A method and device for wideband frequency domain digital beamforming based on FPGA

Abstract

The application relates to a kind of FPGA-based wideband frequency domain digital beam forming method and device, belong to wideband digital beam forming technical field, solve the problem that too much hardware operation resource is used in the wideband beam synthesis method in the prior art, and beam pointing dynamic configuration is not flexible. Including the multi-channel acquisition of the radio frequency data received by antenna, and the radio frequency data collected by each channel is converted into multi-phase time domain data;The multi-phase time domain data of each channel is compensated by integer multiple sampling delay, then parallel fourier transform is carried out, and the multi-phase frequency domain data of each channel is obtained;Based on the frequency domain beam synthesis weight and multi-phase frequency domain data of each channel, multi-channel frequency domain synthesis is carried out, frequency domain multi-channel wideband beam synthesis data is obtained, then parallel inverse fourier transform is carried out, and time domain beam synthesis data is obtained, the synthesis of specified direction beam is realized. The gain of receiving signal is simultaneously enhanced, and the reconnaissance and processing capacity for low interception signal is improved.

Description

Technical Field

[0001] This invention relates to the field of broadband digital beamforming technology, and in particular to a broadband frequency domain digital beamforming method and apparatus based on FPGA. Background Technology

[0002] To improve range resolution and high-precision imaging, radar antennas employ high-bandwidth signals. However, the ultra-wideband instantaneous bandwidth presents a significant challenge to non-cooperative signal reconnaissance. To increase sensitivity in acquiring low-intercept signals and detect radiating targets at greater distances, digital spatial filtering methods are used to increase gain in a specified direction and suppress noise energy in other directions. For wideband digital receivers, the same array delay can lead to different frequency responses, rendering narrowband beamforming methods unsuitable. Therefore, research into wideband beamforming methods is urgent and necessary.

[0003] Currently, filter methods based on Farrow polynomial interpolation structures are simple to reload delays, but for multiphase data processing in ultra-wideband receivers, the computational resources required are too large, and engineering implementation is difficult. Microwave photonics technology relies on optical fibers to achieve photonic physical delays, but it is greatly affected by temperature and humidity. The fractional delay method of FIR filters compensates for the arrival time of signals at each antenna element, aligns the phases between elements, and weights the amplitude to achieve energy superposition in the desired direction. Digital phase shifters compensate for the phase difference caused by the carrier wave, digital delayers use shift registers to compensate for integer multiples of the sampling period delay, and fractional delay filters compensate for fractional multiples of the sampling period delay. By adjusting the above three compensators, the element delay compensation and phase alignment are achieved. RF direct sampling receivers have very high sampling rates. Due to the constraints of FPGA main frequency and timing, the sampled data has a multiphase structure, and the multiplication resources of FIR filters increase linearly with the number of phases, resulting in excessive resource usage.

[0004] Therefore, existing broadband beamforming methods consume excessive hardware computing resources and lack flexibility in dynamic beam pointing configuration. Summary of the Invention

[0005] Based on the above analysis, the embodiments of the present invention aim to provide a broadband frequency domain digital beamforming method and apparatus based on FPGA, in order to solve the problems of excessive hardware computing resources used in existing broadband beamforming methods and inflexible dynamic configuration of beam pointing.

[0006] On one hand, embodiments of the present invention provide a broadband frequency domain digital beamforming method based on FPGA, comprising the following steps:

[0007] The radio frequency data received by the antenna is acquired through multiple channels, and the radio frequency data acquired from each channel is converted into polyphase time domain data.

[0008] The multiphase time domain data of each channel is sampled and delayed by integer multiples, and then parallel Fourier transform is performed to obtain the multiphase frequency domain data of each channel.

[0009] Based on the frequency domain beamforming weights and polyphase frequency domain data of each channel, multi-channel frequency domain synthesis is performed to obtain frequency domain multi-channel broadband beamforming data.

[0010] The frequency domain multi-channel broadband beamforming data is subjected to parallel inverse Fourier transform to obtain time domain beamforming data, thereby achieving beamforming in a specified direction.

[0011] Furthermore, the multiphase time-domain data is represented as an L-row, M-column matrix; the compensated multiphase time-domain data is subjected to a parallel Fourier transform to obtain multiphase frequency-domain data, specifically:

[0012] Perform an M-point Fourier transform on the multiphase time-domain data row by row to obtain L rows of M-point Fourier transform data;

[0013] The M-point Fourier transform data in each row are multiplied by the corresponding N-order rotation factor using complex dot multiplication to obtain L rows and M columns of parallel data; where N is L*M.

