Multi-target radio frequency identification system based on space-time coding metasurface

Abstract

The invention discloses a multi-target radio frequency identification system based on a space-time coding metasurface. The multi-target radio frequency identification system comprises a space-time coding metasurface module, a baseband signal transmitting module, a received signal demodulation module, a signal preprocessing module and a target classification module, the space-time coding metasurface module adjusts the coding module in real time, dynamically regulates and controls electromagnetic waves, generates a plurality of detection channels, and realizes recognition scenes of various different targets; the baseband signal transmitting module generates and transmits a baseband signal, and after the baseband signal is modulated by the space-time coding metasurface module, a multi-order radio frequency signal containing the baseband signal is formed; the receiving signal demodulation module receives a radio frequency signal reflected from a target to be identified, extracts a specified harmonic wave and demodulates the radio frequency signal into a baseband signal; the signal preprocessing module transmits processed data of the demodulated signal to the target classification module; the target classification module uses a machine learning algorithm to classify and identify targets. According to the invention, simultaneous detection of multiple targets is realized, and the detection efficiency and speed are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of radio frequency sensing, and particularly relates to a multi-target radio frequency identification system based on spatio-temporal coding metasurface. Background Art

[0002] There are many limitations in traditional detection means for flammable and explosive liquids. Therefore, the development of remote and non-contact detection technology has become the key. However, existing technologies mostly rely on a single WiFi device, with limited ability to detect multi-target liquids and problems such as subcarrier false alarms. Although traditional microwave detection technology can effectively identify liquid types, the equipment is expensive and difficult to promote on a large scale.

[0003] In the prior art, Ahmad Daud et al. published a paper named "Next-Generation Security: Detecting Suspicious Liquids Through Software Defined Radio Frequency Sensing and Machine Learning" in IEEE Sensors Journal. This paper obtains the channel response information of suspicious and non-suspicious liquids by using a software radio device to transmit and receive orthogonal frequency division multiplexing (OFDM) modulated signals, and uses machine learning to achieve liquid classification. However, the defect of this technology is that it can only identify one liquid at a time.

[0004] Chen Wang et al. published a paper named "Towards In-baggage Suspicious Object Detection Using Commodity WiFi" in the IEEE Conference on Communications and Network Security journal. This paper uses the off-the-shelf WiFi signals of low-cost devices to penetrate restricted luggage and obtains the channel response signals to facilitate the detection and identification of suspicious objects and liquids. However, it still does not overcome the limitation of a single detection channel.

[0005] In summary, the prior art faces many challenges in the field of liquid detection. On the one hand, the high cost of the system caused by the expensive equipment of traditional microwave detection technology makes it difficult to be widely applied in occasions that require large-scale deployment. On the other hand, existing radio frequency sensing technologies, especially detection methods relying on devices such as WiFi, are mostly limited to single-target detection, with low efficiency and long time consumption when dealing with multi-targets. At the same time, these technologies have limited signal processing capabilities in complex multi-target environments and it is difficult to ensure the accuracy and reliability of detection. Summary of the Invention

[0006] To overcome the deficiencies of the above-mentioned existing technologies, the purpose of the present invention is to provide a multi-target radio frequency identification system based on spatio-temporal coding metasurfaces. By adopting cost-effective hardware devices and advanced signal processing algorithms, this system significantly reduces the system cost and realizes the simultaneous detection of multiple targets, improving the detection efficiency and speed. In addition, by utilizing the unique properties of time-coded metasurfaces, the present invention can more precisely control and modulate signals, enhancing the signal processing ability, thereby achieving higher detection accuracy and reliability in complex multi-target environments.

[0007] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0008] A multi-target radio frequency identification system based on spatio-temporal coding metasurfaces, which includes a spatio-temporal coding metasurface module, a baseband signal transmitting module, a received signal demodulation module, a signal preprocessing module, and a target classification module;

[0009] The spatio-temporal coding metasurface module adjusts the spatio-temporal modulation sequence in real time, dynamically regulates electromagnetic waves, generates multiple detection channels, and realizes identification scenarios for various different targets;

[0010] The baseband signal transmitting module generates and transmits a baseband signal, which, after being modulated by the spatio-temporal coding metasurface module, forms a multi-order radio frequency signal containing the baseband signal for irradiating multiple targets to be identified; the baseband signal transmitting module and the spatio-temporal coding metasurface module work together to ensure that the transmitted signal can effectively cover the target area.

[0011] The received signal demodulation module receives the radio frequency signal reflected from the target to be identified, extracts the specified harmonics, and demodulates it into a baseband signal; the received signal demodulation module closely cooperates with the spatio-temporal coding metasurface module and the baseband signal transmitting module to ensure that the received signal can accurately reflect the target feature information.

[0012] The signal preprocessing module performs preprocessing operations such as removing outliers, smoothing, and normalizing on the demodulated signal to extract effective target feature information; the signal preprocessing module receives the signal from the received signal demodulation module and transfers the processed data to the target classification module.

[0013] The target classification module classifies and identifies the target using machine learning algorithms based on the signal characteristics after preprocessing. The target classification module is directly connected to the signal preprocessing module to ensure the accuracy and real-time performance of classification.

[0014] The spatio-temporal coding metasurface module includes a metasurface composed of 12×12 linearly polarized metasurface units and an FPGA circuit board for controlling the coding switching of the metasurface; the FPGA circuit board is used to provide different bias voltage signals for the metasurface units, and the metasurface units generate different reflection phases for the incident electromagnetic wave according to the provided different voltage signals; the reflection phases generated by the metasurface units satisfy the digital discrete requirements within 0° to 360°, that is, the two discrete phases of 1 bit differ approximately by 180°, the four discrete phases of 2 bits differ approximately by 90°, and the 2 n discrete phases of n-bit differ approximately by 360° / 2 n (n = 1, 2, …).

