Breathing detection method based on multi-antenna CSI rapid combination

By combining and processing WiFi signal CSI data in indoor environments, using complex value and principal component analysis technology, the problem of large breath detection error in long distances and large noise environments is solved, and the respiratory mode is accurately extracted and detection perception is improved. Range effect.

CN119055223APending Publication Date: 2024-12-03NANJING UNIV OF POSTS & TELECOMM
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411188375.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-28
Publication Date
2024-12-03

AI Technical Summary

Technical Problem

The prior art cannot accurately detect breathing patterns in long-distance and high-noise environments, resulting in large respiratory detection errors and cannot guarantee the accuracy of breath estimation. Especially under the influence of complex environments and cheap equipment, the respiratory component is easily annihilated by channel noise. .

Method used

By collecting wireless signal data of WiFi transmitters in an indoor environment, setting a complex value of mode to 1, combining CSI data of different antenna pairs, and performing fast Fourier transformation on the combined CSI data to calculate the corresponding frequency range of breathing Energy relative to the percentage of energy sum of all frequencies, find the complex value that maximizes the energy percentage, removes impulse noise and phase shifts, and finally extracts the breathing pattern using principal component analysis.

Benefits of technology

It realizes accurate extraction of breathing mode in long-distance and high-noise environments, improves the perceived range of breath detection by WiFi signal, meets real-time requirements, and solves the problem that dynamic components are flooded with noise and cannot be accurately detected.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119055223A_ABST
    Figure CN119055223A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of respiration monitoring, and discloses a respiration detection method based on multi-antenna CSI rapid combination. The method comprises the following steps: firstly, collecting wireless signal data transmitted by a WiFi transmitter in an indoor environment, and extracting CSI original data from a WiFi signal; secondly, a complex value with a modulus of one is set to combine the CSI data of different antenna pairs, and fast Fourier transform is performed on the amplitude of the combined CSI data; then, calculating the percentage of the energy of the corresponding breathing frequency range relative to the sum of the energy of all the frequencies, and finding out a complex value enabling the energy percentage to reach the maximum; combining the CSI data with the same subcarrier frequency through a complex value, and removing impulse noise and phase offset; and finally, combining the CSI data of all the subcarrier frequencies to enhance the dynamic components so as to extract the breathing mode. According to the invention, the problem of how to effectively combine the dynamic components of the CSI data received by the plurality of antennas is solved, and the sensing range of WiFi signal detection respiration is expanded.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of respiration detection, and in particular relates to a respiration detection method based on multi-antenna CSI fast combination. Background Art

[0002] In recent years, with the rapid development of wireless technology, widely deployed commercial WiFi devices have been used to achieve ubiquitous contactless sensing. Various applications based on WiFi have been developed, such as fine-grained breathing detection, activity recognition, motion tracking, and fall detection. Traditional breathing rate detection methods usually require subjects to wear dedicated equipment, but these devices are often bulky and uncomfortable. In order to overcome the shortcomings of contact sensing, contactless sensing technology has attracted increasing attention in academia and industry.

[0003] Contactless respiratory monitoring systems can detect respiratory rate without contacting the human body, such as frequency modulated continuous wave (FMCW) radar, Doppler radar, and ultra-wideband MIMO radar. Although these methods have been proven to be very effective and accurate, their high deployment costs and the need for dedicated hardware limit their application in daily home monitoring. In order to balance non-invasiveness and cost-effectiveness, widely available and inexpensive commercial WiFi devices have gradually become the focus of research.

