A method for two-dimensional parameter joint estimation of frequency-agile radar based on spatial envelope constraint

Abstract

The application discloses a kind of based on space envelope constraint frequency agile radar two-dimensional parameter joint estimation method, belong to radar signal processing technical field.The method is first to frequency agile radar received echo digital channelization down-conversion and data extraction, and to baseband echo time delay and phase compensation;Then after the echo compensation is pulse compression and non-phase accumulation, extract target rough positioning center, and construct distance-speed two-dimensional local discrete reconstruction grid in its vicinity;Extract rough positioning center and its adjacent distance gate echo, combine adjacent distance gate fast time space envelope feature and slow time Doppler phase feature, establish multi-observation vector joint dictionary matrix;Utilize joint sparse recovery solver to realize target distance and speed two-dimensional parameter joint estimation.The application improves the two-dimensional parameter estimation accuracy and robustness under the condition that rough positioning exists deviation, enhances the false peak suppression ability, and is suitable for complex electromagnetic environment under high-resolution multi-target detection and parameter estimation.

Description

Technical Field

[0001] This invention belongs to the field of radar signal processing technology, specifically relating to a joint estimation method for two-dimensional parameters of frequency-agile radar. Background Technology

[0002] Frequency-agile radar (FAR) exhibits good anti-jamming capabilities and low probability of intercept by hopping the transmitted carrier frequency between pulses, making it highly valuable in complex electromagnetic environments. However, for coherent frequency-agile radar, the slow-time phase of the target echo no longer follows the regular variation relationship under fixed carrier frequency conditions due to the time-varying carrier frequency of each pulse. Traditional coherent accumulation methods based on Doppler filter banks are therefore difficult to apply directly. To address this issue, existing technologies typically employ sparse recovery methods such as compressed sensing, utilizing the sparsity of the target scene in the range-velocity two-dimensional plane to achieve high-resolution estimation of target range and velocity parameters.

[0003] Existing sparse recovery methods for frequency-agile radar typically first compress the single-pulse echo, then determine a coarse localization center based on the peak value of the pulse compression output envelope, and construct a local range-velocity reconstruction dictionary using this coarse localization center as a reference. Under ideal single-target conditions, the above methods can obtain good parameter estimation results; however, in dense multi-target scenarios, when the distance between multiple targets is less than or close to the radar range resolution, aliasing easily occurs between the main lobes of different target echoes, which in turn causes the pulse compression output peak value to deviate from the true target position, resulting in a shift in the coarse localization center.

[0004] When the coarse localization center shifts, sparse recovery methods based on single observation vectors are prone to dictionary mismatch. Especially under frequency-agile waveform conditions, some erroneous range cells may have high correlation with the dictionary atoms corresponding to the true range cells, leading to incorrect support selection during sparse recovery and the formation of spurious peaks in the range-velocity two-dimensional plane, reducing the accuracy of target parameter estimation. Under dense target conditions, these problems are even more pronounced, affecting the robust detection capabilities of frequency-agile radars in complex scenarios.

[0005] Furthermore, in the engineering implementation of broadband frequency-agile radar, to reduce the pressure of high-speed sampling and real-time processing, the receiver front-end typically employs a digital channelization structure to down-convert and decimate the echo signal. While the digital channelization process effectively reduces the back-end processing burden, it also introduces time delays and phase errors related to the filter structure and decimation process. If these are not compensated for in subsequent pulse compression and sparse recovery processing, they may further disrupt the inter-pulse phase consistency, affecting the coarse positioning results and the performance of subsequent joint estimation of range and velocity parameters.

[0006] Therefore, to address the problems of coarse positioning center shift caused by main lobe aliasing in dense target scenarios for frequency-agile radar, the spurious peaks easily generated by traditional single-observation vector sparse recovery methods, and the impact of time delay and phase errors introduced by digital channelization processing on the accuracy of range-velocity two-dimensional parameter estimation, it is urgent to propose a joint estimation method for range-velocity two-dimensional parameters of frequency-agile radar that balances engineering feasibility and anti-multi-target aliasing capabilities. This method should, while retaining the digital channelization receiving and processing flow, combine adjacent range gate spatial envelope information and slow-time Doppler phase information to construct a joint dictionary matrix of multi-observation vectors, and improve the accuracy of target parameter estimation and spurious peak suppression under conditions of coarse positioning bias through joint sparse recovery. Summary of the Invention

