Hls-based broadband signal high-resolution direction finding ip core design method and device

Abstract

This invention discloses a design method and apparatus for a high-resolution direction-finding IP core for broadband signals based on HLS, relating to the field of signal processing technology. By constructing a comprehensive evaluation standard for sub-frequency wavelength and half-wavelength matching and sub-frequency energy—that is, a sub-frequency comprehensive evaluation function—this invention optimizes reference and candidate sub-frequency points that contribute significantly to algorithm performance. This effectively reduces the complexity of broadband signal focusing operations while ensuring algorithm performance. Furthermore, it integrates the high-resolution direction-finding algorithm for broadband coherent signals with a high-level synthesis implementation method, forming a high-resolution direction-finding IP core for broadband coherent signals that can be directly called by FPGAs, possessing strong versatility and reusability.

Description

Technical Field

[0001] This invention belongs to the field of signal processing technology, and particularly relates to a design method and device for a broadband signal high-resolution direction finding IP core based on HLS. Background Technology

[0002] Direction of Arrival (DOA) estimation is one of the fundamental problems in array signal processing. Currently, narrowband DOA algorithms are relatively mature, with the most famous being Multiple Signal Classification (MUSIC). Its core principle is to separate the signal subspace from the noise subspace using eigenvalue decomposition of the covariance matrix, and then construct a spatial spectrum in the angular domain based on the orthogonality of the signal and noise subspaces to achieve high-resolution direction finding. However, when processing broadband signals, the array manifold is inconsistent at different frequencies, resulting in differences in the subspaces generated by the decomposition of signals at different frequencies. Therefore, existing mature narrowband signal localization algorithms are difficult to apply to broadband signal localization. Thus, broadband signal DOA estimation algorithms typically divide the broadband signal into multiple narrowband signals with different frequencies before processing. Incoherent methods usually perform narrowband estimation for each frequency point separately and then fuse the results; coherent methods construct a focusing matrix to "focus" each frequency subspace onto a reference frequency point before performing a high-resolution estimation. However, the focusing process is computationally intensive and sensitive to the selection of the reference frequency point. Therefore, reducing the computational load of the focusing process and optimizing the reference frequency selection strategy are key to improving algorithm performance, robustness, and ease of engineering implementation.

[0003] Currently, existing methods require singular value decomposition of the data covariance matrix and construction of the focusing matrix for all sub-frequency points. This results in high computational complexity and significant hardware resource overhead when the number of sub-frequency points is large. Furthermore, full-angle two-dimensional spatial spectrum search significantly increases the number of on-chip multiplications and additions, storage access pressure, and overall latency, leading to high computational load and making it difficult to implement in engineering. Moreover, most current high-resolution direction-finding methods for broadband coherent signals focus primarily on algorithm performance verification, lacking modular decomposition and unified interface design for high-level synthesis, and lacking reusable IP cores, resulting in deficiencies in scalability, portability, and versatility. Summary of the Invention

[0004] To address the aforementioned problems in the prior art, this invention provides a design method and apparatus for a broadband signal high-resolution direction finding IP core based on HLS.

[0005] The technical problem to be solved by this invention is achieved through the following technical solution: In a first aspect, the present invention provides a design method for a broadband signal high-resolution direction-finding IP core based on HLS, comprising: S10. Acquire the radar signal received by the receiving array in the radar system, and perform frequency domain transformation and sub-band division on the radar signal to obtain multiple sub-frequency points and corresponding frequency domain data. S20. Based on the pre-constructed sub-frequency point comprehensive evaluation function and the frequency domain data corresponding to multiple sub-frequency points, a reference sub-frequency point and candidate sub-frequency points adjacent to the reference sub-frequency point are selected from multiple sub-frequency points to form a focused sub-frequency point set. S30. Based on the frequency domain data corresponding to each sub-frequency point in the set of focused sub-frequency points, determine the covariance matrix corresponding to each sub-frequency point in the set of focused sub-frequency points; S40. Perform eigenvalue decomposition and focus matrix construction on the covariance matrix corresponding to each sub-frequency point in the set of focused sub-frequency points to obtain the focus covariance matrix. S50. Based on the focused covariance matrix, the direction of arrival is estimated, and the direction of arrival of the target corresponding to the radar signal is output. S60 uses advanced synthesis HLS tools to encapsulate the algorithms corresponding to S10 to S50 to obtain the target general-purpose IP core.

