A Multi-Antenna RFID Respiratory Monitoring Method and System with Motion Interference Resistance

Abstract

This invention discloses a multi-antenna RFID respiratory monitoring method and system with resistance to motion interference, comprising the following steps: S1: RFID tags are placed on the respiratory fluctuations of the monitored object, and a multi-antenna array is used for polling and reading to obtain the backscatter phase sequence of multiple antenna channels; S2: Preprocessing is performed to construct a multi-channel phase observation matrix; S3: Principal component analysis is performed on the phase observation matrix to obtain multiple principal component components; S4: Motion interference components are identified and removed; S5: Among the retained principal component components, the target principal component that best matches the respiratory characteristics is selected; S6: The target principal component is filtered, and the frequency of the filtered signal is estimated to obtain the respiratory rate. This invention utilizes multi-antenna spatial diversity to acquire tag backscatter phase information, and enhances the ability to suppress motion interference during respiratory monitoring through multi-channel phase preprocessing, principal component analysis, motion component identification, and target principal component selection.

Description

Technical Field

[0001] This invention relates to the field of respiratory monitoring, and in particular to a multi-antenna RFID respiratory monitoring method and system resistant to motion interference. Background Technology

[0002] Respiratory monitoring has core value in various scenarios such as clinical medicine, intensive care, chronic disease management, sleep medicine, and sports and health, enabling early warning, diagnosis, treatment, and prognosis assessment. However, human movement significantly interferes with respiratory monitoring. Breathing causes millimeter-level micro-displacements, while movements such as turning over, arm swinging, and sitting up produce centimeter-level displacements with energy far exceeding that of respiratory micro-movements, easily masking respiratory characteristics in the phase signal. Therefore, it is necessary to propose a respiratory monitoring method and system that is resistant to motion interference. Summary of the Invention

[0003] Purpose of the invention: In order to overcome the shortcomings of the existing technology, the present invention provides a multi-antenna RFID respiratory monitoring method and system with resistance to motion interference. It utilizes multi-antenna spatial diversity to collect backscatter phase information of tags, and through multi-channel phase preprocessing, principal component analysis, body motion component identification and target principal component screening, it realizes the extraction and continuous monitoring of respiratory-related signals of the monitored object, and enhances the ability to suppress body motion interference during respiratory monitoring.

[0004] Technical Solution: To achieve the above objectives, the present invention provides a multi-antenna RFID respiratory monitoring method with anti-motion interference, comprising the following steps: S1: RFID tags are placed on the respiratory fluctuations of the monitored object, and the multi-antenna array is controlled by an RFID reader to poll and read the RFID tags to obtain the backscatter phase sequences of multiple antenna channels; S2: The phase sequences of each antenna channel are preprocessed to construct a multi-channel phase observation matrix; S3: Principal component analysis is performed on the phase observation matrix to obtain multiple principal component components; S4: The motion interference components are identified based on the variance contribution rate and spectral characteristics of the principal component components, and the identified motion interference components are removed; S5: From the principal component components retained after removing the motion interference components, the target principal component that best matches the respiratory characteristics is selected; S6: The target principal component is filtered, and the frequency of the filtered signal is estimated to obtain the respiratory rate.

[0005] Further, in S1, the polling reading includes the following steps: S1.1: Select one antenna in the multi-antenna array as the current working antenna; S1.2: Control the current working antenna to transmit radio frequency signals to the target area; S1.3: Receive the backscattered signal from the RFID tag and record the timestamp and phase information of the object; S1.4: Switch to the next antenna and repeat the process in S1.2 and S1.3 until a round of acquisition of all antennas is completed; S1.5: Repeat the process from S1.1 to S1.4 according to the set sampling frequency to form the time series phase data of each antenna channel.

[0006] Furthermore, in S2, a multi-channel phase observation matrix is ​​constructed. , for A 3D matrix, denoted as , represented as ;in, To standardize the number of sampling points on the time axis, This refers to the number of antenna channels. For the first The length corresponding to each antenna channel is A time series vector.

