodiment using CSI signals.

[0053] In this scenario, refer to Figure 1 This invention provides a respiratory detection method based on WiFi signals. The implementation of this invention includes the following steps:

[0054] S1. Obtain the CSI signal corresponding to the target object.

[0055] The target object can be the person currently requiring respiratory detection. The CSI signal corresponding to the target object is obtained by collecting several real-time data points from a data acquisition device. These real-time data points include complex values ​​of 128 subcarriers after orthogonal frequency division multiplexing, and the modulus of these complex values ​​is then taken to obtain the signal.

[0056] Specifically, CSI (Channel State Information) is channel state information. In the field of wireless communication, CSI describes the channel attributes of a communication link, describing the signal attenuation factors on each transmission path, i.e., the value of each element in the channel gain matrix H, such as signal scattering, environmental fading (multipath fading or shadowing fading), and power decay of distance. CSI enables communication systems to adapt to current channel conditions, providing a guarantee for high-reliability and high-speed communication in multi-antenna systems.

[0057] The data acquisition device can use a pair of ESP32-WROOM-32U modules, one of which acts as a transmitter and the other as a receiver, each connected to an omnidirectional antenna. The WiFi signal operates on the 2.4GHz band. In the experiment, the line-of-sight distance between the transmitter and receiver was 1.2 meters, and the devices were placed slightly above chest height. This device has advantages such as small size, easy deployment, low cost, and network capability, making it suitable for home environments as an IoT device.

[0058] S2. Perform discrete value removal and wavelet denoising on the CSI signal to obtain the preprocessed signal waveform.

[0059] In the implementation process, step S2 may include discrete value removal and wavelet denoising. Discrete value removal can be achieved by obtaining a first sliding window of a first preset duration, identifying discrete signals whose CSI signals fall within the first sliding window and are greater than the average value, and then removing these discrete signals from the CSI signal. Specifically, the first preset duration can be set to values ​​such as 2 seconds, 5 seconds, or 10 seconds, and can be selected according to actual conditions. The first sliding window refers to the sliding window corresponding to the first preset duration. In this embodiment, due to the instability of the WiFi link, the acquired CSI signal may have abrupt changes in value. This embodiment can use a Hamper filter to identify discrete points in the original CSI signal. The Hamper filter is a decision filter based on median absolute deviation estimation and can be used to identify singular data. For example, the first preset duration is set to 5 seconds, and its corresponding sliding window is used as the first sliding window. Then, points within the sliding window whose CSI signals are significantly greater than the average value are identified, treated as discrete points, and removed.

[0060] After removing discrete values ​​from the CSI signal, this embodiment also requires wavelet denoising. The wavelet denoising process involves performing a wavelet transform on each subcarrier sequence of the CSI signal after discrete value removal, based on a preset decibel level, and then reconstructing the wavelet-transformed signal using approximation coefficients. Specifically, after removing discrete points, the original CSI signal contains a large amount of Gaussian noise that affects the low-frequency breathing signal. Therefore, this embodiment uses a wavelet filter to remove high-frequency noise because wavelet filters can better preserve sharp transitions in the signal than other low-pass filters. For example, a 4-decibel level is selected as the preset decibel level for wavelet transform on each subcarrier sequence, and approximation coefficients are used to "reconstruct" the filtered signal.

[0061] S3. Perform waveform adjustment and data fusion on the preprocessed signal waveform to obtain the principal component analysis-variational mode decomposition signal.

[0062] In this embodiment, due to the diversity of subcarrier frequencies and the influence of the subject's position, the directions of subcarriers obtained from different antennas may be the same or opposite. Although the different directions of a single respiratory signal change do not affect the estimation of respiratory frequency, positive and negative respiratory waveforms may cancel each other out during multi-subcarrier fusion, thereby weakening the effective information in the signal. Therefore, in this embodiment, after preprocessing the signal, the direction of all subcarriers corresponding to the preprocessed signal waveform is adjusted, i.e., the direction of all subcarriers is adjusted. The direction adjustment process can find the coordinates of the first peak and valley points in the first few seconds of respiratory detection, thereby determining whether the respiratory signal change during this time is rising or falling. By traversing the first unidirectional respiratory change of all subcarriers, the signal with the opposite direction of change is multiplied by -1, and the direction of all subcarrier signals is adjusted to be consistent with the direction of the signal with the maximum signal-to-noise ratio. In addition, since the signal may experience "baseline" drift due to factors such as power distribution during transmission, this embodiment obtains a second sliding window of a second preset duration, and then filters out low-frequency components on the direction-adjusted signal through the second sliding window. The second preset duration can be set to different values ​​such as 1 second, 2 seconds, and 3 seconds, which can be determined according to the actual situation. For example, when 1 second is selected as the second preset duration, the 1-second moving window serves as the second sliding window, which is then used to filter out low-frequency components in the signal, thereby removing "baseline" noise. All subcarriers exhibit synchronized fluctuations after waveform adjustment, and different subcarriers have varying sensitivities to breathing perception.

