Human body sensing method and apparatus based on interoceptive integrated routing, device, and medium

Abstract

The application provides a human perception method and device based on a general sense integrated routing, equipment and medium, comprising: building a communication and perception environment; collecting CSIs of different actions in a perception area, and dividing the collected CSI data set into a test set and a training set; preprocessing the collected CSIs; optimizing the preprocessed training set, and removing invalid data streams; and submitting the optimized training set to an SVM for training, and using the training set for predicting the test set. The design scheme and human intelligent perception method based on the general sense integrated routing can realize the human action recognition of the general sense integration based on the upgrade and modification of the conventional router which can support the WiFi networking; the WiFi router is used to carry the perception equipment, and has the advantages of low cost and easy carrying; and the data processing and analysis method is used to make the judgment result more accurate.

Description

Technical Field

[0001] This invention relates to the field of wireless signal behavior recognition technology, specifically to human body perception methods, devices, equipment, and media based on integrated sensing and communication routing. Background Technology

[0002] With the development of communication and sensing technologies, non-contact systems for sensing human movements have given rise to sensing technologies such as WiFi, infrared, radio frequency identification, and computer vision. WiFi sensing technology, due to its advantages in universality, effectiveness, and low cost, has become a research hotspot for human movement recognition using integrated sensing technology.

[0003] Currently, wireless sensing is achieved through dedicated sensor devices, while information transmission within communication devices has little overlap with sensing. Traditional sensing devices suffer from issues such as large size and high power consumption, and other communication-based sensing devices also present problems with portability. In contrast, WiFi routers equipped with sensing devices offer advantages such as easy deployment, enabling human perception through wireless communication. Currently, WiFi signal identification primarily relies on RSSI (Received Signal Strength Indicator), but RSSI is highly susceptible to environmental noise and multipath effects, leading to signal instability and inaccuracy.

[0004] In view of the above, this application is hereby submitted. Summary of the Invention

[0005] In view of this, the purpose of the present invention is to provide a human body sensing method, device, equipment and medium based on integrated sensing routing, which can effectively solve the problem that the identification of WiFi routers equipped with sensing devices in the prior art mainly relies on RSSI, but RSSI is easily affected by environmental noise and multipath effect, resulting in unstable and inaccurate signals.

[0006] This invention discloses a human body perception method based on integrated sensor routing, comprising:

[0007] The integrated sensor router is switched to monitoring mode to monitor the target device, obtain the CSI data file of the target device, store the CSI data file, and upload the CSI data file to the cloud server.

[0008] The integrated sensor router is placed within a preset WiFi signal transmission and reception area. PCAP file groups of preset parent actions of multiple detectors are collected and tagged and classified separately to generate CSI data training set and CSI data test set, which are then uploaded to the cloud server.

[0009] The CSI data file is transmitted according to the IEEE 802.11 protocol under a preset bandwidth using a signal modulated by OFDM with 128 subcarriers, wherein the 128 subcarriers include 14 empty subcarriers and 6 lead subcarriers.

[0010] The CSI data training set is subjected to LDA dimensionality reduction to generate outliers appearing in the labels of the CSI data training set, and the outliers in the CSI data training set are clustered using the DBSCAN algorithm.

[0011] The CSI data training set after clustering by the DBSCAN algorithm is input into the SVM for training. The CSI data test set is then used for prediction. The covariance matrix is ​​reduced to two feature values ​​based on the pre-saved covariance matrix to generate the recognition result.

[0012] Preferably, the integrated sensor router is switched to monitoring mode to monitor the target device, acquire the CSI data file of the target device, store the CSI data file, and simultaneously upload the CSI data file to the server. Specifically:

[0013] Run the script file, use the integrated sensor router as a receiver, and establish a TCP communication link;

[0014] The integrated sensor router monitors the target device, obtains the LTF from the preamble of the target device's WiFi frame, and generates the CSI data file. The integrated sensor router consists of a traditional router and a data acquisition MCU module, which is configured to acquire CSI data packets.

[0015] Preferably, the integrated sensor router monitors the target device, obtains the LTF from the preamble of the target device's WiFi frame, and generates the CSI data file, specifically as follows:

[0016] Install a plugin that can capture WiFi frames on the acquisition MCU module to modify the firmware of the acquisition MCU module, so that the acquisition MCU module has WiFi monitoring function;

[0017] The acquisition MCU acquires WiFi frames of the target MAC address through the physical layer and copies the WiFi frame from the physical layer to RAM;

[0018] The acquisition MCU extracts the CSI from the LTF in the preamble of the WiFi frame, and sends the CSI to the main MCU in the form of UDP packets by installing tcpdump and libpcap to obtain the CSI data packets and generate the CSI data file.

[0019] Preferably, the integrated sensor router is placed within a preset WiFi signal transmission and reception area, and PCAP file groups of preset parent actions from multiple detectors are collected, categorized and labeled separately to generate CSI data training and test sets, which are then uploaded to the cloud server. Specifically:

[0020] The integrated sensor router is placed within a preset WiFi signal transmission and reception area, and PCAP file groups of preset parent actions (standing, sitting, squatting, lying down) of multiple detectors are collected respectively.

