Wall-penetrating gait recognition system based on transmission RIS and wall-penetrating gait recognition method

Abstract

The present application belongs to the technical field of through-wall gait recognition, and particularly relates to a through-wall gait recognition system construction method and a through-wall gait recognition method based on a transmission RIS, comprising: setting an electromagnetic wave transmitter and a receiver in the spatial positions on both sides of a target wall; sequentially setting transmission RIS on different positions of the wall on the side where the receiver is located, and obtaining the optimal codebook of the transmission RIS after each setting, with the strongest received channel state information strength as the target; setting the transmission RIS according to the RIS setting position corresponding to the strongest channel state information and the optimal codebook in multiple settings; wherein the transmitter is used to transmit electromagnetic wave signals to the space, and the receiver is used to receive the channel state information for gait recognition modulated by the transmission RIS; the set transmitter, receiver, transmission RIS and its controller constitute a through-wall gait recognition system. The present application has low cost and high through-wall gait recognition accuracy in poor signal scenarios.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of through-wall gait recognition, and more particularly relates to a through-wall gait recognition system based on a transmissive RIS and a method for building the system and recognizing gait. BACKGROUND

[0002] Through-wall gait recognition can monitor indoor personnel in real time without infringing on personal privacy, and timely detect and warn potential security threats such as illegal intrusion and violent behavior. This is of great significance to the protection of personal and social safety. For example, compared with traditional video monitoring, through-wall gait recognition technology can penetrate walls for detection, avoiding direct shooting of indoor scenes by cameras, thereby protecting personal privacy to some extent; in the medical field, through-wall gait recognition technology can be used to monitor the walking state of the elderly and timely detect gait abnormalities, providing an important basis for medical diagnosis and treatment; in the event of earthquakes, fires and other disasters, through-wall gait recognition technology can help rescue personnel quickly locate trapped personnel and improve rescue efficiency.

[0003] Currently, through-wall gait recognition technology mainly relies on through-wall radar and advanced signal processing algorithms. Through-wall radar uses low-frequency ultra-wideband electromagnetic waves to penetrate building walls to detect indoor personnel. By extracting the micro-Doppler features of human motion and using the information about the speed and acceleration of human motion, the recognition of human gait is achieved. However, the existing through-wall gait recognition method has the following problems due to the applicability of radar and advanced signal processing algorithms: (1) limited recognition accuracy: due to the attenuation and scattering of electromagnetic waves by walls, the signals received by through-wall radar are often weak and fuzzy, limiting the extraction and recognition accuracy of gait features; (2) high algorithm complexity: in order to extract effective gait features from through-wall radar echoes, complex signal processing algorithms such as multi-layer hybrid dynamic probability graph and fast multi-scale intelligent recognition method are required, which have high computational complexity and require high performance hardware devices; (3) large environmental interference: through-wall gait recognition technology is easily affected by indoor environmental factors such as wall material, thickness, indoor layout, etc., which can change the propagation path and attenuation degree of electromagnetic waves, affecting the accuracy of gait recognition.

[0004] Therefore, there is an urgent need for a low-cost, high-precision alternative method to improve the accuracy of gait recognition in closed scenes such as long corridors. SUMMARY

[0005] In view of the above defects or improvement needs of the prior art, the present application provides a transmission RIS-based through-wall gait recognition system building method and a through-wall gait recognition method, which aims to provide a method with low cost and high through-wall gait recognition accuracy in a poor signal scenario.

[0006] To achieve the above-mentioned purpose, according to one aspect of the present application, a transmission RIS-based through-wall gait recognition system building method is provided, comprising:

[0007] An electromagnetic wave transmitter and a receiver are respectively arranged in the spatial positions on both sides of the target wall;

[0008] The transmission RIS is sequentially arranged on different positions of the wall surface on the side where the receiver is located, and after each arrangement, the optimal codebook of the transmission RIS is obtained by optimizing the target of the strongest received channel state information intensity; the transmission RIS is arranged according to the RIS arrangement position corresponding to the strongest channel state information in multiple arrangements, and a controller is configured for the transmission RIS, so that it can control the phase state of the transmission RIS according to the optimal codebook corresponding to the RIS arrangement position;

[0009] Among them, the transmitter is used to transmit electromagnetic wave signals to the space, and the receiver is used to receive the channel state information for gait recognition modulated by the transmission RIS; the arranged transmitter, receiver, and transmission RIS and its controller constitute a through-wall gait recognition system, and the building is completed.