[0063] After waveform adjustment, to obtain a robust breathing waveform, this embodiment proposes a Principal Component Analysis-Variational Mode Decomposition (PCA-VMD) method to fuse all preprocessed subcarriers into a single breathing signal. Specifically, this embodiment calculates the covariance matrix after centering all signals after filtering out low-frequency components; calculates the eigenvalues ​​and eigenvectors of the signal after filtering out low-frequency components based on the truncated singular values ​​and the covariance matrix; calculates the principal component analysis signal based on the largest eigenvalue and eigenvector; decomposes the principal component analysis signal into multiple modal components; and selects the modal component with the largest variance as the optimal solution. For example, this embodiment can employ Principal Component Analysis (PCA) to convert the original eigenvectors extracted from the subcarriers into a set of linearly uncorrelated principal components, eliminating irrelevant noise components in the features, and the transformed principal components have the largest variance. Understandably, the algorithm first centers all signals after filtering out low-frequency components and calculates the covariance matrix. Then, it uses truncated singular value decomposition to solve for the eigenvalues ​​and eigenvectors of the signals after filtering out low-frequency components. Finally, it selects the largest eigenvalue and eigenvector to calculate the one-dimensional principal component analysis (PCA) signal. The PCA method extracts the dominant components from multiple subcarrier signals, removes redundant noise components, improves the signal-to-noise ratio, and has a significant effect on the fusion of signals with different sensitivities.

[0064] After obtaining the PCA signal, this embodiment decomposes the PCA signal into multiple modal components using mode decomposition, extracting the effective respiratory frequency component. In this embodiment, the respiratory frequency fluctuates within a finite range around a center frequency over a period of time, and the respiratory signal is a superposition of the respiratory frequency component and other non-target frequency components. Variational mode decomposition (VMD) is used to perform mode decomposition, selecting the optimal mode from the decomposed components. VMD is an adaptive, fully non-recursive method for variational mode decomposition and signal processing. This technique has the advantage of being able to determine the number of mode decompositions. Its adaptability is manifested in determining the number of mode decompositions for a given sequence based on the actual situation. After manually specifying the number of mode decompositions, the search and solution process can adaptively match the optimal center frequency and finite bandwidth for each mode, and can achieve effective separation of intrinsic modal components and frequency domain partitioning of the signal. This yields the effective decomposed components of the given signal, ultimately obtaining the optimal solution to the variational problem. After mode decomposition, the PCA signal can obtain a given number of K modal components; the mode with the largest variance is considered the optimal solution for the respiratory signal. In the embodiment, when K=3, the obtained principal component analysis-variational mode decomposition signal is very similar to and smoother than the signal before decomposition. The principal component analysis-variational mode decomposition signal is used as the final fusion result.

[0065] S4. Determine the target respiratory rate based on the principal component analysis-variational mode decomposition signal.

[0066] In this embodiment, after obtaining a third sliding window, local maxima of the principal component analysis-variational mode decomposition (PCA-VMD) signal are found within the third sliding window. Multiple peak points are then determined based on these local maxima. The average value of the intervals between all adjacent peak points is calculated. The reciprocal of this average value is used as the target respiratory rate. For example, this embodiment uses a PCA-VMD fused signal to describe the frequency changes of respiratory motion, and the respiratory rate of a single individual is calculated using peak-to-peak intervals. For instance, a moving window is used as the third sliding window, and local maxima are found through this third sliding window to determine peak points. After finding all peaks, the reciprocal of the average value of the intervals between all adjacent peaks is used as the target respiratory rate.

[0067] This invention provides a respiratory detection system based on WiFi signals, comprising:

[0068] The acquisition module is used to acquire the CSI signal corresponding to the target object;

[0069] The data preprocessing module performs discrete value removal and wavelet denoising on the CSI signal to obtain the preprocessed signal waveform;

[0070] The adaptive data fusion module performs waveform adjustment and data fusion on the preprocessed signal waveform to obtain the principal component analysis-variational mode decomposition signal.

[0071] The calculation module determines the target respiratory rate based on the principal component analysis-variational mode decomposition signal.