[0007] This invention addresses the problems of existing two-dimensional range-velocity parameter estimation methods for frequency-agile radar, such as sensitivity to coarse positioning center offset in dense target scenarios, susceptibility to dictionary mismatch and false peaks due to main lobe aliasing, and insufficient consideration of time delay and phase errors introduced by digital channelized receiver structures in existing ideal reconstruction models. To address these issues, this invention proposes a joint estimation method for two-dimensional range-velocity parameters of frequency-agile radar based on spatial envelope constraints.

[0008] The technical problem to be solved by this invention is: under the condition of digital channelization reception of broadband frequency agile radar, in order to address the problems of coarse positioning center drift caused by dense targets, the easy occurrence of incorrect support selection and false peaks in traditional single observation vector sparse recovery, and the impact of time delay and phase error introduced by digital channelization processing on the accuracy of subsequent parameter estimation, this invention provides a processing method that can take into account engineering implementation conditions and improve the accuracy and robustness of joint estimation of range-velocity two-dimensional parameters.

[0009] To solve the above-mentioned technical problems, this invention employs a joint estimation method for two-dimensional parameters of frequency-agile radar based on spatial envelope constraints, characterized by the following steps:

[0010] A joint estimation method for two-dimensional parameters of frequency-agile radar based on spatial envelope constraints, characterized by the following steps:

[0011] Step 1: Acquire the raw echo signal output by the frequency-agile radar receiver, perform digital channelization down-conversion and data extraction processing on the raw echo signal to obtain multi-pulse baseband echo data; at the same time, acquire the frequency hopping point information and channelization extraction factor corresponding to each pulse.

[0012] Step 2: Based on the channelization decimation factor, system sampling rate, and frequency hopping information corresponding to each pulse, calculate the time delay error and phase error introduced during digital channelization, construct a compensation factor, and perform compensation processing on the multi-pulse baseband echo data;

[0013] Step 3: Perform pulse compression on the compensated multi-pulse baseband echo data and non-coherent accumulation on the pulse compression output. Determine the coarse target localization center using constant false alarm rate (CFAR) detection or peak extraction. Using the coarse localization center as a reference, construct a local discrete reconstruction grid in the range-velocity two-dimensional plane. The radar range resolution is [missing information]. , At the speed of light, The radar signal bandwidth; the local discrete reconstruction grid includes: the range search grid set is The speed search grid set is The total number of reconstructed grids is ;

[0014] Step 4: Let the distance gate index corresponding to the coarse positioning center be... The echo data of the coarse positioning center and its multiple adjacent range gates are extracted to form multi-observation vector observation data. :

[0015] ;

[0016] in, Indicates the pulse compression after the first pulse. Fast-time echo data corresponding to each distance gate , This indicates the number of adjacent distance gates selected on one side based on the coarse positioning center;

[0017] Constructing the slow-time Doppler phase vector :

[0018] ;

[0019] in, Indicates the initial carrier frequency of the radar. Indicates the first The frequency point corresponding to each pulse. Indicates the number of pulses. This represents the distance index in the local distance reconstruction mesh. This represents the velocity index in the local velocity reconstruction mesh. The pulse repetition period is . The two-way propagation delay of each pulse is . :

[0020] ;

[0021] in, In the local distance reconstruction mesh, the first... The distance value corresponding to each candidate distance unit In the local velocity reconstruction mesh, the first... The velocity values ​​corresponding to each candidate velocity unit. The pulse repetition period, Represents the speed of light;

[0022] Simultaneously construct a fast time-space envelope vector spanning multiple distance gates. :

[0023] ;

[0024] in, Indicates the bandwidth of the radar transmitted signal. This represents the two-way propagation delay corresponding to the candidate distance. This indicates the reference fast-time sampling moment corresponding to the distance gate of the coarse positioning center;

[0025] A joint atom is constructed from the fast time-space envelope vector and the slow time-Doppler phase vector. :

[0026] ;

[0027] in, For Kronecker product;

[0028] Traverse all candidate range and velocity cells to form a joint dictionary matrix of multi-observation vectors. ;

[0029] ;

[0030] Step 5: Input the multi-observation vector observation data and the multi-observation vector joint dictionary matrix into the joint sparse recovery solver for solving. Determine the target support set based on the correlation between candidate atoms and observation data, and output the estimated results of the target's distance and velocity parameters.