[0010] Further, the transmission RIS is a 1-bit transmission RIS.

[0011] Further, after each arrangement, the optimal codebook of the transmission RIS is obtained by:

[0012] Initializing variables, including: codebook Φ, maximum received signal intensity s max , column number col, row number row, current codebook Φ current ;

[0013] Iterate multiple rounds, and the iteration number i of each round is from 1 to n+m, n represents the number of rows of the transmission RIS, m represents the number of columns of the transmission RIS, when i≤16, col=col+1, otherwise, row=row+1, and the flipping operation is performed row by row and column by column using the greedy algorithm to obtain the next generation of optimized codebook Φ next ; wherein, every time Φ current =Φ next is obtained, and the current received channel state information intensity s current is obtained, if s current >s max , then s max =s current , Φ=Φ current , and the codebook Φcurrent The transmission RIS is assigned.

[0014] Further, the through-wall gait recognition system further comprises a gait recognition model, which is constructed by the following method:

[0015] The channel state information corresponding to the gait of different volunteers is collected by the receiver to construct a training sample set, and a gait recognition model is trained, wherein each training sample comprises a volunteer label.

[0016] Further, the construction method of each training sample is:

[0017] The channel state information is preprocessed by including a sliding average filtering denoising and a data format conversion;

[0018] The preprocessed channel state information is feature extracted to obtain a key information vector;

[0019] The key information vector is feature selected and dimensionally reduced to obtain a low-dimensional feature vector, and the low-dimensional feature vector and the corresponding volunteer label constitute a training sample.

[0020] According to another aspect of the present application, a through-wall gait recognition method based on a transmission RIS is provided, comprising:

[0021] A through-wall gait recognition system based on a target wall body collects channel state information, wherein the through-wall gait recognition system of the target space is constructed by the through-wall gait recognition system construction method based on the transmission RIS.

[0022] Gait recognition is realized based on the channel state information.

[0023] Further, the through-wall gait recognition system based on the target wall body realizes gait recognition, wherein the through-wall gait recognition system of the target space is constructed by the through-wall gait recognition system construction method based on the transmission RIS.

[0024] Overall, compared with the prior art, the above technical solutions conceived by the present application mainly have the following beneficial effects:

[0025] 1.The application proposes a method for building a wall-penetrating gait recognition system based on a transmissive intelligent surface (RIS) in a poor signal scenario (i.e., a low signal-to-noise ratio scenario). The method utilizes the active regulation characteristics of RIS on electromagnetic waves to improve the strength of the received signal and thus improve the signal-to-noise ratio. In the building process, the position of the transmissive RIS on the wall is optimized to maximize the received signal strength. The method can exponentially enhance the received signal power and the gait recognition accuracy in such scenarios without changing the existing hardware structure of the perception network. It can be mainly used for gait recognition in wall-penetrating and laboratory environments, which are closed and have low signal-to-noise ratios, to achieve accurate gait recognition in target scenarios.

[0026] 2.Based on the system built based on electromagnetic waves and transmissive RIS, the application further proposes using a neural network with lower cost and lower computational complexity than existing signal processing algorithms for signal processing. BRIEF DESCRIPTION OF DRAWINGS

[0027] Figure 1 is a structural schematic diagram of a wall-penetrating gait recognition system provided by an embodiment of the application.

[0028] Figure 2 is a schematic diagram of a wall-penetrating gait recognition system provided by an embodiment of the application.

[0029] Figure 3 is a schematic diagram of a transmissive RIS and its controller provided by an embodiment of the application. DETAILED DESCRIPTION

[0030] To make the objectives, technical solutions, and advantages of the application clearer, the application is further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the application and do not limit the application. In addition, the technical features involved in each embodiment of the application described below can be combined with each other as long as they do not conflict with each other.