[0021] The WiFi signal transmission and reception sensing area is a Fresnel region, which is an elliptical region with the transmitter and receiver as foci. The boundary of the nth Fresnel region is defined as...

[0022]

[0023] in, Let wavelength be the wavelength at which M1 transmits radio signals to M2. For the first A Fresnel region boundary point, As the transmitting end As a receiver, the signal transmission process of a WiFi transceiver is concentrated in 8-12 Fresnel zones. More than 70% of the radio frequency energy propagates through the first Fresnel zone. Human activities can cause path differences in signal transmission. The formula is

[0024]

[0025]

[0026]

[0027] in, Centered on the first Fresnel elliptic region. and These are the left and right endpoints of the region, respectively. The major diameter of the elliptical region The intersection of the perpendicular line and the boundary of the region. Left endpoint To the center Distance , Right endpoint To the center Distance , Intersection To the center Distance .

[0028] Preferably, the CSI data file is transmitted according to the IEEE 802.11 protocol under a preset bandwidth using a signal modulated by OFDM with 128 subcarriers, specifically as follows:

[0029] The 14 empty subcarriers and the 6 leading subcarriers are filtered out, and the CSI amplitude waveform is filtered out of glitches by a Hampel filter;

[0030] The remaining 108 subcarriers are used to form the CSI amplitude image. Lagrange interpolation is used to dither the CSI amplitude image, and PCA filter is used to reduce the dimensionality of the CSI amplitude image.

[0031] Wherein, the Lagrange interpolation formula is

[0032]

[0033]

[0034]

[0035] in, For the Lagrange fundamental polynomial, For the existence of jitter subcarrier index, and To construct a known subcarrier index for the fundamental Lagrange polynomial , The subcarrier amplitudes are known in order to construct the Lagrange polynomial.

[0036] Preferably, the CSI data training set is subjected to LDA dimensionality reduction processing to generate outliers appearing in the labels of the CSI data training set, and the outliers in the CSI data training set are clustered using the DBSCAN algorithm, specifically as follows:

[0037] The variance contribution rate of the CSI data training set is calculated, and the remaining 108 subcarriers are replaced by the two main features after dimensionality reduction, so as to store the dimensionality reduction matrix of the CSI data training set for dimensionality reduction of the CSI data test set.

[0038] The DBSCAN algorithm is used to form a cluster of density-connected points in the CSI data training set. Outliers in the cluster are then clustered, with regions of lower density being considered noise or outliers.

[0039] Preferably, the CSI data training set after clustering by the DBSCAN algorithm is input into an SVM for training, and the CSI data test set is used for prediction. The covariance matrix is ​​reduced to two feature values ​​based on a pre-saved covariance matrix to generate the recognition result. Specifically:

[0040] The CSI data training set, after being labeled by the DBSCAN algorithm, is transmitted to the machine learning algorithm SVM for learning in order to solve classification and regression problems;

[0041] The trained SVM is invoked to predict the CSI data test set. The input data is mapped to a higher dimension using a linear kernel function. The recognition result is generated by maintaining the computational power by calculating the inner product of the two input vectors.

[0042] The present invention also discloses a human body sensing device based on integrated sensor routing, comprising:

[0043] The data acquisition unit is used to switch the integrated sensor router to the monitoring mode, monitor the target device, acquire the CSI data file of the target device, store the CSI data file, and upload the CSI data file to the cloud server.

[0044] The data classification unit is used to place the integrated sensor router within a preset WiFi signal transmission and reception area, collect PCAP file groups of preset parent actions of multiple detectors, classify them by labels, generate CSI data training set and CSI data test set, and upload them to the cloud server.

[0045] The data transmission unit is used to transmit the CSI data file in accordance with the IEEE 802.11 protocol under a preset bandwidth using a signal modulated by OFDM with 128 subcarriers, wherein the 128 subcarriers include 14 empty subcarriers and 6 lead subcarriers.

[0046] The data dimensionality reduction unit is used to perform LDA dimensionality reduction processing on the CSI data training set, generate outliers appearing in the labels of the CSI data training set, and perform clustering processing on the outliers of the CSI data training set using the DBSCAN algorithm.

[0047] The result generation unit is used to input the CSI data training set after clustering by the DBSCAN algorithm into the SVM for training, predict the CSI data test set, reduce the dimensionality of the pre-saved covariance matrix into two feature values, and generate the recognition result.

[0048] The present invention also discloses a human body sensing device based on integrated sensor routing, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the human body sensing method based on integrated sensor routing as described above.

[0049] The present invention also discloses a readable storage medium storing a computer program that can be executed by a processor of the device in which the storage medium is located, to implement the human body perception method based on integrated sensor routing as described in any of the above claims.

[0050] In summary, the human body perception method, device, equipment, and medium based on integrated sensing routing provided in this embodiment establish a communication sensing environment; collects CSI (Communication Sensor Indicator) data of different actions within the sensing area, and divides the collected CSI dataset into a test set and a training set; preprocesses the collected CSI data; optimizes the preprocessed training set by removing invalid data streams; and feeds the optimized training set into an SVM (Simultaneous Vector Machine) for training, which is then used to predict the test set. This solves the problem that in existing technologies, the identification of sensing devices on WiFi routers mainly relies on RSSI (Resonance Signal Sensor Indicator), but RSSI is highly susceptible to environmental noise and multipath effects, leading to signal instability and inaccuracy. Attached Figure Description

[0051] Figure 1 This is a flowchart illustrating the human body perception method based on integrated sensor routing provided in the first aspect of the present invention.

