Efficient Amplitude and Phase Extraction Method of Comb Spectrum Signal Based on FPGA

Abstract

The present invention discloses an FPGA-based method for efficiently extracting the amplitude and phase of a comb spectrum signal, belonging to the field of electronic reconnaissance. First, the comb spectrum signal sampled by AD is cyclically accumulated and cached in RAM for segmented accumulation processing. The segmented accumulation improves the signal-to-noise ratio of the comb spectrum signal, thereby improving the accuracy of amplitude and phase extraction. Second, the segmented accumulation result is complex-multiplied by the frequency shift coefficient to perform spectrum shifting processing. The spectrum shifting shifts the spectrum of the comb spectrum signal to zero frequency, so that each frequency point is located at an integer multiple of the DFT frequency resolution. Third, the spectrum shifting result is complex-multiplied by the DFT coefficient to perform DFT calculation. The DFT calculation is used to obtain the spectral information of the comb spectrum signal. Finally, the DFT calculation result is modulo-calculated and LOG-calculated to obtain the amplitude information of all frequency points. The Cordic core is then called to perform an inverse tangent operation to obtain the phase information of all frequency points.

Description

Technical Field

[0001] The present invention belongs to the field of electronic reconnaissance, and in particular to an FPGA-based method for efficiently extracting the amplitude and phase of a comb spectrum signal. Background Art

[0002] In the development of electronic reconnaissance equipment, there is usually inconsistency in the amplitude and phase of signals between channels, and amplitude and phase correction of multi-channel signals is required. Amplitude and phase correction requires extracting the amplitude and phase information of each channel signal. The traditional amplitude and phase extraction method is to perform amplitude and phase extraction on each frequency point by switching the point frequency source. However, as the processing bandwidth increases, the number of extracted frequency points increases, and the processing time also increases, which cannot meet the "fast" requirement of amplitude and phase extraction of multi-frequency point signals. Although the amplitude and phase extraction method based on the correlation method can measure the amplitude and phase of multiple frequency points at the same time, this method has a large amount of redundant calculations and low computational efficiency, which cannot meet the "efficient" requirement of amplitude and phase extraction of multiple frequency point signals. In response to the above situation, the technology of the present invention proposes an efficient amplitude and phase extraction method for comb spectrum signals based on FPGA. This algorithm only uses a small amount of FPGA resources and can quickly and efficiently complete the parallel amplitude and phase extraction of multi-frequency point signals.

[0003] In his paper, "Processing Phase Calibration Signals in VLBI Observations," Liu Yongyou focused on a correlation-based PCAL (phase calibration) signal processing method. This method uses a phase reference signal to perform a correlation operation with the input PCAL signal to extract the amplitude and phase of the reference frequency channel. However, due to the limitations of the correlation principle, this method requires a correlation operation for each extracted frequency point, resulting in a significant amount of computational redundancy. Furthermore, as the observation bandwidth increases, the number of PCAL signal frequencies also increases, significantly increasing the computational complexity.

[0004] In his paper, "An Efficient Parallel Method for PCAL Signal Multi-Frequency Extraction," Jiang Kun focused on a parallel method for PCAL signal multi-frequency extraction based on FFT (Fast Fourier Transform). This method can obtain the amplitude and additive phase of all frequency points with a single FFT operation, achieving high computational efficiency. However, the FFT operation used in this method consumes a lot of resources when implemented on an FPGA, and FFT operations also have some computational redundancy compared to single-point DFT operations. Summary of the Invention

[0005] This paper proposes an FPGA-based method for efficiently extracting the amplitude and phase of comb spectrum signals. Its primary application is amplitude and phase correction for multi-channel signals in electronic reconnaissance equipment. Because traditional amplitude and phase extraction methods fail to meet the requirements for rapid and efficient amplitude and phase extraction of multi-frequency signals, this paper proposes a DFT-based, parallel algorithm for extracting the amplitude and phase of comb spectrum signals and its FPGA implementation. This algorithm achieves rapid and efficient amplitude and phase extraction of multi-frequency signals while utilizing minimal FPGA resources.