[0031] Embodiment One

[0032] A method for building a wall-penetrating gait recognition system based on a transmissive RIS, comprising:

[0033] An electromagnetic wave transmitter and a receiver are respectively arranged in the spatial positions on both sides of the target wall; a transmissive RIS is sequentially arranged on the wall surface on the side where the receiver is located at different positions, and after each arrangement, the strongest received channel state information is taken as the target to optimize the optimal codebook of the transmissive RIS; the transmissive RIS is arranged according to the RIS arrangement position corresponding to the strongest channel state information in multiple arrangements, and a controller is configured for the transmissive RIS to enable the transmissive RIS to control the phase state of the transmissive RIS according to the optimal codebook corresponding to the RIS arrangement position; wherein the transmitter is used to transmit electromagnetic wave signals to the space, and the receiver is used to receive the channel state information modulated by the transmissive RIS for gait recognition; the arranged transmitter, receiver, transmissive RIS and controller thereof constitute a through-wall gait recognition system, as shown in Figure 1 , the construction is completed.

[0034] With the advent of the information age, electromagnetic wave transmission technology has greatly improved the speed and quality of communication. However, in closed spaces such as through-wall and laboratory environments with low signal-to-noise ratio, wireless signals suffer severe attenuation when passing through solid reinforced concrete structures, resulting in extremely poor signal quality and severely affecting the accuracy of gait recognition.

[0035] Reconfigurable Intelligent Surface (RIS) is a cutting-edge 6G technology that represents an innovative electromagnetic surface solution. This artificially constructed electromagnetic surface is composed of numerous carefully designed electromagnetic units, with programmable electromagnetic properties. By finely controlling the phase state of each unit, the transmissive RIS can actively regulate the amplitude, phase, polarization, and frequency of spatial electromagnetic waves, thereby achieving active control of spatial electromagnetic waves.

[0036] Current technology has limitations in addressing the challenges of gait recognition in low signal-to-noise ratio scenarios such as through-wall and laboratory environments. The application of RIS technology provides a possibility for achieving this goal. The present embodiment proposes a method for building a through-wall gait recognition system based on transmissive reconfigurable intelligent surface (RIS), which utilizes the active regulation characteristics of RIS to enhance the transmission effect of through-wall signals and reduce environmental noise interference, significantly improving the signal strength and accuracy of gait recognition in these traditional signal attenuation environments. In addition, this method does not require any modification to existing hardware devices and can be seamlessly integrated into existing perception networks with low construction cost. This technology not only has broad application potential in current low signal-to-noise ratio scenarios, but also provides a new solution for future perception scenarios.

[0037] From a technical principle, the transmission RIS can realize accurate regulation of the through-wall signal by adjusting the electromagnetic properties of its surface units. This regulation capability not only improves the transmission efficiency of the signal, but also enables the signal to better maintain its original characteristic information during the through-wall process, ensuring the identification accuracy. Subsequently, by combining advanced signal processing and deep learning algorithms, effective gait features can be extracted from the through-wall signal.

[0038] The method of the embodiment can also optimize the performance of wireless communication: in a wireless communication system, the transmission RIS can be used to enhance the signal transmission effect and improve the communication quality and coverage range. Especially in complex environments (such as inside buildings, tunnels, etc.), the transmission RIS can significantly improve the performance of the wireless communication system. Therefore, the method of the embodiment can also optimize the propagation path and energy distribution of electromagnetic waves by adjusting the configuration of the surface units of the RIS, thereby improving the signal transmission efficiency and coverage range. The transmission RIS can flexibly regulate the propagation path and energy distribution of electromagnetic waves, enabling the signal to be better transmitted and received in complex environments, thereby improving the communication quality and coverage range.

[0039] In specific implementation, reference can be made to Figure 2 , a computer controls the USRP to control the transmission and reception of signals by the transmitter and receiver.

[0040] As a preferred implementation, the transmission RIS is a 1-bit transmission RIS, and 1-bit means that each unit has two states of phase. As an example, as shown in Figure 3 .

[0041] Regarding specific parameters, the transmission RIS used in the embodiment is 28.8x28.8x0.3 centimeters and includes a two-dimensional electrically controllable RIS element array. The structure of the array is that each row and each column contains 16 RIS elements, totaling 256 elements. Each RIS element is 1.8x1.8x0.3 centimeters and includes four layers of copper layers supported by three substrate layers. It includes two PIN diodes for state modulation: in the 0 state, PIN diode #1 is off and PIN diode #2 is on; for the 1 state, this configuration is reversed, PIN diode #1 is on and PIN diode #2 is off. This binary setting can dynamically control the electromagnetic behavior of the RIS to achieve precise wave control. It should be noted that this is only an example, and the method of the embodiment is not limited to this RIS.

