A multi-target positioning and micro-motion perception method for bistatic multi-channel passive radar

Abstract

This invention discloses a method for multi-target localization and micro-motion sensing using a bistatic multi-channel passive radar. The method includes: first, matrix reconstruction and smoothing preprocessing of the bistatic signals, followed by estimation of the number of signal sources using an improved MDL algorithm; second, addressing multipath and strong interference issues by identifying multipath sources through matrix reconstruction and employing an interference blocking algorithm to suppress strong interference and extract weak target signal angles; third, matching bistatic angles using frequency correlation criteria, combined with direction-finding cross-location and DBSCAN clustering algorithms to achieve accurate multi-target localization; and finally, inverse angle deduction and echo extraction based on the localization results, followed by phase noise elimination through conjugate processing to achieve target micro-motion feature sensing. This invention effectively improves the detection and localization accuracy and feature extraction capability of passive radar for multiple targets in complex environments, and has strong engineering practicality.

Description

Technical Field

[0001] This invention belongs to the field of radar signal processing technology, and in particular, it is a method for multi-target localization and micro-motion sensing of dual-station multi-channel passive radar. Background Technology

[0002] Passive radar, a type of radar that utilizes existing radiation sources in the environment for target detection and parameter estimation, has the core advantage of not consuming additional spectrum resources, making it suitable for electromagnetically congested environments, and possessing good stealth and environmental adaptability. Passive radar systems simultaneously receive direct wave signals and target echo signals, utilizing the correlation characteristics of the signals to achieve functions such as target presence detection and micro-motion information extraction.

[0003] In the process of multi-target localization and perception using bistatic passive radar, the complexity of the environment (such as multipath propagation and strong interference) can affect the accuracy of direction finding and localization. To effectively extract the target's angle information from the radar echo and achieve accurate localization, a complete process needs to be constructed, from angle extraction to micro-motion frequency-angle correlation and then to localization calculation: starting with independent data processing at both stations, a set of target angles is extracted. Then, the angles extracted from both stations are correlated by the consistency of the micro-motion frequency to obtain bistatic angle pairs for the same target (this process not only completes the angle pairing of multiple targets but also simultaneously acquires the micro-motion information of each target). Finally, based on the correlated angle information, localization calculation is completed through direction finding cross-referencing.

[0004] However, current radar signal processing methods still have shortcomings. First, in practical applications of passive radar, since direct waves are usually stronger than reflected waves, although traditional interference blocking algorithms have shown excellent interference suppression capabilities in weak signal direction of arrival (DOA) estimation under strong interference backgrounds, their limitations are also significant—they rely on prior information about the angle and quantity of strong interference. Second, in actual passive radar detection scenarios, the presence of reflective surfaces in the environment (such as walls and tables) causes multipath effects in electromagnetic wave propagation, which poses a significant challenge to the reliability of DOA estimation. Summary of the Invention

[0005] The purpose of this invention is to address the deficiencies or shortcomings of the existing technology by providing a method for multi-target localization and micro-motion sensing using a dual-station multi-channel passive radar.

[0006] The technical solution for achieving the objective of this invention is: a method for multi-target localization and micro-motion sensing using dual-station multi-channel passive radar, the method comprising:

[0007] Step 1: Perform signal enhancement preprocessing on the raw signals acquired independently by the two stations, and use the improved MDL algorithm to estimate the number of signal sources;

[0008] Step 2: Use matrix reconstruction technology to identify multipath signals in single-station angle measurements and extract interference angles containing radiation sources and their multipath components.

[0009] Step 3: Construct a projection matrix based on the estimated interference angle, suppress strong interference signals through interference blocking processing, and estimate the direction of arrival angle of the target reflected signal from the suppressed signal;

[0010] Step 4: Perform frequency correlation matching on the angle data measured by the two stations to determine the angle measurement pairs of the same target, and combine the direction finding cross-positioning and spatial clustering algorithms to achieve spatial positioning of multiple targets;

[0011] Step 5: Based on the target localization results, extract the target echo signal and reference signal, and eliminate the common phase term through conjugate processing to extract phase dynamic features, thereby completing the micro-motion feature perception of multiple targets.

[0012] Furthermore, the signal enhancement preprocessing in step 1 includes:

[0013] Step 1-1: Calculate the forward covariance matrix and the backward covariance matrix of the original signal;

[0014] Steps 1-2: Take the average of the forward covariance matrix and the backward covariance matrix to obtain the joint smoothed covariance matrix;

[0015] Steps 1-3 involve structural repair of the joint smoothing covariance matrix to obtain a reconstructed covariance matrix. Specifically, this includes: utilizing the Hermitian symmetry and Toplitz structural features of the covariance matrix, performing conjugate symmetric position averaging on the off-diagonal elements of the joint smoothing covariance matrix, and retaining the main diagonal elements to obtain the reconstructed covariance matrix.

[0016] Furthermore, the source quantity estimation in step 1 includes:

[0017] Steps 1-4 involve eigenvalue decomposition of the reconstructed covariance matrix to obtain a descending sequence of eigenvalues.

[0018] Steps 1-5: Calculate the ratio of the geometric mean to the arithmetic mean of the eigenvalues ​​of the noise subspace;

[0019] Steps 1-6 involve constructing an improved MDL cost function based on the ratio, and then determining the estimated number of information sources by minimizing the cost function; wherein the improved MDL cost function is expressed as:

[0020]

[0021] In the formula, for Cost function To construct the ratio, For the number of array elements, For the number of information sources, This represents the number of snapshots.

[0022] Furthermore, step 2 involves identifying the multipath signal, specifically including:

[0023] Step 2-1: Perform eigenvalue decomposition on the reconstructed covariance matrix, and determine the number of strong interference sources by checking whether the ratio of adjacent elements in the eigenvalue sequence changes abruptly.