[0031] Furthermore, in step 1, the first The original echo signal of each pulse is :

[0032] ;

[0033] in, Indicates the amplitude of the target echo. Indicates the fast sampling time. Indicates additive noise. Indicates the first The time delay corresponding to each pulse Indicates the pulse repetition period. Indicates the linear frequency modulation slope. Indicates the first The frequency point corresponding to each pulse. This represents the distance to the target at the initial moment. This indicates the radial velocity of the target;

[0034] It is a rectangular window function: ;

[0035] Highest sampling rate of radar system Channelization decimation factor .

[0036] Furthermore, the specific method for step 2 is as follows:

[0037] Step 2.1: Decimate the original echo signal obtained from high-speed sampling by a certain factor. Divided into A parallel, low-speed data branch;

[0038] Step 2.2: Extract the preset low-pass filter coefficients at equal intervals, corresponding to the following divisions: The low-speed data branch is then converted into a sub-filter, and discrete convolution calculations are performed on the corresponding sub-filters.

[0039] Step 2.3: Perform phase rotation on the convolution output of each branch and execute... Point-based inverse fast Fourier transform enables full-band channelization decomposition of broadband signals;

[0040] Step 2.4: Based on the current radar signal issued by the radar system... The frequency hopping point of each pulse, output from step 2.3 In each channel, target channel data containing target echo energy is dynamically extracted to obtain a downsampling rate of... The baseband echo data sequence;

[0041] Step 2.5: In the above polyphase filtering digital channelization process, the hardware decimation structure will cause the baseband envelope to have an inherent low-level sampling time offset in the fast time dimension. In the slow time dimension, it exhibits a carrier phase distortion term that varies with the frequency hopping point. :

[0042] ;

[0043] Using frequency hopping sequences With the extracted inherent time offset Constructing a high-precision pulse-by-pulse inverse physical compensation factor in the digital domain :

[0044]

[0045] Step 2.6: Convert the digitally channelized output to the first... A sequence of pulsed baseband echoes, multiplied by the corresponding inverse compensation factor. Precise phase calibration is performed to obtain the baseband echo data matrix. .

[0046] Furthermore, the multiple distance gates mentioned in step 4 include the distance gate where the coarse localization center is located and the adjacent distance gates on both sides; by introducing fast temporal spatial envelope constraints in the multiple distance gate dimensions, and retaining the relative amplitude envelope ratio characteristics on multiple distance gates during the joint atom normalization process, the correlation between the joint atom corresponding to the real target position and the multiple observation vectors is satisfied;

[0047] ;

[0048] in, This represents the normalized joint atom corresponding to the actual target position. Normalized joint atoms corresponding to spurious distance positions are represented, thereby suppressing erroneous support selection and spurious peaks caused by main lobe aliasing, coarse localization center shift, and high dictionary coherence.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] First, this invention addresses the problem that dense targets located within radar range resolution cells are prone to main lobe aliasing and coarse positioning center shift. By constructing a multi-observation vector sparse recovery model based on joint observation of adjacent range gates, the dependence of range-velocity two-dimensional parameter estimation on a single coarse positioning center is reduced, which is beneficial to improving the accuracy and robustness of target parameter estimation under the condition of deviation in coarse positioning.

[0051] Second, this invention introduces fast temporal spatial envelope information across multiple adjacent range gates into the multi-observation vector joint dictionary matrix construction process, so that the joint sparse recovery process not only utilizes slow-time Doppler phase features, but also utilizes the spatial envelope differences of the target on neighboring range gates, which helps to reduce the degree of confusion between erroneous range positions and true range positions, thereby improving the ability to suppress false peaks and the performance of high-resolution range-velocity two-dimensional parameter extraction.