[0072] The content of the method embodiments of the present invention is applicable to the system embodiments. The specific functions implemented in the system embodiments are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above methods.

[0073] This invention also provides a respiratory detection system based on WiFi signals, comprising:

[0074] At least one memory for storing programs;

[0075] At least one processor is used to load the program to execute such... Figure 1 The aforementioned respiratory detection method based on WiFi signals.

[0076] The content of the method embodiments of the present invention is applicable to the system embodiments. The specific functions implemented in the system embodiments are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above methods.

[0077] This invention also provides a storage medium storing a processor-executable program, which is used to perform actions when executed by a processor. Figure 1 This illustrates a respiratory detection method based on WiFi signals.

[0078] This application also discloses a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium and execute the computer instructions, causing the computer device to perform... Figure 1 The respiratory detection method based on WiFi signals is shown.

[0079] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments, and various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention. Furthermore, the embodiments of the present invention and the features thereof can be combined with each other unless otherwise specified.

Claims

1. A respiratory detection method based on WiFi signals, characterized in that, Includes the following steps: Obtain the CSI signal corresponding to the target object; The CSI signal is subjected to discrete value removal and wavelet denoising to obtain a preprocessed signal waveform; The preprocessed signal waveform is subjected to waveform adjustment and data fusion to obtain the principal component analysis-variational mode decomposition signal; The target respiratory rate is determined based on the principal component analysis-variable mode decomposition signal. The data fusion of the preprocessed signal waveform includes: The covariance matrix is ​​calculated after centering all signals after filtering out low-frequency components. The eigenvalues ​​and eigenvectors of the signal after filtering out low-frequency components are calculated based on the truncated singular values ​​and the covariance matrix. Calculate the principal component analysis signal based on the largest eigenvalue and the eigenvector; The principal component analysis signal is decomposed into multiple modal components; The modal component with the largest variance is taken as the optimal solution; The determination of the target respiratory rate based on the principal component analysis-variable mode decomposition signal includes: Get the third sliding window; The local maximum point of the principal component analysis-variational mode decomposition signal is found according to the third sliding window; Multiple peak points are determined based on the local maximum point; Calculate the average of the intervals between all adjacent peak points; The reciprocal of the average value is taken as the target respiratory rate.

2. The respiratory detection method based on WiFi signals according to claim 1, characterized in that, The acquisition of the CSI signal corresponding to the target object includes: Acquire several real-time data points collected by the acquisition device, wherein the real-time data includes the complex values ​​of the 128 subcarriers after orthogonal frequency division multiplexing; The complex value is moduloed to obtain the corresponding CSI signal.

3. The respiratory detection method based on WiFi signals according to claim 1, characterized in that, The discrete value removal of the CSI signal includes: Get the first sliding window with the first preset duration; Obtain discrete signals whose CSI signals fall within the first sliding window and are greater than the average value; Remove the discrete signal from the CSI signal.

4. The respiratory detection method based on WiFi signals according to claim 3, characterized in that, The wavelet denoising of the CSI signal includes: Perform wavelet transform on each subcarrier sequence corresponding to the CSI signal after removing discrete values ​​according to a preset decibel level; The wavelet-transformed signal is reconstructed based on the approximation coefficients.

5. The method for respiratory detection based on WiFi signals according to claim 1, characterized in that, The waveform adjustment of the preprocessed signal waveform includes: Adjust the direction of all subcarriers corresponding to the preprocessed signal waveform; Get the second sliding window with the second preset duration; The low-frequency components on the direction-adjusted signal are filtered out by the second sliding window.

6. A system for implementing the respiratory detection method based on WiFi signals as described in any one of claims 1-5, characterized in that, include: The acquisition module is used to acquire the CSI signal corresponding to the target object; The data preprocessing module performs discrete value removal and wavelet denoising on the CSI signal to obtain the preprocessed signal waveform; The adaptive data fusion module performs waveform adjustment and data fusion on the preprocessed signal waveform to obtain the principal component analysis-variational mode decomposition signal. The calculation module determines the target respiratory rate based on the principal component analysis-variational mode decomposition signal.

7. A respiratory detection system based on WiFi signals, characterized in that, include: At least one memory for storing programs; At least one processor is configured to load the program to execute the WiFi signal-based respiratory detection method as described in any one of claims 1-5.

8. A storage medium storing a processor-executable program, characterized in that, The program executable by the processor is used, when executed by the processor, to perform a respiratory detection method based on WiFi signals as described in any one of claims 1-5.

Images