[0042] As a preferred implementation, after each setting, the method for obtaining the optimal codebook of the transmission RIS is:

[0043] Initialize variables, including: codebook Φ, maximum received signal strength s max, column number col, row number row, current codebook Φ current ;

[0044] Traverse multiple rounds, and traverse the number of times i from 1 to n+m each round, n represents the number of rows of the transmissive RIS, m represents the number of columns of the transmissive RIS, when i<=16, col=col+1, otherwise, row=row+1, the flipping operation is performed row by row and column by column using the greedy algorithm, and the next generation of optimized codebook Φ next ; wherein, each time Φ current =Φ next is traversed, the current received channel state information strength s current is obtained, if s current >s max , s max =s current , and Φ=Φ current , the codebook Φ current is distributed to the transmissive RIS.

[0045] In order to clarify the optimization process of the RIS configuration (i.e. codebook), the model of the received signal can be described in detail as follows: y=hΦHx+w, wherein x represents the transmission signal (transmission signal), Φ represents the phase shift matrix implemented on the transmissive RIS (i.e. codebook), w represents additive white Gaussian noise with zero mean, and the variance is σ 2 . The matrices H and h represent the channel matrix from the transmitter to the transmissive RIS, and the channel matrix from the transmissive RIS to the receiver, respectively.

[0046] The signal-to-noise ratio is defined by , and the optimization problem can be expressed as:

[0047]

[0048] In the formula, |φ n | represents the phase shift modulus value of the RIS unit n, and N represents the total number of RIS units in the transmissive RIS.

[0049] Using the greedy algorithm, a local (i.e. current) optimal codebook configuration is obtained iteratively, and the received signal strength is gradually enhanced. This process is called configuration alternating optimization algorithm.

[0050] In the specific application of the method of the embodiment, first, according to the application scene type and the site size, the RIS board position is determined by simulation optimization; the transmissive RIS is connected through a computer, the codebook of the transmissive RIS is controlled using matlab, the optimal codebook of the RIS is obtained using the traversal algorithm, thereby completing the beam direction regulation and control of the transmissive RIS board to enhance the perception ability; after the system is built, the transmissive RIS will enhance the signal-to-noise ratio, and the system obtains the channel state information (received signal) to realize accurate gait recognition.

[0051] As a preferred implementation, the wall-penetrating gait recognition system further comprises a gait recognition model, which is constructed by the following method:

[0052] The channel state information corresponding to the gait of different volunteers is collected by the receiver to construct a training sample set, and a gait recognition model is trained, wherein each training sample comprises a volunteer label.

[0053] The construction method of each training sample is as follows:

[0054] The channel state information is preprocessed by including sliding average filtering denoising and data format conversion;

[0055] The preprocessed channel state information is feature extracted to obtain a key information vector;

[0056] The key information vector is subjected to feature selection and dimension reduction operations to obtain a low-dimensional feature vector, and the low-dimensional feature vector and the corresponding volunteer label constitute a training sample.

[0057] The channel state information can be amplitude information or phase information of electromagnetic signals, and exists in the form of a numerical matrix with dimensions of 250(time)*8192(space). First, the preprocessed CSI data is feature extracted to obtain a data vector to capture the key information corresponding to the corresponding gait pattern. Then, the data vector is subjected to feature selection and dimension reduction operations to reduce the data dimension and retain the most relevant features.