[0052] Third, this invention combines a broadband frequency-agile radar digital channelization receiver structure to compensate for the time delay and phase error introduced during the digital channelization process. This helps to improve the phase consistency between pulses and reduce the impact of digital channelization processing on subsequent pulse compression, coarse positioning, and joint estimation of range-velocity two-dimensional parameters, thereby enhancing the applicability of the method in practical radar hardware systems. Attached Figure Description

[0053] Figure 1This is a flowchart illustrating the overall process of the frequency-agile radar range-velocity two-dimensional parameter joint estimation method based on spatial envelope constraints described in this invention.

[0054] Figure 2 This is a schematic diagram of the digital channelization downconversion, data extraction, and compensation processing flow in this invention.

[0055] Figure 3 This is a schematic diagram illustrating the shift in coarse positioning center caused by pulse compression main lobe aliasing in a dense target scene.

[0056] Figure 4 The figure shows the results of the distance-velocity two-dimensional parameter estimation when using the traditional single-observation vector sparse recovery method.

[0057] Figure 5 The figure shows the distance-velocity two-dimensional parameter estimation results when using the multi-observation vector joint sparse recovery method based on spatial envelope constraints of the present invention.

[0058] Figure 6 The graph shows the comparison of normalized mean square errors between the traditional single-observation vector sparse recovery method and the multi-observation vector joint sparse recovery method of this invention.

[0059] Figure 7 The graph shows a comparison of the correct recovery success rates between the traditional single-observation vector sparse recovery method and the multi-observation vector joint sparse recovery method of this invention. Detailed Implementation

[0060] The specific embodiments and working principle of the present invention will be further described in detail below with reference to the accompanying drawings. For ease of description, this embodiment first defines the following: Let the speed of light be . The initial carrier frequency of the radar is The pulse width is The pulse repetition interval is The radar signal bandwidth is The frequency modulation slope is The original sampling rate is The number of digital channelization channels is Single channel width The number of pulses is , No. The frequency hopping point corresponding to each pulse is: The coarse positioning center corresponds to the distance gate index as follows: The number of adjacent doors is Preferably The process involves extracting the range gate at the location of the coarse positioning center and its left and right adjacent range gates. Radar range resolution is defined as:

[0061]

[0062] like Figure 1 As shown, the frequency-agile radar two-dimensional parameter joint estimation method based on spatial envelope constraints according to the present invention includes the following steps.

[0063] Step 1: In this embodiment, the frequency-agile radar transmits a linear frequency modulated pulse signal, the first... The transmitted signal of a pulse can be represented as:

[0064]

[0065] in, A rectangular window function can be represented as:

[0066]

[0067] No. The echo signal of each pulse is:

[0068]

[0069] in, Indicates the amplitude of the target echo. Indicates additive noise. Indicates the first The time delay corresponding to each pulse.

[0070] Step 2: Assume the radar system has the highest sampling rate Channelization decimation factor After the radar receiver intercepts the high-frequency raw echo signal, a polyphase filter bank is used for digital channelization down-conversion and data extraction to reduce the throughput pressure on the back-end digital signal processor. Its underlying hardware processing flow is as follows: Figure 2 As shown, it mainly includes the following parts:

[0071] (1) The original echo signal sampled at high speed is decimated according to the decimation factor. Divided into A parallel, low-speed data branch;

[0072] (2) The preset low-pass filter coefficients are extracted at equal intervals and divided into corresponding groups. The low-speed data branch is then converted into a sub-filter, and discrete convolution calculations are performed on the corresponding sub-filters.

[0073] (3) Perform phase rotation on the convolution output of each branch and execute... Point-based inverse fast Fourier transform is used to achieve full-band channelization decomposition of broadband signals;

[0074] (4) According to the current radar system's notification... The frequency hopping point of each pulse, from the IFFT output. In each channel, target channel data containing target echo energy is dynamically extracted to obtain a downsampling rate of... The baseband echo data sequence.

[0075] In the aforementioned multiphase filter digital channelization process, the hardware decimation structure causes an inherent low-level sampling time offset in the baseband envelope along the fast time dimension. In this embodiment, this minute time offset is precisely quantized into a single sampling period, i.e. .