[0006] The technical solution to realize the present invention is: an FPGA-based method for efficiently extracting the amplitude and phase of a comb spectrum signal, with the following steps:

[0007] Step 1, segmented accumulation processing: cyclically accumulate the comb spectrum signal sampled by AD and cache it in RAM to obtain the segmented accumulation result; determine whether the segmented accumulation is completed. If so, go to step 2, otherwise continue to step 1;

[0008] Step 2: Spectrum shifting: Read the frequency shift coefficient, perform complex multiplication on the segmented accumulation result and the frequency shift coefficient to obtain the spectrum shift result, and then proceed to step 3.

[0009] Step 3, DFT calculation processing: Read the DFT coefficients, perform complex multiplication of the spectrum shift result and the DFT coefficients to obtain the DFT calculation result; determine whether the DFT calculation of all frequency points is completed. If so, proceed to step 4, otherwise continue to step 3;

[0010] Step 4: Amplitude and phase calculation processing: Perform modulus and LOG operation on the DFT calculation results to obtain amplitude information; perform inverse tangent operation on the DFT calculation results to obtain phase information, and obtain amplitude and phase extraction results of all frequency points.

[0011] Compared with the existing technology, the present invention has the following significant advantages: 1) It can parallelly extract the amplitude and phase information of 201 frequency points of a comb spectrum signal with a maximum bandwidth of 1000 MHz and an interval of 5 MHz; 2) Amplitude and phase extraction of the same number of frequency points has higher computational efficiency, and the amplitude and phase characteristics of 201 frequency points can be extracted within a processing time of 400 μs; 3) Through time-sharing multiplexing of multipliers and RAM resources, the FPGA design of the entire algorithm module can be completed using only 3 DSP and 7 BRAM FPGA resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 This is a flow chart of the efficient amplitude and phase extraction method of comb spectrum signals based on FPGA of the present invention. DETAILED DESCRIPTION

[0013] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0014] The following will further introduce the specific implementation methods, as well as the technical difficulties and inventive points of this invention in combination with this design example.

[0015] The FPGA-based efficient amplitude and phase extraction method for comb spectrum signals described in the present invention first performs segmented accumulation on a comb spectrum signal containing multiple frequency points. The segmented accumulation improves the signal-to-noise ratio of the comb spectrum signal, thereby improving the accuracy of amplitude and phase extraction. Second, spectrum shifting is performed on the segmented accumulation results. The spectrum shifting shifts the spectrum of the comb spectrum signal to zero frequency, so that each frequency point is located at an integer multiple of the DFT frequency resolution. Third, DFT processing is performed on the spectrum shifted result. The DFT processing obtains the spectrum information of the comb spectrum signal. Finally, the amplitude and phase information of the desired frequency point are extracted from the DFT processing result.

[0016] The invention realizes the efficient extraction of amplitude and phase of comb spectrum signals, and can extract the amplitude and phase information of multiple frequency points in parallel, without the need for amplitude and phase extraction of a single frequency point one by one, and the processing time is significantly reduced. Compared with the traditional correlation method for amplitude and phase extraction, the algorithm proposed in the present invention has higher computational efficiency. At the same time, through the time-sharing multiplexing of multipliers and RAM resources, the FPGA design of the entire algorithm module can be completed using only a small amount of FPGA resources. It is possible to extract the amplitude and phase information of 201 frequency points of the comb spectrum signal with a maximum bandwidth of 1000MHz and an interval of 5MHz in parallel within a processing time of 400us. Through the time-sharing multiplexing of multipliers and RAM resources, the FPGA design of the entire algorithm module can be completed using only 3 DSP and 7 BRAM FPGA resources.

[0017] The difficulties and innovations of the present invention are as follows: 1) combining the input signal-to-noise ratio of the comb spectrum and the requirements for amplitude and phase extraction accuracy, a reasonable range of parameters such as the number of segmented accumulation cycles, accumulation length, and number of DFT points is given; 2) utilizing FPGA resources of 3 DSPs and 7 BRAMs to achieve high-precision extraction of amplitude and phase features of 201 frequency points within 400 μs.

