Through-the-wall radar clutter suppression method based on spatial frequency diversity

Abstract

The invention provides a through-the-wall radar clutter suppression method based on spatial frequency diversity, and aims to solve the problems of background clutter interference and difficulty in low-speed moving target detection of a through-the-wall radar in a complex indoor environment. The method mainly comprises the following steps: solving a problem P1 (formula (1) in the specification), and generating a signal waveform with low sidelobe and good Doppler tolerance through multiband radar signal waveform optimization design; solving a problem P2 (formula (2) in the specification), and optimizing a beam array according to antenna directional diagrams of different frequency signals in combination with an antenna beam forming technology; according to multi-band radar array signal processing based on bat bionics, effective distinguishing of target signals and background clutters is achieved by comparing echo power ratios of different frequency signals in different directions.

Description

Technical Field

[0001] The present invention relates to the technical field of radar signal processing, and in particular to a background clutter suppression method for spatial frequency diversity (SFD)-assisted complex indoor environment detection in radar target detection. Background Art

[0002] In complex indoor environments, severe radar clutter poses a significant challenge to detecting people behind walls. While through-wall radar technology can penetrate non-metallic obstacles to detect targets behind walls, strong background clutter from walls and floors severely impacts target detection. Therefore, clutter suppression remains a pressing issue in radar target detection. Traditional clutter suppression methods, such as Moving Target Indication (MTI) and Moving Target Detection (MTD), while effective in uniform environments, often suffer from significant performance degradation in complex and changing environments. These methods primarily rely on Doppler shift to distinguish moving targets from stationary clutter, but their ability to distinguish between moving targets and stationary clutter is significantly reduced in the presence of slow-moving targets or strong clutter interference.

[0003] With the continuous advancement of artificial intelligence technology, radar clutter suppression methods have gradually evolved from traditional MTI and MTD to various modern processing methods. For example, X. Wang et al. introduced a deep learning cascade network based on residual networks to suppress random clutter by enhancing feature extraction; a model based on a generative adversarial network (GAN) and gradient point clustering localization (GPCL) achieved clutter suppression through a gradient point clustering algorithm; a sparse recovery space-time adaptive processing (SR-STAP) method exploits the sparsity of the clutter spectrum to improve the clutter suppression performance of radar systems; and a non-uniform detector based on a memory-enhanced autoencoder uses a memory-enhanced autoencoder to enhance clutter suppression capabilities. These methods have improved clutter suppression to a certain extent, but some key issues remain unresolved.

[0004] First, clutter suppression in complex environments remains a challenge. Modern radar systems face multipath effects, non-uniform clutter, and strong interference in complex environments, making traditional clutter suppression methods ineffective. Second, the detection and differentiation of slow-moving targets remains challenging. Slow-moving targets have small Doppler shifts and are easily overwhelmed by background clutter, leading to detection failure. Therefore, how to classify targets and clutter at the source has become a key research point in clutter suppression for modern radar signals.

[0005] It is worth noting that bats in nature can still successfully forage even in environments with strong background clutter, such as the Figure 1 As shown in the figure, this is attributed to bat sonar's ability to transmit multi-band signals. The core of bat sonar is spatial frequency diversity. By transmitting and receiving multi-band signals, bats are able to effectively suppress clutter and extract target signals in complex environments. The bat foraging process is very similar to radar's indoor target detection: both require effective clutter suppression and target signal extraction in complex environments. Therefore, combining through-wall radar with spatial frequency diversity offers a promising approach to complex environment clutter suppression.

[0006] Based on this principle, this paper proposes a clutter suppression method for through-wall radar based on spatial frequency diversity. Through multi-band signal waveform optimization, antenna beamforming optimization, and multi-band array signal processing technology, this method can effectively distinguish target signals from background clutter, significantly improving clutter suppression performance in complex environments and enhancing the ability to detect slow-moving targets. Summary of the Invention

[0007] The present invention proposes a through-wall radar clutter suppression method based on spatial frequency diversity, which aims to solve the problems of strong background clutter interference and difficulty in detecting low-speed moving targets in complex environments. The core of the present invention is to achieve effective distinction between target signals and background clutter through multi-band signal waveform design, antenna beamforming optimization and multi-band array signal processing technology. Sufficient simulation experiments have confirmed the effectiveness of the simulation experiments. Specifically, the multi-band signal waveform design adopts spatial frequency diversity technology, by transmitting multiple radar signals of different frequencies, and utilizing the differences in propagation characteristics of different frequency signals in complex environments to enhance the distinguishability of target echo signals; the antenna beamforming optimization is based on the principle of spatial diversity, by adjusting the radiation pattern of the antenna array, suppressing clutter interference from non-target directions, and at the same time enhancing the signal reception capability in the target direction; the multi-band array signal processing technology combines frequency domain and spatial domain information, and by jointly processing multi-band received signals, achieves deep suppression of background clutter and retains the weak signal characteristics of low-speed moving targets.