[0014] The parallel data is subjected to L-point Fourier transform column by column to obtain M columns of L-point Fourier transform data, which in turn constitute L rows and M columns of polyphase frequency domain data.

[0015] Furthermore, the element X(p,q) in the p-th row and q-th column of the polyphase frequency domain data is represented as:

[0016]

[0017] In the formula, L and M represent the number of rows and columns of the multiphase time-domain data, respectively; N represents the total number of sampling points of the radio frequency data; x(l,m) represents the element in the l-th row and m-th column of the multiphase time-domain data; and W... A Let A denote the A-order twitch factor, where A is N, M, or L.

[0018] Furthermore, the frequency domain beamforming weights include column weights corresponding to each column in the polyphase frequency domain data; the multi-channel frequency domain synthesis based on the frequency domain beamforming weights of each channel and the polyphase frequency domain data to obtain frequency domain multi-channel broadband beamforming data specifically involves:

[0019] Multiply each column of the multiphase frequency domain data of each channel by the column weight in the corresponding frequency domain beamforming weight to obtain the frequency domain beamforming data of each channel.

[0020] The frequency domain beamforming data of each channel are added together to obtain frequency domain multi-channel broadband beamforming data.

[0021] Furthermore, the frequency domain beamforming weights for each channel are obtained in the following manner:

[0022] The control software DSP obtains the arrival delay, carrier phase difference, and beam pointing phase weighting of each channel based on the multi-channel corrected comb spectrum data;

[0023] The arrival delay and carrier phase difference of each channel are mapped to the frequency domain to obtain the compensated phase. The compensated phase is then added to the beam pointing phase of all set frequency points to obtain the column weight of each column in the multiphase frequency domain data, thus forming the frequency domain beamforming weight.

[0024] Furthermore, the step of performing a parallel inverse Fourier transform on the frequency-domain multi-channel broadband beamforming data to obtain time-domain beamforming data specifically involves:

[0025] The frequency domain multi-channel broadband beamforming data is subjected to L-point inverse Fourier transform column by column to obtain M rows of L-point inverse Fourier transform data.

[0026] The L-point inverse Fourier transform data of each column are multiplied by the corresponding N-order conjugate rotation factor by complex number, and then M-point inverse Fourier transform is performed row by row to obtain L rows of M-point Fourier transform data, which in turn constitute L rows of M columns of time-domain beamforming data.

[0027] Furthermore, the Fourier transform is implemented using a radix-2 burst structure butterfly algorithm.

[0028] Furthermore, the sampling delay of multiphase time-domain data is compensated for by a digital delay device, which is an integer multiple of the delay.

[0029] On the other hand, the present invention also provides a broadband frequency domain digital beamforming device based on FPGA, including a data acquisition and conversion module, a data compensation and frequency domain conversion module, a frequency domain multi-channel beamforming module and a specified direction beamforming module;

[0030] The data acquisition and conversion module is used to acquire radio frequency data received by the antenna through multiple channels and convert the radio frequency data acquired from each channel into polyphase time domain data.

[0031] The data compensation and frequency domain conversion module is used to perform integer multiple sampling delay compensation on the multiphase time domain data of each channel, and then perform parallel Fourier transform to obtain the multiphase frequency domain data of each channel.

[0032] The frequency domain multi-channel beamforming module is used to perform multi-channel frequency domain synthesis based on the frequency domain beamforming weights of each channel and polyphase frequency domain data to obtain frequency domain multi-channel broadband beamforming data.

[0033] The specified direction beamforming module is used to perform parallel inverse Fourier transform on the frequency domain multi-channel broadband beamforming data to obtain time domain beamforming data, thereby realizing the synthesis of specified direction beams.

[0034] Furthermore, the polyphase time-domain data in the data compensation and frequency domain conversion module is represented as an L-row, M-column matrix; specifically, the compensated polyphase time-domain data is subjected to a parallel Fourier transform to obtain the polyphase frequency-domain data, as follows:

[0035] Perform an M-point Fourier transform on the multiphase time-domain data row by row to obtain L rows of M-point Fourier transform data;

[0036] The M-point Fourier transform data in each row are multiplied by the corresponding N-order rotation factor using complex dot multiplication to obtain L rows and M columns of parallel data; where N is L*M.

[0037] The parallel data is subjected to L-point Fourier transform column by column to obtain M columns of L-point Fourier transform data, which in turn constitute L rows and M columns of polyphase frequency domain data.

