Respiratory signal enhancement method and system in interference scene

Abstract

The invention provides a breathing signal enhancement method and system in an interference scene, and belongs to the technical field of breathing perception. CSI signals received by a receiving antenna are preprocessed respectively, the optimal projection angle interval of different subcarrier CSI signals is calculated based on a linear projection rule and according to the position of a human body, and the optimal projection angle interval is calculated; projecting each subcarrier under each projection system to obtain a projection waveform, and taking the projection system with the maximum sum of the BNR values of the projection waveforms as an optimal projection system to obtain a position-enhanced respiratory signal; the method comprises the following steps: dividing a respiratory signal into a plurality of segmented waveforms, carrying out short projection transformation and strong filtering on each segmented waveform by utilizing cyclic operation to obtain segmented waveforms after short window transformation, and splicing to obtain a secondarily enhanced respiratory signal; calculating a PAR value and a BNR value of the secondarily enhanced respiratory signal, screening out a plurality of optimal receiving antennas by taking the PAR * BNR value as an index, calculating to obtain an average value of self-correlation estimation of the optimal receiving antennas, and obtaining a respiratory rate estimation value; and the breathing monitoring robustness in the interference scene is improved.

Description

Technical Field

[0001] The present invention relates to the field of respiratory sensing technology, and in particular to a method and system for enhancing respiratory signals in interference scenarios. Background Art

[0002] Medical research shows that breathing is an important indicator of disease progression and health decline, and it mainly includes two indicators: respiratory rate and respiratory pattern. Among them, the respiratory rate reflects the current physical state, whether there are phenomena such as shortness of breath or respiratory failure; the respiratory pattern reflects the individual's breathing state over a period of time, and can observe whether there are phenomena such as apnea over a period of time, and obtain the individual's detailed respiratory pattern over a period of time, which is convenient for direct medical analysis.

[0003] Early respiratory monitoring usually requires wearing pressure sensors or wearable devices for contact measurement, which is easy to cause discomfort for special groups such as the elderly. The wearable method also greatly limits its scope of application and is difficult to popularize. In recent years, non-contact measurement methods based on radar signals, radio frequency signals (such as RFID, Wi-Fi, LoRa, etc.) and acoustic signals have attracted widespread attention from domestic and foreign researchers. Wi-Fi signals have many advantages such as high penetration rate, low cost and small bandwidth, which provides many conveniences for respiratory sensing. Therefore, Wi-Fi sensing has become an effective means of respiratory monitoring.

[0004] Current Wi-Fi breathing monitoring mainly tracks human breathing behavior by deploying a set of commercial Wi-Fi devices in free space. The periodic contraction or expansion of the chest cavity caused by human breathing will cause periodic changes in the path length of the Wi-Fi reflection signal, but this change is smaller than the Wi-Fi wavelength and is reflected as a small arc in the complex plane. By analyzing this arc, the breathing process of the human body during this period of time can be obtained. This tiny chest displacement caused by breathing makes it difficult to accurately estimate breathing. Existing work often focuses on pure breathing scenarios and establishes corresponding theoretical models. However, when the surrounding environment is noisy, the existing breathing sensing system has low monitoring accuracy and large waveform distortion. Therefore, how to achieve effective estimation of breathing rate and accurate acquisition of breathing patterns in interference scenarios is the primary difficulty of current work and the key to deploying breathing sensing systems in actual scenarios.

[0005] In response to the interference problem, some scholars have studied it, and the existing methods can be roughly divided into two categories: 1. Special deployment of transceivers to narrow the perception range. Zeng et al. pointed out that the placement of transceivers will have a non-negligible impact on the perception range and boundaries, which shows that interference can be excluded from perception through the reasonable placement of transceivers. However, this method has high requirements for the deployment environment, and once deployed, it can only perceive the target within a limited perception range, with poor flexibility and limited perception area. For example, at home, it is usually necessary to extend the transceiver distance to 10 meters to avoid interference within the effective perception range. 2. Obtain high-precision ToF and AoA for positioning human breathing and interference. Zhang et al. proposed that the use of large-scale antenna arrays can obtain accurate ToF and AoA for positioning human bodies and interference objects, respectively. However, this method requires strict antenna phase calibration on commercial Wi-Fi devices. Even after calibration, some position estimation errors will still occur after the parameters are obtained. Therefore, it is urgent to explore a robust anti-interference system that is independent of deployment and does not require large-scale antenna arrays. Summary of the invention

[0006] The technical problem to be solved by the present invention is how to improve the robustness of respiratory monitoring in interference scenarios.

[0007] The present invention solves the above technical problems through the following technical solutions: A method for enhancing a respiratory signal in an interference scenario, comprising the following steps:

[0008] Step 1: Preprocessing the CSI signals received by at least three receiving antennas respectively to obtain CSI signals preprocessed by the receiving antennas;

[0009] Step 2: Based on the linear projection rule, the optimal projection angle interval of different subcarrier CSI signals is calculated according to the human body position, and multiple projection systems are obtained by using the space traversal method. Each subcarrier is projected in each projection system to obtain a projection waveform. The projection system with the largest sum of BNR values ​​of the projection waveform is taken as the optimal projection system. In the optimal projection system, the CSI signals of each subcarrier are added to obtain a position-enhanced respiratory signal.

[0010] Step 3, fill in the data, set the transformation window length, divide the position-enhanced respiratory signal into multiple segmented waveforms, use a loop operation to perform short projection transformation and strong filtering on each segmented waveform to obtain a segmented waveform after short window transformation, and splice the segmented waveforms after short window transformation to obtain a secondary enhanced respiratory signal;

[0011] Step 4. Calculate the PAR value and BNR value of the secondary enhanced breathing signal, use the PARxBNR value as an indicator to screen out several optimal receiving antennas, calculate the average value of the autocorrelation estimate of the optimal receiving antenna, and obtain the breathing rate estimate; perform π fuzzy detection and correction on the waveform of the inverted antenna in the receiving antenna, and obtain the enhanced breathing waveform after superposition with other antenna waveforms.