[0006] Secondly, the present invention provides a design apparatus for a broadband signal high-resolution direction finding IP core based on HLS, which obtains a target generalized IP core by executing the broadband signal high-resolution direction finding IP core design method based on HLS of the first aspect.

[0007] This invention provides a design method and apparatus for a high-resolution direction finding IP core for broadband signals based on HLS. By constructing a comprehensive evaluation standard for the matching degree between sub-frequency wavelengths and half-wavelengths and the energy of sub-frequency points, namely the sub-frequency point comprehensive evaluation function, reference sub-frequency points and candidate sub-frequency points that contribute significantly to the algorithm performance are selected. While ensuring the algorithm performance, the complexity of broadband signal focusing operations is effectively reduced. The high-resolution direction finding algorithm for broadband coherent signals is integrated with the high-level synthesis implementation method to form a high-resolution direction finding IP core for broadband coherent signals that can be directly called by FPGA. It supports configuration of the number of array elements, the total number of sub-frequency points, and the maximum number of snapshots, and has strong versatility and reusability.

[0008] The present invention will now be described in further detail with reference to the accompanying drawings. Attached Figure Description

[0009] Figure 1 This is a flowchart illustrating a design method for a broadband signal high-resolution direction finding IP core based on HLS, provided in an embodiment of the present invention. Figure 2 This is a schematic diagram illustrating the principle of the three-level cascaded search method provided in this embodiment of the invention; Figure 3 This is a schematic diagram of the target generalized IP core provided in the embodiments of the present invention. Detailed Implementation

[0010] The present invention will be further described in detail below with reference to specific embodiments, but the implementation of the present invention is not limited thereto.

[0011] In a first aspect, embodiments of the present invention provide a design method for a broadband signal high-resolution direction-finding IP core based on HLS. See also... Figure 1 The method includes the following steps: S10. Acquire the radar signal received by the receiving array in the radar system, and perform frequency domain transformation and sub-band division on the radar signal to obtain multiple sub-frequency points and corresponding frequency domain data.

[0012] For example, the parameters of the receiving array can be obtained in advance, including the number of array elements, the total number of preset sub-frequency points, and the maximum number of snapshots. Taking a uniform planar array as an example, let the number of array elements of the receiving array be... , This represents the total number of rows. The total number of columns is and the element spacing is . ,exist A far-field broadband coherent radar signal from angle Incident, in which, Indicates the first The incident azimuth angle of a radar signal. Indicates the first The incident elevation angle of a radar signal, Represents the speed of light. Then it is located at the... Line 1 Radar signals received by the array elements , is represented as: ; in, For the first One radar signal For noise, The maximum number of snapshots, For the first The radar signal was incident on the first Line 1 The time delay of a column of array elements relative to a reference array element can be expressed as: ; The radar signals received by each array element are subjected to Discrete Fourier Transform and divided into... The number of sub-bands (equal to the number of sub-frequency points) yields the corresponding sub-frequency points. , No. Frequency domain data corresponding to each sub-frequency point , is represented as: ; in, For snapshot matrix, and These are the discrete Fourier transforms of the signal and noise, respectively. For array manifold matrix, guide vector for: .

[0013] S20. Based on the pre-constructed sub-frequency point comprehensive evaluation function and the frequency domain data corresponding to multiple sub-frequency points, a reference sub-frequency point and candidate sub-frequency points adjacent to the reference sub-frequency point are selected from multiple sub-frequency points to form a focused sub-frequency point set.

[0014] Optionally, step S20 may specifically include: S201. Substitute the frequency domain signals corresponding to multiple sub-frequency points into the pre-constructed sub-frequency point comprehensive evaluation function to obtain the evaluation value corresponding to each sub-frequency point.

[0015] For example, in order to reduce the computational load of the focusing process, this embodiment constructs a comprehensive evaluation standard for the matching degree between the sub-frequency wavelength and half-wavelength and the sub-frequency energy, namely the sub-frequency comprehensive evaluation function.