[0007] Furthermore, in S3, the phase observation matrix... When performing principal component analysis, first analyze the phase observation matrix. Centralize the data and construct the covariance matrix:

[0008] ;

[0009] Then, eigenvalue decomposition is performed on the covariance matrix to obtain the eigenvalue matrix. and eigenvector matrix Then, the observation matrix is ​​projected onto the principal component space to obtain the principal component matrix. , The total number of principal components.

[0010] Furthermore, in S4, the first Variance contribution rate of each principal component for:

[0011] ;

[0012] in, For the first Each principal component corresponds to an eigenvalue. The total number of principal components is determined; then, the spectral energy of each principal component within the target breathing frequency band is calculated. And spectral energy outside the target breathing frequency band When a principal component satisfies Higher than the preset value ,and Greater than When this happens, the principal component is identified as a motion interference component.

[0013] Furthermore, in S5, from the principal component components retained after removing the body motion interference components, the principal component component that best matches the respiratory characteristics is selected based on the significance of the spectral peaks, the position of the dominant frequency, and the proportion of frequency band energy within the target respiratory frequency band. As the target principal component signal .

[0014] Furthermore, in S6, when filtering the target principal component, the target principal component is first bandpass filtered to suppress non-breathing frequency band noise; then, the passband range of the bandpass filter is adjusted according to the estimated principal frequency of the filtered signal in the target breathing frequency band; finally, the adjusted filtering result is smoothed to reduce residual glitches and local fluctuations.

[0015] Furthermore, the RFID tag is any one of a passive tag, an active tag, or a semi-active tag, and the multi-antenna array includes at least two antennas with different spatial positions or different polarization directions.

[0016] Furthermore, a multi-antenna RFID respiratory monitoring system resistant to motion interference includes an RFID tag, a multi-antenna array, an RFID reader, and a data processing terminal. The RFID tag is placed on the breathing fluctuation area of ​​the monitored object to receive radio frequency signals and return backscattered signals carrying phase information. The multi-antenna array is used to cover the target monitoring area and receive backscattered signals from the RFID tag. The RFID reader is connected to the multi-antenna array and is used to control the multi-antenna array to poll and acquire raw phase data from multiple antenna channels. The data processing terminal is communicatively connected to the RFID reader and is configured to execute the respiratory monitoring method.

[0017] Furthermore, the data processing terminal is a software processing terminal running a computer program, or a hardware processing terminal implemented using at least one of FPGA, DSP, and ASIC; the data processing terminal includes a preprocessing module, a principal component analysis module, a component discrimination and screening module, and a filtering and solving module; the preprocessing module is used to perform at least one of the following processing on the original phase data: phase unwrapping, outlier removal, interpolation, time alignment, resampling, and normalization, to construct a multi-channel phase observation matrix; the principal component analysis module is used to perform principal component analysis on the multi-channel phase observation matrix; the component discrimination and screening module is used to discriminate the body motion interference components based on the variance contribution rate and spectral characteristics, and to screen the target principal components from the non-body motion components; the filtering and solving module is used to filter the target principal components and output the respiration rate.

[0018] Beneficial effects: The multi-antenna RFID respiratory monitoring method and system of the present invention, which resists motion interference, realizes the extraction and continuous monitoring of respiratory-related signals of the monitored object through multi-antenna polling acquisition, multi-channel time alignment, principal component separation, body motion component discrimination, target principal component screening, filtering processing and respiratory rate calculation; in particular, the method enhances the ability to suppress body motion interference and improves the separation effect of respiratory-related components by using principal component analysis and a body motion component discrimination mechanism based on variance contribution rate and spectral energy distribution. Attached Figure Description

[0019] Appendix Figure 1 This is a schematic diagram of the system scenario of the present invention;

[0020] Appendix Figure 2 This is a flowchart of the method of the present invention;

[0021] Appendix Figure 3 This is a diagram of the original phase signals of the multi-antenna array.