[0012] Beneficial effects: The present invention utilizes information in three dimensions of time, space and frequency to directly or indirectly enhance the respiratory signal, utilizes linear projection rules to perform projection matching enhancement in the frequency domain, qualitatively enhances the respiratory signal, indirectly suppresses the interference signal, and obtains a respiratory signal that matches the human body position. In the time domain, the significant difference between the walking interference signal and the respiratory signal is used to perform short window transformation, and the respiratory signal is indirectly enhanced by suppressing the interference signal. In the spatial domain, selective screening is performed according to the quality of the antenna, and finally an enhanced respiratory waveform is obtained, which better realizes anti-interference performance and improves the robustness of respiratory monitoring in interference scenarios.

[0013] Preferably, the step 1 comprises:

[0014] 1.1. The CSI signals received by at least three receiving antennas are synthesized into a reference signal with a small dynamic component. The calculation formula is:

[0015] H ref =W*H=w1*H1+w2*H2+w3*H3+...+w n *H n (1)

[0016] H n (f, t) = H s，n (f)+H d,n (f,t)+ε n (f,t) (2)

[0017]

[0018] Among them, H ref represents the reference signal, W represents the weight matrix, H represents the receiving antenna signal matrix, and w n represents the weight value of the nth receiving antenna, H n represents the CSI signal received by the nth receiving antenna, n represents the number of receiving antennas, n ≥ 3, H s，n (f) represents the static component of the CSI signal received by the nth receiving antenna, H d，n (f, t) represents the dynamic component of the CSI signal received by the nth receiving antenna, ε n (f, t) represents the noise of the CSI signal received by the nth receiving antenna, represents the breathing signal attenuation coefficient in the CSI signal received by the nth receiving antenna, d t，n (t) represents the chest reflex path of the breather, represents the complex attenuation coefficient of the nth antenna interference signal, d i，n (t) represents the interference source reflection path, f represents the frequency, and c represents the speed of light;

[0019] 1.2. Optimizing the reference signal H ref , and obtain the optimized reference signal H ref and the corresponding weight matrix of n receiving antennas, the optimized reference signal H ref The calculation method is:

[0020] argmin{std(H ref )-0} (4)

[0021] Among them, std() means to find the variance of the variables in the brackets, and argmin() means to find the minimum value of the expression in the brackets;

[0022] 1.3. Using the optimized reference signal H ref As the denominator, the CSI signal received by each receiving antenna is used as the numerator, and the CSI signal after preprocessing of the nth receiving antenna is calculated respectively.

[0023]

[0024] Among them, ε ref (f, t) represents the reference signal H obtained after equivalent signal analysis ref The noise component of (f, t) is due to ε ref The value of (f, t) is very small and formula (5) is approximately equal to the following:

[0025]

[0026] Beneficial effects: The signals of at least three antennas are fully utilized to obtain a synthetic signal with the smallest dynamic fluctuation. Then, the CSI ratio model is used to obtain the preprocessed CSI signal by dividing each antenna signal with the reference signal, which can prevent the direct antenna ratio from destroying the breathing signal characteristics and greatly reduce the degree of signal distortion. The CSI signal obtained after data preprocessing can better reflect the changes of the original antenna signal.

[0027] Preferably, the linear projection rule satisfies the following relationship:

[0028] φ'=2π(f c +mΔf)(τ d -τ s ) (7)

[0029] Among them, τ d and τ s They represent the flight time corresponding to the dynamic path and the direct path respectively, Δf represents the frequency interval of adjacent subcarriers, φ ′ represents the slope, and the value of φ′ reflects the difference between the dynamic reflection path and the direct path, f c Indicates the center frequency of the Wi-Fi signal, and m indicates that the current subcarrier is the mth subcarrier.

[0030] Preferably, the process of calculating the optimal projection angle interval of different subcarrier CSI signals based on the human body position in step 2 includes: calculating the difference between the dynamic reflection path length and the direct path length according to the human body position, substituting it into the linear projection rule relationship to calculate the slope Φ' of the linear projection fitting line, and obtaining the optimal projection angle interval according to the slope Φ'.

[0031] Preferably, the process of obtaining multiple projection systems by using the space traversal method in step 2 includes: traversing in a space of 0-360° with a set step size to obtain corresponding projection systems under different initial angles, thereby realizing the screening of the initial phase.

[0032] Preferably, the method for completing the data in step three is: interpolating and completing the time points according to the timestamps of the CSI signals of different subcarriers to complete the data.

[0033] Preferably, the principle for setting the change window length in step three is to make the breathing signal in the data of the segment smooth and approximately a small arc, while the walking signal changes dramatically and has multiple peaks and troughs.

[0034] Beneficial effect: When strong filtering is performed on the data after short window transformation, the intensity of SG filtering and smoothing filtering can be increased, thereby filtering out the walking signal and retaining the change trend of the breathing signal.

[0035] Preferably, in step 4, the method of selecting several optimal receiving antennas using the PAR×BNR value as an indicator is: using 0.9 times the maximum value of the PAR×BNR value as a threshold, and selecting a receiving antenna with a PAR×BNR value greater than the threshold from multiple receiving antennas as the optimal receiving antenna.

[0036] Beneficial effect: The soft threshold method is used to screen out multiple antennas with an index value greater than 0.9 times the maximum value of PAR×BNR for fusion, that is, the antennas with an index greater than 0.9 times the maximum value of PAR×BNR are screened out, and then the result of the mean autocorrelation is calculated as the estimated value of the respiration rate, making full use of the spatial diversity brought by multiple antennas.