[0016] Optionally, the sub-frequency point comprehensive evaluation function is expressed as: ; in, This represents the sub-frequency point comprehensive evaluation function. Indicating the index of the sub-frequency point, This indicates the total number of subbands. Represents the speed of light. Indicates the carrier wavelength. Indicates the sampling frequency. Indicates the first Frequency domain data corresponding to each sub-frequency point Indicates the first Individual frequency points.

[0017] S202. Select the sub-frequency point with the largest evaluation value as the reference sub-frequency point.

[0018] S203. Based on the reference sub-frequency point and two candidate sub-frequency points adjacent to the reference sub-frequency point, construct a set of focused sub-frequency points.

[0019] For example, in In the middle, select the value that makes the evaluation value reach its maximum value. As the index of the reference sub-frequency point, the reference sub-frequency point is obtained as follows: And select its two adjacent sub-frequency points , As candidate sub-frequency points, construct a set of focused sub-frequency points. , The index of the sub-frequency point in the set of focused sub-frequency points is used to participate in subsequent focusing processing.

[0020] S30. Based on the frequency domain data corresponding to each sub-frequency point in the set of focused sub-frequency points, determine the covariance matrix corresponding to each sub-frequency point in the set of focused sub-frequency points.

[0021] For example, focus on the covariance matrix corresponding to each sub-frequency point in the set of sub-frequency points. , is represented as: ; in, This represents the frequency domain data corresponding to each sub-frequency point in the focused sub-frequency point set. This represents each sub-frequency point in the set of focused sub-frequency points.

[0022] S40. Perform eigenvalue decomposition and focus matrix construction on the covariance matrix corresponding to each sub-frequency point in the focus sub-frequency point set to obtain the focus covariance matrix.

[0023] Optionally, step S40 may specifically include: S401. Perform eigenvalue decomposition on the covariance matrix corresponding to each sub-frequency point in the set of focused sub-frequency points to obtain the eigenvalues ​​corresponding to each sub-frequency point in the set of focused sub-frequency points.

[0024] For example, the eigenvalue decomposition of the covariance matrix corresponding to each sub-frequency point in the focused sub-frequency point set can be expressed as:

[0025] in, express The 1 eigenvalue, express The 1 eigenvector. S402. Calculate the mean and variance of the eigenvalues ​​corresponding to the reference sub-frequency points in the set of focused sub-frequency points to obtain the mean and variance of the eigenvalues ​​corresponding to the reference sub-frequency points.

[0026] For example, the covariance matrix of the reference sub-frequency point is taken. Corresponding eigenvalues Calculate the mean of the eigenvalues and eigenvalue variance , respectively represented as: ; in, Represents the covariance matrix of the reference sub-frequency point The corresponding largest eigenvalue.

[0027] S403. Calculate the threshold factor based on the mean and variance of the eigenvalues ​​corresponding to the reference sub-frequency points.

[0028] Optionally, the threshold factor is expressed as: ; in, Indicates the threshold factor. Indicates the number of elements in the receiving array. Represents the mean of the eigenvalues. This represents the variance of the eigenvalues.

[0029] S404. Determine the number of eigenvalues ​​greater than the threshold factor among the eigenvalues ​​corresponding to the reference sub-frequency point, and use this number as the target quantity.

[0030] For example, eigenvalue set By comparing with a threshold factor, the number of feature values ​​in the feature set that are greater than the threshold factor is determined, thus obtaining the target number. .

[0031] S405. Select the target number of target feature values ​​from the feature values ​​corresponding to each sub-frequency point in the set of focused sub-frequency points in descending order, and construct the initial matrix corresponding to each candidate sub-frequency point based on the feature vector corresponding to the target feature value.

[0032] For example, the covariance matrix corresponding to each sub-frequency point in the set of focused sub-frequency points is taken from largest to smallest. Construct the initial matrix corresponding to each candidate sub-frequency point from the eigenvectors corresponding to the eigenvalues. : ; in, The reference sub-frequency point is the first The feature vector corresponding to each target feature value The first sub-frequency point represents the candidate sub-frequency point. The eigenvectors corresponding to the target eigenvalues.