[0022] Appendix Figure 4 Principal component analysis results;

[0023] Appendix Figure 5 The respiratory signal image after principal component filtering;

[0024] Appendix Figure 6 This is a spectrum of respiratory signals. Detailed Implementation

[0025] The invention will now be further described with reference to the accompanying drawings.

[0026] As attached Figures 1 to 6 The aforementioned multi-antenna RFID respiratory monitoring method for resisting motion interference includes the following steps.

[0027] S1: Multi-channel data acquisition. RFID tags are placed on the breathing undulations of the monitored object. An RFID reader controls a multi-antenna array to poll and read the RFID tags, acquiring the backscatter phase sequence of multiple antenna channels.

[0028] The RFID tag is any one of a passive tag, an active tag, or a semi-active tag, and is placed on the chest, abdomen, or other parts of the monitored object that rise and fall significantly with breathing.

[0029] The multi-antenna array includes at least two antennas with different spatial locations or polarization directions to acquire phase perturbation information of the monitored object under different propagation paths through spatial diversity. (See attached image) Figure 3 In the illustrated embodiment, the multi-antenna array includes three antennas, but is not limited to this; it can also include two, four, eight, or other numbers. Each antenna can be positioned in a different spatial location and / or employ different polarization directions to receive backscattered signals returned by the RFID tag from different propagation paths.

[0030] In S1, the RFID reader activates each antenna sequentially in a time-division multiplexing manner through an internal radio frequency switching mechanism. For example... Figure 3 The image shows the raw phase data collected. The polling reading process includes the following steps: S1.1: Select one antenna from the multi-antenna array as the current working antenna; S1.2: Control the current working antenna to transmit radio frequency signals towards the target area, i.e., the breathing fluctuation area of ​​the monitored object; S1.3: Receive the backscattered signal from the RFID tag and record the object's timestamp and phase information; S1.4: Switch to the next antenna and repeat steps S1.2 and S1.3 until one round of acquisition from all antennas is completed; S1.5: Repeat steps S1.1 to S1.4 according to the set sampling frequency to form time-series phase data for each antenna channel.

[0031] S2: Data Preprocessing. The phase sequences of each antenna channel are preprocessed to construct a multi-channel phase observation matrix.

[0032] In the data preprocessing stage of S2, preprocessing includes phase dewinding, time alignment, and resampling. It also includes outlier removal, interpolation, and normalization of the phase sequences of each antenna channel to reduce channel imbalance caused by interleaved polling timestamps and differences in antenna gain and path loss, and to align the data of each channel on a unified time axis. Outlier removal includes identifying and removing phase transition points, read / write failure points, or data points exceeding a preset statistical range. Time alignment includes interpolating the original time sequences of each channel and mapping them to a unified time axis. Resampling includes generating a multi-channel synchronization sequence at a unified sampling frequency.

[0033] In actual preprocessing, the original phase sequence of each antenna channel is first unwrapped to eliminate phase mode distortion. The impact of jumps on continuity analysis is analyzed; then, outliers are removed from data points with missing, abrupt, or abnormal fluctuations; next, data from each channel is interpolated based on the original timestamps and mapped to a unified time axis; then, data is resampled at a unified sampling frequency to form a synchronized multi-channel phase sequence. Finally, the data from each channel is normalized to reduce the impact of differences in gain and propagation loss between different channels.

[0034] In S2, assume that the preprocessed multi-channel phase data has a total of [number] channels on the same time axis. There are sampling points in the system, with a total of _ sampling points. If there are multiple antenna channels, a multi-channel phase observation matrix is ​​constructed. , for A 3D matrix, denoted as , represented as .in, To standardize the number of sampling points on the time axis, This refers to the number of antenna channels. For the first The length corresponding to each antenna channel is A time series vector.

[0035] S3: Principal Component Analysis. Principal component analysis is performed on the phase observation matrix to obtain multiple principal component components and their corresponding eigenvalues.