[0037] Preferably, the method of performing π ambiguity detection and correction on the waveform of the inverted antenna in the receiving antenna in step 4 is: by inverting the waveform of each receiving antenna in turn, and then superimposing it with other CSI signals, calculating the index value of each waveform, and screening out the one with the largest value, that is, the antenna with correctly corrected inversion.

[0038] The present invention also provides a system based on the method for enhancing the respiratory signal in an interference scenario, comprising:

[0039] A data preprocessing module, used to preprocess the CSI signals received by at least three receiving antennas respectively to obtain CSI signals preprocessed by the receiving antennas;

[0040] A position enhancement module is used to calculate the optimal projection angle interval of different subcarrier CSI signals according to the position of the human body based on the linear projection rule, obtain multiple projection systems by using the space traversal method, project each subcarrier in each projection system to obtain a projection waveform, take the projection system with the largest sum of BNR values ​​of the projection waveform as the optimal projection system, add each subcarrier CSI signal in the optimal projection system, and obtain a position-enhanced respiratory signal;

[0041] The short window transformation module is used to fill in the data, set the transformation window length, divide the position-enhanced respiratory signal into multiple segmented waveforms, perform short projection transformation and strong filtering on each segmented waveform by loop operation, obtain the segmented waveform after short window transformation, and splice the segmented waveform after short window transformation to obtain the secondary enhanced respiratory signal;

[0042] The multi-antenna fusion screening module is used to calculate the PAR value and BNR value of the secondary enhanced breathing signal, screen out several optimal receiving antennas with PAR×BNR value as the indicator, calculate the average value of the autocorrelation estimate of the optimal receiving antenna, and obtain the breathing rate estimate; perform π fuzzy detection and correction on the waveform of the inverted antenna in the receiving antenna, and obtain the enhanced breathing waveform after superposition with other antenna waveforms. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A schematic diagram of the Fresnel zone cutting of the walking and breathing signals of the method for enhancing the breathing signal in an interference scenario provided by an embodiment of the present invention;

[0044] Figure 2 A schematic diagram of respiratory CSI signal perception of a method for enhancing respiratory signals in interference scenarios provided by an embodiment of the present invention;

[0045] Figure 3 A schematic diagram of a moving CSI signal of a method for enhancing a breathing signal in an interference scenario provided by an embodiment of the present invention;

[0046] Figure 4A human body position relationship diagram under a linear projection rule of a method for enhancing a respiratory signal in an interference scenario provided by an embodiment of the present invention;

[0047] Figure 5 A schematic diagram of a method for enhancing a respiratory signal in an interference scenario provided by an embodiment of the present invention;

[0048] Figure 6 A flowchart of a method for enhancing a respiratory signal in an interference scenario provided by an embodiment of the present invention;

[0049] Figure 7 A schematic diagram of a system for enhancing a respiratory signal in an interference scenario provided by an embodiment of the present invention;

[0050] Figure 8 A cdf comparison diagram of the method for enhancing the respiratory signal in an interference scenario provided by an embodiment of the present invention and the existing method;

[0051] Fig. 9 This is a comparison chart of the detection rates of the breathing signal enhancement method in interference scenarios provided by an embodiment of the present invention and the existing methods. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the technical solution of the present invention is clearly and completely described below in combination with specific embodiments and with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0053] Explanation of relevant terms involved in the present invention

[0054] Wireless sensing is a technology that uses ubiquitous wireless sensor signals in the environment to obtain channel state information to perceive the surrounding environment and human activities. It is one of the most critical technologies in the field of the Internet of Things. There are many types of signals that can be used for wireless sensing, such as Wi-Fi, sound, light signals, Bluetooth, RFID, etc.

[0055] Wi-Fi breathing sensing refers to deploying a set of commercial Wi-Fi devices in the space to collect signals related to the user's breathing, analyze and process the signals, and ultimately achieve contactless measurement of the target person's breathing rate in a dynamic indoor environment.

[0056] Channel State Information (CSI) describes the propagation process of wireless signals between the transmitter and the receiver. It contains two types of information: amplitude and phase. It is a channel characteristic that reflects the frequency response of the wireless communication link, including multipath effects, fading, shadowing, and delay distortion.

[0057] The Fresnel zone refers to the middle area in free space formed by the elliptical boundaries of half the Wi-Fi wavelength that are integer multiples of the transceiver. It describes a series of elliptical areas formed when the electromagnetic wave signal propagates along paths of different distances, such as Figure 1 As shown, the construction of this area is derived from the Fresnel theoretical model. When an object moves in space, passing through a series of Fresnel zones will cause corresponding enhancement and attenuation changes in the CSI signal, thereby generating a waveform similar to a sine wave.

[0058] The spatial projection mechanism refers to a unique projection mechanism for respiratory signals, so that the received respiratory CSI signal can avoid the "blind spot" problem in space after combining the amplitude and phase, and obtain the corresponding periodic respiratory waveform after projection.

[0059] The linear projection mechanism refers to the linear variation between the optimal projection angles under different subcarriers caused by the frequency domain diversity technology adopted by Wi-Fi devices and the spatial projection processing method of the respiratory signal. This law contains position information, which can help achieve static human body positioning and enhance human respiratory signals.