[0058] In summary, this embodiment uses an algorithm to calculate the codebook configuration of the transmission RIS to optimize the signal-to-noise ratio. In particular, in a low signal-to-noise ratio environment such as a closed laboratory, an efficient network architecture is designed and deployed to ensure accurate identification. Regarding data collection and preprocessing, the transmitter and receiver in the CSI monitoring environment are used to collect CSI data of different volunteers' gait patterns and perform preprocessing, including denoising and data format conversion, to prepare a high-quality data set for subsequent processing. Regarding non-invasive gait recognition, CSI data is used to achieve non-invasive gait recognition, avoiding interference with the identified individual and improving the naturalness and accuracy of the identification. Regarding the recognition ability in complex scenarios, gait features in CSI data are extracted to enhance the recognition ability in complex scenarios, such as closed spaces and low signal-to-noise ratio environments. Regarding gait recognition model training and optimization, neural network models are trained and optimized to improve the accuracy and generalization ability of gait recognition, ensuring identification performance in different individuals and environments. Regarding parameter adjustment and structure optimization, the model performance is evaluated based on the identification results, and the parameters are adjusted and the neural network structure is optimized to achieve accurate recognition of complex gait patterns.

[0059] Embodiment Two

[0060] A wall-penetrating gait recognition method based on a transmissive RIS, comprising:

[0061] A wall-penetrating gait recognition system based on a target wall, collects channel state information, wherein the wall-penetrating gait recognition system of the target space is built by the wall-penetrating gait recognition system building method based on a transmissive RIS as described above.

[0062] Gait recognition is realized based on the channel state information.

[0063] As a preferred embodiment, the wall-penetrating gait recognition system based on a target wall realizes gait recognition, wherein the wall-penetrating gait recognition system of the target space is built by the wall-penetrating gait recognition system building method based on a transmissive RIS as described above.

[0064] The related technical solutions are the same as above, and will not be repeated here.

[0065] Those skilled in the art will easily understand that the above description is only a preferred embodiment of the present application, and is not intended to limit the present application, and any modifications, equivalent replacements and improvements made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for building a wall-penetrating gait recognition system based on a transmissive RIS, comprising: setting an electromagnetic wave transmitter and receiver in the spatial positions on both sides of a target wall; sequentially setting a transmissive RIS on different positions of the wall on the side where the receiver is located, and after each setting, taking the strongest received channel state information as the target, and optimizing the optimal codebook of the transmissive RIS; setting the transmissive RIS according to the RIS setting position corresponding to the strongest channel state information in multiple settings, and configuring a controller for the transmissive RIS so that it can control the phase state of the transmissive RIS according to the optimal codebook corresponding to the RIS setting position; wherein the transmitter is used to transmit electromagnetic wave signals into space, and the receiver is used to receive channel state information modulated by the transmissive RIS for gait recognition; the set transmitter, receiver, and transmissive RIS and its controller constitute a wall-penetrating gait recognition system, and the building is completed; the wall-penetrating gait recognition system further comprises a gait recognition model, which is obtained by the following method: collecting channel state information corresponding to different volunteers' gaits through the receiver, constructing a training sample set, and training to obtain a gait recognition model, wherein each training sample includes a volunteer label; the construction method of each training sample is: performing pre-processing operations including sliding average filtering denoising and data format conversion on the channel state information; performing feature extraction on the pre-processed channel state information to obtain a key information vector; performing feature selection and dimensionality reduction operations on the key information vector to obtain a low-dimensional feature vector, and the low-dimensional feature vector and the corresponding volunteer label constitute a training sample.

2. The wall-penetrating gait recognition system setup method of claim 1, wherein, The transmissive RIS is a 1-bit transmissive RIS.

3. The wall-penetrating gait recognition system setup method of claim 2, wherein, After each setting, the optimal codebook of the transmissive RIS is obtained by the following method: Initialization of variables, including: codebook , maximum received signal strength , column number col, row number row, current codebook ; Traverse multiple rounds, traverse each round a number of times i From 1 to n+m, n represents the number of rows of the transmission RIS, m represents the number of columns of the transmission RIS, col = col + 1 when i ≤ 16, otherwise, row = row + 1, and the flipping operation is performed row by row and column by column using the greedy algorithm to obtain the next generation of optimized codebook ; wherein, each time the codebook is traversed, = , and the current received channel state information strength is obtained , if > , then = , = , the codebook is assigned to the transmission RIS.

4. A through-the-wall gait recognition method based on transmission RIS, characterized in that, comprising: a wall-penetrating gait recognition system based on a target wall, which collects channel state information, wherein the wall-penetrating gait recognition system of the target space is built by a method for building a wall-penetrating gait recognition system based on a transmissive RIS according to any one of claims 1 to 3; realizing gait recognition based on the channel state information.

Images