[0015] The spatio-temporal coding metasurface module uses the binary particle swarm optimization (BPSO) algorithm to generate a spatio-temporal coding (STC) sequence, and through iterative optimization, finds the optimal coding combination that makes the specific harmonic beam accurately point to a predetermined angle.

[0016] The specific harmonic beam is N harmonics selected from (-5, +5) orders, the predetermined angle is N angles selected from (-90°, 90°), the angle corresponds to the harmonic one by one, and it is required that other non-target harmonics are significantly suppressed at the target harmonic position.

[0017] In the multi-target radio frequency identification system of the spatio-temporal coding metasurface, the placement positions of multiple different targets are at the set harmonic angles, that is, at N predetermined angle positions, where the number of targets is less than or equal to N, and one type of target is placed in each predetermined angle direction;

[0018] The modulation period of the spatio-temporal coding (STC) sequence is set to be greater than 0.04 μs, the corresponding modulation frequency is 0 to 25 MHz, and the frequencies of the N generated harmonics are the radio frequency transceiver frequency plus and minus the modulation frequency multiplied by the harmonic order.

[0019] The spatio-temporal coding (STC) sequence is a periodic array of (M, M, L), where M is the number of rows and columns of the metasurface array, and L is the length of the time sequence.

[0020] The clock rate of the FPGA circuit board is 50 MHz. In order to match the modulation period, the modulation period of the time coding sequence of the FPGA circuit board is the same as that of the spatio-temporal coding (STC) sequence, which is set to be greater than 0.04 μs. At the same time, it is ensured that a single time interval is greater than 0.02 μs. Therefore, the PIN diode switching rate is limited within the range of 0 to 50 MHz to ensure the stable operation of the system.

[0021] The baseband signal transmitting module and the received signal demodulating module are implemented by a Linux operating system computer equipped with GNU radio software. Subsequently, a software-defined radio device USRP B210 is used to complete the modulation, transmission, and reception processes of radio frequency signals. Among them, USRP B210 is connected to a feed horn antenna through a radio frequency cable at the baseband signal transmitting port, and the receiving port is connected to an omnidirectional dipole antenna through a radio frequency cable;

[0022] The USRP B210 is a radio frequency transceiver device, and its radio frequency transceiver frequency is designed to be adjustable. The radio frequency transceiver frequency is set to 70 MHz to 10 GHz as needed.

[0023] The baseband signal in the baseband signal transmitting module is an OFDM modulated signal, where the modulation bandwidth of the OFDM modulated signal is less than the modulation frequency of the space-time coding (STC) sequence to prevent subsequent harmonic frequency band aliasing.

[0024] In the GNU radio software, a flowchart is built. Multiple random data are generated at the transmitting end of the flowchart and mapped to quadrature phase shift keying (QPSK) modulation symbols;

[0025] In particular, according to the input payload data and optional meta-information (such as frame length, sequence number, etc.), a frame header conforming to the OFDM protocol format is generated, and the payload is combined with the frame header to adapt to the multi-carrier characteristics of OFDM and generate a complete protocol frame.

[0026] In the transmission scheme of the OFDM modulated signal, multiple sub-carriers are set, namely data sub-carriers, pilot sub-carriers, and zero sub-carriers; the data sub-carriers are used to transmit valid data, the pilot sub-carriers are used for channel estimation and correction, and the zero sub-carriers are used for spectrum protection;

[0027] The QPSK modulation symbols are assigned to the data sub-carriers to generate a parallel data stream. At the same time, zeros are inserted at the positions of the pilot sub-carriers and the zero sub-carriers to generate an OFDM data frame; subsequently, an inverse fast Fourier transform (IFFT) operation is performed on each frame signal of the OFDM data frame to convert the frequency-domain signal into a time-domain signal; to prevent inter-symbol interference (ISI) and inter-carrier interference (ICI) caused by multipath propagation, a cyclic prefix (CP) is introduced into the time-domain signal, that is, 1 / 4 of the data is read from the end of each frame signal and placed at the frame head to form a cyclic structure.

[0028] The USRP B210 ensures that the omnidirectional dipole antenna can effectively receive the OFDM signal spectrum containing N target harmonics.

[0029] The receiving end of the flowchart is divided into N parallel paths according to different extraction frequency bands of the N target harmonics;

[0030] For each of the N channels of signals, first, the target angle harmonics are shifted to the center frequency point through a frequency shift operation; then, a low-pass filter with a bandwidth equal to that of the OFDM modulated signal is used to obtain a pure target OFDM signal; to further demodulate the signal, first, the cyclic prefix (CP) of each frame of the signal is removed, and the time and frequency offsets are estimated and eliminated using the Van de Beek algorithm with the CP data; afterwards, the time-domain OFDM sample data is converted to the frequency domain using a multi-point fast Fourier transform (FFT); in the frequency domain, the data subcarrier signals of each frame of OFDM, i.e., the modulation symbols, are extracted, and their amplitude information is obtained, providing data support for subsequent analysis of the target liquid channel state response.

[0031] The signal preprocessing module selects a 4-level syms5 wavelet for filtering, adopts a soft heuristic SURE threshold technique when processing the detail coefficients, and particularly considers the influence of scale noise, not only effectively removing outliers but also ensuring clear data transitions; a moving average filter is applied to the wavelet-filtered signal to smooth the data to further suppress high-frequency noise; the smoothed signal data is normalized to adjust the data within the target range, providing reliability for subsequent target feature extraction and classification of the data.