[0038] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects:

[0039] This invention provides a broadband frequency domain digital beamforming method and apparatus based on FPGA. It involves multi-channel acquisition of radio frequency (RF) data received by an antenna, converting the RF data acquired from each channel into polyphase time-domain data, performing integer-multiple sampling delay compensation on the polyphase time-domain data of each channel, and then performing a parallel Fourier transform to obtain the polyphase frequency-domain data of each channel. Based on the frequency-domain beamforming weights of each channel and the polyphase frequency-domain data, multi-channel frequency-domain synthesis is performed to obtain frequency-domain multi-channel broadband beamforming data. Finally, the frequency-domain multi-channel broadband beamforming data is subjected to a parallel inverse Fourier transform to obtain time-domain beamforming data, thus achieving the synthesis of a beam in a specified direction and transmitting broadband signals. The signal is decomposed into a narrowband signal, and the multi-channel data is aligned according to frequency points. The phase and amplitude weighting of the multi-channel specific directional beam is completed in the frequency domain, which solves the problem of the aperture effect of the ultra-wideband instantaneous bandwidth antenna array and simulates the beamforming pointing deviation problem. It enhances the gain of the desired beam direction and suppresses interference and noise in other directions. Furthermore, the broadband frequency domain digital beamforming method based on parallel Fourier transform has the advantages of low computational load, low hardware resource consumption and high real-time performance. At the same time, it can dynamically configure multiple beam pointing and can carry out direction finding and positioning for multiple directions and multiple targets, improve the detection and processing capabilities of low intercept signals, and has important application value in the fields of electronic countermeasures and phased array antennas.

[0040] In this invention, the above-described technical solutions can be combined with each other to achieve more preferred combinations. Other features and advantages of this invention will be set forth in the following description, and some advantages may become apparent from the description or be learned by practicing the invention. The objects and other advantages of this invention can be realized and obtained from what is particularly pointed out in the description and drawings. Attached Figure Description

[0041] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts.

[0042] Figure 1 This is a flowchart illustrating the FPGA-based broadband frequency domain digital beamforming method provided in Embodiment 1 of the present invention.

[0043] Figure 2 This is a schematic diagram of the antenna array element arrangement structure provided in Embodiment 1 of the present invention;

[0044] Figure 3 This is a block diagram illustrating the principle of the parallel discrete Fourier transform algorithm provided in Embodiment 1 of the present invention.

[0045] Figure 4 This is a comparison diagram of the frequency domain digital beam normal direction synthesis results before and after the present invention, provided in Embodiment 1 of the present invention. Detailed Implementation

[0046] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, which form part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not intended to limit the scope of the present invention.

[0047] Example 1

[0048] A specific embodiment of the present invention discloses a broadband frequency domain digital beamforming method based on FPGA, such as... Figure 1 As shown, it includes the following steps:

[0049] S1. Multi-channel acquisition of radio frequency data received by the antenna, and conversion of radio frequency data acquired from each channel into polyphase time domain data.

[0050] Specifically, an ultra-wideband digital receiver is used to acquire radio frequency data. An AD converter with a JESD204B interface is used as the receiver's analog-to-digital converter, which can achieve a sampling rate of gigahertz frequency, realize the direct acquisition of radio frequency data, and complete the coverage of ultra-large instantaneous bandwidth.

[0051] Specifically, since multiphase data is parallel, the serial high sampling rate data is converted into parallel multi-channel low sampling rate data through FIFO, so that the original sampling rate remains unchanged.

[0052] It is understandable that the sampling rate of broadband RF sampling data reaches gigahertz. Clocks of this order of magnitude cause FPGA timing to fail to converge. Therefore, considering the constraints of FPGA main frequency clock and timing convergence requirements, the high sampling rate digital signal directly acquired by RF data needs to be converted into a multiphase low sampling rate data structure to ensure that the RF sampling rate remains unchanged and reduce the main frequency clock. This allows the FPGA to perform signal processing in the multiphase domain, which is more conducive to the stability of FPGA timing and function.

[0053] S2. Perform integer multiple sampling delay compensation on the multiphase time domain data of each channel, and then perform parallel Fourier transform to obtain the multiphase frequency domain data of each channel.

[0054] Specifically, the antenna array receives non-cooperative signals in the radiation space. The path difference of the electromagnetic waves arriving at each array element is different, resulting in different time delays of the electrical signals output by the array elements. In order to ensure the phase consistency of the output signals of the array elements, a digital delay unit is used to compensate for the integer multiple sampling delay of the multiphase time domain data.

[0055] More specifically, the electromagnetic wave of the non-cooperative signal is incident parallel to the antenna array, with an angle θ between the incident wave and the normal direction of the array element, and a spacing D between adjacent array elements. Taking the first array element as a reference, the time difference τ between the arrival wave and the i-th array element is calculated. i for:

[0056]

[0057] In the formula, n′ is the number of antenna array elements, and c is the speed of light.