[0060] Example 1

[0061] Most of the current methods focus on how to better separate dynamic and static signals. However, the multipath effect is an inevitable problem in indoor wireless sensing. When interference is introduced, the separation of multipath signals becomes more difficult. Therefore, in order to improve the robustness of respiratory monitoring in interference scenarios, the present invention mainly faces the following two challenges:

[0062] Challenge 1: How to better separate the walking signal from the breathing signal? To achieve accurate detection of the breathing signal in an interference scenario, an intuitive idea is to directly separate the interference signal, so that the separated signal is closer to the pure breathing signal in an interference-free scenario. However, we found that the walking signal is irregular. When it is mixed with the regular breathing signal, we cannot separate it cleanly and completely. Therefore, we use the method of enhancing the breathing signal and suppressing noise to eliminate the impact of walking interference on the breathing signal as much as possible.

[0063] Challenge 2: How to screen out antennas or subcarrier signals with better signal quality to achieve the maximum signal gain? Due to the frequency diversity technology used in commercial Wi-Fi devices, the breathing signals under different subcarriers contain position information. We further discovered the linear projection rule, which became an important theoretical basis for achieving breathing positioning. Using this position information, we can enhance the breathing signal in the mixed signal at a fixed point to maximize the breathing characteristics to achieve the purpose of one enhancement. Therefore, the present invention makes full use of all subcarrier signals. For antennas, due to different spatial positions, even after preprocessing, the antenna CSI still has a large difference in quality. For this reason, we propose a new indicator PAR×BNR value to achieve antenna quality screening, combined with the use of antennas with better quality to estimate the human breathing rate.

[0064] The present invention is described in detail below:

[0065] like Figure 2-6 As shown, this embodiment provides a method for enhancing a breathing signal in an interference scenario, comprising the following steps:

[0066] Step 1: Preprocess the CSI signals received by at least three receiving antennas respectively to obtain CSI signals preprocessed by the receiving antennas.

[0067] The present invention uses two commercial Wi-Fi devices equipped with Intel 5300 network cards, one of which is equipped with a transmitting antenna as a transmitter, and the other is equipped with three receiving antennas as a receiver. The three receiving antennas are arranged in a linear array, and the distance between adjacent receiving antennas is set to half the wavelength of Wi-Fi, specifically 2.5 cm. A CSI tool is installed in the commercial Wi-Fi device to collect CSI readings. The transmitter is set to injection mode, the receiver is set to monitoring mode, the channel is 64, the center frequency is 5.32 GHz, the bandwidth is 20 MHz, and the data packet transmission rate is set to 100 Hz.

[0068] The present invention deploys experiments by recruiting 10 volunteers in two scenarios, open space and indoor office. The experimental settings used to control the precise distance to obtain parameter performance are carried out in the open space, while the indoor office is used to explore whether the system performance can be stable in other scenarios (especially indoor scenarios with significant multipath effects). The true value of the experiment is recorded by the Neulog NUL236 respiratory monitoring belt and connected to the PC through a matching USB module (USB-200).

[0069] It should be noted that the commercial Wi-Fi device used as a transmitter in the present invention can be configured with multiple transmitting antennas, and the commercial Wi-Fi device used as a receiver can be configured with multiple receiving antennas, with one transmitting antenna and three receiving antennas being the minimum configuration. The open space of the present invention is actually also an indoor environment, but it is a smaller enclosed space compared to an indoor office. The precise distance can be controlled according to the size of the brick (the side length of a brick is 0.8m) to accurately control the distance between the transmitting and receiving ends and the distance of the human body. The parameter performance is mainly reflected in: by controlling the human body to remain unchanged in a certain position, and controlling the interference distance according to the size of the brick in turn, the influence of the interference distance on the experiment can be studied. By changing the position of the breathing human body while controlling the interference distance to remain unchanged, the influence of the distance of the breathing signal on the experimental results can be studied.

[0070] Due to the inherent defects of commercial Wi-Fi devices and the time asynchrony of transceivers, additional random phase noise is usually introduced to the signal, such as carrier frequency offset (CFO), sampling frequency offset (SFO), packet detection delay (PDD), etc., so these random phase offsets need to be removed. According to the CSI Ratio model proposed in the paper "FarSense: Pushing the Range Limit of WiFi-based Respiration Sensing with CSI Ratio of Two Antennas", antennas from the same receiver often share the same oscillator, so a simple division operation of the signal between the receiving antennas can remove the random phase offset. Simple division can easily distort the signal. Although the Farsense paper points out that the shape and direction of the signal are not changed after division, this still affects the degree of change of the signal. And considering that there are six combinations of different combinations of three receiving antennas, an effective screening algorithm is needed to obtain the optimal antenna ratio to minimize the division distortion of the signal. The step one specifically includes the following processes:

[0071] 1.1. The CSI signals received by at least three receiving antennas are synthesized into a reference signal with a small dynamic component. The calculation formula is:

[0072] H ref =W*H=w1*H1+w2*H2+w3*H3+…+w n *H n (1)

[0073] H n (f, t) = Hs，n (f)+H d，n (f,t)+ε n (f, t) (2)

[0075]

[0076] Among them, H ref represents the reference signal, W represents the weight matrix, H represents the receiving antenna signal matrix, and w n represents the weight value of the nth receiving antenna, H n represents the CSI signal received by the nth receiving antenna, n represents the number of receiving antennas, n ≥ 3, H s，n (f) represents the static component of the CSI signal received by the nth receiving antenna, H d，n (f, t) represents the dynamic component of the CSI signal received by the nth receiving antenna, ε n (f, t) represents the noise of the CSI signal received by the nth receiving antenna, represents the breathing signal attenuation coefficient in the CSI signal received by the nth receiving antenna, d t,n (t) represents the chest reflex path of the breather, represents the complex attenuation coefficient of the nth antenna interference signal, d i , n(t) represents the interference source reflection path, f represents the frequency, and c represents the speed of light.