[0032] The signal preprocessing module eliminates singular values by performing a 4-level syms5 wavelet transform on the received signal to ensure clear data transitions; afterwards, to eliminate high-frequency noise, a moving average filter is used to smooth the signal; further, the signal data is placed within the target range through normalization.

[0033] The multi-target radio frequency identification system based on a spatio-temporal coding metasurface has the ability to detect N targets;

[0034] The target classification module classifies multiple target objects at N target angles and is realized by training with machine learning algorithms;

[0035] The target classification module is an application of an advanced machine learning algorithm with data classification capabilities.

[0036] A multi-target radio frequency identification method based on a spatio-temporal coding metasurface includes the following steps;

[0037] First, the spatio-temporal coding metasurface module is used to adjust the coding module in real time, dynamically control electromagnetic waves, generate multiple detection channels, and realize multiple different target recognition scenarios;

[0038] Secondly, the baseband signal transmitting module is used to generate and transmit a baseband signal. After being modulated by the spatio-temporal coding metasurface module, a multi-order radio frequency signal containing the baseband signal is formed for irradiating the multi-targets to be recognized. This module works in cooperation with the spatio-temporal coding metasurface module to ensure that the transmitted signal can effectively cover the target area;

[0039] The received signal demodulation module is used to receive the radio frequency signal reflected from the target and extract the specified harmonics and then demodulate it into a baseband signal. This module closely cooperates with the spatio-temporal coding metasurface module and the baseband signal transmitting module to ensure that the received signal can accurately reflect the target feature information;

[0040] After that, the signal preprocessing module is used to perform preprocessing operations such as removing outliers, smoothing, and normalizing on the demodulated signal to extract effective target feature information. This module receives the signal from the received signal demodulation module and transfers the processed data to the target classification module;

[0041] Finally, according to the signal characteristics after preprocessing, the target classification module uses machine learning algorithms to classify and identify the targets. This module is directly connected to the signal preprocessing module to ensure the accuracy and real-time performance of classification.

[0042] Advantages of the present invention:

[0043] 1. Improve efficiency: The present invention realizes a leapfrog improvement from single-target recognition in traditional liquid detection to multi-target recognition. By parallel processing multiple signals and adopting efficient signal processing algorithms such as frequency shift, low-pass filtering, and FFT transformation, the present invention can simultaneously process the signals reflected by multiple harmonics pointing to multiple targets, thus greatly improving the recognition efficiency. By configuring multiple harmonic receiving channels through the spatio-temporal coding metasurface module and the received signal demodulation module, synchronous processing of harmonic signals in different frequency bands is achieved. This multi-target detection ability makes the present invention have significant advantages in scenarios where multiple liquids need to be monitored simultaneously or a large number of samples need to be screened quickly.

[0044] 2. Customizable channels: It is achieved through the beam steering characteristics of the time-coded metasurface. By introducing the BPSO algorithm to optimize the time modulation coding sequence of the spatio-temporal coding metasurface and using FPGA to make the states of the metasurface units switch periodically with time to generate specific harmonic components in the far field, the present invention can precisely control the phase distribution of the harmonic components and achieve the regulation of the pointing of the specified harmonic beam, that is, the beam angle can be adjusted according to actual needs. This characteristic not only provides more possibilities for the channel type but also enables the present invention to more flexibly adapt to different detection environments and requirements, thereby further enhancing the flexibility and practicality of the system. Description of the Drawings

[0045] Figure 1Illustration of multi - liquid target detection based on spatio - temporal coding metasurface in Embodiment 1 of the present invention. (a) The PC controls the USRP to transmit an OFDM - modulated signal. Meanwhile, the FPGA generates a series of OFDM harmonic beams through periodic coding switching, and the +2 and - 2 order harmonics point to the +30° and - 30° directions respectively. (b) Scattering pattern of the spatio - temporal coding metasurface under the time - coding matrix. (c) OFDM harmonic signals at the receiving end.

[0046] Figure 2 Spatio - temporal modulation sequence of the spatio - temporal coding metasurface optimized by BPSO in Embodiment 1 of the present invention.

[0047] Figure 3 Receiving spectrum of the receiving end of the RF device in Embodiment 1 of the present invention at a sampling rate of 640 kHz.

[0048] Figure 4 Demodulated signal constellation diagram of the receiving signal demodulation module at one of the harmonics in Embodiment 1 of the present invention.

[0049] Figure 5 Demonstration of the processing process of the signal pre - processing module for the amplitude of the primary received signal in Embodiment 1 of the present invention, which are the original signal in the time domain, the smoothed signal, and the normalized signal respectively.

[0050] Figure 6 Classification confusion matrices of three models during multi - target detection in Embodiment 1 of the present invention. (a) SVM; (b) Decision tree; (c) CNN. Detailed implementation mode

[0051] The present invention will be further described in detail below with reference to the accompanying drawings.

[0052] As Figure 1 shown, a multi - target radio frequency identification system based on spatio - temporal coding metasurface includes a spatio - temporal coding metasurface module, a baseband signal transmission module, a receiving signal demodulation module, a signal pre - processing module, and a target classification module;

[0053] The spatio-temporal coding metasurface module consists of a metasurface composed of 12×12 linearly polarized metasurface units and an FPGA circuit board that controls the coding switching of the metasurface; the baseband signal transmission module and the received signal demodulation module are implemented by a Linux operating system computer equipped with GNUradio software. Subsequently, the software-defined radio device USRP B210 is used to complete the modulation, transmission, and reception processes of radio frequency signals. Among them, USRP B210 is connected to a feedhorn antenna through a radio frequency cable at the baseband signal transmission port, and the receiving port is connected to an omnidirectional dipole antenna through a radio frequency cable; the signal preprocessing module consists of steps such as removing outliers, smoothing, and normalizing; the target classification module is the application of an advanced machine learning algorithm with data classification capabilities.