[0058] As can be seen from the above formula, the delay between each array element can be decomposed into integer multiple sampling delay and fractional multiple sampling delay. Integer multiple delay can be implemented using the shift register of a digital delay unit, while fractional multiple sampling delay is mapped as a phase shift in the frequency domain and completed in the frequency domain through subsequent steps, which can reduce the amount of computation and resource usage.

[0059] Phase difference of the i-th element with reference to the first element for:

[0060]

[0061] In the formula, f c For carrier waves.

[0062] For example, the antenna array adopts a uniform four-element linear array, and the schematic diagram of the array element arrangement structure is shown in Figure 1. Figure 2 As shown.

[0063] When implementing, such as Figure 2As shown, the multiphase time-domain data is represented as an L-row, M-column matrix; the compensated multiphase time-domain data is subjected to a parallel Fourier transform to obtain the multiphase frequency-domain data, specifically:

[0064] Perform an M-point Fourier transform on the multiphase time-domain data row by row to obtain L rows of M-point Fourier transform data;

[0065] The M-point Fourier transform data in each row are multiplied by the corresponding N-order rotation factor using complex dot multiplication to obtain L rows and M columns of parallel data; where N is L*M.

[0066] The parallel data is subjected to L-point Fourier transform column by column to obtain M columns of L-point Fourier transform data, which in turn constitute L rows and M columns of polyphase frequency domain data.

[0067] It is understandable that the rotation factor is of order M in the M-point Fourier transform and order L in the L-point Fourier transform. It should be noted that the rotation factor can be stored in ROM memory and retrieved during calculations.

[0068] Specifically, the element X(p,q) in the p-th row and q-th column of the polyphase frequency domain data is represented as:

[0069]

[0070] In the formula, L and M represent the number of rows and columns of the multiphase time-domain data, respectively; N represents the total number of sampling points of the radio frequency data; x(l,m) represents the element in the l-th row and m-th column of the multiphase time-domain data; and W... A Let A denote the A-order twitch factor, where A is N, M, or L.

[0071] More specifically, the A-order twisting factor W A Represented as:

[0072]

[0073] It should be noted that the elements in the polyphase frequency domain data are derived through the following derivation:

[0074] The Discrete Fourier Transform (DFT) of a radio frequency sampled data sequence x(n) of length N is:

[0075]

[0076] In the formula, x(n) is the nth value in the radio frequency sampling data sequence, X(k) is the kth value after the discrete Fourier transform of the sampling data sequence, and W N This represents the Nth order rotation factor.

[0077] The polyphase structure of radio frequency data can be understood as the concept of rows and columns in a matrix array. A data sampling sequence of length N points can be decomposed into the following formula:

[0078] N = M * L (6)

[0079] In the formula, M and L are positive integers, L is the number of phases in the multiphase structure (i.e., the number of rows in the matrix), and M is the number of columns in the matrix. Therefore, the multiphase structure of the data is mapped to the coordinate form in the matrix, and the coordinate indices of the data elements in the multiphase time-domain data matrix are:

[0080] n=l+m*L (7)

[0081] In the formula, l is the row index label of the multiphase time-domain data matrix, 0≤l≤L-1, and m is the column index label of the multiphase time-domain data matrix, 0≤m≤M-1.

[0082] Similarly, the frequency domain polyphase data matrix X(k) calculated by DFT can be mapped to the form (p,q), as shown in the following equation:

[0083] k=M*p+q (8)

[0084] In the formula, p is the row index of the polyphase frequency domain data matrix, 0≤p≤L-1, and q is the column index of the polyphase frequency domain data matrix, 0≤q≤M-1.

[0085] Therefore, combining equations (5) to (8), the multiphase frequency domain data matrix after DFT can be further expressed as:

[0086]

[0087] Further simplification:

[0088]

[0089] Based on the reducibility, symmetry, and periodicity of the rotation factor, the rotation factor in equation (10) simplifies to:

[0090]

[0091] Therefore, the elements in the polyphase frequency domain data can be simplified to:

[0092]

[0093] Equation (12) shows that the multiphase parallel DFT breaks down the serial long sequence DFT into multiple parallel short DFTs, and the FPGA-based parallel DFT algorithm is implemented as follows: Figure 3As shown. The N-point L-phase RF sampling time-domain data is allocated as L rows and N columns of multi-phase time-domain data. The rotation factor in equation (12) is stored in the ROM memory. Specifically, the parallel Fourier transform is first performed according to the M-point Fourier transform (DFT). The rotation factor vector in the ROM is indexed by the lookup table method. The M-point DFT data and the rotation factor are multiplied by a complex number. The pulsating parallel data is then subjected to the L-point DFT, thus completing the algorithm implementation of the parallel DFT.