[0077] 1.2. Optimizing the reference signal H ref , and obtain the optimized reference signal H ref And the corresponding weight matrix of n receiving antennas, which is the weight matrix that generates the reference signal H ref The set of weight values ​​with the smallest dynamic fluctuation, the optimized reference signal H ref The calculation method is:

[0078] argmin{std(H ref )-0} (4)

[0079] Among them, std() means to find the variance of the variables in the brackets, and argmin() means to find the minimum value of the expression in the brackets.

[0080] 1.3. Using the optimized reference signal H ref As the denominator, the CSI signal received by each receiving antenna is used as the numerator, and the CSI signal after preprocessing of the nth receiving antenna is calculated respectively.

[0081]

[0082] Among them, εref (f, t) represents the reference signal H obtained after equivalent signal analysis ref The noise component of (f, t) is a combination of the noise components of the original antenna, expressed as:

[0083] ε ref (f, t)=w1*ε1(f,t)+w2*ε2(f,t)+w3*ε3(f,t)+…+w n *ε n (f, t)

[0084] Due to ε ref The value of (f, t) is very small, which is consistent with the static signal H of the reference signal. s，ref (f) can be neglected, so the equation can be approximated as follows:

[0085]

[0086]

[0087] After step 1, the CSI signal received by the receiving antenna is preprocessed to obtain a CSI signal without random phase noise such as CFO, SFO, PDD, etc. In the preprocessing process, the reference signal is obtained by equivalent signal analysis, that is, appropriate weights are selected to make full use of the signals of at least three antennas to obtain a synthetic signal with the smallest dynamic fluctuation. Then, using the CSI ratio model, the preprocessed CSI signal is obtained by dividing each antenna signal with the reference signal, which can prevent the direct antenna ratio from destroying the breathing signal characteristics and greatly reduce the degree of signal distortion. The CSI signal obtained after data preprocessing can better reflect the changes in the original antenna signal.

[0088] Step 2: Based on the linear projection rule, the optimal projection angle interval of different subcarrier CSI signals is calculated according to the human body position, and multiple projection systems are obtained by using the space traversal method. Each subcarrier is projected in each projection system to obtain a projection waveform. The projection system with the largest sum of BNR values ​​of the projection waveform is taken as the optimal projection system. In the optimal projection system, the CSI signals of each subcarrier are added to obtain a position-enhanced breathing signal.

[0089] Each receiving antenna has multiple subcarriers. The CSI signal after preprocessing by the receiving antenna is a three-dimensional matrix of the number of time acquisition points × the number of antennas × the number of subcarriers. According to the CSI signal of the receiving antenna, the CSI signals of different subcarriers in the third dimension can be obtained. According to the above-mentioned Farsense paper, the CSI signals under different subcarriers correspond to an optimal projection waveform in the complex plane. The optimal projection angle corresponding to this waveform reflects the optimal spatial resolution under the current carrier, that is, the coordinate axis that can maximize the periodic change characteristics of breathing. If you want to enhance the respiratory signal in the mixed signal, you need to resolve the signal in the optimal projection space to maximize the characteristics of the respiratory signal and randomize the walking signal.

[0090] The commercial Wi-Fi device used in the present invention adopts frequency domain diversity technology. The optimal projection angles of the signal under frequency domain diversity under different subcarriers satisfy the linear variation law. By fitting the curves of these optimal projection angles, a straight line whose slope reflects the length of the reflection path can be obtained, such as Figure 4 As shown, based on the optimal projection system, the signal under each subcarrier is projected to obtain the optimal spatial resolution of the respiratory signal. Therefore, the present invention is based on the linear projection rule and reversely utilizes the linear projection rule to design the position enhancement algorithm.

[0091] The linear projection rule satisfies the following relationship:

[0092] φ'=2π(f c +mΔf)(τ d -τ s ) (7)

[0093] Among them, τ d and τ s represents the flight time corresponding to the dynamic path and the direct path respectively, Δf represents the frequency interval of adjacent subcarriers, φ' represents the slope, and the value of φ' reflects the difference between the dynamic reflection path and the direct path (the flight time difference multiplied by the speed of light c is the difference between the dynamic reflection path and the direct path), f c Indicates the center frequency of the Wi-Fi signal, and m indicates that the current subcarrier is the mth subcarrier.

[0094] The process of calculating the optimal projection angle interval of different subcarrier CSI signals based on the human body position includes: calculating the difference (RPL) between the dynamic reflection path length and the direct path length according to the human body position, substituting it into the linear projection rule relationship to calculate the slope φ' of the linear projection fitting line, and obtaining the optimal projection angle interval according to the slope φ'.

[0095] The process of obtaining multiple projection systems by using the spatial traversal method includes: traversing in the space of 0-360° with a set step size, obtaining the corresponding projection systems under different initial angles, and implementing the screening of the initial phase based on the principle of maximizing the sum of BNR. The step size of the present invention is set to 5°. After projecting each subcarrier under each projection system to obtain a projection waveform, FFT calculations are performed in sequence to obtain a spatial spectrum diagram under each projection, and the waveform in the current space is estimated by calculating the BNR value of each waveform to obtain a group with the largest BNR value under the spatial projection system, thereby obtaining the best spatial projection system for determining the initial phase, and this group of projection systems is a spatial projection system that can satisfy the maximization of the characteristic value of the respiratory signal. Because only when the target projection space matches the current position space, the respiratory component in the mixed signal can be maximized and the purpose of respiratory enhancement can be achieved.

[0096] Breath-to-Noise Ratio (BNR) reflects the quality of the respiratory signal. BNR is used as an indicator to measure the periodic characteristics of the respiratory signal. The larger the BNR value, the more obvious the periodicity of the respiratory signal, and the closer the projection angle is to the optimal projection angle. Therefore, the present invention uses the BNR indicator to screen out a group of projection systems with the highest BNR sum, which corresponds to the optimal spatial resolution of the respiratory signal, and obtains the optimal projection angle of the respiratory signal under different subcarriers.