[0054] The multi-target radio frequency identification system based on spatio-temporal coding metasurface has the ability to detect N targets. By optimizing the spatio-temporal coding with BPSO, the energy distribution of harmonic beams in the spatial domain and frequency domain is controlled, so that the beam energies of N different harmonics are concentrated in N different directions. Furthermore, the generated harmonic beams are used to detect targets in the area of interest, and the channel state information is extracted from the demodulated received signals to detect N targets.

[0055] The software-defined radio device USRP B210 is a radio frequency transceiver device, and its radio frequency transceiver frequency is designed to be adjustable, enabling it to flexibly adapt to the wireless communication requirements of different frequency bands. The radio frequency transceiver frequency is set to 70 MHz to 10 GHz as needed.

[0056] The metasurface unit is composed of three dielectric layers, two metal layers, three metal posts, and a DC bias layer. Its characteristics are that the dielectric layer is composed of the first dielectric layer, the second dielectric layer, and the third dielectric layer distributed from top to bottom; the metal layer is composed of a metal patch layer distributed from top to bottom and a complete metal layer serving as the reference ground plane for radio frequency signals and DC bias lines. Among them, the metal patch layer is located on the upper surface of the first dielectric layer, and the complete metal layer serving as the reference ground plane for radio frequency signals and DC bias lines is located between the first dielectric layer and the second dielectric layer.

[0057] The thickness of the first dielectric layer is 3 mm, and its material is F4BM with a relative dielectric constant of 3.5 and a loss tangent of 0.003; the thickness of the second dielectric layer is 0.075 mm, and its material is a laminated material Tu872_1080 with a relative dielectric constant of 3.65; the thickness of the third dielectric layer is 0.5 mm, and its material is FR4 with a relative dielectric constant of 4.4 and a loss tangent of 0.026. The three dielectric layers and the complete metal layer serving as the reference ground plane for radio frequency signals and DC bias lines are characterized by having a length and width of 25 mm.

[0058] The metal patch layer is composed of two metal rectangular patches of the same size and a PIN diode. It is characterized in that the two metal rectangular patches of the same size have a length of 19 mm and a width of 8.8 mm, and the geometric centers of the metal rectangular patches coincide with the geometric center of the metasurface unit in the vertical direction. The distance between the two metal rectangular patches is 0.4 mm. The PIN diode model is Skyworks SMP1340-040LF. The PIN diode is located at the center of the metasurface unit, and its anode and cathode face the directions of the two metal patches respectively and are connected to the two metal patches.

[0059] Among the three metal columns, two of them connect the metal rectangular patch connected to the cathode of the PIN diode to the complete metal layer serving as the reference ground plane for the radio frequency signal and the DC bias line. The radii of the two metal columns are 0.25 mm, and they are distributed on both sides of the inner long side of the metal rectangular patch, 0.2 mm away from the wide side. The other metal column connects the metal rectangular patch connected to the anode of the PIN diode to the DC bias layer and is placed 0.2 mm outside the wide side of the metal rectangular patch. The DC bias layer includes a DC bias line and a fan-shaped stub. It is characterized in that the DC bias line is connected to one of the metal columns connected to the anode of the PIN diode. A fan-shaped stub is integrated on the DC bias line. The fan-shaped opening angle is 60°, and the radius is 4.76 mm, which effectively filters out high-frequency signals and reduces the coupling effect between the radio frequency signal and the DC signal.

[0060] The FPGA circuit board is composed of an FPGA core control board and a control drive circuit. It is characterized in that the FPGA core control board uses the FPGA chip XC7K325TFFG900, which is used to convert the spatio-temporal coding (STC) sequence into digital control signals bit by bit and output them through 144 I / O ports configured thereon. The control drive circuit includes 18 74HC245TS driver chips. The input ends of each driver chip are respectively connected to 8 I / O ports of the FPGA core control board to receive the digital control signals. The output ends of each driver chip are connected in parallel to form 144 output ends of the control drive circuit, and the digital control signals are transmitted to the DC bias lines of 144 metasurface units to control the working state switching of the PIN diodes.

[0061] The spatio-temporal coding (STC) sequence of the spatio-temporal coding metasurface module is generated by using the binary particle swarm optimization (BPSO) algorithm. Through iterative optimization, the optimal coding combination that makes the specific harmonic beam accurately point to the predetermined angle is found.

[0062] The specific harmonic beams are N harmonics selected from the (-5, +5) order, and the predetermined angles are N angles selected from (-90°, 90°), and it is required that significant suppression is achieved at each angular position to reduce harmonic interference. In the multi-target radio frequency identification system of the space-time coding metasurface, the placement positions of multiple targets are at the set harmonic angles, that is, at N predetermined angular positions.

[0063] The modulation period of the space-time coding (STC) sequence of the space-time coding metasurface module is set to be greater than 0.04 μs, and the corresponding modulation frequency is less than 25 MHz. The frequencies of the N generated harmonics are the radio frequency transceiver frequency plus or minus the modulation frequency multiplied by the harmonic order. The space-time coding (STC) sequence of the space-time coding metasurface module is a periodic array of (M, M, L), where M is the number of rows and columns of the metasurface array, and L is the length of the time sequence.

[0064] The clock rate of the FPGA circuit board is 50 MHz. To match the modulation period, the modulation period of the time coding sequence of the FPGA circuit board is the same as the modulation period of the space-time coding (STC) sequence of the space-time coding metasurface module, which is set to be greater than 0.04 μs. At the same time, it is ensured that a single time interval is greater than 0.02 μs. Therefore, the PIN diode switching rate is limited to a range less than 50 MHz to ensure the stable operation of the system.