[0076] In traditional fixed-carrier-frequency radar, this tiny delay only produces a negligible constant initial phase; however, in coherent frequency-agile radar systems, due to the carrier frequency between pulses... A pseudo-random jump occurs within a wide bandwidth, with a fixed time offset. It will couple with the carrier frequency of the frequency jump, exhibiting a carrier phase distortion term that varies with the frequency jump point in the slow time dimension. :

[0077]

[0078] This non-ideal phase distortion completely destroys the inter-pulse coherence of the frequency-hopping echo. Without intervention, it will directly lead to basis mismatch in subsequent two-dimensional parameter reconstruction algorithms, generating numerous spurious distance peaks. This invention utilizes frequency-hopping sequences... With the extracted inherent time offset Constructing a high-precision pulse-by-pulse inverse physical compensation factor in the digital domain :

[0079]

[0080] Subsequently, the digitally channelized output of the first A sequence of pulsed baseband echoes, multiplied by the corresponding inverse compensation factor. Precise phase calibration is performed to obtain the baseband echo data matrix. .

[0081] Step 3: Extract the coarse localization center using pulse compression and define the two-dimensional local reconstruction mesh. For the compensated... Perform pulse compression to obtain the pulse compression output matrix, and then... The pulse compression envelope of each pulse is modulated and noncoherent energy is accumulated. A constant false alarm rate (CFAR) is used to extract the peak index from the accumulated energy envelope.

[0082] In dense target scenarios, when the distance between multiple targets is less than or close to the radar range resolution, the main lobes of the echoes from different targets will alias, causing the pulse compression output peak to deviate from any true target position, thus resulting in a coarse positioning center offset. See [link to documentation]. Figure 3To address this problem, this invention does not directly rely on a single coarse positioning center for single-observation vector recovery. Instead, it uses the coarse positioning center as a reference and constructs a two-dimensional local discrete reconstruction grid for range and velocity in its vicinity, providing a candidate search space for subsequent joint sparse recovery of multiple observation vectors. Using the drifting center as a reference, a two-dimensional local reconstruction grid covering the radar range resolution is defined. Let the range search grid set be... The speed search grid set is The total number of reconstructed grids is .

[0083] Step 4: Construct a joint dictionary matrix of multi-observation vectors with spatial envelope constraints using the Kronecker product. Extract the coarse localization center and its neighbors. A distance gate (preferably, take a distance gate) The pulse compression echo data is used to concatenate multi-range gate data column by column to form a multi-observation vector. .

[0084] For each discrete grid point Constructing a joint dictionary of spacetime multi-observation vectors ,in By utilizing Kronecker's properties, the crossing... Atoms are constructed by combining the fast temporal spatial envelope morphological features of a distance gate with the slow-time Doppler phase features:

[0085]

[0086] Among them, the slow-time Doppler phase vector Defined as:

[0087]

[0088] Fast time-space envelope vector Includes radar The function is in The amplitude broadening and attenuation patterns at a distance gate are defined as follows:

[0089]

[0090] By traversing all distance and velocity grid combinations, and concatenating all generated joint atoms as column vectors, an unnormalized multi-observation vector joint dictionary with spatial envelope prior constraints is constructed. Its block matrix expansion form is as follows:

[0091]

[0092] Wherein, corresponding to the first Sub-dictionary matrix of distances from grid points This includes all velocity assumptions for that distance, expanded as follows:

[0093]

[0094] To meet the computational requirements of sparse solvers such as orthogonal matching pursuit, the constructed two-dimensional joint atoms need to be subjected to... Norm normalization. Based on the properties of the Kronecker product, the norm normalization of atoms... Norm calculation can be performed as follows:

[0095]

[0096] Since the norm of the Doppler phase vector is always Therefore, the normalized joint atoms It has the following parsing form:

[0097]

[0098] Step 5: Suppression of spurious peaks based on envelope non-collinearity and joint solution of two-dimensional parameters. The multi-measurement vector observation vector extracted in Step 4... With the MMV joint dictionary matrix The input is fed into a sparse recovery solver such as orthogonal matching pursuit for iterative optimization. The joint sparse recovery solver can be an orthogonal matching pursuit solver, a synchronous orthogonal matching pursuit solver, a staged orthogonal matching pursuit solver, or other joint sparse recovery solvers suitable for multi-observation vector models.