[0052] Figure 2 This is a flowchart illustrating the human body perception method based on integrated sensor routing provided in the second aspect of the present invention.

[0053] Figure 3 This is a schematic diagram of the transceiver communication model of the human body perception method based on integrated sensing routing provided in this embodiment of the invention.

[0054] Figure 4 This is a schematic diagram of a subcarrier containing the original CSI amplitude information in the human perception method based on integrated sensing routing provided in this embodiment of the invention.

[0055] Figure 5 This is a schematic diagram of the CSI acquisition process of the human perception method based on integrated sensor routing provided in this embodiment of the invention.

[0056] Figure 6 This is a schematic diagram of a subcarrier containing CSI amplitude information after Hampel filtering in the human perception method based on integrated sensing routing provided in this embodiment of the invention.

[0057] Figure 7This is a schematic diagram of a Lagrange interpolated subcarrier in the human perception method based on integrated sensing routing provided in this embodiment of the invention.

[0058] Figure 8 This is a schematic diagram of a subcarrier containing CSI amplitude information after PCA filtering in the human perception method based on integrated sensing routing provided in this embodiment of the invention.

[0059] Figure 9 This is a schematic diagram of the variance contribution rate of three subcarrier points containing principal components in the LDA dimensionality reduction of the human perception method based on integrated sensing routing provided in this embodiment of the invention.

[0060] Figure 10 This is a two-dimensional planar diagram of LDA dimensionality reduction points and a schematic diagram of subcarrier points (red circles) displaying abnormal label values ​​by DBSCAN clustering for the human perception method based on integrated sensing and routing provided in this embodiment of the invention.

[0061] Figure 11 This is a schematic diagram of a human body sensing device based on integrated sensing routing provided in an embodiment of the present invention. Detailed Implementation

[0062] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to represent selected embodiments of the invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0063] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0064] This invention discloses a human body perception method, device, equipment, and medium based on integrated sensor routing, which at least partially solves the shortcomings of the prior art.

[0065] Please see Figures 1 to 2 The first embodiment of this invention provides a human body sensing method based on integrated sensing routing, which can be executed by a human body sensing device (hereinafter referred to as the sensing device), specifically by one or more processors within the sensing device, to achieve the following steps:

[0066] S101, switch the integrated sensor router to monitoring mode, monitor the target device, obtain the CSI data file of the target device, store the CSI data file, and upload the CSI data file to the cloud server.

[0067] Specifically, step S101 includes: running a script file, using the integrated sensor router as a receiver, and establishing a TCP communication link;

[0068] The integrated sensor router monitors the target device, obtains the LTF from the preamble of the target device's WiFi frame, and generates the CSI data file. The integrated sensor router consists of a traditional router and a data acquisition MCU module, which is configured to acquire CSI data packets.

[0069] The integrated sensor router monitors the target device, obtains the LTF from the preamble of the target device's WiFi frame, and generates the CSI data file. Specifically, it installs a plugin that can collect WiFi frames on the acquisition MCU module to modify the firmware of the acquisition MCU module, so that the acquisition MCU module has WiFi monitoring function.

[0070] The acquisition MCU acquires WiFi frames of the target MAC address through the physical layer and copies the WiFi frame from the physical layer to RAM;

[0071] The acquisition MCU extracts the CSI from the LTF in the preamble of the WiFi frame, and sends the CSI to the main MCU in the form of UDP packets by installing tcpdump and libpcap to obtain the CSI data packets and generate the CSI data file.

[0072] In this embodiment, the sensing device may be a user terminal device (such as a smartphone, smart computer, or other smart device), which can establish a communication connection with a cloud server to achieve data interaction.

[0073] Currently, WiFi signal identification mainly relies on RSSI (Received Signal Strength Indicator), but RSSI is highly susceptible to environmental noise and multipath effects, resulting in unstable and inaccurate signals.

[0074] Please see Figures 3 to 4Specifically, in this embodiment, based on the human intelligent perception process under the CSI acquisition system, an integrated sensing router environment is built. The integrated sensing router is switched to monitoring mode based on the 5GHz frequency band to monitor the target device, obtain CSI data files, store the CSI data files, and transmit them to the server. The CSI acquisition system consists of a main MCU (Microcontroller Unit) that transmits data and an integrated sensing router that receives data. The integrated sensing router extracts CSI information from the LTF (Long Training Field) in the preamble of the WiFi frame, thus not affecting the uplink data transmission speed. Furthermore, the integrated sensing router consists of a traditional router and an acquisition MCU (M2) module, with the acquisition MCU responsible for acquiring CSI data packets. The specific steps are as follows: The MCU firmware is modified by installing a plugin that can capture WiFi frames, enabling WiFi monitoring. The MCU's network card is a FullMAC chip (the wireless network MAC layer is implemented in hardware). First, WiFi frames at the target MAC address are captured through the physical layer. Then, the network card core copies the WiFi frames from the physical layer to RAM (Random Access Memory) and extracts the CSI from the LTF in the WiFi frame's preamble. Finally, by installing tcpdump (dump the traffic on a network) and libpcap (Packet Capture Library), the CSI is sent to the MCU host in UDP (User Datagram Protocol) format to obtain the CSI data packet. Before the acquisition process, the main MCU runs a script file, and the sensor router acts as the receiver, establishing a TCP communication link between them. The acquisition MCU module connected to the router collects packets sent by the main MCU; the collected packets are in PCAP format. The subject completes the test within the sensing area formed by the main MCU and the sensor router.