[0094] Understandably, the parallel Fourier transform breaks down long data sequences into multiple short data sequences, accelerating frequency domain conversion. It has the advantages of fast computation speed, low latency, and high real-time performance, while reducing the computational load of the Fourier transform, reducing hardware resource usage, and being easy to engineer, thus having wide application value.

[0095] S3. Based on the frequency domain beamforming weights and polyphase frequency domain data of each channel, perform multi-channel frequency domain synthesis to obtain frequency domain multi-channel broadband beamforming data.

[0096] In implementation, the frequency domain beamforming weights include the column weights corresponding to each column in the polyphase frequency domain data; the multi-channel frequency domain synthesis based on the frequency domain beamforming weights of each channel and the polyphase frequency domain data to obtain frequency domain multi-channel broadband beamforming data is specifically as follows:

[0097] Multiply each column of the multiphase frequency domain data of each channel by the column weight in the corresponding frequency domain beamforming weight to obtain the frequency domain beamforming data of each channel.

[0098] The frequency domain beamforming data of each channel are added together to obtain frequency domain multi-channel broadband beamforming data.

[0099] Specifically, the frequency domain beamforming weights for each channel are obtained in the following way:

[0100] The control software DSP obtains the arrival delay, carrier phase difference, and beam pointing phase weighting of each channel based on the multi-channel corrected comb spectrum data;

[0101] The arrival delay and carrier phase difference of each channel are mapped to the frequency domain to obtain the compensated phase. The compensated phase is then added to the beam pointing phase of all set frequency points to obtain the column weight of each column in the multiphase frequency domain data, thus forming the frequency domain beamforming weight.

[0102] It is understandable that the frequency points in each column of the polyphase frequency domain data are the same, and the beam pointing phase weight of each frequency point corresponds to the column in the polyphase frequency domain data.

[0103] More specifically, the correction process in the control software DSP obtains the arrival delay, carrier phase difference, and beam pointing phase weighting of each channel by switching the built-in standard correction source of the digital receiver to transmit the comb spectrum signal.

[0104] More specifically, the beam pointing phase weighting of all frequency points is set according to actual needs to achieve beam synthesis in a specific direction. The physical antenna array can be dynamically configured with multiple digitally synthesized beams to achieve parallel reconnaissance and processing analysis of multiple directions and multiple targets, further enhancing the application value of the array antenna in the field of electronic warfare.

[0105] It should be noted that arbitrary beam pointing can be synthesized as needed by adjusting the beam pointing phase weighting of all set frequency points. This enables the antenna array to simultaneously detect multiple radiation sources in multiple directions, and to carry out functions such as multi-target, multi-directional direction finding and positioning, thereby enhancing the reconnaissance capabilities of the electronic countermeasures system.

[0106] S4. Perform parallel inverse Fourier transform on the frequency domain multi-channel broadband beamforming data to obtain time domain beamforming data, thereby achieving beamforming in a specified direction.

[0107] In practice, the process of performing parallel inverse Fourier transform on the frequency-domain multi-channel broadband beamforming data to obtain time-domain beamforming data specifically involves:

[0108] The process of performing parallel inverse Fourier transform on the frequency domain multi-channel broadband beamforming data to obtain time domain beamforming data specifically involves:

[0109] The frequency domain multi-channel broadband beamforming data is subjected to L-point inverse Fourier transform column by column to obtain M rows of L-point inverse Fourier transform data.

[0110] The L-point inverse Fourier transform data of each column are multiplied by the corresponding N-order conjugate rotation factor by complex number, and then M-point inverse Fourier transform is performed row by row to obtain L rows of M-point Fourier transform data, which in turn constitute L rows of M columns of time-domain beamforming data.

[0111] Understandably, in order to perform subsequent signal analysis tasks such as signal detection, direction finding, and positioning, the frequency domain multi-channel broadband beamforming data is converted into time domain beamforming data using a parallel inverse Fourier transform algorithm, thereby achieving the synthesis of beams in a specified direction.

[0112] Specifically, the Fourier transform uses the Fast Fourier Transform, and the inverse Fourier transform uses the Fast Inverse Fourier Transform.

[0113] More specifically, the Fourier transform employs a radix-2 burst structure butterfly algorithm.