[0004] In recent years, many breathing monitoring systems based on WiFi signals have been proposed. For example, Dongheng Zhang et al. from the University of Science and Technology of China used the phase change of CSI to track human breathing in a paper titled Breath status tracking using commodity WiFi published in IEEE Conference on Global Communications. A peak detection-based method was used to establish a multidimensional clustering of the evolving signal subspace to improve the performance of practical sleep monitoring applications. Kamran Ali from Middlesex University used a peak detection-based method to establish a multidimensional clustering of the signal subspace to improve the performance of practical breathing monitoring applications in a paper titled On goodness of WiFi based monitoring of sleepvital signs in the wild published in IEEE Transactions on Mobile Computing. Li Yang et al. from Peking University combined the sensing signals received from all subcarriers and all antennas in the array, making full use of spatial and frequency diversity to improve the range of WiFi sensing breathing in a paper titled Diversense: Maximizing Wi-Fi sensing range leveraging signal diversity published in Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies.

[0005] However, when the object is far from the WiFi device and the channel noise is large, the above-mentioned prior art method will produce a large breathing detection error and cannot guarantee the accuracy of breathing estimation. In the actual home environment, affected by the complex environment and cheap equipment, the breathing component will be obliterated by the channel noise in most cases. Summary of the invention

[0006] In order to overcome the defects of the prior art and accurately extract the breathing pattern under long distance and high noise conditions, the present invention first collects wireless signal data transmitted by a WiFi transmitter in an indoor environment, then sets a complex value with modulus 1 to combine the CSI data of different antenna pairs and performs a fast Fourier transform on the amplitude of the combined CSI data, then calculates the percentage of the energy in the frequency range corresponding to the breathing relative to the energy sum of all frequencies and finds the complex value that maximizes the energy percentage, then combines the CSI data of the same subcarrier frequency through the complex value and removes the pulse noise and phase offset, and finally combines the CSI data of all subcarrier frequencies to enhance the dynamic component and extract the breathing pattern.

[0007] In order to achieve the above object, the present invention is achieved through the following technical solutions:

[0008] The present invention is a breathing detection method based on multi-antenna CSI fast combination, which specifically includes the following steps:

[0009] Step 1: Collect wireless signal data transmitted by the WiFi transmitter in an indoor environment, and extract the original data of the Channel State Information (CSI) from the WiFi signal;

[0010] Step 2: The channel state information data of the first antenna pair is used as reference data, the signal of the pth antenna pair is multiplied by a complex value with a modulus of 1 and added to the channel state information data of the first antenna pair, and then a fast Fourier transform is performed on the amplitude of the added channel state information data;

[0011] Step 3: Calculate the percentage of the energy in the frequency range corresponding to the breathing relative to the sum of the energies of all frequencies, and find the complex value that maximizes the energy percentage. The dynamic component of the channel state information of the pth antenna pair multiplied by the complex value is aligned with the dynamic component of the reference data.

[0012] Step 4: The channel state information of the same subcarrier frequency of each antenna pair is multiplied by the corresponding complex value and added together to remove the impulse noise in the CSI amplitude and the random offset in the CSI phase. Then, the channel state information of all subcarrier frequencies without impulse noise and phase offset is added together, and finally the breathing pattern is extracted using principal component analysis.

[0013] In step 1, the wireless signal data transmitted by the transmitter is collected in an indoor environment, and the channel state information CSI raw data is extracted from the WiFi signal, specifically:

[0014] The communication equipment of this method includes a transmitter router and a receiver router. The transmitter has one antenna and the receiver has three antennas. The operating frequency of the two routers is 5.24GHz and the sampling rate is 2000Hz. The receiver extracts the original channel state information CSI data from the collected WiFi signal. The channel state information CSI represents the link change state of the wireless signal in the space propagation, which can reflect the highly sensitive changes in the surrounding environment.