[0024] Step 2-2: Construct the interference subspace and calculate the spatial spectrum, and detect the angles corresponding to the spectral peaks to obtain a set of strong interference angles;

[0025] Steps 2-3: Based on the difference in direction of arrival between the multipath signal and the direct signal from the radiation source on the array aperture, classify and assign the identified signals as interference sources or target multipaths.

[0026] Furthermore, step 3 specifically includes:

[0027] Step 3-1: Construct the strong interference array manifold matrix based on the estimated set of strong interference angles. ;

[0028] Step 3-2, using the identity matrix With the strong interference array manifold matrix Constructing the blocking projection matrix , represented as:

[0029]

[0030] Step 3-3: Apply the blocking projection matrix to the original signal to obtain the output signal after suppressing strong interference, as shown below:

[0031]

[0032] In the formula, The signal matrix is ​​designed to suppress strong interference.

[0033] Steps 3-4: Calculate the covariance matrix after suppressing strong interference. :

[0034]

[0035] In the formula, This refers to the number of snapshots;

[0036] Steps 3-5, for Perform eigenvalue decomposition and sort the eigenvalues ​​to extract the noise subspace. And search the spatial spectrum to estimate the direction of arrival angle of the weak target signal, expressed as:

[0037]

[0038] In the formula, For MUSIC spectrum, For array manifold guide vectors, Indicates the angle of strong interference.

[0039] Further, in step 4, frequency association matching is performed, specifically including:

[0040] Step 4-1: Demodulate the target echo signal measured by the dual stations by performing conjugate multiplication with the radiation source reference signal respectively;

[0041] Step 4-2: Perform spectrum analysis on the demodulated signal to extract the target characteristic frequency, and then establish a mapping table between the target angle and the characteristic frequency for both sides.

[0042] Step 4-3: Using frequency as the matching key, select bi-station angle combinations with frequency errors within a preset threshold range from the mapping table as valid observation angle pairs.

[0043] Further, step 4 involves direction finding and cross-location, specifically including:

[0044] Step 4-1: Obtain the coordinates of the dual-station observation points and the effective observation angle pairs, and construct a system of linear equations about the target location coordinates, expressed as:

[0045]

[0046] In the formula, The target location coordinates, and They are dual-station observation stations and coordinates and The target signal arrives at the observation station. and Obtained angle of arrival information;

[0047] Step 4-2: Obtain the initial spatial position solution of the target by solving the system of linear equations.

[0048] Furthermore, step 4 employs a spatial clustering algorithm to optimize localization, specifically including:

[0049] Step 4-3: The DBSCAN algorithm is used to define the density correlation between samples based on the neighborhood radius and the minimum number of samples in the neighborhood.

[0050] Step 4-4: Divide the calculated initial spatial location samples into core points, boundary points, or noise points;

[0051] Steps 4-5 involve iteratively traversing the unclassified core points and their densities to form clusters of samples, eliminating abnormal positioning points, and determining the final target positioning coordinates.

[0052] Furthermore, before extracting the phase dynamic features in step 5, the signal is separated in the following way:

[0053] Step 5-1: Based on the incident angle determined by the target positioning coordinates, construct the target reflected wave steering vector and the external radiation source direct wave steering vector;

[0054] Step 5-2: Use the guide vector to separate the target reflected wave signal and the direct wave signal from the original signal.

[0055] Further, step 5 extracts phase dynamic features, including:

[0056] Step 5-3: Perform orthogonal down-conversion on the direct wave from the external radiation source and the reflected wave from the target to obtain the corresponding baseband signal;

[0057] Step 5-4: Multiply the baseband signal of the direct wave from the external radiation source with the baseband signal of the reflected wave from the target by conjugate multiplication to eliminate the common phase term introduced by the frequency drift of the local oscillator.

[0058] Step 5-5: Extract the time-varying phase from the signal after conjugate multiplication. The periodic fluctuation of the time-varying phase corresponds to the micro-displacement characteristics of the target.

[0059] Compared with the prior art, the significant advantages of this invention are:

[0060] (1) By using the “improved forward and backward smoothing algorithm with matrix reconstruction enhancement” to preprocess the signal (which can adapt to the hardware channel number limit of 4-element passive radar and solve the problem of fuzzy or even failed angle estimation in the case of 4 elements of traditional algorithm), and combined with the “improved MDL source estimation algorithm” (which eliminates the rank deficiency of multipath coherence on the covariance matrix through the decoherence link, so that the algorithm can still maintain high estimation accuracy under complex coherence conditions), the number of sources can be estimated more accurately in complex environments, laying a solid foundation for subsequent angle estimation.

[0061] (2) In response to the common "multipath problem" in single-station angle measurement, this invention can use matrix reconstruction technology to identify the source of multipath and accurately extract the angle of strong interference (including direct waves from radiation sources and their multipath components), thus avoiding the misleading of target positioning by multipath signals.

[0062] (3) An "improved interference blocking algorithm" is proposed. By identifying the abrupt changes in the eigenvalues ​​of the signal covariance matrix, the number of strong interferences can be automatically estimated. This not only eliminates the dependence of the traditional interference blocking algorithm on the prior information of the number and angle of strong interferences, but also solves the problem of the fixed projection blocking algorithm's dependence on the prior information of the number and the gap in the extraction of interference angle.

[0063] (4) By matching the angles measured by dual stations with frequency consistency, valid observation angle pairs belonging to the same target are screened out, effectively eliminating false positioning points and ensuring the reliability of cross-positioning. Combined with segmented processing and DBSCAN clustering algorithm, abnormal positioning data caused by direction finding errors or failures can be automatically identified and eliminated, further optimizing the spatial positioning results of multi-target targets.