[0099] Because this invention introduces adjacent distance gate spatial envelope constraints into the joint dictionary, even if the coarse localization center shifts and some erroneous distance positions show high similarity to the true target position in the slow-time Doppler phase dimension, the amplitude distributions of the true target position and the false distance positions on multiple adjacent distance gates still exhibit stable differences. Therefore, the joint sparse recovery solver can utilize this difference to suppress erroneous support selection, reduce the probability of false peak generation, and improve the accuracy and robustness of the joint estimation of range-velocity two-dimensional parameters. The mathematical mechanism for suppressing range ambiguity is as follows:

[0100] In traditional single-observation vector scenarios, the range-Doppler complete dictionary is obtained by normalizing the pulse compression output of a single range-gate echo signal. Let the distance difference between the actual target position and a certain erroneous grid position in the dictionary be... The corresponding two-way delay difference is .

[0101] In discrete frequency agile systems, the first... The carrier frequency of each pulse is (in For frequency hopping codewords, (For minimum frequency hopping steps). The phase difference term of the dictionary atoms corresponding to the true position and the erroneous position in the slow time dimension depends on... .

[0102] When the distance difference between the recovered erroneous point and the true point satisfies a specific equation, that is... ( When the integer is non-zero, the time delay difference At this point, the phase difference caused by the frequency term that jumps with the pulse is:

[0103]

[0104] As can be seen from the above mathematical derivation, the random pseudo-random phase introduced by frequency hopping is canceled out, and the slow-time Doppler vectors of the erroneous point and the true point atom differ by only a constant phase. Because the single-observation model, when constructing its dictionary, performs column normalization operations on the fast time-space envelope (… The magnitude of the function is canceled out as a constant. This leads to a high correlation between the inner products of atoms at true positions and those at incorrect positions during sparse optimization, which the sparse solver cannot distinguish, thus causing errors when atoms are far apart. The false target was restored at its original location.

[0105] Within the multi-observation vector reconstruction framework constructed in this invention, by extracting the span A joint sparse recovery model is constructed using data from neighboring distance gates. This joint dictionary... In norm normalization, the denominator term fully preserves the relative amplitude envelope proportions of the target echo across multiple range gate dimensions, overcoming the loss of envelope information in single-observation vector models. Let the normalized joint atom corresponding to the range-velocity grid where the real target is located be... Then the atom at the true position and the multi-measurement vector observation signal The inner product matching degree between them can be characterized as:

[0106]

[0107] For distance ambiguity locations (i.e., spurious grid points where the distance difference satisfies a specific integer multiple relationship) that are prone to basis mismatch in traditional single-observation vector models. Since the joint atom of the present invention fully maps the multi-distance gate... According to the Cauchy-Schwarz inequality, the inner product of the spurious position atom and the observed signal strictly satisfies the following constraint relationship regarding the envelope broadening and attenuation morphology:

[0108]

[0109] Because the fast temporal spatial envelope distribution characteristics of the real target observation signal and the envelope vector of the spurious location joint atom are significantly misaligned in physical space, they exhibit strict non-collinearity in the joint vector space. Therefore, the Cauchy-Schwarz inequality cannot be strictly equalized. Furthermore, even if the real target position and the spurious range position show high similarity in the slow-time Doppler phase dimension, the fast-time spatial envelope broadening shape and relative amplitude distribution of the two positions at multiple adjacent range gates still have stable differences, which can still be completely preserved during the joint atom normalization process. Therefore, the spatial envelope constraint introduced in this invention can provide the joint sparse recovery solver with additional discrimination criteria besides slow-time phase information, thereby improving the rejection capability for erroneous supports and enhancing the stability of two-dimensional parameter recovery under the condition of coarse positioning center offset.

[0110] This demonstrates that even if the extremely high mutual coherence of the frequency-agile system causes the slow-time phase difference term to degenerate into a constant, the inner product of the multi-observation vector joint dictionary constructed in this invention at the real target grid is still strictly greater than the inner product at the absolute distance ambiguity location. This invention imposes a strong constraint on the iterative optimization trajectory of the sparse recovery solver by introducing a mathematical penalty mechanism based on the spatial envelope morphology difference of the multi-distance gate, causing it to converge to the real target coordinates.