[0015] A further improvement of the present invention is that: in step 2, the channel state information data of the first antenna pair is used as reference data, the signal of the pth antenna pair is multiplied by a complex value with a modulus of 1 and added to the channel state information data of the first antenna pair, and then a fast Fourier transform is performed on the amplitude of the added channel state information data, specifically:

[0016] Step 2-1: The channel state information data of the first antenna pair is used as reference data, and the signal of the pth antenna pair is multiplied by a complex value with a modulus of 1 and added to the channel state information data of the first antenna pair. The calculation formula is:

[0017]

[0018] Among them, f i represents the ith subcarrier frequency, t represents time, p represents the pth antenna pair, H1(f i ,t) represents the CSI signal data of the first antenna pair, H p (f i ,t) represents the CSI signal data of the pth antenna pair, is a complex value modulo 1, θ p,i The initial value of H is 0. 1,p (f i ,t) is the signal of the pth antenna pair multiplied by The data after adding the channel state information data of the first antenna pair, A noise is the impulse noise in the CSI amplitude, is the random offset in the CSI phase;

[0019] Step 2-2: Perform fast Fourier transform on the amplitude of the added channel state information data, and the calculation formula is:

[0020] S(w)=FFT{|H 1,p (f i ,t)|}

[0021] Among them, |H 1,p (f i ,t)| is the signal of the pth antenna pair multiplied by The amplitude of the data after adding the channel state information data of the first antenna pair, FFT is fast Fourier transform, S(w) is |H 1,p (f i ,t)|The spectrum obtained after fast Fourier transform in the time dimension, w is the frequency in the spectrum.

[0022] A further improvement of the present invention is that in step 3, the percentage of the energy in the frequency range corresponding to breathing relative to the energy sum of all frequencies is calculated, and a complex value that maximizes the energy percentage is found. The dynamic component of the channel state information of the pth antenna pair multiplied by the complex value is aligned with the dynamic component of the reference data, specifically:

[0023] Step 3-1: First, calculate the percentage of the energy of the frequency range corresponding to the breathing relative to the energy of all frequencies. The calculation formula is:

[0024]

[0025] Among them, the corresponding frequency range of breathing is 0.1Hz to 0.5Hz. is the energy of the frequency range corresponding to breathing, ∑ w S(w) is the energy sum of all frequencies, and ER is the percentage of the energy in the frequency range corresponding to breathing relative to the energy sum of all frequencies;

[0026] Step 3-2: Then, loop through the complex values θ p,i The value range of is 0 to 2π, and find the complex value that maximizes the energy percentage.

[0027] A further improvement of the present invention is that in step 4, the channel state information of the same subcarrier frequency of each antenna pair is multiplied by the corresponding complex value and then added, the impulse noise in the CSI amplitude and the random offset in the CSI phase are removed, and then the channel state information of all subcarrier frequencies without impulse noise and phase offset is added, and finally the breathing pattern is extracted using principal component analysis, specifically:

[0028] Step 4-1: First, the channel state information of the same subcarrier frequency of each antenna pair is multiplied by the corresponding complex value and then added. The calculation formula is:

[0029]

[0030] Where, M is the number of antenna pairs;

[0031] Step 4-2: Then, when θ p,iWhen the rotation angle π is increased, the signal of the pth antenna pair and the signal of the first antenna pair are added to cancel the dynamic component. The signal after the same subcarrier frequency combination is divided by the signal after the dynamic component is canceled to remove the pulse noise and phase offset. The calculation formula is:

[0032]

[0033] in, It is the signal after the same subcarrier frequency combination after removing the pulse noise and phase offset;

[0034] Step 4-3: Combine all signals without impulse noise and phase shift, and the calculation formula is:

[0035]

[0036] Where N is the number of subcarriers, is the combination of all signals with impulse noise and phase offset removed;

[0037] Step 4-4: Finally, The breathing pattern was obtained by principal component analysis.

[0038] The beneficial effects of the present invention are as follows: the present invention sets complex value combination CSI data, traverses the complex values ​​and finds the complex value that makes the percentage of the energy in the frequency range corresponding to the breathing relative to the energy of all frequencies the largest, and can also use the complex values ​​to remove pulse noise and phase offset, so as to realize fast alignment of dynamic components and extract breathing waveforms, and meet real-time requirements; at the same time, the CSI data of all subcarriers of all antennas are fully utilized, and the problem that the target is far away from the WiFi device, resulting in a small dynamic component being submerged by noise and unable to accurately detect breathing is solved, thereby improving the perception range of WiFi signal detection breathing. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a schematic diagram of the detection method of the present invention.