[0064] (5) The target reflected wave is accurately separated from the original signal by using the positioning coordinates to deduce the precise angle. By performing conjugate multiplication with the reference signal, the influence of local oscillator frequency drift is eliminated, thereby extracting the phase dynamic characteristics that reflect the micro-motion law of the target and realizing the perception of the target's subtle movements.

[0065] (6) This method starts with independent processing of two stations, does not require additional synchronization equipment or complex prior information, and can automatically complete interference identification, angle estimation and positioning optimization, which is more in line with the actual application needs of passive radar.

[0066] The present invention will now be described in further detail with reference to the accompanying drawings. Attached Figure Description

[0067] Figure 1 This is a flowchart of a dual-station four-channel passive radar multi-target localization and micro-motion sensing process in one embodiment.

[0068] Figure 2 This is a flowchart of an improved interference blocking algorithm in one embodiment.

[0069] Figure 3 This is a scenario diagram of dual-station localization for multiple targets in one embodiment.

[0070] Figure 4 In one embodiment, an improved jamming blocking algorithm for two radars estimates the spatial spectrum of the jamming source and target, where... Figure 4 (a) in the image is the radar spectrum of radar 1. Figure 4 (b) in the image is the radar spectral diagram 2.

[0071] Figure 5 The image shows the demodulated signal and spectrum of radar 1 at -30° and 5.6° in one embodiment. Figure 5 (a) in the diagram is the phase diagram. Figure 5 (b) in the diagram is a spectrum diagram.

[0072] Figure 6 The image shows the demodulated signal and spectrum of radar 1 at -30° and 47.6° in one embodiment. Figure 6 (a) in the diagram is the phase diagram. Figure 6 (b) in the diagram is a spectrum diagram.

[0073] Figure 7 The image shows the demodulated signals and spectrum of radar 2 at -32.1° and -5.4° in one embodiment. Figure 7 (a) in the diagram is the phase diagram. Figure 7 (b) in the diagram is a spectrum diagram.

[0074] Figure 8 The image shows the demodulated signals and spectrum of radar 2 at -32.1° and 46° in one embodiment. Figure 8 (a) in the diagram is the phase diagram. Figure 8 (b) in the diagram is a spectrum diagram.

[0075] Figure 9 This is a spatial distribution map of dual-station multi-target positioning in one embodiment. Figure 9 (a) in the image is the location map of Target 1. Figure 9 (b) is the location map of target 2. Figure 9 (c) in the diagram is the location map of the external radiation source.

[0076] Figure 10 The image shows the demodulated signal and spectrum from radar 2 at -32° and -4.5° in one embodiment. Figure 10 (a) in the diagram is the phase diagram. Figure 10 (b) in the diagram is a spectrum diagram. Detailed Implementation

[0077] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0078] It should be noted that if the embodiments of the present invention involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of the components in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.

[0079] Furthermore, if the embodiments of this invention involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0080] In one embodiment, combined Figure 1 A method for multi-target localization and micro-motion sensing using a dual-station multi-channel passive radar is provided, the method comprising:

[0081] Step 1: Perform signal enhancement preprocessing on the raw signals acquired independently by the two stations, and use the improved MDL algorithm to estimate the number of signal sources;

[0082] Step 2: Use matrix reconstruction technology to identify multipath signals in single-station angle measurements and extract interference angles containing radiation sources and their multipath components.

[0083] Step 3: Construct a projection matrix based on the estimated interference angle, suppress strong interference signals through interference blocking processing, and estimate the direction of arrival angle of the target reflected signal from the suppressed signal;

[0084] Step 4: Perform frequency correlation matching on the angle data measured by the two stations to determine the angle measurement pairs of the same target, and combine the direction finding cross-positioning and spatial clustering algorithms to achieve spatial positioning of multiple targets;

[0085] Step 5: Based on the target localization results, extract the target echo signal and reference signal, and eliminate the common phase term through conjugate processing to extract phase dynamic features, thereby completing the micro-motion feature perception of multiple targets.

[0086] Furthermore, in one embodiment, the signal enhancement preprocessing in step 1 includes:

[0087] Step 1-1: Calculate the forward covariance matrix and the backward covariance matrix of the original signal;

[0088]

[0089]

[0090] In the formula, Let be the covariance matrix of the first n terms. Let be the covariance matrix of the latter term. For the number of snapshots, It is a commutative matrix. This is a swap matrix used to reverse the order of matrix elements. ;

[0091] Steps 1-2: Average the forward and backward covariance matrices to obtain the joint smoothed covariance matrix. :

[0092]

[0093] Steps 1-3 involve structural repair of the joint smoothing covariance matrix to obtain a reconstructed covariance matrix. Specifically, this includes: utilizing the Hermitian symmetry and Toplitz structural features of the covariance matrix, performing conjugate symmetric position averaging on the off-diagonal elements of the joint smoothing covariance matrix, and retaining the main diagonal elements to obtain the reconstructed covariance matrix.

[0094] Here, the inherent structure of the array signal covariance matrix (Hermitian and Toeplitz) is used to repair the smoothed matrix.

[0095] Here, steps 1-3 specifically include: constructing a symmetric matrix. To ensure that the reconstructed matrix strictly satisfies Hermitian symmetry, for off-diagonal elements, the average value of the elements at their conjugate symmetric positions is taken.

[0096]

[0097] For diagonal elements, the extracted main diagonal element is used directly. .in, For the reconstructed matrix, the first... Line number Column elements, The covariance matrix after joint smoothing is the first... Line number Column elements, This indicates taking the complex conjugate of the matrix.