[0075] In this embodiment, the target device is monitored to obtain CSI data files, which are then stored and transmitted to the server. The MCU in the sensor router also runs two scripts: one to enable its listening mode, and the other to begin collecting packets from a specified number of target MAC addresses. The PCAP packet header contains information such as the file identifier, version, timestamp, and capture length. Each data packet header contains information such as the packet length and timestamp, while the packet content is the raw network layer or link layer data. PCAP packets can be used to analyze network traffic, detect anomalies, and debug protocols.

[0076] S102, the integrated sensor router is placed in a preset WiFi signal transmission and reception sensing area, and PCAP file groups of preset parent actions of multiple detectors are collected respectively, and labeled and classified respectively to generate CSI data training set and CSI data test set, which are then uploaded to the cloud server.

[0077] Specifically, step S102 includes: placing the integrated sensor router in a preset WiFi signal transmission and reception area, and collecting PCAP file groups of preset mother actions of multiple detectors, such as standing, sitting, squatting, and lying down;

[0078] The WiFi signal transmission and reception sensing area is a Fresnel region, which is an elliptical region with the transmitter and receiver as foci. The boundary of the nth Fresnel region is defined as...

[0079]

[0080] in, Let wavelength be the wavelength at which M1 transmits radio signals to M2. For the first A Fresnel region boundary point, As the transmitting end As a receiver, the signal transmission process of a WiFi transceiver is concentrated in 8-12 Fresnel zones. More than 70% of the radio frequency energy propagates through the first Fresnel zone. Human activities can cause path differences in signal transmission. The formula is

[0081]

[0082]

[0083]

[0084] in, Centered on the first Fresnel elliptic region. and These are the left and right endpoints of the region, respectively. The major diameter of the elliptical region The intersection of the perpendicular line and the boundary of the region. Left endpoint To the center Distance , Right endpoint To the center Distance , Intersection To the center Distance .

[0085] Specifically, in this embodiment, the integrated sensor router is placed in the sensing area for WiFi signal transmission and reception. The subject completes four parent actions in sequence—standing, sitting, squatting, and lying down—within the sensing area formed by the MCU and the sensor router. For each action, 40 PCAP (Packet Capture) files are collected. The corresponding actions are labeled and classified to obtain 160 PCAP files. The CSI dataset is divided into training and testing sets and then transmitted back to the cloud for processing.

[0086] In this embodiment, the CSI acquisition process mainly consists of two parts: the subject sequentially completes four basic actions—standing, sitting, squatting, and lying down—within the sensing area formed by the MCU and the sensor router. This sensing area is a Fresnel region. Within the Fresnel region, it is an elliptical region with the transmitter and receiver as its focal points. The boundary of the nth Fresnel region is defined as follows: ,in, As the transmitting end As a receiver, the signal transmission process of a WiFi transceiver is concentrated in 8-12 Fresnel zones. More than 70% of the radio frequency energy propagates through the first Fresnel zone. Human activities can cause path differences in signal transmission. The formula is

[0087]

[0088]

[0089]

[0090] in, Centered on the first Fresnel elliptic region. and These are the left and right endpoints of the region, respectively. The major diameter of the elliptical region The intersection of the perpendicular line and the boundary of the region. Left endpoint To the center Distance , Right endpoint To the center Distance , Intersection To the center Distance .

[0091] S103, the CSI data file is transmitted using a signal modulated by OFDM with 128 subcarriers under a preset bandwidth according to the IEEE 802.11 protocol, wherein the 128 subcarriers include 14 empty subcarriers and 6 lead subcarriers;

[0092] Specifically, step S103 includes: filtering out the 14 empty subcarriers and the 6 leading subcarriers, and filtering out glitch points in the CSI amplitude waveform through a Hampel filter;

[0093] The remaining 108 subcarriers are used to form the CSI amplitude image. Lagrange interpolation is used to dither the CSI amplitude image, and PCA filter is used to reduce the dimensionality of the CSI amplitude image.

[0094] Wherein, the Lagrange interpolation formula is

[0095]

[0096]

[0097]

[0098] in, For the Lagrange fundamental polynomial, For the existence of jitter subcarrier index, and To construct a known subcarrier index for the fundamental Lagrange polynomial , The subcarrier amplitudes are known in order to construct the Lagrange polynomial.