[0114] Compared with existing technologies, this embodiment provides a broadband frequency domain digital beamforming method based on FPGA. It involves multi-channel acquisition of radio frequency (RF) data received by the antenna, converting the RF data acquired from each channel into polyphase time-domain data, performing integer-multiple sampling delay compensation on the polyphase time-domain data of each channel, and then performing parallel Fourier transform to obtain polyphase frequency-domain data for each channel. Based on the frequency-domain beamforming weights of each channel and the polyphase frequency-domain data, multi-channel frequency-domain synthesis is performed to obtain frequency-domain multi-channel broadband beamforming data. Finally, the frequency-domain multi-channel broadband beamforming data is subjected to parallel inverse Fourier transform to obtain time-domain beamforming data, thus achieving beamforming in a specified direction. Broadband signals are decomposed into narrowband signals, and multi-channel data are aligned according to frequency points. Phase and amplitude weighting of multi-channel specific directional beams is completed in the frequency domain, solving the problem of antenna aperture effect in ultra-wideband instantaneous bandwidth antenna arrays and simulating beamforming pointing deviation. This enhances the gain of the desired beam direction and suppresses interference and noise in other directions. Furthermore, the broadband frequency domain digital beamforming method based on parallel Fourier transform has the advantages of low computational load, low hardware resource consumption, and high real-time performance. It can also dynamically configure multiple beam pointing, enabling direction finding and positioning for multiple directions and multiple targets, improving the detection and processing capabilities of low intercept signals. It has significant application value in the fields of electronic warfare and phased array antennas.

[0115] Example 2

[0116] A specific embodiment of the present invention discloses a broadband frequency domain digital beamforming device based on FPGA, including a data acquisition and conversion module, a data compensation and frequency domain conversion module, a frequency domain multi-channel beamforming module, and a specified direction beamforming module;

[0117] The data acquisition and conversion module is used to acquire radio frequency data received by the antenna through multiple channels and convert the radio frequency data acquired from each channel into polyphase time domain data.

[0118] The data compensation and frequency domain conversion module is used to perform integer multiple sampling delay compensation on the multiphase time domain data of each channel, and then perform parallel Fourier transform to obtain the multiphase frequency domain data of each channel.

[0119] The frequency domain multi-channel beamforming module is used to perform multi-channel frequency domain synthesis based on the frequency domain beamforming weights of each channel and polyphase frequency domain data to obtain frequency domain multi-channel broadband beamforming data.

[0120] The specified direction beamforming module is used to perform parallel inverse Fourier transform on the frequency domain multi-channel broadband beamforming data to obtain time domain beamforming data, thereby realizing the synthesis of specified direction beams.

[0121] In implementation, the polyphase time-domain data in the data compensation and frequency domain conversion module is represented as an L-row, M-column matrix; specifically, the compensated polyphase time-domain data is subjected to a parallel Fourier transform to obtain the polyphase frequency-domain data, as follows:

[0122] Perform an M-point Fourier transform on the multiphase time-domain data row by row to obtain L rows of M-point Fourier transform data;

[0123] The M-point Fourier transform data in each row are multiplied by the corresponding N-order rotation factor using complex dot multiplication to obtain L rows and M columns of parallel data; where N is L*M.

[0124] The parallel data is subjected to L-point Fourier transform column by column to obtain M columns of L-point Fourier transform data, which in turn constitute L rows and M columns of polyphase frequency domain data.

[0125] The specific implementation process of this invention can be found in the above method embodiments, and will not be repeated here.

[0126] Since this embodiment is based on the same principle as the above method embodiments, this system also has the corresponding technical effects of the above method embodiments.

[0127] Example 3

[0128] This invention provides a specific embodiment to describe and verify the method proposed in Embodiment 1. In this embodiment, the digital receiver has 4 analog channels, the RF sampling data is 8 phases, and a 256-point DFT is performed in the frequency domain. Therefore, the parallel DFT algorithm can be converted into a Fast Fourier Transform (FFT) implementation, further reducing the DFT computational load. The principle implementation block diagram of the broadband frequency domain digital beamforming device is shown below. Figure 1 As shown, the RF direct sampling data with a sampling rate of 1.2GHz is converted into 8-phase 150MHz sampling data after multiphase conversion.