[0033] S406. Perform singular value decomposition on the initial matrix corresponding to each candidate sub-frequency point to obtain the left singular component and right singular component corresponding to each candidate sub-frequency point.

[0034] For example, performing singular value decomposition on the initial matrix yields: ; in, and These are the left singular component and the right singular component, respectively.

[0035] S407. Based on the left singular component and right singular component corresponding to each candidate sub-frequency point, construct the focusing matrix corresponding to each candidate sub-frequency point.

[0036] For example, the focusing matrix corresponding to each candidate sub-frequency point Represented as: .

[0037] S408. Calculate the focusing covariance matrix based on the focusing matrix corresponding to each candidate sub-frequency point and the covariance matrix corresponding to each sub-frequency point in the set of focused sub-frequency points.

[0038] Alternatively, the focus is on the covariance matrix, expressed as: ; in, This indicates the focus covariance matrix. This represents the covariance matrix corresponding to the reference sub-frequency point. Indicates the reference sub-frequency point. This indicates the index corresponding to the reference sub-frequency point. This represents the focusing matrix corresponding to the candidate sub-frequency points. This represents the covariance matrix corresponding to the candidate sub-frequency points. Indicates candidate sub-frequency points, Indicates the index of the candidate sub-frequency point.

[0039] S50. Based on the focused covariance matrix, the direction of arrival is estimated, and the direction of arrival of the target corresponding to the radar signal is output.

[0040] Optionally, step S50 may specifically include: S501. Perform eigenvalue decomposition on the focusing covariance matrix to obtain the eigenvalues ​​corresponding to the focusing covariance matrix.

[0041] For example, eigenvalue decomposition is performed on the focusing covariance matrix: ; in, express The 1 eigenvalue, express The 1 eigenvector.

[0042] S502. Select a preset number of eigenvalues ​​from the eigenvalues ​​corresponding to the focusing covariance matrix in ascending order, and construct a noise subspace based on the eigenvectors corresponding to the selected eigenvalues.

[0043] For example, the preset quantity is Noise subspace representation for: ; S503. Based on the noise subspace and the predetermined steering vector, the spatial spectrum estimate is calculated.

[0044] Alternatively, the spatial spectrum estimation is expressed as: ; in, Indicates spatial spectrum estimation, Indicates the first The incident azimuth angle of a radar signal. Indicates the first The incident elevation angle of a radar signal, This represents the steering vector corresponding to the reference sub-frequency point. Indicates the reference sub-frequency point. This indicates the index corresponding to the reference sub-frequency point. This represents the noise subspace.

[0045] S504. A three-level cascaded search method is used to search the spatial spectrum to obtain the direction of arrival of the target wave corresponding to the radar signal.

[0046] For example, a three-level cascaded search method from coarse to fine is used to achieve two-dimensional spectral peak search, the basic principle of which is shown in the appendix. Figure 2 As shown. The first stage uses a larger first preset step size for the entire angular domain. A coarse search is performed to quickly eliminate more than 90% of the no-signal areas; subsequent levels only use smaller steps to perform fine searches within the window retained from the previous level.

[0047] The first-level local search window retained after the first-level coarse search. It can be represented as: ; in, , These represent the incident elevation angle and incident azimuth angle under the first-level search, respectively. This is the threshold coefficient for the first-level search. This represents the maximum spectral value for the candidate region at this level. Subsequently, [the following is done / conducted / etc.]. Local maxima filtering is performed, retaining the point with the largest spectral value in its neighborhood. Then, based on the angular region corresponding to each local maxima, the eight adjacent angular grid points are merged to construct a second-level local search window. : ; in, , These represent the incident elevation angle and incident azimuth angle under the second-level search, respectively. and Let represent the incident elevation angle and incident azimuth angle, respectively, corresponding to the local maxima of the first-level search. Then, for... Use a smaller second preset step size A preliminary fine search is performed, followed by local maxima filtering. Points with the largest spectral values ​​in their neighborhoods are retained. The eight adjacent angle grid points of each local maximum are merged based on the angle domain corresponding to that local maximum, thus constructing a third-level local search window. : ; in, , These represent the incident elevation angle and incident azimuth angle under the third-level search, respectively. , denoted as the incident elevation angle and incident azimuth angle, respectively, corresponding to the local maxima of the second-level search.