[0111] To make the objectives, technical solutions, and technical effects of this invention clearer, a detailed description of this invention is provided through specific simulation experiments:

[0112] Simulation experimental conditions: Set the number of transmitted pulses for the coherent frequency agile radar system. signal bandwidth Pulse width Pulse repetition period Initial carrier frequency System sampling rate Channelization decimation factor At this time, the radar's range resolution is... .

[0113] Two closely spaced moving targets are set up in the observation scene, with their distance and velocity being respectively... and The distance between the two targets is approximately Within the radar range resolution range, main lobe aliasing occurs. The pre-pulse compression signal-to-noise ratio is set to... .

[0114] In the aforementioned dense target scenario, the main lobe coherent interference causes a severe shift in the pulse compression output peak, and the coarse localization center is selected on a non-real grid point.

[0115] The single-observation vector sparse recovery method using existing technology has achieved the following results: Figure 4 The simulation diagram of two-dimensional parameter estimation. Figure 4 In the diagram, the x-axis represents the high-resolution range cell, the y-axis represents the high-resolution velocity cell, and the z-axis represents the target amplitude. Due to envelope loss caused by column normalization, the reconstruction results not only produce deviations near the true location but also exhibit multiple false targets with extremely high energy at a distance from the true target, leading to the failure of target detection and parameter estimation.

[0116] The distance-velocity two-dimensional parameter estimation results obtained by the multi-observation vector joint sparse recovery method based on spatial envelope constraints proposed in this invention are as follows: Figure 5 As shown, this method compensates for the time delay and phase error introduced by digital channelization processing in step 2, and extracts the coarse localization center and its adjacent range gates to construct a joint dictionary matrix of multi-observation vectors in step 4. Simulation results show that this invention can significantly suppress spurious peaks, eliminate the ambiguity caused by the main lobe aliasing of dense targets, and estimate the range and velocity parameters of two real targets relatively accurately. This demonstrates that this invention has good tolerance for coarse localization center offset and resistance to range ambiguity.

[0117] Furthermore, in order to quantitatively compare the performance of the method of the present invention with that of the traditional single-observation vector sparse recovery method, the normalized mean square error and the success rate of correct recovery of the two methods were statistically analyzed under different signal-to-noise ratio conditions. Figure 6 The normalized mean square error comparison curves of the two methods are given. Figure 7 Comparison curves of the correct recovery success rates of the two methods are presented. Simulation results show that, under the same simulation conditions, the multi-observation vector joint sparse recovery method based on spatial envelope constraints described in this invention can achieve a lower normalized mean square error and a higher correct recovery success rate, indicating that this method has better performance in joint estimation of range-velocity two-dimensional parameters under dense target conditions.

Claims

1. A joint estimation method for two-dimensional parameters of frequency-agile radar based on spatial envelope constraints, characterized in that, Includes the following steps: Step 1: Acquire the raw echo signal output by the frequency-agile radar receiver, perform digital channelization down-conversion and data extraction processing on the raw echo signal to obtain multi-pulse baseband echo data; at the same time, acquire the frequency hopping point information and channelization extraction factor corresponding to each pulse. Step 2: Based on the channelization decimation factor, system sampling rate, and frequency hopping information corresponding to each pulse, calculate the time delay error and phase error introduced during digital channelization, construct a compensation factor, and perform compensation processing on the multi-pulse baseband echo data; Step 3: Perform pulse compression on the compensated multi-pulse baseband echo data and non-coherent accumulation on the pulse compression output. Use constant false alarm rate detection or peak extraction to determine the coarse target positioning center. Based on the coarse positioning center, construct a local discrete reconstruction grid in the range-velocity two-dimensional plane. Radar range resolution is , At the speed of light, The radar signal bandwidth; the local discrete reconstruction grid includes: the range search grid set is The speed search grid set is The total number of reconstructed grids is ; Step 4: Let the distance gate index corresponding to the coarse positioning center be... The echo data of the coarse positioning center and its multiple adjacent range gates are extracted to form multi-observation vector observation data. : ； in, Indicates the pulse compression after the first pulse. Fast-time echo data corresponding to each distance gate , This indicates the number of adjacent distance gates selected on one side based on the coarse positioning center; For each candidate range-velocity element in the locally discrete reconstructed mesh Constructing the slow-time Doppler phase vector : ； in, Indicates the initial carrier frequency of the radar. Indicates the first The frequency point corresponding to each pulse. Indicates the number of pulses. This represents the distance index in the local distance reconstruction mesh. This represents the velocity index in the local velocity reconstruction mesh. The pulse repetition period is . The two-way propagation delay of each pulse is . : ； in, In the local distance reconstruction mesh, the first... The distance value corresponding to each candidate distance unit In the local velocity reconstruction mesh, the first... The velocity values ​​corresponding to each candidate velocity unit. The pulse repetition period, Represents the speed of light; Simultaneously construct a fast time-space envelope vector spanning multiple distance gates. : ； in, Indicates the bandwidth of the radar transmitted signal. This represents the two-way propagation delay corresponding to the candidate distance. This indicates the reference fast-time sampling moment corresponding to the distance gate of the coarse positioning center; A joint atom is constructed from the fast time-space envelope vector and the slow time-Doppler phase vector. : ； in, For Kronecker product; Traverse all candidate range and velocity cells to form a joint dictionary matrix of multi-observation vectors. ; Step 5: Input the multi-observation vector observation data and the multi-observation vector joint dictionary matrix into the joint sparse recovery solver for solving. Determine the target support set based on the correlation between candidate atoms and observation data, and output the estimated results of the target's distance and velocity parameters.