[0098] Furthermore, in one embodiment, the source quantity estimation in step 1 includes:

[0099] Steps 1-4 involve eigenvalue decomposition of the reconstructed covariance matrix to obtain a descending sequence of eigenvalues. , Let M be the Mth eigenvalue, and M be the total number of eigenvalues;

[0100] Steps 1-5: Calculate the ratio of the geometric mean to the arithmetic mean of the eigenvalues ​​of the noise subspace;

[0101] Steps 1-6 involve constructing an improved MDL cost function based on the ratio, and then determining the estimated number of information sources by minimizing the cost function; wherein the improved MDL cost function is expressed as:

[0102]

[0103] in,

[0104]

[0105] In the formula, for Cost function To construct the ratio, For the number of array elements, For the number of information sources, This is the number of snapshots. According to the AM-GM inequality... The condition is true only if all the eigenvalues ​​of the noise subspace are equal. .

[0106] Source number estimation results 'Minimize the corresponding cost function :

[0107] .

[0108] Furthermore, in one embodiment, step 2, which involves identifying the multipath signal, specifically includes:

[0109] Step 2-1: Perform eigenvalue decomposition on the reconstructed covariance matrix. Determine the number of strong interference sources by checking for abrupt changes in the ratio of adjacent elements in the eigenvalue sequence. Here, if the following conditions are met... Then determine the previous Each eigenvalue corresponds to a strong disturbance, i.e., the number of strong disturbances. ;

[0110] Step 2-2: Construct the interference subspace and calculate the spatial spectrum, and detect the angles corresponding to the spectral peaks to obtain a set of strong interference angles;

[0111] Here, in determining the number of strong interferences Then, extract the former from the eigenvector matrix. The eigenvectors constitute the interference subspace. Construct the noise subspace for interference estimation The interference MUSIC spectrum is calculated, and the angles corresponding to the spectral peaks are extracted to obtain the final result. A strong interference angle;

[0112] Here, the strong interference feature vector matrix is ​​extracted. (the first in the eigenvector matrix) (Eigenvectors) and the noise subspace of the interference estimation (Extracting the first from the eigenvector matrix) To the (Eigenvectors).

[0113] Calculate the MUSIC spectrum for interference estimation And extract the angles corresponding to the spectral peaks to finally obtain strong interference angle .

[0114]

[0115] In the formula, For the noise subspace of the interference estimation, It is an array manifold matrix.

[0116] Steps 2-3: Based on the difference in direction of arrival between the multipath signal and the direct signal from the radiation source on the array aperture, the identified signals are classified as either interference sources or target multipath sources (in this case, both the direct wave from the radiation source and its multipath components are identified as independent interference sources. Conversely, if the multipath component originates from the target signal, the algorithm will still primarily identify the radiation source as the interference source). Here, since the intensity of the direct wave from the external radiation source differs significantly from the intensity of the reflected wave from the target, the external radiation source is considered an interference source.

[0117] Furthermore, in one embodiment, combined with Figure 2 Step 3 specifically includes:

[0118] Step 3-1: Construct the strong interference array manifold matrix based on the estimated set of strong interference angles. , represented as:

[0119]

[0120] in, It is an array manifold;

[0121] Step 3-2, using the identity matrix With the strong interference array manifold matrix Constructing the blocking projection matrix , represented as:

[0122]

[0123] Step 3-3: Apply the blocking projection matrix to the original signal to obtain the output signal after suppressing strong interference, as shown below:

[0124]

[0125] In the formula, The signal matrix is ​​designed to suppress strong interference.

[0126] Steps 3-4: Calculate the covariance matrix after suppressing strong interference. :

[0127]

[0128] In the formula, This refers to the number of snapshots;

[0129] Steps 3-5, for Perform eigenvalue decomposition and sort the eigenvalues ​​to extract the noise subspace. And search the spatial spectrum to estimate the direction of arrival angle of the weak target signal, expressed as:

[0130]

[0131] In the formula, For MUSIC spectrum, For array manifold guide vectors, Indicates the angle of strong interference.

[0132] Furthermore, in one embodiment, step 4 involves frequency association matching, specifically including:

[0133] Step 4-1: Demodulate the target echo signal measured by the dual stations by performing conjugate multiplication with the radiation source reference signal respectively;

[0134] Step 4-2: Perform spectrum analysis on the demodulated signal to extract the target characteristic frequency, and then establish a mapping table between the target angle and the characteristic frequency for both sides.

[0135] Step 4-3: Using frequency as the matching key, select bi-station angle combinations with frequency errors within a preset threshold range from the mapping table as valid observation angle pairs.

[0136] Furthermore, in one embodiment, step 4 involves direction finding and cross-location, specifically including:

[0137] Step 4-1: Obtain the coordinates of the dual-station observation points and the effective observation angle pairs, and construct a system of linear equations about the target location coordinates, expressed as:

[0138]

[0139] in,

[0140]

[0141] The matrix form is as follows:

[0142]

[0143] in, , , ;

[0144] In the formula, The target location coordinates, and They are dual-station observation stations and coordinates and The target signal arrives at the observation station. and Obtained angle of arrival information;

[0145] Step 4-2: By solving the system of linear equations, the initial spatial position solution of the target is obtained:

[0146]

[0147] Furthermore, in one embodiment, step 4 employs a spatial clustering algorithm to optimize localization, specifically including:

[0148] Step 4-3: The DBSCAN algorithm is used to define the density correlation between samples based on the neighborhood radius and the minimum number of samples in the neighborhood.

[0149] Here, the data sample set is used DBSCAN is defined as follows:

[0150] (1) - Neighborhood: Based on samples Centered on, A region with radius , containing all elements in the dataset that are related to... Distance less than or equal to The sample.