[0008] The indoor personnel detection scene modeling of the present invention is shown in the attached Figure 2 As shown, the process mainly includes steps 200 to 220, wherein steps 200 to 210 are radar signal beam design steps, and step 220 is a power ratio calculation step.

[0009] Step 200 is to solve the following problem P1 (Formula (1)), complete the multi-band radar signal waveform optimization design, and generate a signal waveform with low side lobes and good Doppler tolerance.

[0010] Bat sonar simultaneously transmits multiple frequency signals with varying frequency differences, forming a unique multi-band nonlinear frequency modulation signal over time. This signal design leverages the principle of frequency diversity, enhancing the robustness and environmental adaptability of target detection through the synergistic effect of multiple frequency signals. By analyzing the power ratio of echo signals from different frequency bands, bats can accurately determine the spatial location of echo signals, effectively suppressing various background clutter. This frequency diversity-based signal processing mechanism enables bats to successfully forage in complex, clutter-filled environments, such as forests, demonstrating their exceptional target detection capabilities in natural environments.

[0011] Frequency diversity technology is an advanced radar signal processing method that uses multiple sub-pulses of different frequencies in the radar's transmitted signal to detect and identify targets. This method exploits the differences in spatial propagation characteristics of signals of different frequencies, such as frequency-dependent attenuation, multipath effects, and scattering. It effectively distinguishes targets from clutter, improving radar system detection performance in complex environments. At the receiver, by analyzing the power ratio of echo signals at different frequencies, frequency diversity technology can identify the spatial location of the target signal, thereby suppressing background clutter and enhancing detection of actual targets.

[0012] Inspired by the working principle of bat sonar and frequency diversity technology, this project intends to design an instantaneous multi-frequency radar signal (similar to the attached Figure 3 The instantaneous multi-frequency radar signal is composed of multiple sub-pulses with a certain frequency difference. Its frequency distribution and modulation characteristics are optimized and designed to cover a wide frequency range at the moment of transmission. By analyzing the power ratio of echo signals in different frequency bands, this method can accurately identify the spatial position of the target signal and effectively distinguish the target signal from background clutter. Similar to the design of traditional radar signals, the instantaneous multi-frequency signal also needs to meet the requirements of low sidelobes, good Doppler tolerance, etc., in order to reduce mutual interference between signals and adapt to the frequency offset caused by target motion. The specific design can be achieved by solving the following optimization problems:

[0013]

[0014] In the above formula, AF(·) is the ambiguity function, PSL(·) represents the peak-to-sidelobe ratio, d(·) is the signal duration, BW(·) is the bandwidth, TF(·) is the time-frequency spectrum, and δ(·) is the Dirac function. Considering constraints such as system bandwidth and signal duration, the ambiguity function (AF) is optimized by maximizing the peak-to-sidelobe ratio (PSL), thereby improving signal resolution. Solving this problem using an optimization algorithm can yield a multiband signal that meets the requirements.

[0015] Step 210 is to solve the following problem P2 (Formula (2)), perform antenna beamforming optimization design, and optimize the beam array according to the antenna radiation pattern of different frequency signals by combining antenna beamforming technology.

[0016] Antenna beamforming is a key technology in radar systems. Its core goal is to achieve accurate target signal detection and effective background clutter suppression by optimizing the antenna array's radiation pattern. In multi-band signal processing, antenna beamforming must simultaneously consider the propagation characteristics of signals at different frequencies to ensure robustness and stability in complex environments.

[0017] The transmission and reception of the multi-band signal in step 200 requires a corresponding radar antenna array to be implemented. For different frequency components of the multi-band signal, the antenna beam main lobe width, side lobe amplitude (such as the attached Figure 4 As shown in the figure), dynamic beam scanning, etc. all put forward different requirements for antenna array beamforming. In multi-band signal processing, antenna beamforming needs to meet the following requirements:

[0018] 1. Main lobe width optimization: By optimizing the main lobe width, the beam resolution capability is improved and the accuracy of target detection is enhanced.