[0048] Finally, Use the minimum preset third step size A fine-grained search is performed, outputting the incident elevation and azimuth angles corresponding to each maxima to obtain the target arrival direction of the radar signal estimated by DOA. Typically, the number of local search windows retained from the previous stage is much smaller than the number of global grid points; therefore, the three-stage cascaded search method is suitable for implementing low-latency two-dimensional spatial spectrum search in hardware systems.

[0049] S60 uses advanced synthesis HLS tools to encapsulate the algorithms corresponding to S10 to S50 to obtain the target general-purpose IP core.

[0050] For example, the structure of the target generalized IP core is shown in the appendix. Figure 3As shown in the figure, this embodiment employs a four-layer collaborative optimization strategy of "algorithm decomposition—inter-module data flow—loop-level pipeline—memory-level parallel access" to ensure that the broadband coherent signal high-resolution direction finding algorithm corresponding to the aforementioned steps S10 to S50 can be stably mapped into a target general-purpose IP core for engineering deployment. The target general-purpose IP core is divided into sub-modules corresponding to parameter configuration interface, frequency domain transformation (S10), reference frequency selection (S20), covariance matrix calculation (S30), focusing matrix construction and focusing covariance fusion (S40), and independently schedulable MUSIC spatial spectrum estimation and three-level cascaded peak search (S50). Furthermore, a data flow optimization instruction DATAFLOW (task-level data pipeline instruction) is applied at the top level to decouple the front-end and back-end modules through the flow interface and first-in-first-out buffer, transforming the traditional serial software execution link into an overlapping execution link in hardware, significantly reducing processing latency. Within each submodule, PIPELINE (data pipeline instruction), LOOP_FLATTEN (nested loop unrolling instruction), and UNROLL (loop unrolling instruction) optimization instructions are applied to the loops of operations such as snapshot traversal, matrix multiplication and addition, vector inner product, and angle grid search. The loops are unrolled to achieve parallel processing and reduce processing latency.

[0051] Furthermore, the target general-purpose IP core employs a dual-buffered storage structure to achieve cross-frame overlapping processing. While the current buffer serves the calculation of the current frame, the backup buffer synchronously receives data from the next frame, thereby achieving parallel overlap between the input buffer and kernel calculation, improving overall throughput. For sub-modules intensive in matrix operations, complex multiplication, and multiply-addition operations, such as covariance matrix calculation, focused matrix construction, and focused covariance fusion, the complex multiplication and multiply-addition operations are preferentially mapped to the dedicated computing unit DSP48 through BIND_OP or RESOURCE (resource mapping instruction) optimization instructions, improving computational efficiency. The covariance matrix calculation sub-module only calculates the upper triangular elements of the matrix and uses conjugate symmetry to backfill the lower triangular elements to reduce the number of complex multiplications and additions. The ARRAY_PARTITION (array partitioning instruction) or ARRAY_RESHAPE (array reorganization instruction) optimization instructions are applied to the local matrix buffer and the guided vector buffer, enabling concurrent access to multiple data streams in the same clock cycle and avoiding pipeline pauses caused by single-port memory access conflicts.

[0052] This embodiment achieves a high clock frequency, stable real-time processing capability, and good scalability through algorithm and hardware co-optimization design. It is suitable for deployment as a general-purpose broadband high-resolution direction finding IP core on an FPGA-based array signal processing platform.

[0053] The present invention provides a design method for a broadband signal high-resolution direction-finding IP core based on HLS, which has the following significant advantages compared with the prior art: 1. This invention constructs a comprehensive evaluation standard for the matching degree between sub-frequency wavelength and half-wavelength and the energy of sub-frequency points, and selects reference sub-frequency points and candidate sub-frequency points that contribute significantly to the algorithm performance. While ensuring the algorithm performance, it effectively reduces the complexity of broadband signal focusing operations.