[0018] Combine Figure 1 , an efficient amplitude and phase extraction method for comb spectrum signals based on FPGA, the method comprising the following steps:

[0019] Step 1: Segmented accumulation processing: cyclically accumulate the comb spectrum signal sampled by AD and cache it in RAM to obtain the segmented accumulation result; and determine whether the segmented accumulation is completed:

[0020] Step 11, set the loop depth and minimum data width of the comb spectrum signal buffer RAM. The RAM loop depth is f s f m , where f s is the sampling rate of the comb spectrum signal, f m is the fundamental frequency interval of the comb spectrum signal; the minimum data width of RAM is DW+log2(M), where DW is the data width of the comb spectrum signal and M is the number of segmented accumulations.

[0021] Step 12: The pipeline receives the comb spectrum signal data sampled by AD, accumulates the comb spectrum signal data of a single cycle with the data in the cache RAM, and writes the accumulated result into the cache RAM, thereby realizing single segmented accumulation processing.

[0022] Step 13: Determine whether the segment accumulation process is completed based on the set segment accumulation times. If completed, proceed to step 2; otherwise, continue to step 1.

[0023] Step 2: Spectrum shifting: Read the frequency shift coefficient and perform complex multiplication of the segmented accumulation result and the frequency shift coefficient to obtain the spectrum shift result:

[0024] Step 21: Generate frequency shift coefficients. The coefficient generation formula is exp[-j2π(f offset / f s )n], where f offset is the offset frequency of the comb spectrum signal, n=0,,N-1 is the nth sampling point, and N is the period of a single comb spectrum signal, which is also the number of points in the DFT calculation.

[0025] Step 22: quantize and perform two's complement conversion operations on the frequency shift coefficients, convert the two's complement coefficients into a ROM core coefficient loading file, and add it to the frequency shift coefficient ROM core.

[0026] Step 23: Control the read address of the frequency shift coefficient ROM according to the read enable signal generated at the start time of the spectrum shifting process to read out the frequency shift coefficient.

[0027] Step 24: After the segmented accumulation process is completed, the N-point segmented accumulation results are stored in different RAMs in multiple phases, and the N-point segmented accumulation results are read out from these RAMs in sequence.

[0028] Step 25: Perform a complex multiplication operation on the read segment accumulation result and the frequency shift coefficient to complete the spectrum shifting process, and then write the frequency shift result into the buffer RAM.

[0029] The mathematical formula for spectrum shifting is s′(n)=s(n)·exp[-j2π(f offset / f s )n], where s(n) is the signal after segment accumulation and s′(n) is the signal after spectrum shifting.

[0030] Go to step 3.

[0031] Step 3: DFT calculation and processing: Read the DFT coefficients, perform complex multiplication of the spectrum shift result and the DFT coefficients to obtain the DFT calculation result:

[0032] Step 31: Generate DFT coefficients. The coefficient generation formula is: Wherein, k=0, K-1 is the kth DFT frequency point, and K is the number of comb spectrum signal frequency points.

[0033] Step 32: quantize and perform two's complement conversion on the DFT coefficients of the first frequency point (k=1), convert the two's complement coefficients into a ROM core coefficient loading file, and add the file to the DFT coefficient ROM core.

[0034] Step 33: Coefficient extraction can be achieved by controlling the read address stepping of the DFT coefficient ROM, thereby reading out the DFT coefficients of K frequency points.

[0035] Considering the mutual correlation between the DFT coefficients of different frequency points, the DFT coefficients of the first frequency point are decimated by 0 to K-1 times respectively to obtain the DFT coefficients of K frequency points. When this method is implemented in FPGA, only the DFT coefficients of the first frequency point need to be stored, saving a lot of BRAM resources.

[0036] Step 34: Read the N-point spectrum shifting results from the frequency shift result RAM for K cycles, and read the N-point DFT coefficients from the DFT coefficient ROM for K cycles for complex multiplication. This complex multiplier is time-multiplexed with the spectrum shifting multiplier.

[0037] Step 35: Accumulate the results of the complex multiplication at each N point to obtain the DFT result of a single frequency point. Perform K N-point complex multiplication and accumulation operations to obtain the DFT results of K frequency points.

[0038] The mathematical formula for DFT calculation is Wherein, S(k) is the DFT calculation result of the k-th frequency point.

[0039] Step 36: The real part and the imaginary part of the DFT calculation result are stored in RAM respectively. The DFT calculation result buffer RAM can be time-shared with the segmented accumulation result buffer RAM.