[0019] 2. Sidelobe suppression: By reducing the sidelobe amplitude, the mutual interference between signals is reduced and the signal-to-noise ratio of the target signal is improved.

[0020] 3. Dynamic scanning capability: By optimizing the dynamic scanning performance of the beam, the coverage of target detection is expanded and the target motion characteristics in complex environments are adapted.

[0021] In this project, antenna beamforming needs to consider multi-band signals simultaneously, which can be abstractly described as the following mathematical model:

[0022]

[0023] In the above formula, A(θ; f i ,θ c , x) is the center frequency f i , the maximum beam gain points to θ c Antenna pattern, BM 3dB(·) is the 3dB beamwidth function, SL m (·) is the maximum sidelobe value. i This is a hyperparameter used to compromise the mainlobe width of antenna patterns in different frequency bands. K is the number of frequency bands, typically K = 2. This model solves the weighted vector for antenna beamforming by optimizing the mainlobe width and sidelobe amplitude of the antenna pattern. The specific steps are as follows:

[0024] 1. Objective function: Minimize the error between the antenna pattern and the weighting vector to ensure optimal beamforming performance.

[0025] 2. Constraints: Limit the sidelobe amplitude of any frequency signal to less than a preset value to ensure the sidelobe suppression effect.

[0026] 3. Solution: Use a numerical optimization algorithm (such as gradient descent or genetic algorithm) to solve the weight vector x. Solving the above optimization problem yields the weight vector for antenna beamforming.

[0027] Step 220, multi-band array signal processing clutter suppression, based on bat-inspired multi-band radar array signal processing, effectively distinguishes target signals from background clutter by comparing the echo power ratios of different frequency signals in different directions.

[0028] Strong background clutter caused by walls, floors, etc. seriously affects the target detection and imaging quality of through-wall radar. Effective background clutter suppression can significantly improve the target detection performance of through-wall radar. Distributed background clutter that is in the same range gate as the tangentially moving target cannot be effectively suppressed by traditional methods such as Doppler filtering and beamforming spatial nulling. Doppler filtering technology relies on the difference in velocity between the target and the background clutter, while beamforming spatial nulling technology relies on the difference in spatial position between the target and the background clutter. However, these traditional methods have limited effectiveness for distributed background clutter that is in the same range gate as the tangentially moving target. The present invention can effectively overcome this limitation by analyzing the echo power ratio of multi-band signals. Since signals of different frequencies have different propagation characteristics in space, the components of their echo signals in different directions will show significant differences. By comparing these differences, the system can accurately identify and suppress background clutter, thereby improving the accuracy of target detection and imaging quality.

[0029] Bats detect their environment by emitting dual-band signals. As these signals propagate through space, they are affected by the radiation pattern, resulting in different gains for echo signal components reflected from different directions. Specifically, the power of each frequency component in the echo signal received by bats varies significantly depending on the direction of propagation. By calculating the power ratio of the two-band echo signals and combining it with a preset threshold, bats can effectively distinguish between the different directions of echo signals, thereby achieving clutter suppression. For example, when the power ratio of the echo signal in a certain direction exceeds the threshold, the bat can determine that the signal comes from the target direction, while signals with a power ratio below the threshold are identified as clutter and suppressed. This method not only improves the accuracy of target detection, but also significantly enhances the bat's perception ability in complex environments, providing reliable technical support for its efficient navigation and hunting.

[0030] Inspired by this method, this project intends to implement spatial frequency diversity based on the bat sonar principle, using the multi-band signal waveform and beamforming method obtained by the optimization design of steps 200-210 to achieve background clutter suppression for through-wall radar. Figure 5 shown.

[0031] First, according to the basic radar equation, the echo power ratio of a single target can be expressed as

[0032]

[0033] Among them, P t Represents the transmit signal power, G t Represents the transmit antenna gain, G r represents the receiving antenna gain, represents the wavelength of the transmitted signal, σ is the radar cross-section (RCS), and R is the distance between the target and the radar. Considering that RCS is related to the relative dielectric constant, frequency, and incident angle θ, and assuming that the transmit antenna and receive antenna have equal gains, the above formula can be further expressed as

[0034]

[0035] According to the principle of bat bionics, the echo power ratio of a single target is calculated and can be expressed as

[0036]

[0037] Taking into account

[0038]

[0039] The echo power ratio can finally be simplified to

[0040]

[0041] By comparing the above ratio with a certain threshold, different echoes are determined to be target signals or background clutter, ultimately achieving effective distinction between clutter and targets.