[0097] The present invention uses the linear projection rule satisfied by the signal under Wi-Fi diversity technology for the first time to enhance the position of the breathing signal, so as to enhance the breathing signal in the mixed signal with interference, so that the weak breathing signal can be improved in the mixed signal, and the breathing signal is enhanced once. Position enhancement of the breathing signal is essentially the reverse use of the linear projection rule. In the scene where only the human body breathes, the slope of the straight line fitted according to the optimal projection angle under different subcarriers reflects the position information of the human body, and the optimal projection angle corresponds to the optimal projection spatial resolution under each subcarrier. In the scene with interference, since it is impossible to directly separate the breathing signal from the interference signal, the method of enhancing the breathing signal is adopted to indirectly suppress the interference signal. The idea of ​​enhancing the breathing signal here is to maximize the spatial resolution of the breathing signal and comprehensively utilize multiple subcarrier signals, that is, to find the optimal projection spatial resolution of the breathing CSI signal under different subcarriers.

[0098] Step 3: fill in the data, set the transformation window length, divide the position-enhanced respiratory signal into multiple segmented waveforms, use a loop operation to perform short projection transformation and strong filtering on each segmented waveform to obtain a segmented waveform after short window transformation, and splice the segmented waveforms after short window transformation to obtain a secondary enhanced respiratory signal.

[0099] Due to the rapid changes and irregularity of walking signals, it is very difficult to separate the breathing signal when processing mixed signals over a long period of time. Even if the breathing signal is enhanced by position enhancement, the interference of the walking signal over a long period of time cannot be ignored. Therefore, it is also necessary to suppress the walking interference in the mixed signal through the short-time window transformation method, so as to indirectly enhance the breathing signal and obtain a secondary enhanced waveform. In this part, firstly, timestamp padding is performed to fill the lost data packets to the number of packets sent during the experimental setting according to cubic spline interpolation, and then position matching projection is performed according to the overall waveform to obtain the projection angle that maximizes the breathing feature and record it, and then operate on this projection angle during the subsequent short window transformation and filtering operations. Specifically, it includes the following processes:

[0100] The method of padding the data is to interpolate and pad the time points according to the timestamps of the CSI signals of different subcarriers to pad the data. The specified number of packets set in the experiment will usually be lost during the sending and receiving process. If it is not padded, the analyzed data points will be less than the actual number of points, resulting in a shortened time and a large error in the final analysis of the breathing rate. After padding, the calculated breathing rate is closer to the actual value.

[0101] The principle of setting the transformation window length is to make the breathing signal in the data of the segment smooth, approximately a small arc, while the walking signal changes dramatically, with multiple peaks and troughs. In this way, when strong filtering is performed on this part of the data, the strength of SG filtering and smoothing filtering can be increased, thereby filtering out the walking signal and retaining the change trend of the breathing signal.

[0102] After cyclic processing of each small segment of data, the data segments are stitched together to obtain a complete data after secondary enhancement. Since there are small overlaps between different segments during cyclic processing, there are also overlaps when stitching. The total number of data points in the experiment is 6000 points, and the number of points in each small segment of data is 3 times the number of sampling points, that is, 300 points. The number of points between two adjacent windows is 50, and the overlap rate is 1 / 6.

[0103] The present invention finds that there is a significant difference between the characteristics of walking signals and breathing signals, that is, walking signals change significantly while breathing signals change slowly. In response to this, a short window strong filtering method is used to splice multiple segments of data to replace the traditional bandpass filtering process, which will maximize the retention of breathing signal characteristics while effectively suppressing walking signals.

[0104] Step 4: Calculate the PAR value and BNR value of the secondary enhanced breathing signal, select several optimal receiving antennas based on the PAR×BNR value, calculate the average value of the autocorrelation estimate of the optimal receiving antenna, and obtain the breathing rate estimate; perform π fuzzy detection and correction on the waveform of the inverted antenna in the receiving antenna, and obtain the enhanced breathing waveform after superposition with other antenna waveforms.

[0105] PAR (Peak-to-Average Ratio) indicator: an indicator used to measure the relationship between the peak value and the average value of a signal. The peak amplitude refers to the maximum amplitude value in the signal, while the average amplitude refers to the average value of the signal amplitude. The PAR value can tell us whether there is an instantaneous peak in the signal and the dynamic range of the signal.

[0106] BNR (Breath Signal-to-Noise Ratio) indicator: According to the definition in the Farsense paper, BNR is a specific indicator that describes the signal-to-noise ratio of the respiratory signal. It reflects the percentage of the respiratory signal energy in the spectrum to the total energy, reflecting the quality of the waveform. It is defined as the ratio of the sum of the energy of the respiratory signal unit on the spectrum to the total energy of the spectrum unit.

[0107] The method of selecting several best receiving antennas with the PAR×BNR value as an indicator is as follows: 0.9 times of the maximum value of the PAR×BNR value is used as a threshold, and a receiving antenna with a PAR×BNR value greater than the threshold is selected from multiple receiving antennas as the best receiving antenna.

[0108] We use the new index PAR×BNR to select several antennas with larger indexes. We found that if we only use this index to select the best antenna to estimate the human respiratory rate, it is easy to lose the antenna with inferior quality. Although the indexes of the two antennas are close, the estimated respiratory rate is also different. If we consider the two antennas comprehensively and average the estimated human respiratory rate, we can get a more accurate respiratory rate. For the antennas with better quality, we use the soft threshold method to select multiple antennas with an index value of more than 0.9 times the maximum value of PAR×BNR, that is, we select antennas with an index greater than 0.9 times the maximum value of PAR×BNR, and then calculate the result of the mean autocorrelation as the estimated respiratory rate, making full use of the spatial diversity brought by multiple antennas.