[0036] In S3, to extract the main variation components from different sources, the phase observation matrix is... Principal component analysis was performed. During the analysis, the phase observation matrix was first analyzed. Centralize the data and construct the covariance matrix:

[0037] ;

[0038] Then, eigenvalue decomposition is performed on the covariance matrix to obtain the eigenvalue matrix. and eigenvector matrix ,in, For the first The eigenvalues ​​corresponding to each principal component. For the first The eigenvectors corresponding to each principal component are then projected onto the principal component space to obtain the principal component matrix. , For the first Principal component components The total number of principal components.

[0039] S4: Identification of motion interference components. Based on the variance contribution rate and spectral characteristics of the principal component components, motion interference components are identified and then discarded.

[0040] In S4, the first Variance contribution rate of each principal component for:

[0041] ;

[0042] in, For the first The eigenvalues ​​corresponding to each principal component. The total number of principal components.

[0043] Then, the spectral energy of each principal component within the target respiratory frequency band is calculated. And spectral energy outside the target breathing frequency band In the specific calculation, for the i-th principal component... Perform a fast Fourier transform or power spectrum estimation to obtain its power spectrum. Using the target respiratory frequency band B as the integration or summation interval, for Integral or summation of the energy at each frequency point within B. for The integration or summation of energy at each frequency point outside of B and within the analysis frequency range. Typically, the target breathing frequency band is 0.1 Hz to 0.6 Hz. When a principal component satisfies... Higher than the preset value ,and Greater than When this occurs, the principal component is identified as a motion interference component and removed. The preset value is... It can be a fixed threshold or an adaptive threshold.

[0044] S5: Target Principal Component Screening. From the principal component components retained after removing motion interference components, the target principal components that best match the respiratory characteristics are screened.

[0045] In S5, after removing the body motion interference component, spectral analysis is performed on the remaining principal components. From the principal components retained after removing the body motion interference component, each principal component is compared based on the significance of the spectral peaks, the position of the dominant frequency, and the proportion of frequency band energy within the target respiratory frequency band. The principal component component that best matches the respiratory characteristics is then selected. As the target principal component signal The target principal component can be used as the input signal for subsequent respiratory signal extraction and respiratory rate estimation. (See attached diagram) Figure 4 The diagram shown is a principal component plot after decomposition.

[0046] S6: Filtering and Respiratory Rate Calculation. The target principal component is filtered, and the frequency of the filtered signal is estimated to obtain the respiratory rate.

[0047] In S6, when filtering the target principal component, a bandpass filter is first applied to suppress non-breathing band noise. Then, based on the estimated principal frequency of the filtered signal within the target breathing band, the passband range of the bandpass filter is adjusted. The passband can be set to 0.1Hz to 0.5Hz or 0.1Hz to 0.6Hz depending on the application scenario. Finally, the adjusted filtering result is smoothed to reduce residual glitches and local fluctuations. Figure 5 The image shows the filtered respiratory signal.

[0048] During the respiratory rate calculation stage, spectral analysis is performed on the filtered respiratory signal, such as... Figure 6 The image shown is a respiratory spectrogram; the main peak frequency was extracted. The respiratory rate is calculated based on the following relationship: .in, The respiratory rate is the number of breaths per minute. It can also be estimated using methods such as time-domain peak interval statistics, short-time Fourier analysis, or autocorrelation analysis.

[0049] This invention also provides a motion-interference-resistant multi-antenna RFID respiratory monitoring system, comprising RFID tags, a multi-antenna array, an RFID reader, and a data processing terminal. The RFID tags are positioned on the breathing fluctuations of the monitored object, receiving radio frequency signals and returning backscattered signals carrying phase information. The multi-antenna array covers the target monitoring area and receives backscattered signals from the RFID tags. The RFID reader is connected to the multi-antenna array and controls it to poll and acquire raw phase data from multiple antenna channels. The data processing terminal is communicatively connected to the RFID reader and configured to execute the aforementioned motion-interference-resistant multi-antenna RFID respiratory monitoring method.