[0040] Step 37: After the DFT calculation of a single frequency point is completed, determine whether the DFT calculation of K frequency points is completed. If completed, go to step 4; otherwise, switch to the next frequency point and re-execute step 3.

[0041] Step 4: Amplitude and phase calculation: Perform modulus calculation and LOG operation on the DFT calculation results to obtain amplitude information; perform inverse tangent operation on the DFT calculation results to obtain phase information, and obtain the amplitude and phase extraction results of all frequency points:

[0042] Step 41: After completing the DFT calculation of K frequency points, read the DFT calculation results from the RAM;

[0043] Step 42: Call the Cordic kernel to perform an inverse tangent operation on the DFT calculation result to obtain phase information, and write the phase result into RAM.

[0044] Step 43: Perform modulus and LOG operations on the DFT calculation results to obtain amplitude information, and write the amplitude results to RAM. To save FPGA resources, the squaring operation of the amplitude calculation continues to reuse the complex multiplier, and the LOG operation is implemented using a lookup table. The amplitude and phase result cache RAM and the spectrum shift cache RAM are time-shared.

[0045] Step 44: The read enable and read address input signals of the module input port are used to control the read enable and read address of the amplitude and phase result cache RAM, so that the amplitude and phase data of all frequency points can be read out and output to the module port.

[0046] Through the above method, the amplitude and phase information of the comb spectrum signal can be extracted in parallel and efficiently, multi-channel rapid amplitude and phase correction of the equipment can be achieved, and the working capability and efficiency of the electronic reconnaissance equipment can be improved.

[0047] Examples:

[0048] Suppose a device includes a multi-channel amplitude and phase real-time correction FPGA and other programs, with an AD sampling rate of 2.4GHz, a bit width of 16bit, AD data divided into 8 phases, a data rate of 300M per phase, a comb spectrum signal band range of 1.3~2.3GHz, a bandwidth of 1GHz, and a baseband interval of 5MHz. In order to achieve a comb spectrum signal signal-to-noise ratio of 10dB, the amplitude extraction accuracy is 0.1dBm, the phase extraction accuracy is 1°, and the number of segmented integrations is set to 512. This multi-channel amplitude and phase real-time correction FPGA enters the FPGA-based comb spectrum signal amplitude and phase efficient extraction method disclosed in this patent, and its specific steps are as follows:

[0049] Step 1, segmented accumulation processing: cyclically accumulate the comb spectrum signal sampled by AD and cache it in RAM to obtain the segmented accumulation result; determine whether the segmented accumulation is completed. If so, go to step 2, otherwise continue to execute step 1.

[0050] Step 11: Set the loop depth and minimum data width of the comb spectrum signal buffer RAM. The RAM loop depth is 2.4 GHz / 5 MHz = 480 points; the RAM minimum data width is 16 + log2 (512) = 25 bits.

[0051] Step 12: The pipeline receives the comb spectrum signal data sampled by AD, accumulates the 8-phase comb spectrum signal data of a single cycle with the data in 8 cache RAMs (RAM_0 to RAM_7), and writes the accumulated results into the cache RAM to achieve single segmented accumulation processing.

[0052] Step 13: Determine whether 512 segment accumulations have been completed. If so, proceed to step 2; otherwise, continue to step 1.

[0053] Step 2: Spectrum shifting processing: Read the frequency shift coefficient, perform complex multiplication on the segmented accumulation result and the frequency shift coefficient to obtain the spectrum shift result, and then proceed to step 3.

[0054] Step 21: Generate frequency shift coefficients. The coefficient generation formula is exp[-j2π(f offset / f s )n], where n=0,…,N-1 is the nth sampling point, and the offset frequency f offset The frequency is 1.3 GHz, and the period N of the comb spectrum signal is 480;

[0055] Step 22: quantize and perform two's complement conversion operations on the frequency shift coefficients, convert the two's complement coefficients into a ROM core coefficient loading file, and add it to the frequency shift coefficient ROM core.

[0056] Step 23: Control the read address of the frequency shift coefficient ROM according to the read enable signal generated at the start time of the spectrum shifting process to read out the frequency shift coefficient.