[0109] For subcarrier i, breathing pattern y i The autocorrelation r of the time series when it moves by k samples i (k) is defined as:

[0110]

[0111] in, Represents breathing mode y i The average value over the time series, k represents moving k sample points, and T represents the breathing pattern y i The total length of , k = 0, .., T-1 is the number of samples of translation.

[0112] The method for performing π ambiguity detection and correction on the waveform of the inverted antenna in the best receiving antenna is: by inverting the waveform of each receiving antenna in turn, and then superimposing it with the CSI signals of other antennas, the PAR×BNR value of each waveform is calculated, and the one with the largest value is screened out, that is, the antenna with correctly corrected inversion.

[0113] An algorithm is used to detect the antenna waveform that produces inversion. The waveform after correcting the π blur is superimposed on the normal waveform to obtain the total waveform after the multi-antenna combination. The peaks and troughs of the waveform are made more obvious, thereby obtaining a waveform that clearly reflects the respiratory fluctuations. The respiratory waveform is the breathing pattern, which can clearly observe a person's breathing status over a period of time.

[0114] The present invention is a method for enhancing the respiratory signal in an interference scenario by using position information and signal differences. It is independent of deployment and does not require a large-scale antenna array. The method uses the frequency domain diversity of Wi-Fi signals as the background to use the multi-subcarrier projection rule to perform position enhancement to achieve the enhancement of the respiratory signal in the mixed signal. The information of the three dimensions of time, space and frequency is used to directly or indirectly enhance the respiratory signal. The linear projection rule is used in the frequency domain to perform projection matching enhancement, and the respiratory signal is qualitatively enhanced to achieve indirect suppression of the interference signal, and a respiratory signal matching the human body position is obtained. In the time domain, the significant difference between the walking interference signal and the respiratory signal is used to perform short window transformation, and the indirect enhancement of the respiratory signal is achieved by suppressing the interference signal. In the spatial domain, selective screening is performed according to the quality of the antenna, and finally the enhanced respiratory waveform is obtained, which better achieves anti-interference performance and improves the robustness of respiratory monitoring in interference scenarios.

[0115] Example 2

[0116] like Figure 7 As shown, this embodiment provides a system based on a method for enhancing a breathing signal in an interference scenario, including:

[0117] A data preprocessing module, used to preprocess the CSI signals received by at least three receiving antennas respectively to obtain CSI signals preprocessed by the receiving antennas;

[0118] A position enhancement module is used to calculate the optimal projection angle interval of different subcarrier CSI signals according to the position of the human body based on the linear projection rule, obtain multiple projection systems by using the space traversal method, project each subcarrier in each projection system to obtain a projection waveform, take the projection system with the largest sum of BNR values ​​of the projection waveform as the optimal projection system, add each subcarrier CSI signal in the optimal projection system, and obtain a position-enhanced respiratory signal;

[0119] The short window transformation module is used to fill in the data, set the transformation window length, divide the position-enhanced respiratory signal into multiple segmented waveforms, perform short projection transformation and strong filtering on each segmented waveform by loop operation, obtain the segmented waveform after short window transformation, and splice the segmented waveform after short window transformation to obtain the secondary enhanced respiratory signal;

[0120] The multi-antenna fusion screening module is used to calculate the PAR value and BNR value of the secondary enhanced breathing signal, screen out several optimal receiving antennas with PAR×BNR value as the indicator, calculate the average value of the autocorrelation estimate of the optimal receiving antenna, and obtain the breathing rate estimate; perform π fuzzy detection and correction on the waveform of the inverted antenna in the receiving antenna, and obtain the enhanced breathing waveform after superposition with other antenna waveforms.

[0121] The present invention conducts a large number of experiments in two scenarios by controlling variables to prove the anti-interference perception boundary of the proposed system. Figure 8-9 As shown, the experimental results show that compared with the existing pure breathing plus simple filtering solution, the system can achieve effective perception of breathing at 6.4m under the condition of 2.4m anti-interference, and the perception accuracy reaches 80%.

[0122] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for enhancing a respiratory signal in an interference scenario, characterized in that: The following steps are involved: Step 1: Preprocessing the CSI signals received by at least three receiving antennas respectively to obtain CSI signals preprocessed by the receiving antennas; Step 2: Based on the linear projection rule, the optimal projection angle interval of different subcarrier CSI signals is calculated according to the human body position, and multiple projection systems are obtained by using the space traversal method. Each subcarrier is projected in each projection system to obtain a projection waveform. The projection system with the largest sum of BNR values ​​of the projection waveform is taken as the optimal projection system. In the optimal projection system, the CSI signals of each subcarrier are added to obtain a position-enhanced respiratory signal. Step 3, fill in the data, set the transformation window length, divide the position-enhanced respiratory signal into multiple segmented waveforms, use a loop operation to perform short projection transformation and strong filtering on each segmented waveform to obtain a segmented waveform after short window transformation, and splice the segmented waveforms after short window transformation to obtain a secondary enhanced respiratory signal; Step 4: Calculate the PAR value and BNR value of the secondary enhanced breathing signal, select several optimal receiving antennas based on the PAR×BNR value, calculate the average value of the autocorrelation estimate of the optimal receiving antenna, and obtain the breathing rate estimate; perform π fuzzy detection and correction on the waveform of the inverted antenna in the receiving antenna, and obtain the enhanced breathing waveform after superposition with other antenna waveforms.

2. The method for enhancing the respiratory signal in an interference scenario according to claim 1, characterized in that: The step one comprises: 1.