[0042] Beneficial effects

[0043] The present invention's through-wall radar clutter suppression method based on spatial frequency diversity innovatively addresses the ineffectiveness of current radar clutter suppression methods in complex environments and low-speed target discrimination. It successfully applies the target discrimination method used by bats in nature to through-wall radar clutter suppression. Specifically, through multi-band radar signal waveform optimization, a signal waveform with low sidelobes and good Doppler tolerance is generated. Antenna beamforming optimization is used to optimize the beam array. Furthermore, bat-inspired multi-band radar array signal processing effectively distinguishes target signals from background clutter by comparing the echo power ratios of signals of different frequencies in different directions.

[0044] Furthermore, we constructed detailed simulation scenarios based on real-world indoor scene parameters and conducted detailed simulation and comparison experiments. We conducted simulation experiments by varying the number of human targets and the relative positions of human targets and clutter. The simulation results showed that echo ratio differentiation using spatial frequency diversity was applicable in all simulation scenarios, validating the superiority of the proposed method for indoor human target differentiation. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] To clearly explain the technical steps of the present invention, all the drawings used in the description of the present invention are briefly described below. It should be noted that the drawings described below are only some examples of the implementation of the present invention, and other persons skilled in the art can still obtain other drawings in different scenarios based on these drawings.

[0046] Attachment Figure 1 This is the bat bionics scene that this invention draws on;

[0047] Attachment Figure 2 It is an indoor radar detection scene modeling of the present invention;

[0048] Attachment Figure 3 The present invention draws on the instantaneous multi-frequency signal of bats;

[0049] Attachment Figure 4 is the antenna pattern of the radar antenna in the present invention;

[0050] Attachment Figure 5 Schematic diagram of the present invention imitating bats to suppress indoor clutter;

[0051] Attachment Figure 6 -Attached Figure 9 These are four examples of indoor simulation scenarios of the present invention;

[0052] Attachment Figure 10 -Attached Figure 13 These are the simulation results of four indoor scenes of the present invention. DETAILED DESCRIPTION

[0053] The following is a complete and clear description of the steps and processes of the present invention in conjunction with the drawings in this application. It is obvious that the example described in this application is only an example application scenario of the present invention, and other results based on the content of the present invention without making substantial changes all fall within the scope of protection of the present invention.

[0054] Attachment Figure 6-9 (The four simulation scenes in the paper) are the four typical scenes of indoor human target detection of the present invention, Figure 10-13 The simulation results for each scenario are shown. The only difference between the scenarios is the number of human targets and the relative positions of the targets and clutter. The human body in the scenario is the target to be detected, and the table and chairs are stationary clutter. The simulation tool used is gprMax.

[0055] Radar target detection can be divided into two categories: single target detection and multi-target detection. The indoor single target simulation scene is attached. Figure 6-8 , the indoor multi-target simulation scene is attached Figure 9 We will attach Figure 2 The general scenario modeling is described to illustrate the specific implementation of the present invention. According to the relevant references, the simulation parameters are set as follows:

[0056] 1. Simulation frequency band: 300MHz and 2400MHz

[0057] 2. Relative dielectric constant of the human body: 49.8 (300MHz), 38.1 (2400MHz)

[0058] 3. Human body conductivity: 0.641 (300MHz), 1.44 (2400MHz)

[0059] 4. Relative dielectric constant of table and chair clutter: 4.5 (300MHz), 3.5 (2400MHz)

[0060] 5. Conductivity of table and chair clutter: 4.58×10 -3 (300MHz), 2.29×10 -2 (2400MHz)

[0061] The specific steps of the indoor clutter suppression method of through-wall radar based on spatial frequency diversity are as follows:

[0062] Step 300: Write a .in file and set the simulation domain, grid resolution, and time window.

[0063] First, use the #domain keyword to define the size of the simulation domain, including the dimensions in the x, y, and z directions, in meters (m). The size of the simulation domain should be set according to the actual scenario requirements to ensure that it can fully cover the target area.

[0064] Next, use the #dx_dy_dz keyword to define the simulation grid resolution, which is the grid spacing in the x, y, and z directions, in meters (m). The grid resolution directly affects the accuracy and computational effort of the simulation and should be set appropriately based on the simulation requirements.