[0151] (2) Core object: If a certain sample of - The number of samples in the neighborhood is greater than or equal to ,but As the core object, it is the basic unit that constitutes the cluster;

[0152] Here, according to All samples are divided into three categories: core points (i.e., core objects), boundary points (those located at the core points), and boundary points (those located at the core points). - Within the neighborhood, but not a core object itself), noise points (neither core points nor boundary points);

[0153] Step 4-4: Divide the calculated initial spatial location samples into core points, boundary points, or noise points;

[0154] Steps 4-5 involve iteratively traversing the unclassified core points and their densities to form clusters of samples, eliminating abnormal positioning points, and determining the final target positioning coordinates.

[0155] Here, the algorithm flow is as follows:

[0156] Step 4-4-1, Sample Classification. Based on... All samples are divided into three categories: core points (i.e., core objects), boundary points (those located at the core points), and boundary points (those located at the core points). - Within the neighborhood, but not a core object itself), noise points (neither core points nor boundary points);

[0157] Step 4-4-2, Cluster Generation. Starting from any unclassified core point, traverse all samples (including core points and boundary points) whose density is reachable to form a cluster;

[0158] Step 4-4-3, Iterative Clustering. Repeat the above steps, select the next unassigned cluster core point, and continue to expand its density-reachable sample set until all core points are assigned to their corresponding clusters, at which point the algorithm terminates.

[0159] Furthermore, in one embodiment, before extracting the phase dynamic features in step 5, the signal is separated in the following manner:

[0160] Step 5-1: Based on the incident angle determined by the target positioning coordinates, construct the target reflected wave steering vector and the external radiation source direct wave steering vector;

[0161] Step 5-2: Use the guide vector to separate the target reflected wave signal and the direct wave signal from the original signal.

[0162] Further, step 5 extracts phase dynamic features, including:

[0163] Step 5-3: Perform orthogonal down-conversion on the direct wave from the external radiation source and the reflected wave from the target to obtain the corresponding baseband signal;

[0164] Step 5-4: Multiply the baseband signal of the direct wave from the external radiation source with the baseband signal of the reflected wave from the target by conjugate multiplication to eliminate the common phase term introduced by the frequency drift of the local oscillator.

[0165] Step 5-5: Extract the time-varying phase from the signal after conjugate multiplication. The periodic fluctuation of the time-varying phase corresponds to the micro-displacement characteristics of the target.

[0166] In detail, step 5 is as follows:

[0167] Step 5-1: Based on the angle information calculated in Step 4, establish the steering vectors of the target reflected wave and the direct wave from the external radiation source. Define the signal incident angle. Let be the angle between the signal propagation direction and the array axis. For a 4-element uniform linear array (element spacing is ), The carrier wavelength is ), No. Phase difference between each array element and the reference array element (the first element) for:

[0168]

[0169] When the target reflects the wave from the angle At incident, the phase difference between each array element is as follows: the first array element (reference array element) The second array element The third array element The 4th array element .

[0170] Therefore, the steering vector of the target reflected wave for:

[0171]

[0172] Direct waves from external radiation sources from angle Similarly, its guiding vector can be obtained at the incident moment. for:

[0173]

[0174] Assume the original received signal of the array is Use step 4 to determine and Then, the corresponding steering vector is used to achieve precise separation of the target reflected wave and the direct wave.

[0175] Target reflected wave for:

[0176]

[0177] direct wave for:

[0178]

[0179] Step 5-2: Demodulation of target micro-motion information. Single-frequency continuous wave signal emitted by the external radiation source. It can be represented as:

[0180]

[0181] in, It is the carrier frequency.

[0182] The signal received by passive radar is the superposition of signals from two paths, the direct wave signal. It can be represented as:

[0183]

[0184] in, This refers to the fixed distance from the external radiation source to the radar. This is the path attenuation coefficient.

[0185] The target reflected wave signal can be represented as:

[0186]

[0187] in, The distance from the external radiation source to the target is time-varying. The time-varying distance from the target to the radar. The path attenuation coefficient is the emission path attenuation coefficient.

[0188] The receiver uses a local oscillator (LO) to perform quadrature downconversion on the received radio frequency signal, converting it to baseband for processing.

[0189] Baseband signal after direct wave downconversion for:

[0190]

[0191] in, The LO signal generated by the local oscillator has a frequency deviation. The low-pass filter (LPF) filters out the sum frequency component and retains the difference frequency component.

[0192] Baseband signal after down-conversion of the target reflected wave for:

[0193]

[0194] The baseband signal of the direct wave (reference signal) is conjugately multiplied with the baseband signal of the target reflected wave signal to eliminate the local oscillator (LO) frequency drift. The introduced common phase term yields the signal after conjugate multiplication. :

[0195]

[0196] phase ,in This is the time-varying path difference caused by the slight movement of the target. Phase With the target micro-displacement They exhibit a linear correspondence, with periodic changes in displacement directly manifesting as periodic fluctuations in phase.

[0197] This means the phase Dynamic characteristics can directly reflect the pattern of target micro-motions; the periodic fluctuation of the phase corresponds to the periodicity of the target micro-motions. Therefore, from the signal after conjugate multiplication... The micro-motion information of the target can be extracted from the phase, enabling passive radar to perceive the micro-motion characteristics of the target.

[0198] In one embodiment, a multi-target localization and micro-motion sensing system for dual-station multi-channel passive radar is provided, the system comprising:

[0199] The first module is used to perform signal enhancement preprocessing on the raw signals independently acquired by the two stations, and to estimate the number of signal sources using the improved MDL algorithm.