[0065] The baseband signal in the baseband signal transmission module is an OFDM modulation signal, where the modulation bandwidth of OFDM is less than the modulation frequency of the space-time coding (STC) sequence to prevent subsequent harmonic frequency band aliasing. The flowchart is built in the GNU radio software. First, multiple random data are generated at the transmitting end of the flowchart and mapped to quadrature phase shift keying (QPSK) modulation symbols. In particular, a frame header is added to improve the transmission accuracy of the signal. In the transmission scheme of the OFDM modulation signal, multiple subcarriers are set, including data subcarriers for transmitting valid data, pilot subcarriers for channel estimation and correction, and zero subcarriers for spectrum protection; the QPSK modulation symbols are assigned to the data subcarriers to generate a parallel data stream, that is, an OFDM data frame; subsequently, an inverse fast Fourier transform (IFFT) operation is performed on each frame of signal at multiple points to convert the frequency-domain signal into a time-domain signal; to prevent inter-symbol interference ISI and inter-carrier interference ICI caused by multipath propagation, a cyclic prefix (CP) is introduced into the time-domain signal, that is, 1 / 4 of the data is read from the end of each frame of signal and placed at the frame head to form a cyclic structure.

[0066] The received signal demodulation module first sets the sampling rate of the received signal to ensure that the omnidirectional dipole antenna can effectively receive the OFDM signal spectrum containing N target harmonics.

[0067] The receiving end of the signal flow chart is divided into N parallel paths according to the different extraction frequency bands of N target harmonics; for each path of signal, first, the target angle harmonics are shifted to the center frequency point through a frequency shift operation; then, a low-pass filter with a bandwidth of the OFDM modulation signal bandwidth is used to obtain a pure target OFDM signal; to further demodulate the signal, first, the cyclic prefix (CP) of each frame of the signal is removed, and the time and frequency deviations are estimated and eliminated by using the CP data through the Van de Beek algorithm; then, the time-domain OFDM sample data is converted to the frequency domain by using a multi-point fast Fourier transform (FFT); in the frequency domain, the data sub-carrier signals of each frame of OFDM, that is, the modulation symbols, can be extracted, and their amplitude information is obtained, providing data support for the subsequent analysis of the target liquid channel state response.

[0068] The signal preprocessing module eliminates singular values by performing the 4th-level syms5 wavelet transform on the received signal to ensure clear data transitions; then, in order to eliminate high-frequency noise, a moving average filter is used to smooth the signal; further, the signal data is normalized to be within the target range.

[0069] The target classification module classifies multiple liquid types at N target angles, which is realized by training with machine learning algorithms.

[0070] The working principle of the multi-target radio frequency identification system based on the spatio-temporal coding metasurface is as follows:

[0071] First, the spatio-temporal coding metasurface module is used to adjust the coding module in real time, dynamically control electromagnetic waves, generate multiple detection channels, and realize multiple different target recognition scenarios;

[0072] Secondly, the baseband signal transmitting module is used to generate and transmit a baseband signal, which forms a multi-order radio frequency signal containing the baseband signal after being modulated by the spatio-temporal coding metasurface module and is used to irradiate the multi-targets to be recognized. This module works in cooperation with the spatio-temporal coding metasurface module to ensure that the transmitted signal can effectively cover the target area;

[0073] Further, the received signal demodulation module is used to receive the radio frequency signal reflected from the target and extract the specified harmonics and then demodulate it into a baseband signal. This module works closely with the spatio-temporal coding metasurface module and the baseband signal transmitting module to ensure that the received signal can accurately reflect the target feature information;

[0074] After that, the signal preprocessing module is used to perform preprocessing operations such as removing outliers, smoothing, and normalizing on the demodulated signal to extract effective target feature information. This module receives the signal from the received signal demodulation module and transfers the processed data to the target classification module;

[0075] Finally, based on the preprocessed signal features, the target classification module uses machine learning algorithms to classify and identify the targets. This module is directly connected to the signal preprocessing module to ensure the accuracy and real-time performance of classification.

[0076] Embodiment 1

[0077] Referring to Figure 1 , a multi-target radio frequency identification system based on a spatio-temporal coding metasurface, which includes a spatio-temporal coding metasurface module, a baseband signal transmitting module, a received signal demodulating module, a signal preprocessing module, and a target classification module. The spatio-temporal coding metasurface module is composed of a metasurface consisting of 12×12 linearly polarized metasurface units and an FPGA circuit board for controlling the coding switching of the metasurface; the baseband signal transmitting module and the received signal demodulating module are implemented by a Linux operating system computer equipped with GNU radio software. Subsequently, a software-defined radio device USRP B210 is used to complete the modulation, transmission, and reception processes of radio frequency signals. Among them, USRP B210 is connected to a feed horn antenna through a radio frequency cable at the baseband signal transmitting port, and the receiving port is connected to an omnidirectional dipole antenna through a radio frequency cable; the signal preprocessing module consists of steps such as removing outliers, smoothing, and normalizing; the target classification module is an application of advanced machine learning algorithms with data classification capabilities.

[0078] Referring to Figure 1 , the multi-target radio frequency identification system based on the spatio-temporal coding metasurface of this embodiment has the ability to detect 2 targets. The software-defined radio device USRP B210 is a radio frequency transceiver device, and its radio frequency transceiver frequency is designed to be adjustable. The radio frequency transceiver frequency is set to 5.8 GHz according to needs.