[0065] Finally, use the #time_window keyword to define the simulation time window in seconds. The time window should be set to fully capture the signal propagation process. For example, #time_window:5e-8 indicates a time window of 50 nanoseconds (ns).

[0066] Through the above steps, the simulation domain, grid resolution, and time window settings can be completed, providing basic parameter support for subsequent simulation calculations.

[0067] Step 310: Define material properties.

[0068] Use the #material keyword to define the material properties used in the simulation, mainly including relative dielectric constant and conductivity. Accurate setting of material properties is the key to the reliability of simulation results. The core of this invention is to use multi-frequency signals for target differentiation and clutter suppression, and the key to differentiation lies in the difference in electromagnetic properties between different targets. By accurately setting the relative dielectric constant and conductivity of the material, it is possible to simulate the reflection, transmission and absorption characteristics of electromagnetic waves of different materials in real scenes, thereby providing a reliable data basis for subsequent target differentiation and clutter suppression.

[0069] Step 320: Define the excitation source waveform and perform waveform optimization.

[0070] Use the #waveform keyword to define the excitation source waveform type and parameters. The excitation source waveform selection directly affects the characteristics of the simulation results. During waveform optimization, waveform parameters should be adjusted based on the specific application scenario and target characteristics to maximize signal effectiveness and detection accuracy. This optimized waveform can significantly improve target detection and clutter suppression capabilities of through-wall radar in complex environments.

[0071] When defining the waveform, it is important to ensure that the waveform signal satisfies the constraints of formulas (1) and (2). Problems P1 and P2 are solved using an optimization algorithm, which ensures low sidelobes and good Doppler tolerance. Low sidelobes can effectively reduce signal interference in non-target areas, while good Doppler tolerance ensures that the signal maintains stable detection performance even when the target is moving.

[0072] Step 330: Set the excitation source position and the receiver position.

[0073] Use the #hertzian_dipole or #hortzialpipe keyword to define the location and direction of the stimulus source. The location of the stimulus source should be consistent with the location of the actual detection device to ensure the authenticity of the simulation scenario and the reliability of the simulation results.

[0074] The #rx keyword also defines the receiver's location. The receiver's location should match the actual detection equipment. By properly positioning the source and receiver, we can accurately simulate the signal transmission and reception process in a real-world detection scenario, providing reliable data support for subsequent signal processing and target detection.

[0075] Step 340: Define geometric shapes and generate geometric views.

[0076] Use #cylinder or other geometric keywords to define objects in the simulation, such as walls, furniture, or other obstacles. Accurate modeling of geometry is key to reliable simulation results, so it is important to set it precisely based on the shape and size of objects in the real scene.

[0077] After defining the geometry, use the #geometry_view or #geometry_write keywords to generate a geometry view file for the simulation domain. This file can be used to visualize the simulation model and verify its accuracy. Generating a geometry view file allows users to visually inspect the model's geometry, ensuring consistency with the real-world scenario, thereby improving the reliability of simulation results.

[0078] Step 350: Run forward modeling to generate A-scan or B-scan images.

[0079] After the user defines the simulation scenario (underground structure, material properties, etc.) and radar parameters (antenna frequency, sampling rate, etc.), gprMax simulates the propagation of electromagnetic waves in the scenario using the FDTD method, records the electromagnetic field strength of the receiving antenna at different time steps, and generates an A-scan (the time domain signal of a single transmission and reception); by moving the antenna along the survey line and repeating the simulation, multiple A-scans are arranged in spatial order to generate a B-scan (a two-dimensional radar profile).

[0080] The following explains some of the result graphs obtained in each simulation scenario.

[0081] In the attached Figure 6 In the simulation scenario shown, we assume that the human target and the table and chair clutter are on the same line, and the human target is in front of the table and chair clutter. Figure 10 As shown), we find that the human body echo power ratio is The echo power ratio of the table and chair clutter is Therefore, the echo power ratio can be used to distinguish between human targets and table and chair clutter.

[0082] Similarly, Figure 7 and attached Figure 6 The only difference is the front-to-back position relationship between the human target and the table and chair clutter. Figure 11 ), the echo power ratio of the human body is 4.51, and the echo power ratio of the table and chair clutter is 2.62, which can also distinguish the two.

[0083] Attachment Figure 8 The description is about the scene where the human target and the table and chair clutter are located on both sides of the radar. Figure 12 ), the echo power ratio of the human body is 4.55, and the echo power ratio of the table and chair clutter is 2.74, which can also distinguish the two.