[0200] The second module is used to: identify multipath signals in single-station angle measurements using matrix reconstruction technology, and extract interference angles containing radiation sources and their multipath components;

[0201] The third module is used to: construct a projection matrix based on the estimated interference angle, suppress strong interference signals through interference blocking processing, and estimate the direction of arrival angle of the target reflected signal from the suppressed signal;

[0202] The fourth module is used to: perform frequency correlation matching on the angle data measured by the two stations to determine the angle measurement pairs of the same target, and combine the direction finding cross-positioning and spatial clustering algorithms to achieve spatial positioning of multiple targets;

[0203] The fifth module is used to: extract the target echo signal and reference signal based on the target positioning result, eliminate the common phase term through conjugate processing to extract phase dynamic features, and complete the micro-motion feature perception of multiple targets.

[0204] Specific limitations regarding the multi-target localization and micro-motion sensing system for bistatic multi-channel passive radar can be found in the limitations of the multi-target localization and micro-motion sensing method for bistatic multi-channel passive radar mentioned above, and will not be repeated here. The modules in the aforementioned multi-target localization and micro-motion sensing system for bistatic multi-channel passive radar can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in hardware or independently of the processor in a computer device, or stored in software in the memory of a computer device, so that the processor can call and execute the corresponding operations of each module.

[0205] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements:

[0206] Step 1: Perform signal enhancement preprocessing on the raw signals acquired independently by the two stations, and use the improved MDL algorithm to estimate the number of signal sources;

[0207] Step 2: Use matrix reconstruction technology to identify multipath signals in single-station angle measurements and extract interference angles containing radiation sources and their multipath components.

[0208] Step 3: Construct a projection matrix based on the estimated interference angle, suppress strong interference signals through interference blocking processing, and estimate the direction of arrival angle of the target reflected signal from the suppressed signal;

[0209] Step 4: Perform frequency correlation matching on the angle data measured by the two stations to determine the angle measurement pairs of the same target, and combine the direction finding cross-positioning and spatial clustering algorithms to achieve spatial positioning of multiple targets;

[0210] Step 5: Based on the target localization results, extract the target echo signal and reference signal, and eliminate the common phase term through conjugate processing to extract phase dynamic features, thereby completing the micro-motion feature perception of multiple targets.

[0211] For specific limitations on each step, please refer to the limitations on multi-target localization and micro-motion sensing methods for bistatic multi-channel passive radar mentioned above, which will not be repeated here.

[0212] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program being implemented when executed by a processor:

[0213] Step 1: Perform signal enhancement preprocessing on the raw signals acquired independently by the two stations, and use the improved MDL algorithm to estimate the number of signal sources;

[0214] Step 2: Use matrix reconstruction technology to identify multipath signals in single-station angle measurements and extract interference angles containing radiation sources and their multipath components.

[0215] Step 3: Construct a projection matrix based on the estimated interference angle, suppress strong interference signals through interference blocking processing, and estimate the direction of arrival angle of the target reflected signal from the suppressed signal;

[0216] Step 4: Perform frequency correlation matching on the angle data measured by the two stations to determine the angle measurement pairs of the same target, and combine the direction finding cross-positioning and spatial clustering algorithms to achieve spatial positioning of multiple targets;

[0217] Step 5: Based on the target localization results, extract the target echo signal and reference signal, and eliminate the common phase term through conjugate processing to extract phase dynamic features, thereby completing the micro-motion feature perception of multiple targets.

[0218] For specific limitations on each step, please refer to the limitations on multi-target localization and micro-motion sensing methods for bistatic multi-channel passive radar mentioned above, which will not be repeated here.

[0219] As a specific example, the invention is illustrated in one embodiment.

[0220] In this embodiment, the experiment is set up as a passive localization scenario with two targets and two radars, such as... Figure 3 As shown, the specific parameters are as follows:

[0221] Target parameters: Two micro-moving targets were designed, both driven by motors to achieve low-frequency reciprocating motion, with a fixed unidirectional movement distance of 15mm per cycle. The motion frequency of target 1 is 0.24Hz, and that of target 2 is 0.16Hz. These differentiated frequency parameters provide feature basis for subsequent multi-target association.

[0222] Radar parameters: Radar 1 is deployed at (45cm, 0cm), Radar 2 is deployed at (0cm, 10cm), and the angle between Radar 2 and the y-axis is 65°. The passive radar angle is corrected to a positive angle to the left of the normal.

[0223] Angular parameters: From the perspective of radar 1, the direction of target 1 is 6°, the direction of target 2 is 46°, and the direction of the radiation source is -31°; From the perspective of radar 2, the direction of target 1 is -4°, the direction of target 2 is 47°, and the direction of the radiation source is -32°.

[0224] (2) Angle estimation

[0225] During the experiment, there was a problem of strong interference suppressing the target spectral peaks. To address this, an improved interference blocking algorithm was used to estimate the angle of the measured data. Figure 4 The spatial spectral lines of the interference source and target are estimated for the improved jamming blocking algorithm. The spectral results clearly identify the angular positions of the interference source (radiation source) and the target. Specific estimation results are as follows: Radar 1 estimates the interference source angle to be -30°, and the angles of the two targets to be 5.6° and 47.6°, respectively; Radar 2 estimates the interference source angle to be -32.1°, and the angles of the two targets to be -5.4° and 46°, respectively. Based on the comparison between the estimated angles and the actual angles, the root mean square error (RMSE) of the angle estimation is calculated to be 1.06°.

[0226] (3) Angle correlation

[0227] After each radar completes its angle estimation, the question of how the two target angles estimated by radar 1 correspond to the two target angles estimated by radar 2 needs to be addressed. This experiment uses frequency consistency to achieve cross-radar target association, that is, by demodulating the frequency of the micro-motion signal corresponding to each angle, angles with consistent frequency characteristics are identified as the same target.