[0050] The data processing terminal includes a preprocessing module, a principal component analysis module, a component discrimination and filtering module, and a filtering and solving module. The preprocessing module performs at least one of the following processing steps on the raw phase data: phase unwrapping, outlier removal, interpolation, time alignment, resampling, and normalization, to construct a multi-channel phase observation matrix. The principal component analysis module performs principal component analysis on the multi-channel phase observation matrix. The component discrimination and filtering module identifies body motion interference components based on variance contribution rate and spectral characteristics, and filters target principal components from non-body motion components. The filtering and solving module filters the target principal components and outputs the respiration rate.

[0051] The data processing terminal is a software processing terminal that runs computer programs, or a hardware processing terminal implemented with at least one of FPGA, DSP, and ASIC.

[0052] Traditional mainstream contact-based devices often suffer from drawbacks such as strong wearing constraints, poor comfort, low long-term durability, easy detachment, susceptibility to interference, and cumbersome operation. This invention can be implemented using commercial RFID readers and passive, semi-active, or active tags, offering advantages such as lower cost, convenient deployment, and suitability for long-term non-contact vital sign monitoring.

[0053] Existing single-antenna monitoring suffers from limited viewing angles. When the line-of-sight path between the monitored object and the antenna is obstructed, or when the object's posture changes, causing the tag to be at an unfavorable angle, a single antenna often struggles to stably read respiratory signals. This invention utilizes multi-antenna spatial diversity to acquire tag phase information from multiple viewing angles, reducing the impact of limited single-antenna viewing angles and localized obstructions on the continuity of respiratory monitoring and improving monitoring coverage.

[0054] While multi-antenna polling can extend coverage, the data from different channels naturally have interleaved timestamps. Direct joint analysis without time alignment and unified resampling can easily lead to the failure of multi-channel eigenvalue decomposition. In this invention, by interpolating, aligning, and resampling the multi-channel phase sequences, the synchronization of the multi-channel observation matrix is ​​improved, providing a stable data foundation for subsequent principal component analysis.

[0055] Human body motion significantly interferes with signals. Respiration causes millimeter-level micro-displacements, while movements like turning over, swinging arms, and sitting up produce centimeter-level displacements with energy far exceeding that of respiration, easily obscuring respiratory features in the phase signal. This invention enhances the suppression of motion interference and improves the separation of respiratory-related components through principal component analysis and a motion component discrimination mechanism based on variance contribution rate and spectral energy distribution. Furthermore, bandpass filtering, passband adjustment, and smoothing are applied to the target principal components, further suppressing some motion interference and enabling applications for respiratory rate estimation and continuous monitoring.

[0056] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A multi-antenna RFID respiratory monitoring method with resistance to motion interference, characterized in that: Includes the following steps: S1: Place the RFID tag on the breathing fluctuation area of ​​the monitored object, and use the RFID reader to control the multi-antenna array to poll and read the RFID tag to obtain the backscatter phase sequence of multiple antenna channels; S2: Preprocess the phase sequence of each antenna channel to construct a multi-channel phase observation matrix; S3: Perform principal component analysis on the phase observation matrix to obtain multiple principal component components; S4: Identify the motion interference components based on the variance contribution rate and spectral characteristics of the principal component components, and remove the identified motion interference components. S5: From the principal components retained after removing the body motion interference components, select the target principal components that best match the respiratory characteristics; S6: Filter the target principal component and estimate the frequency of the filtered signal to obtain the respiratory rate.

2. The multi-antenna RFID breathing monitoring method with anti-motion interference according to claim 1, characterized in that: In S1, polling reads include the following steps: S1.1: Select one antenna from the multi-antenna array as the current working antenna; S1.2: Control the currently operating antenna to transmit radio frequency signals to the target area; S1.3: Receive the backscattered signal from the RFID tag and record the timestamp and phase information of the object; S1.4: Switch to the next antenna and repeat the process in S1.2 and S1.3 until a round of data acquisition from all antennas is completed; S1.5: Repeat steps S1.1 to S1.4 according to the set sampling frequency to form time-series phase data for each antenna channel.