[0079] Referring to Figure 1 and Figure 2 , the spatio-temporal coding (STC) sequence of the spatio-temporal coding metasurface module in this embodiment is generated by using the binary particle swarm optimization (BPSO) algorithm. Through iterative optimization, the optimal coding combination that makes a specific harmonic beam accurately point to a predetermined angle is found. The specific harmonic beam is 2 harmonics selected from (-5, +5) orders, that is, +2 and -2 order harmonics. The predetermined angles are 2 angles selected from (-90°, 90°), that is, +30° and -30°, and it is required that they are significantly suppressed at each other's angular positions. The placement positions of various target objects in the multi-target radio frequency identification system based on the spatio-temporal coding metasurface are at the set harmonic angles, that is, at 2 predetermined angles (+30°, -30°) positions.

[0080] Referring to Figure 1 and Figure 2, the modulation period of the spatio-temporal coding (STC) sequence of the spatio-temporal coding metasurface module in this embodiment is 10 μs, and the corresponding modulation frequency is 0.1 MHz. Then, the frequencies of the +2 and -2 order harmonics are the radio frequency transceiver frequency plus 2 times the modulation frequency and the radio frequency transceiver frequency minus 2 times the modulation frequency, respectively. The spatio-temporal coding (STC) sequence of the spatio-temporal coding metasurface module is a periodic array of (12, 12, 10), where 12 is the number of rows and columns of the metasurface array, and 10 is the length of the time sequence.

[0081] Referring to Figure 1 , the clock rate of the FPGA circuit board in this embodiment is 50 MHz. To match the modulation period, the modulation period of the time coding sequence of the FPGA circuit board is the same as that of the spatio-temporal coding (STC) sequence of the spatio-temporal coding metasurface module, which is set to 10 μs, and the single time interval is 1 μs. Therefore, the PIN diode switching rate is limited to 1 MHz to ensure the stable operation of the system.

[0082] Referring to Figure 1 , the baseband signal in the baseband signal transmitting module of this embodiment is an OFDM modulated signal, where the modulation bandwidth of OFDM is 80 kHz, which is less than the modulation frequency of the spatio-temporal coding (STC) sequence of 0.1 MHz, effectively preventing subsequent harmonic frequency band aliasing. The flowchart is built in the GNU radio software. First, 108 random data are generated at the transmitting end of the flowchart and mapped to quadrature phase shift keying (QPSK) modulation symbols. In particular, a frame header is added to improve the transmission accuracy of the signal. In the transmission scheme of the OFDM modulated signal, 64 subcarriers are set, including 48 data subcarriers for transmitting valid data, 4 pilot subcarriers for channel estimation and correction, and 12 zero subcarriers for spectrum protection; the QPSK modulation symbols are assigned to the data subcarriers to generate a parallel data stream, that is, an OFDM data frame; subsequently, a 64-point inverse fast Fourier transform (IFFT) operation is performed on each frame of the signal to convert the frequency-domain signal into a time-domain signal; to prevent inter-symbol interference ISI and inter-carrier interference ICI caused by multipath propagation, a cyclic prefix (CP) is introduced into the time-domain signal, that is, 1 / 4 of the data is read from the end of each frame of the signal and placed at the frame head to form a cyclic structure; the settings of the OFDM modulated signal in the above experimental process are shown in Table 1.

[0083] Table 1 - OFDM Configuration

[0084]

[0085] Referring to Figure 3 and Figure 4, the receiving signal demodulation module of this embodiment first sets the signal sampling rate of the received signal to 640 kHz to ensure that the omnidirectional dipole antenna can effectively receive the OFDM signal spectrum under the +2 and -2 order harmonics. The receiving end of the signal flow chart is divided into two parallel paths according to the different extraction frequency bands of the +2 and -2 order harmonics; for each path of the signal, first, the target angle harmonic is shifted to the center frequency point through a frequency shift operation; then, a low-pass filter with a bandwidth of 80 kHz is used to obtain a pure target OFDM signal; to further demodulate the signal, first, the cyclic prefix (CP) of each frame of the signal is removed, and the time and frequency offsets are estimated and eliminated using the Van de Beek algorithm with the CP data; then, a 64-point fast Fourier transform (FFT) is used to convert the time-domain OFDM sample data to the frequency domain; in the frequency domain, the data subcarrier signals of each frame of OFDM, that is, the modulation symbols, can be extracted, and their amplitude information is obtained, providing data support for the subsequent analysis of the target liquid channel state response.

[0086] Refer to Figure 5 , the signal preprocessing module of this embodiment performs a 4th-level syms5 wavelet transform on the received signal to eliminate singular values and ensure clear data transitions; then, to eliminate high-frequency noise, a moving average filter is used to smooth the signal; further, the signal data is normalized to be within the target range.

[0087] Refer to Figure 6 , the target classification module of this embodiment classifies three types of liquids (plastic bottled alcohol, plastic bottled cooking oil, and plastic bottled tap water) at two target angles (30°, -30°), and is trained by three machine learning algorithms: support vector machine (SVM), decision tree, and convolutional neural network (CNN).

[0088] Experimental content and results:

[0089] Detect and classify the liquid data under the system model in Embodiment 1 of this example to obtain the model confusion matrix. The results are as Figure 6 shown. The recognition accuracy of the SVM model reaches 91%, the recognition accuracy of the decision tree model reaches 82%, and particularly, the CNN model can reach a recognition accuracy of 99%, maintaining an advantage in the complex scenario of multi-target detection.

[0090] 1. Selection of the signal reflection array: Although the spatio-temporal coding metasurface is a preferred embodiment of the present invention, it is not the only option. Any reflection array technology that can meet the requirement of regulating multiple beams to the specified direction angle can be used as an alternative, as long as it can achieve similar signal regulation and reflection functions.