[0099] Specifically, in this embodiment, a human intelligent perception system is used to collect channel state information. The collected CSI data is preprocessed and stored in the cloud. Each pcap file contains 10 packets, and 128 complex numbers are extracted from each packet. The amplitude of each subcarrier is obtained by modulo the complex numbers. According to the IEEE 802.11 protocol, in the 5GHz band, channel 157, 40MHz bandwidth, a signal modulated by OFDM (Orthogonal frequency-division multiplexing) using 128 subcarriers is used for transmission. Since the 128 subcarriers include 14 empty subcarriers and 6 leading subcarriers, these subcarriers will cause waveform jitter to the CSI amplitude, such as... Figure 5 As shown. First, these 20 subcarriers are removed, and the final result is a CSI amplitude graph with the subcarrier index on the horizontal axis and the CSI amplitude graph on the vertical axis. A CSI amplitude graph can be plotted for each packet. Then, the CSI amplitude waveform is filtered out for glitches using a Hampel filter, as shown. Figure 6 As shown. After the above process, the CSI amplitude image composed of the remaining 108 subcarriers still has jitter issues in the middle 8 subcarriers (indexes 50, 51, 52, 53, 54, 55, 56, 57). To restore the waveform of the original signal, Lagrange interpolation is used to process the jitter. Lagrange interpolation is performed using subcarriers with indices 5, 10, 15, 20, 25, 30, 35, 40, 49, 60, 65, 70, 75, 80, 85, 90, 100, 105 as known points to calculate the Lagrange polynomial. The interpolation point values ​​are then obtained through the polynomial, as shown below. Figure 7 As shown. After performing Lagrange interpolation, dimensionality reduction is finally achieved using a PCA (Principal Component Analysis) filter, reducing the 10 CSI amplitude maps to one. The filtering result is shown below. Figure 8 As shown.

[0100] In this embodiment, the number of spatial streams depends on the transmitter antenna, and the number of subcarriers depends on the bandwidth of the selected WiFi. The number of CSI streams in a WiFi frame is expressed as: Where sts represents the transmit spatial stream and rx represents each receive antenna. Since traditional routers have only one receive antenna and the MCU transmitting data has one transmit antenna, the WiFi bandwidth used in this invention is 40MHz. The number of CSI complex values ​​collected in a single session is 128. Because these 128 subcarriers include 20 empty subcarriers and 8 leading subcarriers, these subcarriers can cause many outliers in the CSI amplitude waveform. Therefore, the preprocessing process aims to remove these 28 abnormal subcarriers using a Hampel filter.

[0101] The specific process for anomalous subcarrier filtering is as follows: Based on the CSI amplitude waveform composed of 128 subcarriers, after a preprocessing process to remove anomalous subcarriers, 8 subcarriers in the remaining 108 subcarriers of the CSI amplitude image still exhibit jitter. Therefore, Lagrange interpolation is used to optimize the amplitude waveform. When using the Hampel filter to optimize the initial data, the Hampel filter makes decisions based on median absolute deviation estimation and can be used to identify outlier data. Its discrimination process is similar to that of the median filter, but the Hampel filter only replaces values ​​that differ from the local median value by a few times the standard deviation. It can be applied to time series to identify outliers and use more representative values ​​for numerical replacement, thereby achieving the purpose of restoring the original signal waveform and optimizing the data.

[0102] The specific process of using Lagrange interpolation to solve the intermediate subcarrier jitter problem is as follows: The CSI amplitude waveform after Hampel filtering consists of the remaining 108 subcarriers, of which 8 subcarriers still exhibit jitter. Since this jitter condition is not suitable for correlation filtering, the Lagrange interpolation formula is used to restore the original signal waveform.

[0103]

[0104]

[0105]

[0106] Data preprocessing is achieved through Lagrange interpolation. Since M2 collects 10 packets each time, the PCA filter is used to filter the 10 CSI amplitude streams collected in the same time into 1. This allows us to determine the impact of different actions on the CSI transceiver CSIM1 between the channel transmitter and receiver.

[0107] S104, perform LDA dimensionality reduction on the CSI data training set to generate outliers appearing in the labels of the CSI data training set, and perform clustering on the outliers of the CSI data training set using the DBSCAN algorithm.

[0108] Specifically, step S104 includes: calculating the variance contribution rate of the CSI data training set, and replacing the remaining 108 subcarriers with the two main features after dimensionality reduction, so as to store the dimensionality reduction matrix of the CSI data training set for dimensionality reduction of the CSI data test set;

[0109] The DBSCAN algorithm is used to form a cluster of density-connected points in the CSI data training set. Outliers in the cluster are then clustered, with regions of lower density being considered noise or outliers.

[0110] Please see Figure 9 Specifically, in this embodiment, based on the sensing environment of the integrated sensing and communication routing, each of the four actions collected includes 40 PCAP files as a training set, and these 160 PCAP files are classified and labeled. After preprocessing, each PCAP file forms a CSI amplitude stream. LDA (Linear Discriminant Analysis) is performed on each labeled CSI amplitude stream for dimensionality reduction, and the corresponding covariance matrix is ​​saved. Based on the variance contribution rate, each CSI amplitude stream is reduced to two eigenvalues, and the dimensionality reduction result is represented as a point on a two-dimensional plane. However, since outliers identified by dimensionality reduction of abnormal packets cannot be considered as valid training set points, outliers appearing in the labeled training set after dimensionality reduction are clustered using DBSCAN (Density-Based Spatial Clustering of Applications with Noise) to identify and remove them from the training set.