[0054] 2. This invention uses a three-stage cascaded filtering search method from coarse to fine to transform the search complexity from a single-stage traversal of a fixed number of grid points across the entire angle domain to a hierarchical structure of "coarse search + local fine search". This can significantly reduce the number of multiply-accumulate operations and on-chip memory access required for spectral peak search, making it particularly suitable for real-time implementation in engineering.

[0055] 3. This invention boasts strong versatility and good scalability, facilitating IP core reuse in engineering applications. Its input / output interfaces, control registers, and intermediate buffer methods are unified and clearly defined, allowing configuration of array size, sub-frequency points, and snapshot number. Through algorithm and hardware co-design, while ensuring algorithm performance, it simultaneously considers data flow, double buffering, and matrix operation resource sharing issues in high-level synthesis, reducing algorithm complexity, hardware resource consumption, and power consumption, thus exhibiting strong portability.

[0056] Secondly, embodiments of the present invention also provide a design apparatus for a broadband signal high-resolution direction finding IP core based on HLS. This apparatus obtains a target generalized IP core by executing the broadband signal high-resolution direction finding IP core design method based on HLS of the first aspect.

[0057] It should be noted that, as the apparatus is basically similar to the method embodiment of the first aspect, the description is relatively simple, and relevant parts can be referred to in the description of the method embodiment.

[0058] It should be noted that the terms "first," "second," etc., are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the invention.

[0059] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Furthermore, those skilled in the art can combine and integrate the different embodiments or examples described in this specification.

[0060] Although the invention has been described herein in conjunction with various embodiments, those skilled in the art will understand and implement other variations of the disclosed embodiments by reviewing the accompanying drawings and the disclosure in carrying out the claimed invention. In the description of the invention, the word "comprising" does not exclude other components or steps, "a" or "an" does not exclude a plurality, and "a plurality" means two or more, unless otherwise explicitly specified. Furthermore, while different embodiments may describe certain measures, this does not mean that these measures cannot be combined to produce good results.

[0061] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of the present invention, and all such modifications and substitutions should be considered within the scope of protection of the present invention.

Claims

1. A design method for a broadband signal high-resolution direction-finding IP core based on HLS, characterized in that, include: S10. Acquire the radar signal received by the receiving array in the radar system, and perform frequency domain transformation and sub-band division on the radar signal to obtain multiple sub-frequency points and corresponding frequency domain data. S20. Based on the pre-constructed sub-frequency point comprehensive evaluation function and the frequency domain data corresponding to the multiple sub-frequency points, a reference sub-frequency point and a candidate sub-frequency point adjacent to the reference sub-frequency point are selected from the multiple sub-frequency points to form a focused sub-frequency point set. S30. Based on the frequency domain data corresponding to each sub-frequency point in the set of focused sub-frequency points, determine the covariance matrix corresponding to each sub-frequency point in the set of focused sub-frequency points; S40. Perform eigenvalue decomposition and focus matrix construction on the covariance matrix corresponding to each sub-frequency point in the set of focused sub-frequency points to obtain the focus covariance matrix. S50. Based on the focusing covariance matrix, perform direction of arrival estimation and output the target arrival direction corresponding to the radar signal; S60 uses advanced synthesis HLS tools to encapsulate the algorithms corresponding to S10 to S50 to obtain the target general-purpose IP core.

2. The design method for a broadband signal high-resolution direction-finding IP core based on HLS according to claim 1, characterized in that, S20 includes: Substitute the frequency domain signals corresponding to the multiple sub-frequency points into the pre-constructed sub-frequency point comprehensive evaluation function to obtain the evaluation value corresponding to each sub-frequency point; Select the sub-frequency point with the highest evaluation value as the reference sub-frequency point; Based on the reference sub-frequency point and two candidate sub-frequency points adjacent to the reference sub-frequency point, a set of focused sub-frequency points is constructed.