1. The CSI signals received by at least three receiving antennas are synthesized into a reference signal with a small dynamic component. The calculation formula is: H ref =W*H=w1*H1+w2*H2+w3*H3+...+w n *H n (1) H n (f,t)=H s,n (f)+H d,n (f,t)+ε n (f,t) (2) Among them, H ref represents the reference signal, W represents the weight matrix, H represents the receiving antenna signal matrix, and w n represents the weight value of the nth receiving antenna, H n represents the CSI signal received by the nth receiving antenna, n represents the number of receiving antennas, n ≥ 3, H s,n (f) represents the static component of the CSI signal received by the nth receiving antenna, H d,n (f, t) represents the dynamic component of the CSI signal received by the nth receiving antenna, ε n (f, t) represents the noise of the CSI signal received by the nth receiving antenna, represents the breathing signal attenuation coefficient in the CSI signal received by the nth receiving antenna, d t,n (t) represents the respiratory chest reflex path, represents the complex attenuation coefficient of the nth antenna interference signal, d i,n (t) represents the interference source reflection path, f represents the frequency, and c represents the speed of light; 1.

2. Optimizing the reference signal H ref , and obtain the optimized reference signal H ref and the corresponding weight matrix of n receiving antennas, the optimized reference signal H ref The calculation method is: argmin{std(H ref )-0} (4) Among them, std() means to find the variance of the variables in the brackets, and argmin() means to find the minimum value of the expression in the brackets; 1.

3. Using the optimized reference signal H ref As the denominator, the CSI signal received by each receiving antenna is used as the numerator, and the CSI signal after preprocessing of the nth receiving antenna is calculated respectively. Among them, ε ref (f, t) represents the reference signal H obtained after equivalent signal analysis ref The noise component of (f, t) is due to ε ref The value of (f, t) is very small and formula (5) is approximately equal to the following:

3. The method for enhancing the respiratory signal in an interference scenario according to claim 1, characterized in that: The linear projection rule satisfies the following relationship: Φ'=2π(f c +mΔf)(τ d -t s ) (7) Among them, τ d and τ s They represent the flight time corresponding to the dynamic path and the direct path respectively, Δf represents the frequency interval of adjacent subcarriers, Φ' represents the slope, and the value of Φ' reflects the difference between the dynamic reflection path and the direct path, f c Indicates the center frequency of the Wi-Fi signal, and m indicates that the current subcarrier is the mth subcarrier.

4. The method for enhancing the respiratory signal in an interference scenario according to claim 3, characterized in that: The process of calculating the optimal projection angle interval of different subcarrier CSI signals according to the human body position in step 2 includes: calculating the difference between the dynamic reflection path length and the direct path length according to the human body position, substituting it into the linear projection rule relationship to calculate the slope Φ' of the linear projection fitting line, and obtaining the optimal projection angle interval according to the slope Φ'.

5. The method for enhancing the respiratory signal in an interference scenario according to claim 1, characterized in that: The process of obtaining multiple projection systems by using the space traversal method in step 2 includes: traversing in the space of 0-360° with a set step size to obtain corresponding projection systems under different initial angles, and screening the initial phase based on the principle of maximizing the total BNR.

6. The method for enhancing the respiratory signal in an interference scenario according to claim 1, characterized in that: The method for completing the data in step three is: interpolating and completing the time points according to the timestamps of the CSI signals of different subcarriers to complete the data.

7. The method for enhancing the respiratory signal in an interference scenario according to claim 1, characterized in that: The principle of setting the change window length in step three is to make the breathing signal in the data of the segment smooth and approximately a small arc, while the walking signal changes dramatically and has multiple peaks and troughs.

8. The method for enhancing the respiratory signal in an interference scenario according to claim 1, characterized in that: In the step 4, the method of selecting several optimal receiving antennas using the PAR×BNR value as an indicator is as follows: using 0.9 times of the maximum value of the PAR×BNR value as a threshold, selecting a receiving antenna having a PAR×BNR value greater than the threshold from multiple receiving antennas as the optimal receiving antenna.

9. The method for enhancing the respiratory signal in an interference scenario according to claim 1, characterized in that: The method of performing π ambiguity detection and correction on the waveform of the inverted antenna in the receiving antenna in step 4 is: by inverting the waveform of each receiving antenna in turn, and then superimposing it with the CSI signal of other antennas, calculating the PAR×BNR value of each waveform, and screening out the one with the largest value, that is, the antenna with correctly corrected inversion.

10. A system based on the method for enhancing a respiratory signal in an interference scenario according to any one of claims 1 to 9, characterized in that: include: A data preprocessing module, used to preprocess the CSI signals received by at least three receiving antennas respectively to obtain CSI signals preprocessed by the receiving antennas; A position enhancement module is used to calculate the optimal projection angle interval of different subcarrier CSI signals according to the position of the human body based on the linear projection rule, obtain multiple projection systems by using the space traversal method, project each subcarrier in each projection system to obtain a projection waveform, take the projection system with the largest sum of BNR values ​​of the projection waveform as the optimal projection system, add each subcarrier CSI signal in the optimal projection system, and obtain a position-enhanced respiratory signal; The short window transformation module is used to fill in the data, set the transformation window length, divide the position-enhanced respiratory signal into multiple segmented waveforms, perform short projection transformation and strong filtering on each segmented waveform by loop operation, obtain the segmented waveform after short window transformation, and splice the segmented waveform after short window transformation to obtain the secondary enhanced respiratory signal; The multi-antenna fusion screening module is used to calculate the PAR value and BNR value of the secondary enhanced breathing signal, screen out several optimal receiving antennas with PAR×BNR value as the indicator, calculate the average value of the autocorrelation estimate of the optimal receiving antenna, and obtain the breathing rate estimate; perform π fuzzy detection and correction on the waveform of the inverted antenna in the receiving antenna, and obtain the enhanced breathing waveform after superposition with other antenna waveforms.

Images