[0111] In this embodiment, LDA is a linear dimensionality reduction technique that interprets data using linear combinations of features. It is a supervised dimensionality reduction technique that can be used for classification tasks. By considering class label information, the aim is to find a linear discriminant method, thereby maximizing the variance between different classes and minimizing the variance within the same class. By calculating the variance contribution rate, the initial 108 features can be replaced by two principal features after dimensionality reduction, and this is used to store the dimensionality reduction matrix of the training set for dimensionality reduction of the test set. Since outlier packets cannot be classified into a cluster in the two-dimensional plane after dimensionality reduction by LDA, DBSCAN is used to remove outliers to optimize the training set.

[0112] DBSCAN is a density-based clustering algorithm that distinguishes clusters of arbitrary shapes and can identify noise points. DBSCAN implements clusters of density-connected points, while regions with lower density are considered noise or outliers. Its two main parameters are ε and minPts, where ε represents the radius of the neighborhood and minPts represents the minimum number of points in the neighborhood. The values ​​of ε and minPts need to be adjusted according to the specific LDA dimensionality reduction graph used in different scenarios.

[0113] S105, the CSI data training set after clustering by the DBSCAN algorithm is input into the SVM for training, the CSI data test set is predicted, and the covariance matrix is ​​reduced to two feature values ​​according to the pre-saved covariance matrix to generate the recognition result.

[0114] Specifically, step S105 includes: transmitting the CSI data training set labeled by the DBSCAN algorithm to the machine learning algorithm SVM for learning in order to solve classification and regression problems;

[0115] The trained SVM is invoked to predict the CSI data test set. The input data is mapped to a higher dimension using a linear kernel function. The recognition result is generated by maintaining the computational power by calculating the inner product of the two input vectors.

[0116] Specifically, in this embodiment, the training set labeled by DBSCAN is fed into the machine learning algorithm SVM for learning, using a binary classification model to solve classification and regression problems. SVM possesses good generalization ability, the ability to handle high-dimensional data, and the ability to avoid local optima. The trained SVM is used to predict the test set, using a linear kernel function to map the input data to a higher dimension, while maintaining constant computational power by calculating the inner product of the two input vectors. The DBSCAN-labeled training set is input into the SVM to predict the test set, and the dimensionality is reduced to two feature values ​​based on the saved covariance matrix to obtain the recognition result, such as... Figure 10 As shown. Based on the characteristics of the SVM kernel function, data in the original space is mapped to a higher-dimensional feature space, thus making linearly inseparable data linearly separable. Assume the kernel function is... The input samples of the kernel function are The mapped samples are If the kernel function that needs to satisfy the conditions is K, then K satisfies:

[0117]

[0118] in, and They are respectively and transpose, right The result of the calculation is equivalent to performing a dot product operation on the mapped result, thus completing the mapping without changing the computational complexity.

[0119] In summary, the human body perception method based on integrated sensing routing includes: establishing a communication sensing environment; collecting CSI (Content Indicator Signals) of different actions within the sensing area, and dividing the collected CSI dataset into a test set and a training set; preprocessing the collected CSI; optimizing the preprocessed training set and removing invalid data streams; and feeding the optimized training set into an SVM for training and using it to predict the test set. The design scheme and intelligent human body perception method based on integrated sensing routing of this invention can achieve integrated sensing human action recognition by upgrading and modifying existing traditional routers that support WiFi networking; utilizing a WiFi router equipped with sensing devices offers advantages such as low cost and portability; and employing data processing and analysis methods makes the judgment results more accurate.

[0120] Please see Figure 11 A second embodiment of the present invention provides a human body sensing device based on integrated sensor routing, comprising:

[0121] The data acquisition unit 201 is used to switch the integrated sensor router to the monitoring mode, monitor the target device, acquire the CSI data file of the target device, store the CSI data file, and upload the CSI data file to the cloud server.

[0122] The data classification unit 202 is used to place the integrated sensor router in a preset WiFi signal transmission and reception area, collect PCAP file groups of preset parent actions of multiple detectors, classify them with labels, generate CSI data training set and CSI data test set, and upload them to the cloud server.

[0123] The data transmission unit 203 is used to transmit the CSI data file in accordance with the IEEE 802.11 protocol under a preset bandwidth using a signal modulated by OFDM with 128 subcarriers, wherein the 128 subcarriers include 14 empty subcarriers and 6 lead subcarriers.

[0124] The data dimensionality reduction unit 204 is used to perform LDA dimensionality reduction processing on the CSI data training set, generate outliers appearing in the labels of the CSI data training set, and perform clustering processing on the outliers of the CSI data training set using the DBSCAN algorithm.

[0125] The result generation unit 205 is used to input the CSI data training set after clustering by the DBSCAN algorithm into the SVM for training, predict the CSI data test set, reduce the dimensionality of the pre-saved covariance matrix into two feature values, and generate the recognition result.

[0126] A third embodiment of the present invention provides a human body sensing device based on integrated sensor routing, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the human body sensing method based on integrated sensor routing as described above.

[0127] The fourth embodiment of the present invention provides a readable storage medium storing a computer program that can be executed by a processor of the device in which the storage medium is located, so as to implement the human body perception method based on integrated sensor routing as described in any of the above embodiments.