[0129] The correction process involves switching the receiver's built-in standard correction source to transmit a comb spectrum signal, calculating the signal arrival delay between array elements, the carrier phase difference, and the beam pointing phase weighting across all broadband frequencies. These parameters are then issued by the control unit (DSP) during the beamforming process. The integer delay of the path difference between array elements is implemented using FPGA shift register resources, enabling arbitrary integer sampling delays at a 1.2GHz sampling rate, fractional delays of the path difference, and phase compensation mapping the carrier phase difference to the frequency domain. To improve the timeliness of beamforming in the frequency domain, parallel Fourier transform is used for accelerated processing. 256 sampled data points are constructed into an 8-row, 32-column multiphase time-domain data matrix. Fast Fourier transform is performed according to the 32-point sampled data points per row of the multiphase time-domain data matrix. A lookup table method is used to index the address of the twitch factor memory, and the corresponding twitch factor vector slides through each row of the fast Fourier transform, performing complex dot multiplication. Then, fast Fourier transform is performed according to the 8-point sampled data points per column of the matrix, finally reaching the result of the serial Fourier transform. The computational complexity and time consumption results of the serial and parallel Fourier transforms in this embodiment are shown in Table 1. The Fourier transform uses radix-2... The burst structure butterfly algorithm implementation, with 256 sampled data, requires 1024 additions and 2048 multiplications in the serial Fourier transform, consuming 1677 clock cycles. The parallel Fourier transform requires 92 additions and 184 multiplications, consuming 326 clock cycles. Regardless of the computational load or time, the parallel Fourier transform significantly outperforms the serial Fourier transform. The more Fourier transform points, the stronger the advantages of the parallel Fourier transform in terms of computational load and time. Therefore, the parallel Fourier transform algorithm is more suitable for scenarios with higher requirements for beamforming direction and synchronization correction accuracy.

[0130] Table 1. Comparison of computational complexity and time consumption between serial Fourier transform and parallel Fourier transform.

[0131]

[0132] The DSP control software calculates the phase weights of the frequency domain synthesis of the beam pointing of the four channels based on the corrected comb spectrum data. In the frequency domain, the real-time sampling stream and beam pointing weights of each channel are multiplied. The frequency domain beamforming results of all channels are summed. By adjusting the phase between channels to ensure their amplitudes are in phase, the multi-channel beamforming with a specific direction is finally achieved. For further data processing and analysis of signals in a fixed direction, the frequency domain beamforming results are converted to the time domain using a parallel inverse Fourier transform. Following the inverse Fourier transform of the 8-point beamforming matrix, the rotation factor corresponding to the lookup table index is conjugate. Then, the inverse Fourier transform is performed using the 32-point data of the matrix rows to obtain the time domain results of the 4-channel beamforming with a specified direction. A comparison of the frequency domain beamforming results before and after beamforming in the normal direction of the 4-channel broadband receiver is shown in the figure below. Figure 4As shown, the straight waveform represents the original data amplitude, and the dashed waveform represents the amplitude of the frequency domain digital beamforming result. It can be seen from the figure that the background noise is basically the same before and after beamforming. The receiver's noise floor gain is not changed; only the signal gain is altered. This ensures that increasing the signal gain does not cause the data saturation and overflow problem that was originally represented by the bit width. A local magnified view of the signal gain is shown below. Figure 4 As shown on the left, the signal amplitude before beamforming is 17.03 dB, and the signal amplitude after beamforming is 23.02 dB. The peak amplitude in the normal direction differs by 6 dB. The beamforming method improves the signal gain and conforms to the theoretical gain of 4-channel beamforming.

[0133] Those skilled in the art will understand that all or part of the processes of the methods described in the above embodiments can be implemented by a computer program instructing related hardware, and the program can be stored in a computer-readable storage medium. The computer-readable storage medium may be a disk, optical disk, read-only memory, or random access memory, etc.

[0134] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A broadband frequency domain digital beamforming method based on FPGA, characterized in that, Includes the following steps: The radio frequency data received by the antenna is acquired through multiple channels, and the radio frequency data acquired from each channel is converted into polyphase time domain data; the polyphase time domain data is represented as an L-row M-column matrix; The multiphase time-domain data of each channel is subjected to integer multiple sampling delay compensation, followed by parallel Fourier transform to obtain the multiphase frequency-domain data of each channel; the compensated multiphase time-domain data is then subjected to parallel Fourier transform to obtain the multiphase frequency-domain data, specifically as follows: The multiphase time-domain data is subjected to an M-point Fourier transform row by row to obtain L rows of M-point Fourier transform data; the Fourier transform is implemented using a radix-2 burst structure butterfly algorithm; The M-point Fourier transform data in each row are multiplied by the corresponding N-order rotation factor using complex dot multiplication to obtain L rows and M columns of parallel data; where N is the length of L. M; The parallel data is subjected to L-point Fourier transform column by column to obtain M columns of L-point Fourier transform data, which in turn constitute L rows and M columns of polyphase frequency domain data. Based on the frequency domain beamforming weights and polyphase frequency domain data of each channel, multi-channel frequency domain synthesis is performed to obtain frequency domain multi-channel broadband beamforming data; the frequency domain beamforming weights include the column weights corresponding to each column in the polyphase frequency domain data; the frequency domain beamforming weights of each channel are obtained in the following way: The DSP obtains the arrival delay, carrier phase difference, and beam pointing phase weighting of each channel based on the multi-channel corrected comb spectrum data; The arrival delay and carrier phase difference of each channel are mapped to the frequency domain to obtain the compensated phase. The compensated phase is then added to the beam pointing phase of all set frequency points to obtain the column weight of each column in the multiphase frequency domain data, which constitutes the frequency domain beamforming weight. The frequency domain multi-channel broadband beamforming data is subjected to parallel inverse Fourier transform to obtain time domain beamforming data, thereby achieving beamforming in a specified direction.