[0091] 2. Modulation method of radio frequency detection fundamental wave: Although the OFDM modulation method is applied in the present invention, it is not necessary. As long as the modulated transmitted signal can generate a channel response containing characteristic signals for the target object, it can also be used as an alternative solution.

[0092] 3. Number of subcarriers in OFDM modulation: Setting 64 subcarriers is a specific implementation manner of the present invention, but it is not fixed. As long as the OFDM modulation principle is satisfied, the number of subcarriers can be adjusted according to actual needs. Whether it is increased or decreased, it will not affect the core idea and implementation effect of the present invention.

[0093] 4. Harmonic selection in multi-target detection: The present invention uses the +2 and -2 order harmonics to detect two target objects only as a specific example, not a restrictive choice. In fact, the selection of harmonics can be flexible and variable. Not only can the orders be different, but the number can also be adjusted according to needs. As long as it is ensured that each of the multiple harmonic signals can be separated to a specified angle and can be effectively distinguished from the received signal containing multiple target characteristics, then whether other orders of harmonics are selected, or the number of harmonics is increased or decreased, the purpose of multi-target detection can also be achieved. This flexibility enables the present invention to more widely adapt to different detection scenarios and requirements.

[0094] 5. Types of detection targets: Although the present invention is described by taking a static liquid as an example, the detection target is not limited to this. Whether it is other static or dynamic objects, as long as they produce different characteristic responses to signal reception, they can be used as the detection targets of the present invention.

[0095] 6. Target classification algorithms and models: Support Vector Machine (SVM), decision tree, and Convolutional Neural Network (CNN) are several target classification algorithms and models mentioned in the present invention, but they are not the only choices. Any algorithms and models that can effectively classify the received characteristic signals, such as random forest, K-nearest neighbor, neural network, etc., can be used as alternative solutions.

[0096] 7. Radio frequency transceiver device and transmitting and receiving antennas: In the embodiments of the present invention, although the configuration of the radio frequency transceiver device and the transmitting and receiving antennas is specifically mentioned, these are not necessary limitations of the present invention. As long as the signal transceiver function can be achieved, no matter what type of radio frequency transceiver device or transmitting and receiving antenna is used, it can be regarded as meeting the technical requirements of the present invention. This flexibility enables the present invention to more widely adapt to different hardware environments and application scenarios.

Claims

1. A multi-target radio frequency identification system based on spatiotemporal coding metasurface, characterized in that: The system includes a space-time coding metasurface module, a baseband signal transmission module, a received signal demodulation module, a signal preprocessing module and a target classification module; The spatiotemporal coding metasurface module adjusts the spatiotemporal modulation sequence in real time, dynamically regulates electromagnetic waves, generates multiple detection channels, and realizes the recognition scenarios of multiple different targets; The baseband signal transmission module generates and transmits a baseband signal, which, after being modulated by the space-time coding metasurface module, forms a multi-order radio frequency signal containing the baseband signal, which is used to illuminate multiple targets to be identified; The received signal demodulation module receives the radio frequency signal reflected from the target to be identified, extracts the specified harmonics and demodulates them into a baseband signal; The signal preprocessing module performs preprocessing operations such as removing outliers, smoothing, and normalizing on the demodulated signal to extract effective target feature information, and transmits the processed data to the target classification module; The target classification module classifies and identifies targets using a machine learning algorithm based on preprocessed signal features.

2. The multi-target radio frequency identification system based on spatiotemporal coding metasurface according to claim 1 is characterized in that: The space-time coding metasurface module includes a metasurface and an FPGA circuit board; the metasurface is composed of linear polarization metasurface units; The FPGA circuit board is used to control the switching of the metasurface coding, thereby providing different bias voltage signals for the metasurface units, and the metasurface units generate different reflection phases for the incident electromagnetic waves according to the different voltage signals provided; The reflection phase generated by the metasurface unit is between 0° and 360°, which meets the discrete requirements of digitalization. That is, the difference between the two discrete phases of 1 bit is approximately 180°, the difference between the four discrete phases of 2 bit is approximately 90°, and the difference between the two discrete phases of n-bit is approximately 90°. n The discrete phase difference is approximately 360° / 2 n (n=1,2,…).

3. The multi-target radio frequency identification system based on spatiotemporal coding metasurface according to claim 2 is characterized in that: The space-time coding metasurface module uses a binary particle swarm optimization (BPSO) algorithm to generate a space-time coding (STC) sequence, and through iterative optimization, finds the optimal coding combination that allows a specific harmonic beam to accurately point to a predetermined angle; The specific harmonic beams are N harmonics selected from the (-5, +5) order, the predetermined angles are N angles selected from (-90°, 90°), the angles correspond to the harmonics one by one, and other non-target harmonics are required to be significantly suppressed at the target harmonic position; The placement positions of various different targets in the multi-target RFID system of the space-time coded metasurface are at set harmonic angles, that is, at N predetermined angle positions, where the number of targets is less than or equal to N, and one target is placed in each predetermined angle direction.

4. The multi-target radio frequency identification system based on spatiotemporal coding metasurface according to claim 3 is characterized in that: The modulation period of the space-time coding (STC) sequence is set to be greater than 0.04 μs, the corresponding modulation frequency is 0 to 25 MHz, and the frequencies of the N harmonics generated are the modulation frequencies of the radio frequency transceiver frequency plus or minus harmonic orders; The space-time coding (STC) sequence is a periodic array of (M, M, L), where M is the number of rows and columns of the metasurface array, and L is the length of the time series; The clock rate of the FPGA circuit board is 50 MHz. In order to match the modulation period, the modulation period of the time coding sequence of the FPGA circuit board is the same as the modulation period of the space-time coding (STC) sequence of the space-time coding metasurface module, which is set to be greater than 0.04 μs. At the same time, it is ensured that the single time interval is greater than 0.02 μs, and the PIN diode switching rate is limited to the range of 0 to 50 MHz.