[0128] Exemplary examples show that the computer program described in the third and fourth embodiments of the present invention can be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules can be a series of computer program instruction segments capable of performing specific functions, which describe the execution process of the computer program in the human perception device implementing sensor-integrated routing. For example, the apparatus described in the second embodiment of the present invention.

[0129] The processor referred to can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor. This processor is the control center of the human body perception method based on integrated sensor routing, connecting various parts of the method via various interfaces and lines.

[0130] The memory can be used to store the computer program and / or modules. The processor, by running or executing the computer program and / or modules stored in the memory, and by calling the data stored in the memory, realizes various functions of the human perception method based on integrated sensor routing. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, text conversion function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, text message data, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0131] If the implemented module is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0132] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0133] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions that fall within the scope of the present invention are within the scope of protection of the present invention.

Claims

1. A human perception method based on interoceptive integrated routing, characterized in that, The application relates to a method for realizing WiFi signal transmission and reception sensing, and belongs to the field of WiFi signal transmission and reception sensing. The method comprises the following steps: placing a WiFi signal transmission and reception sensing integrated router in a preset WiFi signal transmission and reception sensing area, collecting PCAP file groups of preset mother actions of multiple testers, respectively classifying the PCAP file groups, and generating a CSI data training set and a CSI data test set, and uploading the CSI data training set and the CSI data test set to a cloud server. The CSI data file is transmitted under a preset wideband according to an IEEE 802.11 protocol, and a signal modulated by 128 subcarriers through OFDM is used, wherein 14 empty subcarriers and 6 pilot subcarriers are contained in the 128 subcarriers. The CSI data training set is subjected to LDA dimension reduction processing, and abnormal values appearing in labels of the CSI data training set are generated, and the abnormal values of the CSI data training set are subjected to clustering processing through a DBSCAN algorithm. The CSI data training set subjected to the clustering processing through the DBSCAN algorithm is input into an SVM for training, the CSI data test set is predicted, and two characteristic values are generated by dimension reduction according to a pre-stored covariance matrix, and an identification result is generated. The WiFi signal transmission and reception sensing integrated router is placed in a preset WiFi signal transmission and reception sensing area, PCAP file groups of preset mother actions of multiple testers are collected, the PCAP file groups are respectively classified, a CSI data training set and a CSI data test set are generated, and the CSI data training set and the CSI data test set are uploaded to the cloud server. The WiFi signal transmission and reception sensing integrated router is placed in a preset WiFi signal transmission and reception sensing area, PCAP file groups of preset mother actions of multiple testers are collected, the PCAP file groups are respectively classified, a CSI data training set and a CSI data test set are generated, and the CSI data training set and the CSI data test set are uploaded to the cloud server. The WiFi signal transmission and reception sensing area is a Fresnel area, the Fresnel area is an oval area with a transmitter and a receiver as focal points, and the boundary of an n-th Fresnel area is defined as The WiFi signal transmission and reception sensing integrated router is placed in a preset WiFi signal transmission and reception sensing area, PCAP file groups of preset mother actions of multiple testers are collected, the PCAP file groups are respectively classified, a CSI data training set and a CSI data test set are generated, and the CSI data training set and the CSI data test set are uploaded to the cloud server. wherein, the wavelength of the radio signal emitted by M1 towards M2, the first Fresnel region boundary point, as the transmitting end, as the receiving end, the signal transmission process of WiFi transceiver concentrates on 8-12 Fresnel zones, more than 70% of the radio frequency energy is transmitted through the first Fresnel zone, and human activities will cause path difference of signal transmission The formula is wherein is the center of the first Fresnel elliptic area, and are the left and right end points of the area, respectively, is the long diameter of the elliptic area is the intersection of the perpendicular to the long diameter and the boundary of the area, is the distance from the left end point to the center of the circle i.e. , is the distance from the right end point to the center of the circle i.e. , is the distance from the intersection point to the center of the circle i.e. . 2.The human perception method based on the integrated route of common sense according to claim 1, characterized in that, A script file is run, the WiFi signal transmission and reception sensing integrated router is used as a receiver, and a TCP communication link is established. The WiFi signal transmission and reception sensing integrated router monitors a target device, acquires LTF from a preamble section of a WiFi frame of the target device, and generates a CSI data file, wherein the WiFi signal transmission and reception sensing integrated router is composed of a traditional router and a collection MCU module, and the collection MCU module is configured to collect a CSI data packet. The WiFi signal transmission and reception sensing integrated router monitors a target device, acquires LTF from a preamble section of a WiFi frame of the target device, and generates a CSI data file. 3.The human perception method based on the integration of common sense and routing according to claim 2, characterized in that, ​ The plug-in capable of collecting WiFi frames is installed on the collection MCU module to modify the firmware of the collection MCU module, so that the collection MCU module has the WiFi monitoring function; The collection MCU collects the WiFi frame of the target MAC address through the physical layer and copies the WiFi frame from the physical layer to the RAM; The collection MCU extracts the CSI from the LTF in the preamble section of the WiFi frame, sends the CSI in the form of a UDP message to the main MCU by installing tcpdump and libpcap, obtains the CSI data packet, and generates the CSI data file. 4.The human perception method based on the integration of common sense and routing according to claim 1, wherein, The CSI data file is transmitted under a preset wideband according to the IEEE 802.11 protocol, using a signal modulated by 128 subcarriers through OFDM, specifically: The 14 empty subcarriers and the 6 pilot subcarriers are filtered out, and the CSI amplitude waveform is filtered through a Hampel filter to remove burr points; The CSI amplitude image composed of the remaining 108 subcarriers is subjected to jitter processing using Lagrange interpolation, and dimensionality reduction processing is performed on the CSI amplitude image using a PCA filter; The Lagrange interpolation formula is wherein, is a Lagrange basis polynomial, is a subcarrier index of a dithering, and is a known subcarrier index for constructing the Lagrange basis polynomial , is a known subcarrier amplitude for constructing the Lagrange polynomial.