[0040] Figure 2 It is a schematic diagram of CSI data multiplied by a complex value combination according to the present invention.

[0041] Figure 3 It is a schematic diagram comparing the detection effects of the present invention and other respiratory detection methods. DETAILED DESCRIPTION

[0042] The following will disclose the embodiments of the present invention with drawings. For the purpose of clear description, many practical details will be described together in the following description. However, it should be understood that these practical details should not be used to limit the present invention. That is to say, in some embodiments of the present invention, these practical details are not necessary.

[0043] like Figure 1 As shown, after WiFi transmits to the human body and receives the reflected signal, it generates CSI raw data. The present invention first sets the complex value modulus to 1 to combine the CSI data of different antenna pairs and performs a fast Fourier transform on the amplitude of the combined CSI data. Then the present invention calculates the percentage of the energy in the frequency range corresponding to the breathing relative to the energy sum of all frequencies and finds the complex value that maximizes the energy percentage. Next, the present invention combines the CSI data of the same subcarrier frequency through the complex value and removes the pulse noise and phase offset. Finally, the present invention combines the CSI data of all subcarrier frequencies to enhance the dynamic component and thus extract the breathing pattern. The breathing detection method of the present invention specifically includes the following steps:

[0044] In step 1: In an indoor environment, the wireless signal data transmitted by the WiFi transmitter is collected through the communication equipment, and the original channel state information (CSI) data is extracted from the WiFi signal. The specific steps are as follows:

[0045] The communication equipment of this method includes a transmitter router and a receiver router. The transmitter has one antenna and the receiver has three antennas. The operating frequency of the two routers is 5.24GHz and the sampling rate is 2000Hz. The receiver extracts the original channel state information CSI data from the collected WiFi signal. The channel state information CSI represents the link change state of the wireless signal in the space propagation, which can reflect the highly sensitive changes in the surrounding environment.

[0046] Step 2: The channel state information data of the first antenna pair is used as reference data, the signal of the pth antenna pair is multiplied by a complex value with a modulus of 1 and added to the channel state information data of the first antenna pair, and then a fast Fourier transform is performed on the amplitude of the added channel state information data. The specific steps are:

[0047] Step 2-1: The channel state information data of the first antenna pair is used as reference data, the signal of the pth antenna pair is multiplied by a complex value with a modulus of 1 and added to the channel state information data of the first antenna pair, and the calculation formula is:

[0048]

[0049] Among them, f i represents the ith subcarrier frequency, t represents time, p represents the pth antenna pair, H1(f i ,t) represents the CSI signal data of the first antenna pair, H p (f i ,t) represents the CSI signal data of the pth antenna pair, is a complex value modulo 1, θp,i The initial value of H is 0. 1,p (f i ,t) is the signal of the pth antenna pair multiplied by The data after adding the channel state information data of the first antenna pair, A noise is the impulse noise in the CSI amplitude, is the random offset in the CSI phase;

[0050] Step 2-2: Perform fast Fourier transform on the amplitude of the added channel state information data, and the calculation formula is:

[0051] S(w)=FFT{|H 1,p (f i ,t)|}

[0052] Among them, |H 1,p (f i ,t)| is the signal of the pth antenna pair multiplied by The amplitude of the data after adding the channel state information data of the first antenna pair, FFT is fast Fourier transform, S(w) is |H 1,p (f i ,t)|The spectrum obtained after fast Fourier transform in the time dimension, w is the frequency in the spectrum.