2. The joint estimation method for two-dimensional parameters of frequency-agile radar based on spatial envelope constraints according to claim 1, characterized in that, In step 1 The original echo signal of each pulse is : ； in, Indicates the amplitude of the target echo. Indicates the fast sampling time. Indicates additive noise. Indicates the first The time delay corresponding to each pulse Indicates the pulse repetition period. Indicates the linear frequency modulation slope. Indicates the first The frequency point corresponding to each pulse. This represents the distance to the target at the initial moment. This indicates the radial velocity of the target; It is a rectangular window function: ; Highest sampling rate of radar system Channelization decimation factor .

3. The method for joint estimation of two-dimensional parameters of frequency-agile radar based on spatial envelope constraints according to claim 1, characterized in that, The specific method for step 2 is as follows: Step 2.1: Decimate the original echo signal obtained from high-speed sampling by a certain factor. Divided into A parallel, low-speed data branch; Step 2.2: Extract the preset low-pass filter coefficients at equal intervals, corresponding to the following divisions: The low-speed data branch is then converted into a sub-filter, and discrete convolution calculations are performed on the corresponding sub-filters. Step 2.3: Perform phase rotation on the convolution output of each branch and execute... Point-based inverse fast Fourier transform enables full-band channelization decomposition of broadband signals; Step 2.4: Based on the current radar signal issued by the radar system... The frequency hopping point of each pulse, output from step 2.3 In each channel, target channel data containing target echo energy is dynamically extracted to obtain a downsampling rate of... The baseband echo data sequence; Step 2.5: In the above polyphase filtering digital channelization process, the hardware decimation structure will cause the baseband envelope to have an inherent low-level sampling time offset in the fast time dimension. ; The carrier phase distortion term varies with the frequency hopping point in the slow time dimension. : ； Using frequency hopping sequences With the extracted inherent time offset Constructing a high-precision pulse-by-pulse inverse physical compensation factor in the digital domain : Step 2.6: Convert the digitally channelized output to the first... A sequence of pulsed baseband echoes, multiplied by the corresponding inverse compensation factor. Precise phase calibration is performed to obtain the baseband echo data matrix. .

4. The method for joint estimation of two-dimensional parameters of frequency-agile radar based on spatial envelope constraints according to claim 1, characterized in that, The multiple distance gates mentioned in step 4 include the distance gate where the coarse localization center is located and the adjacent distance gates on both sides; by introducing fast temporal spatial envelope constraints in the multiple distance gate dimensions, and retaining the relative amplitude envelope ratio characteristics on multiple distance gates during the joint atom normalization process, the correlation between the joint atom corresponding to the real target position and the multiple observation vectors is satisfied; ； in, This represents the normalized joint atom corresponding to the actual target position. Normalized joint atoms corresponding to spurious distance positions are represented, thereby suppressing erroneous support selection and spurious peaks caused by main lobe aliasing, coarse localization center shift, and high dictionary coherence.

Images