[0228] The micro-motion information was demodulated for each angle estimated by radar 1 and radar 2 respectively, and the corresponding phase waveforms and spectrum diagrams were obtained. Figures 5-8 Based on demodulated spectrum analysis: the micro-motion frequency corresponding to the 5.6° angle of radar 1 is approximately 0.238Hz, and the micro-motion frequency corresponding to the -5.4° angle of radar 2 is approximately 0.247Hz, both close to the true motion frequency of target 1 (0.24Hz); the micro-motion frequency corresponding to the 47.6° angle of radar 1 is approximately 0.166Hz, and the micro-motion frequency corresponding to the 46° angle of radar 2 is approximately 0.162Hz, both close to the true motion frequency of target 2 (0.16Hz). Therefore, target angle matching is completed: target 1 corresponds to 5.6° of radar 1 and -5.4° of radar 2, and target 2 corresponds to 47.6° of radar 1 and 46° of radar 2.

[0229] (4) Positioning

[0230] To improve positioning accuracy, two data processing schemes were designed for comparison: Scheme 1 extracts the angle from the entire data packet and then directly locates the target; Scheme 2 performs angle estimation independently in 5-second time segments, and then uses DBSCAN clustering to remove outliers before positioning. Segmented processing can reduce the impact of noise accumulation over long time series, while DBSCAN clustering is used to remove outlier data caused by angle estimation failures (such as...). Figure 9 (Abnormal angles in two segments of the radar).

[0231] After processing in 5-second segments, the angle estimation results for each segment of the dual radar are shown in Table 1 below. It should be noted that the second set of data from radar 2 (22.9°, only one angle was estimated) and the fifth set of data (20.6°, -75.5°) were outliers in the angle estimation and were removed by DBSCAN clustering to ensure the reliability of the positioning data.

[0232] Table 1. Estimation results of external radiation sources and target angles in multi-target localization.

[0233]

[0234] The localization results of the two schemes are summarized in Table 2 below. The spatial distribution of DBSCAN clustering localization after segmentation is shown in Table 2 below. Figure 9 As shown.

[0235] Table 2 Results of multi-target localization

[0236]

[0237] After segmentation, DBSCAN clustering localization achieved an RMSE of 2.14 cm for the radiation source location and 0.69 cm and 2.54 cm for the two target locations. In contrast, direct angle localization of the entire package achieved an RMSE of 2.28 cm for the radiation source location and 1.76 cm and 5.62 cm for the two target locations. The comparison shows that segmented DBSCAN clustering localization has a smaller error.

[0238] (5) Closed-loop verification of positioning results

[0239] To further verify the reliability of the positioning results, based on the positioning positions obtained by DBSCAN clustering after segmentation, the angles of each target and radiation source under the dual radar perspectives were derived in reverse: the reverse angles of the external radiation source of radar 1, target 1, and target 2 were -30°, 5.3°, and 47.3°, respectively; the reverse angles of the external radiation source of radar 2, target 1, and target 2 were -32°, -4.5°, and 47.5°, respectively.

[0240] Comparing the inverted angle with the true angle, the calculated RMSE is 0.78°, a reduction compared to the 1.06° error obtained by direct processing of the entire packet. Simultaneously, micro-motion demodulation verification was conducted based on the inverted angle: the micro-motion waveforms of target 1 and target 2 obtained by demodulation at -30°, 5.3°, and 47.3° from radar 1 were compared with... Figure 5 , Figure 6 The demodulation results are consistent; the signal and spectrum obtained by conjugate multiplication and demodulation of the signals separated from radar 2 at -32° (radiation source) and -4.5° (target 1) are as follows: Figure 10 As shown, compared to demodulation of the entire data packet angles (-32.1°, -5.4°), Figure 7 result, Figure 10 The waveform is closer to the true micro-motion characteristics of target 1; the signal and spectrum obtained by conjugate multiplication and demodulation of the signals separated from radar 2 at -32° (radiation source) and 47.5° (target 2) are compared with... Figure 9 The results were consistent.

[0241] In summary, compared with traditional signal source angle estimation and position calculation methods, this invention proposes an improved MDL source estimation algorithm, based on an improved forward and backward smoothing algorithm enhanced by matrix reconstruction, and an improved interference blocking algorithm to solve the problem. The calculation method is effective, feasible, reliable, and has higher accuracy.

[0242] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention without departing from its spirit and scope should be included within the protection scope of the present invention.

Claims

1. A method for multi-target localization and micro-motion sensing using a dual-station multi-channel passive radar, characterized in that, The method includes: Step 1: Perform signal enhancement preprocessing on the raw signals acquired independently by the two stations, and use the improved MDL algorithm to estimate the number of signal sources; Step 2: Use matrix reconstruction technology to identify multipath signals in single-station angle measurements and extract interference angles containing radiation sources and their multipath components. Step 3: Construct a projection matrix based on the estimated interference angle, suppress strong interference signals through interference blocking processing, and estimate the direction of arrival angle of the target reflected signal from the suppressed signal; Step 4: Perform frequency correlation matching on the angle data measured by the two stations to determine the angle measurement pairs of the same target, and combine the direction finding cross-positioning and spatial clustering algorithms to achieve spatial positioning of multiple targets; Step 5: Based on the target localization results, extract the target echo signal and reference signal, and eliminate the common phase term through conjugate processing to extract phase dynamic features, thereby completing the micro-motion feature perception of multiple targets.