[0053] Step 3: Calculate the percentage of the energy in the frequency range corresponding to breathing relative to the sum of the energies of all frequencies, and find the complex value that maximizes the energy percentage. The dynamic component of the channel state information of the pth antenna pair multiplied by the complex value is aligned with the dynamic component of the reference data. The specific steps are:

[0054] Step 3-1: First, calculate the percentage of the energy of the frequency range corresponding to the breathing relative to the energy of all frequencies. The calculation formula is:

[0055]

[0056] Among them, the corresponding frequency range of breathing is 0.1Hz to 0.5Hz. is the energy of the frequency range corresponding to breathing, ∑ w S(w) is the energy sum of all frequencies, and ER is the percentage of the energy in the frequency range corresponding to breathing relative to the energy sum of all frequencies;

[0057] Step 3-2: Iterate over the complex values θ p,i The value range of is 0 to 2π, and find the complex value that maximizes the energy percentage.

[0058] Step 4: Figure 2 As shown, Figure 2(a) shows the CSI data of the first antenna pair and the CSI data of the pth antenna pair. Direct combination will lead to dynamic component cancellation. Figure 2 (b) indicates that when the CSI data of the pth antenna pair is multiplied by a complex value, the dynamic components are aligned, which can enhance the perception effect. The channel state information of the same subcarrier frequency of each antenna pair is multiplied by the corresponding complex value and added, the impulse noise in the CSI amplitude and the random offset in the CSI phase are removed, and then the channel state information of all subcarrier frequencies without impulse noise and phase offset is added, and finally the breathing pattern is extracted using principal component analysis. The specific steps are as follows:

[0059] Step 4-1: Multiply the channel state information of the same subcarrier frequency of each antenna pair by the corresponding complex value and add them together. The calculation formula is:

[0060]

[0061] Where, M is the number of antenna pairs;

[0062] Step 4-2: When the signal of the pth antenna pair is multiplied by The dynamic component after can be aligned with the dynamic component of the signal of the first antenna pair, then when θ p,i When the rotation angle π is increased, the signal of the pth antenna pair and the signal of the first antenna pair are added to cancel the dynamic component. The signal after the same subcarrier frequency combination is divided by the signal after the dynamic component is canceled to remove the pulse noise and phase offset. The calculation formula is:

[0063]

[0064] in, It is the signal after the same subcarrier frequency combination after removing the pulse noise and phase offset;

[0065] Step 4-3: Combine all signals without impulse noise and phase shift, and the calculation formula is:

[0066]

[0067] Where N is the number of subcarriers, is the combination of all signals with impulse noise and phase offset removed;

[0068] Step 4-4: The breathing pattern was obtained by principal component analysis.

[0069] In order to verify the effect of the present invention, Figure 3As shown, the present invention compares the breathing detection effects of two breathing detection systems, namely, CSI amplitude and CSI ratio. The environment in which the present invention collects data is an indoor environment. During the experiment, the distance between the experimental subject and the WiFi transceiver device was 6m. While measuring the CSI data, the experimental subject wore a breathing belt to obtain the real breathing of the human body to further verify the accuracy of the system designed by the present invention. Figure 3 (a) represents the true value of human breathing obtained using a breathing belt. Figure 3 (b) shows the waveform of the respiratory structure detected using the CSI amplitude. Figure 3 (c) shows the waveform of the respiratory structure detected using the CSI ratio. Figure 3 (d) shows the waveform constructed by the present invention. When the distance between the experimental subject and the WiFi transceiver is 6m, only the present invention can achieve accurate breathing detection.

[0070] The above description is only an embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.

Claims

1. A respiration detection method based on multi-antenna CSI fast combination, characterized by: The breathing detection method specifically comprises the following steps: Step 1: In an indoor environment, wireless signal data transmitted by a WiFi transmitter is collected through a communication device, and the raw channel state information (CSI) data is extracted from the WiFi signal. Step 2: Using the channel state information data of the first antenna pair as reference data, multiplying the signal of the pth antenna pair by a complex value modulo 1 and adding the resultant signal to the channel state information data of the first antenna pair, and then performing a fast Fourier transform on the amplitude of the added channel state information data. Step 3: Calculate the percentage of the energy in the frequency range corresponding to the breathing relative to the sum of the energies of all frequencies, and find the complex value that maximizes the energy percentage. The dynamic component of the channel state information of the pth antenna pair multiplied by this complex value is aligned with the dynamic component of the reference data. Step 4: Multiply the channel state information of the same subcarrier frequency of each antenna pair by the corresponding complex value and add them together to remove the impulse noise in the CSI amplitude and the random offset in the CSI phase. Then, add the channel state information of all subcarrier frequencies without impulse noise and phase offset, and finally use principal component analysis to extract the breathing pattern.