[0057] Step 24: After the segmented accumulation process is completed, the 480-point segmented accumulation results are stored in 8 phases in RAM_0 to RAM_7 respectively, and the segmented accumulation results are read out from the 8 RAMs in sequence.

[0058] Step 25: Perform a complex multiplication operation on the read segmented accumulation result and the frequency shift coefficient to complete the spectrum shifting process, and then write the real and imaginary parts of the spectrum shifting result into RAM_8 and RAM_9 respectively;

[0059] Step 3, DFT calculation processing: Read the DFT coefficients, perform complex multiplication of the spectrum shift result and the DFT coefficients to obtain the DFT calculation result; determine whether the DFT calculation of all frequency points is completed. If so, go to step 4, otherwise continue to step 3.

[0060] Step 31: Generate DFT coefficients. The coefficient generation formula is: Where k = 0, K-1 is the kth DFT frequency point, K is the number of comb spectrum signal frequency points, which is 1GHz / 5MHz+1=201;

[0061] Step 32: quantize and perform two's complement conversion on the DFT coefficients of the first frequency point (k=1), convert the two's complement coefficients into a ROM core coefficient loading file, and add the file to the DFT coefficient ROM core.

[0062] Step 33: Coefficient extraction can be achieved by controlling the read address stepping of the DFT coefficient ROM, thereby reading out the DFT coefficients of K frequency points.

[0063] Considering the mutual correlation between the DFT coefficients of different frequency points, the DFT coefficients of the first frequency point are decimated by 0 to 200 times respectively to obtain the DFT coefficients of 201 frequency points.

[0064] Step 34: Read the 480-point spectrum shifting result from RAM_8 and RAM_9 201 times in a loop, and read the 480-point DFT coefficients from the DFT coefficient ROM 201 times in a loop for complex multiplication. This complex multiplier is time-multiplexed with the spectrum shifting multiplier.

[0065] Step 35: Accumulate the results of every 480-point complex multiplication to obtain the DFT result of a single frequency point. Perform 201 480-point complex multiplication and accumulation operations to obtain the DFT results of 201 frequency points.

[0066] Step 36: The real part and the imaginary part of the DFT calculation result are stored in RAM_0 and RAM_1 respectively. The DFT calculation result buffer RAM and the segment accumulation result buffer RAM are time-shared multiplexed.

[0067] Step 37: After the DFT calculation of a single frequency point is completed, determine whether the DFT calculation of 201 frequency points is completed. If completed, go to step 6; otherwise, switch to the next frequency point and go to step 4.

[0068] Step 4: Amplitude and phase calculation processing: Perform modulus and LOG operation on the DFT calculation results to obtain amplitude information; perform inverse tangent operation on the DFT calculation results to obtain phase information, and obtain amplitude and phase extraction results of all frequency points;

[0069] Step 41: After completing the DFT calculation of 201 frequency points, read the DFT calculation results from RAM_0 and RAM_1.

[0070] Step 42: Call the Cordic kernel to perform an inverse tangent operation on the DFT calculation result to obtain phase information, and write the phase result to RAM_9.

[0071] Step 43: Perform modulus and LOG operations on the DFT calculation results to obtain amplitude information, and write the amplitude results to RAM_9. To save FPGA resources, the squaring operation of the amplitude calculation continues to reuse the complex multiplier, and the LOG operation is implemented using a lookup table. The amplitude and phase result cache RAM and the spectrum shift cache RAM are time-shared.

[0072] Step 44: The read enable and read address input signals of the module input port are used to control the read enable and read address of the amplitude and phase result cache RAM, so that the amplitude and phase data of all frequency points can be read out and output to the module port.

[0073] The entire algorithm module consumes only a small amount of FPGA resources, such as 7 BRAMs and 3 DSPs. The processing delay for efficient amplitude and phase extraction of 201 frequency points within the 1GHz band is approximately 430us (300M working clock).