[0084] Attachment Figure 9 It describes the scene of multiple human target detection. Figure 13 ), the echo power ratios of the two human bodies are 4.57 and 4.46 respectively, and the echo power ratio of the table and chair clutter is 2.54. The two can also be distinguished.

[0085] In summary, the simulation experiments in various scenarios have fully demonstrated that the proposed method can effectively distinguish targets from clutter in single-target and multi-target situations, regardless of the relative positions of the target to be detected and the clutter.

Claims

1. A method for suppressing clutter in a through-wall radar based on spatial frequency diversity, characterized in that: include: The multi-band radar signal waveform design is completed by solving the optimization problem P1. The optimization problem P1 is shown in the specification formula (1). Its objective function is to minimize the sidelobe level of the signal autocorrelation function. The constraints include signal energy constraint and Doppler frequency shift tolerance constraint, and generate an instantaneous multi-frequency signal with low sidelobe characteristics and Doppler tolerance. The antenna beamforming optimization design is performed by solving the optimization problem P2. The optimization problem P2 is shown in the specification formula (2). Its objective function is to jointly optimize the mainlobe width and sidelobe level of the beam pattern in multiple frequency bands. Clutter suppression is performed based on multi-band array signal processing, by calculating the power ratio P of echo signals at different frequencies. ratio (ε r ) to distinguish targets from clutter, the power ratio is shown in formula (7), where f1 and f2 are the center frequencies of different frequency bands, σ is the radar cross-section, and ε r is the relative dielectric constant, θ is the incident angle, and the calculation of the power ratio includes performing frequency domain transformation, sub-band separation, and energy integration processing on the received signal.

2. The method according to claim 1, characterized in that The instantaneous multi-frequency signal in step 200 is composed of K sub-pulses s with frequency differences k (t), the time-frequency spectrum of each sub-pulse satisfies the orthogonality condition δ(S tf =TF(s k ))δ(S tf =TF(s l ))=δ kl TF(·) represents the time-frequency transform operator, δ kl It is a Kronecker delta function, and the center frequency interval of each sub-pulse is greater than 1.5 times the signal bandwidth. The sub-pulse duration is adjustable in the range of 10ns to 100ns.

3. The method according to claim 1, characterized in that The antenna beam forming in step 210 must simultaneously meet the main lobe width optimization condition and the side lobe suppression condition. The main lobe width optimization condition is: Where A(θ;f i ,θ c , x) is the center frequency f i , the maximum beam gain points to θ c Antenna pattern, BM 3dB (·) is the 3dB beamwidth function, SL m (·) is the maximum sidelobe value, λ i It is a hyperparameter used to compromise the main lobe width of antenna patterns in different frequency bands, K is the number of frequency bands; the sidelobe suppression condition is: for all frequency bands f1 to f K , satisfying S Lm (A(θ;f i ,θ c ,x) / A(θ c ;f i ,θ c , x))<Γ SL , where Γ SL The sidelobe level threshold is set to -20dB to -30dB. m (·) represents the maximum value function of the sidelobe region.

4. The method according to claim 1, wherein The specific method of distinguishing the target from the clutter in step 220 is to calculate the dual-band power ratio: Set the material discrimination threshold Γ th , when P ratio (ε r )>Γ th It is judged as the target signal when P ratio (ε r )<Γ th It is judged as background clutter.

5. The method according to any one of claims 1 to 4, characterized in that The center frequency of the multi-band signal includes a dual-band combination of 300 MHz and 2400 MHz. The bandwidth of each frequency band is 10%-20% of the center frequency, and the interval between adjacent frequency bands is not less than 500 MHz.

6. A through-wall radar system, characterized in that: include: A multi-band radar transmitting module for generating the instantaneous multi-frequency signal described in any one of claims 1-5; a reconfigurable antenna array for implementing the beamforming optimization described in claim 3; a multi-channel receiver for synchronously receiving echo signals in each frequency band; a signal processing unit for executing the clutter suppression algorithm described in claim 1; a target display terminal for visualizing processing results; the system achieves life form detection and background clutter suppression in complex building environments through spatial frequency diversity processing.

7. The through-wall radar system according to claim 6, characterized in that: The signal processing unit includes: an adaptive filtering module for suppressing static clutter; a time-frequency analysis module for extracting multi-band signal features; a power ratio calculation module for realizing the material discrimination described in claim 4; and a target tracking module for associating multi-frame detection results. Each module uses an FPGA or DSP hardware platform to achieve parallel processing.

Images