2. The respiration detection method based on multi-antenna CSI fast combination according to claim 1, characterized in that: In step 1, the communication device includes a WiFi transmitter router and a receiver router. The WiFi transmitter router has one antenna, and the receiver router has three antennas. The operating frequency of the WiFi transmitter router and the receiver router is 5.24 GHz, and the sampling rate of the WiFi transmitter router and the receiver router is 2000 Hz. The receiver router extracts the original channel state information (CSI) data from the collected WiFi signal.

3. The respiration detection method based on multi-antenna CSI fast combination according to claim 1, characterized in that: The step 2 specifically includes the following steps: Step 2-1: Using the channel state information data of the first antenna pair as reference data, the signal of the pth antenna pair is multiplied by a complex value modulo 1 and added to the channel state information data of the first antenna pair. The calculation formula is: Among them, f i represents the ith subcarrier frequency, t represents time, p represents the pth antenna pair, H1(f i , t) represents the CSI signal data of the first antenna pair, H p (f i , t) represents the CSI signal data of the pth antenna pair, is a complex value modulo 1, θ p,i The initial value of H is 0. 1,p (f i , t) is the signal of the pth antenna pair multiplied by The data after adding it to the channel state information data of the first antenna pair, A noise is the impulse noise in the CSI amplitude, is the random offset in the CSI phase; Step 2-2: Perform fast Fourier transform on the amplitude of the added channel state information data. The calculation formula is: S(w)=FFT{|H 1,p (f i ,t)|} Among them, |H 1,p (f i , t)| is the signal of the pth antenna pair multiplied by The amplitude of the data after adding it to the channel state information data of the first antenna pair, FFT is fast Fourier transform, S(w) is |H 1,p (f i , t)|The spectrum obtained after fast Fourier transform in the time dimension, w is the frequency in the spectrum.

4. The respiration detection method based on multi-antenna CSI fast combination according to claim 1, characterized in that: The step 3 specifically includes the following steps: Step 3-1. Calculate the percentage of energy in the frequency range corresponding to breathing relative to the sum of energy of all frequencies. The calculation formula is: Among them, the breathing frequency range is 0.1Hz to 0.5Hz. is the energy of the breathing frequency range, ∑ w S(w) is the energy sum of all frequencies, and ER is the percentage of energy in the frequency range corresponding to breathing relative to the energy sum of all frequencies; Step 3-2, traverse the complex value θ p,i The value of range is 0 to 2π, find the complex value that maximizes the energy percentage.

5. The respiration detection method based on multi-antenna CSI fast combination according to claim 1, characterized in that: The step 4 specifically includes the following steps: Step 4-1: Multiply the channel state information of the same subcarrier frequency of each antenna pair by the corresponding complex value and add them together. The calculation formula is: Where M is the number of antenna pairs; Step 4-2: When the signal of the pth antenna pair is multiplied by The dynamic component after θp,i can be aligned with the dynamic component of the signal of the first antenna pair. Then, when θp,i is further rotated by an angle π, the signal of the pth antenna pair and the signal of the first antenna pair are added to cancel the dynamic component. The signal after the same subcarrier frequency combination is divided by the signal after the dynamic component is canceled to remove the impulse noise and phase offset. The calculation formula is: in, It is the signal after the same subcarrier frequency combination after removing the pulse noise and phase offset; Step 4-3: Combine all signals after removing impulse noise and phase offset. The calculation formula is: Where N is the number of subcarriers, is the combination of all signals with impulse noise and phase offset removed; Step 4-4, The breathing pattern was obtained by principal component analysis.