3. The multi-antenna RFID breathing monitoring method with anti-motion interference according to claim 1, characterized in that: In S2, a multi-channel phase observation matrix is ​​constructed. , for A 3D matrix, denoted as , represented as ;in, To standardize the number of sampling points on the time axis, This refers to the number of antenna channels. For the first The length corresponding to each antenna channel is A time series vector.

4. The multi-antenna RFID breathing monitoring method with anti-motion interference according to claim 3, characterized in that: In S3, the phase observation matrix When performing principal component analysis, first analyze the phase observation matrix. Centralize the data and construct the covariance matrix: ； Then, eigenvalue decomposition is performed on the covariance matrix to obtain the eigenvalue matrix. and eigenvector matrix Then, the observation matrix is ​​projected onto the principal component space to obtain the principal component matrix. , The total number of principal components.

5. The multi-antenna RFID breathing monitoring method with anti-motion interference according to claim 4, characterized in that: In S4, the first Variance contribution rate of each principal component for: ； in, For the first Each principal component corresponds to an eigenvalue. The total number of principal components; Then, the spectral energy of each principal component within the target respiratory frequency band is calculated. And spectral energy outside the target breathing frequency band When a principal component satisfies Higher than the preset value ,and Greater than When this happens, the principal component is identified as a motion interference component.

6. The multi-antenna RFID breathing monitoring method with anti-motion interference according to claim 5, characterized in that: In S5, from the principal component components retained after removing the body motion interference components, the principal component component that best matches the respiratory characteristics is selected based on the spectral peak significance, dominant frequency position, and frequency band energy proportion within the target respiratory frequency band. As the target principal component signal .

7. The multi-antenna RFID breathing monitoring method with anti-motion interference according to claim 1, characterized in that: In S6, when filtering the target principal component, the target principal component is first bandpass filtered to suppress non-breathing band noise; then, the passband range of the bandpass filter is adjusted according to the estimated principal frequency of the filtered signal in the target breathing band; finally, the adjusted filtering result is smoothed to reduce residual glitches and local fluctuations.

8. The multi-antenna RFID breathing monitoring method with anti-motion interference according to claim 1, characterized in that: The RFID tag is any one of a passive tag, an active tag, or a semi-active tag, and the multi-antenna array includes at least two antennas with different spatial positions or different polarization directions.

9. A multi-antenna RFID respiratory monitoring system with anti-motion interference according to any one of claims 1 to 8, characterized in that: This includes RFID tags, multi-antenna arrays, RFID readers, and data processing terminals; RFID tags are placed on the breathing patterns of the monitored object to receive radio frequency signals and return backscattered signals carrying phase information. A multi-antenna array is used to cover the target monitoring area and receive backscattered signals from RFID tags; The RFID reader is connected to a multi-antenna array and is used to control the multi-antenna array to poll and acquire raw phase data from multiple antenna channels. The data processing terminal is communicatively connected to the RFID reader and is configured to execute the respiratory monitoring method.

10. A multi-antenna RFID respiratory monitoring system with anti-motion interference according to claim 9, characterized in that: The data processing terminal is a software processing terminal that runs computer programs, or a hardware processing terminal implemented with at least one of FPGA, DSP, and ASIC; the data processing terminal includes a preprocessing module, a principal component analysis module, a component discrimination and screening module, and a filtering and solving module; The preprocessing module is used to perform at least one of the following processes on the raw phase data: phase unwrapping, outlier removal, interpolation, time alignment, resampling, and normalization, in order to construct a multi-channel phase observation matrix; The principal component analysis module is used to perform principal component analysis on the multi-channel phase observation matrix; The component discrimination and screening module is used to discriminate the body motion interference components based on the variance contribution rate and spectral characteristics, and to screen the target principal components from the non-body motion components. The filtering and solving module is used to filter the target principal components and output the respiration rate.

Images