5. The multi-target radio frequency identification system based on spatiotemporal coding metasurface according to claim 4 is characterized in that: The baseband signal transmission module and the received signal demodulation module are implemented by a Linux operating system computer equipped with GNU radio software, and then the modulation, transmission and reception process of the radio frequency signal are completed by using the software defined radio device USRP B210; Among them, the USRP B210 is connected to the feed horn antenna through an RF cable at the baseband signal transmission port, and the receiving port is connected to the omnidirectional dipole antenna through an RF cable; The USRP B210 is a radio frequency transceiver device, and its radio frequency transceiver frequency is in an adjustable mode, and the radio frequency transceiver frequency is set to 70MHz to 10GHz; The baseband signal in the baseband signal transmission module is an OFDM modulated signal, wherein the modulation bandwidth of the OFDM modulated signal is smaller than the modulation frequency of the space-time coding (STC) sequence; The flow chart is constructed in the GNU radio software, and a plurality of random data are generated at the transmitting end of the flow chart and mapped into quadrature phase shift keying (QPSK) modulation symbols.

6. The multi-target radio frequency identification system based on spatiotemporal coding metasurface according to claim 5, characterized in that: The baseband signal transmission module generates a frame header that conforms to the OFDM protocol format based on the input payload data and optional meta-information, combines the payload with the frame header to adapt to the multi-carrier characteristics of OFDM, and generates a complete protocol frame; In the transmission scheme of the OFDM modulated signal, a plurality of subcarriers are set, which are data subcarriers, pilot subcarriers and zero subcarriers; The data subcarrier is used to transmit valid data, the pilot subcarrier is used for channel estimation and correction, and the zero subcarrier is used for spectrum protection; The QPSK modulation symbols are assigned to the data subcarriers to generate parallel data streams, and at the same time, zero values ​​are inserted at the positions of the pilot subcarriers and the zero subcarriers to generate OFDM data frames; Subsequently, a multi-point inverse fast Fourier transform (IFFT) operation is performed on each frame signal of the OFDM data frame to convert the frequency domain signal into a time domain signal; A cyclic prefix (CP) is introduced into the time domain signal, that is, 1 / 4 of the data is read from the end of each frame and placed at the head of the frame to form a cyclic structure.

7. The multi-target radio frequency identification system based on spatiotemporal coding metasurface according to claim 5, characterized in that: The USRP B210 ensures that the omnidirectional dipole antenna can receive the OFDM signal spectrum containing N target harmonics; The receiving end of the flow chart is divided into N parallel paths according to different extraction frequency bands of the N target harmonics, and each path performs a frequency shift filtering operation on a specific target harmonic to obtain an OFDM receiving signal at a corresponding angle.

8. The multi-target radio frequency identification system based on spatiotemporal coding metasurface according to claim 7, characterized in that: For each signal in N channels; First, the target angle harmonic is moved to the center frequency through the frequency shift operation; Then, the signal is passed through a low-pass filter with a bandwidth equal to the bandwidth of the OFDM modulation signal to obtain a pure target OFDM signal; Remove the cyclic prefix (CP) of each frame signal and use the CP data to estimate and eliminate time and frequency deviations through the Van de Beek algorithm; Finally, multi-point fast Fourier transform (FFT) is used to convert the time domain OFDM sample data into the frequency domain; in the frequency domain, the data subcarrier signal of each frame of OFDM, that is, the modulation symbol, is extracted, and its amplitude information is obtained to provide data support for the subsequent target liquid channel state response analysis.

9. The multi-target radio frequency identification system based on spatiotemporal coding metasurface according to claim 1, characterized in that: The signal preprocessing module is to filter the received signal through the 4th level syms5 wavelet transform to eliminate singular values, and use the soft heuristic SURE threshold technology when processing detail coefficients to ensure clear data transition; then, the moving average filter is used to smooth the signal after wavelet filtering to eliminate high-frequency noise; the signal data after smoothing is normalized, and the signal data is placed within the target range through normalization; The target classification module is to classify multiple targets under N target angles, which is realized by training of machine learning algorithm; The target classification module is an application of an advanced machine learning algorithm with data classification capabilities.

10. A multi-target radio frequency identification method based on spatiotemporal coding metasurface, characterized in that: The steps include: First, the space-time coding metasurface module is used to adjust the coding module in real time, dynamically control the electromagnetic waves, generate multiple detection channels, and realize the recognition scenarios of various different targets; Secondly, the baseband signal transmission module is used to generate and transmit the baseband signal. After being modulated by the space-time coding metasurface module, a multi-order RF signal containing the baseband signal is formed to illuminate multiple targets to be identified. This module works in conjunction with the space-time coding metasurface module to ensure that the transmitted signal can effectively cover the target area. The received signal demodulation module is used to receive the RF signal reflected from the target, extract the specified harmonics and demodulate them into baseband signals. This module works closely with the space-time coding metasurface module and the baseband signal transmission module to ensure that the received signal can accurately reflect the target feature information. Afterwards, the signal preprocessing module is used to perform preprocessing operations on the demodulated signal by removing outliers, smoothing, and normalizing to extract effective target feature information. This module receives the signal from the received signal demodulation module and passes the processed data to the target classification module; Finally, according to the preprocessed signal characteristics, the target classification module uses machine learning algorithms to classify and identify targets. This module is directly connected to the signal preprocessing module to ensure the accuracy and real-time performance of the classification.

Images