3. The design method for a broadband signal high-resolution direction-finding IP core based on HLS according to claim 1, characterized in that, S40 includes: Eigenvalue decomposition is performed on the covariance matrix corresponding to each sub-frequency point in the set of focused sub-frequency points to obtain the eigenvalues ​​corresponding to each sub-frequency point in the set of focused sub-frequency points. The mean and variance of the eigenvalues ​​corresponding to the reference sub-frequency points in the set of focused sub-frequency points are calculated to obtain the mean and variance of the eigenvalues ​​corresponding to the reference sub-frequency points. The threshold factor is calculated based on the mean and variance of the eigenvalues ​​corresponding to the reference sub-frequency points; The number of feature values ​​greater than the threshold factor among the feature values ​​corresponding to the reference sub-frequency point is determined as the target quantity; Select the target number of target feature values ​​from the feature values ​​corresponding to each sub-frequency point in the set of focused sub-frequency points in descending order, and construct the initial matrix corresponding to each candidate sub-frequency point based on the feature vector corresponding to the target feature values; Singular value decomposition is performed on the initial matrix corresponding to each candidate sub-frequency point to obtain the left singular component and right singular component corresponding to each candidate sub-frequency point; Based on the left singular component and right singular component corresponding to each candidate sub-frequency point, construct the focusing matrix corresponding to each candidate sub-frequency point; The focusing covariance matrix is ​​calculated based on the focusing matrix corresponding to each candidate sub-frequency point and the covariance matrix corresponding to each sub-frequency point in the set of focused sub-frequency points.

4. The design method for a broadband signal high-resolution direction-finding IP core based on HLS according to claim 1, characterized in that, The S50 includes: Eigenvalue decomposition is performed on the focusing covariance matrix to obtain the eigenvalues ​​corresponding to the focusing covariance matrix; A predetermined number of eigenvalues ​​are selected from the eigenvalues ​​corresponding to the focused covariance matrix in ascending order, and a noise subspace is constructed based on the eigenvectors corresponding to the selected eigenvalues. Based on the noise subspace and the predetermined steering vector, the spatial spectrum estimate is calculated; A three-level cascaded search method is used to search the spatial spectrum estimation to obtain the direction of arrival of the target wave corresponding to the radar signal.

5. The design method for a broadband signal high-resolution direction-finding IP core based on HLS according to claim 2, characterized in that, The sub-frequency point comprehensive evaluation function is expressed as follows: ； in, This represents the sub-frequency point comprehensive evaluation function. Indicating the index of the sub-frequency point, This indicates the total number of subbands. Represents the speed of light. Indicates the carrier wavelength. Indicates the sampling frequency. Indicates the first Frequency domain data corresponding to each sub-frequency point Indicates the first Individual frequency points.

6. The design method for a broadband signal high-resolution direction-finding IP core based on HLS according to claim 3, characterized in that, The threshold factor is expressed as: ； in, Indicates the threshold factor. This indicates the number of elements in the receiving array. This represents the mean of the eigenvalues. This represents the variance of the eigenvalues.

7. The design method for a broadband signal high-resolution direction-finding IP core based on HLS according to claim 3, characterized in that, The focused covariance matrix is ​​expressed as: ； in, This indicates the focus covariance matrix. This represents the covariance matrix corresponding to the reference sub-frequency point. Indicates the reference sub-frequency point. This indicates the index corresponding to the reference sub-frequency point. This represents the focusing matrix corresponding to the candidate sub-frequency points. This represents the covariance matrix corresponding to the candidate sub-frequency points. Indicates candidate sub-frequency points, Indicates the index of the candidate sub-frequency point.

8. The design method for a broadband signal high-resolution direction-finding IP core based on HLS according to claim 4, characterized in that, The spatial spectrum estimation is expressed as: ； in, Indicates spatial spectrum estimation, Indicates the first The incident azimuth angle of a radar signal. Indicates the first The incident elevation angle of a radar signal, This represents the steering vector corresponding to the reference sub-frequency point. Indicates the reference sub-frequency point. This indicates the index corresponding to the reference sub-frequency point. Represents the noise subspace.

9. A design device for a broadband signal high-resolution direction-finding IP core based on HLS, characterized in that, The device obtains a target generalized IP core by executing the HLS-based broadband signal high-resolution direction finding IP core design method as described in any one of claims 1-8.

Images