2. The method for multi-target localization and micro-motion sensing for dual-station multi-channel passive radar according to claim 1, characterized in that, Step 1, signal enhancement preprocessing, includes: Step 1-1: Calculate the forward covariance matrix and the backward covariance matrix of the original signal; Steps 1-2: Take the average of the forward covariance matrix and the backward covariance matrix to obtain the joint smoothed covariance matrix; Steps 1-3 involve structural repair of the joint smoothing covariance matrix to obtain a reconstructed covariance matrix. Specifically, this includes: utilizing the Hermitian symmetry and Toplitz structural features of the covariance matrix, performing conjugate symmetric position averaging on the off-diagonal elements of the joint smoothing covariance matrix, and retaining the main diagonal elements to obtain the reconstructed covariance matrix.

3. The multi-target localization and micro-motion sensing method for dual-station multi-channel passive radar according to claim 2, characterized in that, Step 1, which involves estimating the number of information sources, includes: Steps 1-4 involve eigenvalue decomposition of the reconstructed covariance matrix to obtain a descending sequence of eigenvalues. Steps 1-5: Calculate the ratio of the geometric mean to the arithmetic mean of the eigenvalues ​​of the noise subspace; Steps 1-6 involve constructing an improved MDL cost function based on the ratio, and then determining the estimated number of information sources by minimizing the cost function; wherein the improved MDL cost function is expressed as: In the formula, for Cost function To construct the ratio, For the number of array elements, For the number of information sources, This represents the number of snapshots.

4. The method for multi-target localization and micro-motion sensing for dual-station multi-channel passive radar according to claim 2, characterized in that, Step 2 involves identifying multipath signals, specifically including: Step 2-1: Perform eigenvalue decomposition on the reconstructed covariance matrix, and determine the number of strong interference sources by checking whether the ratio of adjacent elements in the eigenvalue sequence changes abruptly. Step 2-2: Construct the interference subspace and calculate the spatial spectrum, and detect the angles corresponding to the spectral peaks to obtain a set of strong interference angles; Steps 2-3: Based on the difference in direction of arrival between the multipath signal and the direct signal from the radiation source on the array aperture, classify and assign the identified signals as interference sources or target multipaths.

5. The multi-target localization and micro-motion sensing method for dual-station multi-channel passive radar according to claim 4, characterized in that, Step 3 specifically includes: Step 3-1: Construct the strong interference array manifold matrix based on the estimated set of strong interference angles. ; Step 3-2, using the identity matrix With the strong interference array manifold matrix Constructing the blocking projection matrix , is represented as: Step 3-3: Apply the blocking projection matrix to the original signal to obtain the output signal after suppressing strong interference, as shown below: In the formula, The signal matrix is ​​designed to suppress strong interference. Steps 3-4: Calculate the covariance matrix after suppressing strong interference. : In the formula, This refers to the number of snapshots; Steps 3-5, for Perform eigenvalue decomposition and sort the eigenvalues ​​to extract the noise subspace. And search the spatial spectrum to estimate the direction of arrival angle of the weak target signal, expressed as: In the formula, For MUSIC spectrum, For array manifold guide vectors, Indicates the angle of strong interference.

6. The method for multi-target localization and micro-motion sensing for dual-station multi-channel passive radar according to claim 5, characterized in that, Step 4 involves frequency association matching, specifically including: Step 4-1: Demodulate the target echo signal measured by the dual stations by performing conjugate multiplication with the radiation source reference signal respectively; Step 4-2: Perform spectrum analysis on the demodulated signal to extract the target characteristic frequency, and then establish a mapping table between the target angle and the characteristic frequency for both sides. Step 4-3: Using frequency as the matching key, select bi-station angle combinations with frequency errors within a preset threshold range from the mapping table as valid observation angle pairs.

7. The method for multi-target localization and micro-motion sensing for dual-station multi-channel passive radar according to claim 6, characterized in that, Step 4 involves direction finding and cross-location, specifically including: Step 4-1: Obtain the coordinates of the dual-station observation points and the effective observation angle pairs, and construct a system of linear equations about the target location coordinates, expressed as: In the formula, The target location coordinates, and They are dual-station observation stations and coordinates and The target signal arrives at the observation station. and Obtained angle of arrival information; Step 4-2: Obtain the initial spatial position solution of the target by solving the system of linear equations.

8. The method for multi-target localization and micro-motion sensing for dual-station multi-channel passive radar according to claim 7, characterized in that, Step 4 employs a spatial clustering algorithm to optimize localization, specifically including: Step 4-3: The DBSCAN algorithm is used to define the density correlation between samples based on the neighborhood radius and the minimum number of samples in the neighborhood. Step 4-4: Divide the calculated initial spatial location samples into core points, boundary points, or noise points; Steps 4-5 involve iteratively traversing the unclassified core points and their densities to form clusters of samples, eliminating abnormal positioning points, and determining the final target positioning coordinates.

9. The method for multi-target localization and micro-motion sensing for dual-station multi-channel passive radar according to claim 8, characterized in that, Before extracting the phase dynamic features in step 5, the signal is separated in the following way: Step 5-1: Based on the incident angle determined by the target positioning coordinates, construct the target reflected wave steering vector and the external radiation source direct wave steering vector; Step 5-2: Use the guide vector to separate the target reflected wave signal and the direct wave signal from the original signal.

10. The method for multi-target localization and micro-motion sensing for dual-station multi-channel passive radar according to claim 9, characterized in that, Step 5 extracts phase dynamic features, including: Step 5-3: Perform orthogonal down-conversion on the direct wave from the external radiation source and the reflected wave from the target to obtain the corresponding baseband signal; Step 5-4: Multiply the baseband signal of the direct wave from the external radiation source with the baseband signal of the reflected wave from the target by conjugate multiplication to eliminate the common phase term introduced by the frequency drift of the local oscillator. Step 5-5: Extract the time-varying phase from the signal after conjugate multiplication. The periodic fluctuation of the time-varying phase corresponds to the micro-displacement characteristics of the target.

Images