2. The FPGA-based broadband frequency domain digital beamforming method according to claim 1, characterized in that, The first of the multiphase frequency domain data Line 1 Column elements Represented as: In the formula, L , M These represent the number of rows and columns of the multiphase time-domain data, respectively. N This indicates the total number of sampling points for radio frequency data. In multiphase time-domain data, the first... Line 1 Column elements, Let A be the rotation factor of order A, where A is... N , M or L .

3. The FPGA-based broadband frequency domain digital beamforming method according to claim 1, characterized in that, The process involves multi-channel frequency domain synthesis based on the frequency domain beamforming weights and polyphase frequency domain data of each channel, resulting in frequency domain multi-channel broadband beamforming data. Specifically: Multiply each column of the multiphase frequency domain data of each channel by the column weight in the corresponding frequency domain beamforming weight to obtain the frequency domain beamforming data of each channel. The frequency domain beamforming data of each channel are added together to obtain frequency domain multi-channel broadband beamforming data.

4. The FPGA-based broadband frequency domain digital beamforming method according to claim 1, characterized in that, The process of performing parallel inverse Fourier transform on the frequency domain multi-channel broadband beamforming data to obtain time domain beamforming data specifically involves: The frequency domain multi-channel broadband beamforming data is subjected to L-point inverse Fourier transform column by column to obtain M rows of L-point inverse Fourier transform data. The L-point inverse Fourier transform data of each column are multiplied by the corresponding N-order conjugate rotation factor by complex number, and then M-point inverse Fourier transform is performed row by row to obtain L rows of M-point Fourier transform data, which in turn constitute L rows of M columns of time-domain beamforming data.

5. The FPGA-based broadband frequency domain digital beamforming method according to claim 1, characterized in that, The sampling delay of multiphase time domain data is compensated by an integer multiple of the delay using a digital delay device.

6. A broadband frequency domain digital beamforming device based on FPGA, characterized in that, It includes a data acquisition and conversion module, a data compensation and frequency domain conversion module, a frequency domain multi-channel beamforming module, and a specified direction beamforming module; The data acquisition and conversion module is used to acquire radio frequency data received by the antenna through multiple channels and convert the radio frequency data acquired from each channel into polyphase time domain data; the polyphase time domain data is represented as an L-row M-column matrix. The data compensation and frequency domain conversion module is used to perform integer-multiple sampling delay compensation on the multiphase time domain data of each channel, followed by parallel Fourier transform to obtain the multiphase frequency domain data of each channel; and to perform parallel Fourier transform on the compensated multiphase time domain data to obtain the multiphase frequency domain data, specifically: The multiphase time-domain data is subjected to an M-point Fourier transform row by row to obtain L rows of M-point Fourier transform data; the Fourier transform is implemented using a radix-2 burst structure butterfly algorithm; The M-point Fourier transform data in each row are multiplied by the corresponding N-order rotation factor using complex dot multiplication to obtain L rows and M columns of parallel data; where N is the length of L. M; The parallel data is subjected to L-point Fourier transform column by column to obtain M columns of L-point Fourier transform data, which in turn constitute L rows and M columns of polyphase frequency domain data. The frequency domain multi-channel beamforming module is used to perform multi-channel frequency domain synthesis based on the frequency domain beamforming weights of each channel and polyphase frequency domain data to obtain frequency domain multi-channel broadband beamforming data; the frequency domain beamforming weights include the column weights corresponding to each column in the polyphase frequency domain data; the frequency domain beamforming weights of each channel are obtained in the following way: The DSP obtains the arrival delay, carrier phase difference, and beam pointing phase weighting of each channel based on the multi-channel corrected comb spectrum data; The arrival delay and carrier phase difference of each channel are mapped to the frequency domain to obtain the compensated phase. The compensated phase is then added to the beam pointing phase of all set frequency points to obtain the column weight of each column in the multiphase frequency domain data, which constitutes the frequency domain beamforming weight. The specified direction beamforming module is used to perform parallel inverse Fourier transform on the frequency domain multi-channel broadband beamforming data to obtain time domain beamforming data, thereby realizing the synthesis of specified direction beams.

Images