5. The human perception based on the integration of the sense of the route according to claim 4, characterized in that, The CSI data training set is subjected to LDA dimensionality reduction processing to generate abnormal values appearing in the label of the CSI data training set, and the abnormal values of the CSI data training set are subjected to clustering processing through a DBSCAN algorithm, specifically: The variance contribution rate of the CSI data training set is calculated, the remaining 108 subcarriers are replaced by two main features after dimensionality reduction, and a dimensionality reduction matrix of the CSI data training set is stored for dimensionality reduction of the CSI data test set; The DBSCAN algorithm is used to group points with high density in the CSI data training set into a cluster, and the abnormal values in the cluster are clustered, wherein a region with low density is regarded as noise or an abnormal point. 6.The human perception method based on the integration of common sense and routing according to claim 1, wherein, The CSI data training set after clustering processing by the DBSCAN algorithm is input into an SVM for training, and the CSI data test set is predicted, and the covariance matrix is dimensionally reduced into two eigenvalues according to the pre-stored covariance matrix, to generate an identification result, specifically: The CSI data training set after labeling by the DBSCAN algorithm is transmitted to the machine learning algorithm SVM for learning to solve classification and regression problems; The trained SVM is called to predict the CSI data test set, the input data is mapped to a higher dimension using a linear kernel function, the calculation power of the inner product of two input vectors is maintained, and an identification result is generated.

7. A human perception device based on synesthesia integrated routing, characterized by, It includes: The data acquisition unit is used to adjust the all-sensing integrated router to a monitoring mode, monitor a target device, acquire a CSI data file of the target device, store the CSI data file, and upload the CSI data file to a cloud server. The data classification unit is configured to place the general perception integrated router in a preset WiFi signal transceiving perception area, collect PCAP file groups of preset mother actions of multiple testers respectively, and perform respective label classification to generate a CSI data training set and a CSI data test set, and upload the CSI data training set and the CSI data test set to the cloud server. The data transmission unit is configured to transmit the CSI data file according to the IEEE 802.11 protocol under a preset wideband, using a signal modulated by 128 subcarriers through OFDM, wherein the 128 subcarriers include 14 empty subcarriers and 6 pilot subcarriers. The data dimension reduction unit is configured to perform LDA dimension reduction processing on the CSI data training set, generate abnormal values present in a label of the CSI data training set, and perform clustering processing on the abnormal values of the CSI data training set through a DBSCAN algorithm. The result generation unit is configured to input the CSI data training set after the clustering processing of the DBSCAN algorithm into an SVM for training, perform prediction on the CSI data test set, reduce dimensions of a covariance matrix saved in advance into two characteristic values, and generate an identification result. The general perception integrated router is placed in a preset WiFi signal transceiving perception area, PCAP file groups of preset mother actions of multiple testers are collected respectively, and respective label classification is performed to generate a CSI data training set and a CSI data test set, and the CSI data training set and the CSI data test set are uploaded to the cloud server. The general perception integrated router is placed in a preset WiFi signal transceiving perception area, PCAP file groups of preset mother actions of multiple testers are collected respectively, and respective label classification is performed to generate a CSI data training set and a CSI data test set, and the CSI data training set and the CSI data test set are uploaded to the cloud server. The WiFi signal transceiving perception area is a Fresnel area, the Fresnel area is an elliptical area with a transmitter and a receiver as focal points, and the boundary of the nth Fresnel area is defined as wherein, the wavelength of the radio signal transmitted by M1 to M2, the first Fresnel region boundary point, as the transmitting end, as the receiving end, the signal transmission process of the WiFi transceiver is concentrated in 8-12 Fresnel zones, more than 70% of the radio frequency energy is transmitted through the first Fresnel zone, and human activities will cause path difference of signal transmission The formula is wherein is the center of the first Fresnel ellipse region, and are the left and right end points of the region, respectively, is the long diameter of the ellipse region is the intersection of the perpendicular to the long diameter and the boundary of the region, is the distance from the left end point to the center of the circle i.e. , is the distance from the right end point to the center of the circle i.e. , is the distance from the intersection point to the center of the circle i.e. .

8. A human perception device based on interoceptive integrated routing, characterized in that, The processor, the memory, and the computer program stored in the memory and configured to be executed by the processor are included, and the processor implements the human perception method based on the general perception integrated routing according to any one of claims 1 to 6 when the computer program is executed.

9. A readable storage medium, characterized by, The computer program is stored in the storage medium, and the computer program can be executed by the processor of the device where the storage medium is located to implement the human perception method based on the general perception integrated routing according to any one of claims 1 to 6.

Images