Claims

1. An efficient method for extracting amplitude and phase of comb spectrum signals based on FPGA, characterized in that: The following steps are involved: Step 1, segmented accumulation processing: cyclically accumulate the comb spectrum signal sampled by AD and cache it in RAM to obtain the segmented accumulation result; determine whether the segmented accumulation is completed. If so, go to step 2, otherwise continue to step 1; Step 2: Spectrum shifting: Read the frequency shift coefficient, perform complex multiplication on the segmented accumulation result and the frequency shift coefficient to obtain the spectrum shift result, and then proceed to step 3. Step 3, DFT calculation processing: Read the DFT coefficients, perform complex multiplication of the spectrum shift result and the DFT coefficients to obtain the DFT calculation result; determine whether the DFT calculation of all frequency points is completed. If so, proceed to step 4, otherwise continue to step 3; Step 4: Amplitude and phase calculation processing: Perform modulus and LOG operation on the DFT calculation results to obtain amplitude information; perform inverse tangent operation on the DFT calculation results to obtain phase information, and obtain amplitude and phase extraction results of all frequency points.

2. The FPGA-based efficient amplitude and phase extraction method for comb spectrum signals according to claim 1, characterized in that: The segmented accumulation process in step 1 is as follows: Step 11: Set the loop depth and minimum data width of the comb spectrum signal buffer RAM. The RAM loop depth is f s / f m , where f s is the sampling rate of the comb spectrum signal, f m is the fundamental frequency interval of the comb spectrum signal; the minimum data width of RAM is DW+log2(M), where DW is the data width of the comb spectrum signal and M is the number of segmented accumulations; Step 12: The pipeline receives the comb spectrum signal data sampled by AD, accumulates the comb spectrum signal data of a single cycle with the data in the cache RAM, and writes the accumulated result to the cache RAM, thus realizing single segment accumulation processing; Step 13: Determine whether the segment accumulation process is completed based on the set segment accumulation times. If completed, proceed to step 2; otherwise, continue to step 1.

3. The FPGA-based efficient amplitude and phase extraction method for comb spectrum signals according to claim 2, characterized in that: The frequency shift processing in step 2 is as follows: Step 21: Generate frequency shift coefficients. The frequency shift coefficient generation formula is exp[-j2π(f offset / f s )n], where f offset is the offset frequency of the comb spectrum signal, n=0,…,N-1 is the nth sampling point, N is the period of a single comb spectrum signal, which is also the number of points in the DFT calculation; j is a complex number; Step 22: quantize and perform two's complement conversion on the frequency shift coefficients, convert the two's complement coefficients into a ROM core coefficient loading file, and add the file to the frequency shift coefficient ROM core. Step 23: Control the read address of the frequency shift coefficient ROM according to the read enable signal generated at the start time of the spectrum shifting process to read out the frequency shift coefficient; Step 24: After the segmented accumulation process is completed, the N-point segmented accumulation results are stored in different RAMs in multiple phases, and the N-point segmented accumulation results are read out from these RAMs in sequence. Step 25: Perform a complex multiplication operation on the read segment accumulation result and the frequency shift coefficient to complete the spectrum shifting process, and then write the frequency shift result into the buffer RAM; The mathematical formula for spectrum shifting is s′(n)=s(n)·exp[-j2π(f offset / f s )n], where s(n) is the signal after segment accumulation and s′(n) is the signal after spectrum shifting.

4. The FPGA-based efficient amplitude and phase extraction method for comb spectrum signals according to claim 3, characterized in that: The DFT calculation process in step 3 is as follows: Step 31: Generate DFT coefficients. The DFT coefficient generation formula is: Wherein, k=0,…,K-1 is the kth DFT frequency point, and K is the number of comb spectrum signal frequency points; Step 32: quantize and perform two's complement conversion on the DFT coefficients of the first frequency point, convert the two's complement coefficients into a ROM core coefficient loading file, and add the file to the DFT coefficient ROM core. Step 33: Coefficient extraction can be achieved by controlling the read address stepping of the DFT coefficient ROM, thereby reading the DFT coefficients of K frequency points; Considering the mutual correlation between the DFT coefficients of different frequency points, the DFT coefficients of the first frequency point are extracted from 0 to K-1 cycles to obtain the DFT coefficients of K frequency points; Step 34: Read the N-point spectrum shift results from the frequency shift result RAM for K cycles, and read N points from the DFT coefficient ROM for K cycles: The DFT coefficients are complex multiplied, and this complex multiplier is time-multiplexed with the spectrum shifting multiplier; Step 35: Accumulate the results of the complex multiplication at every N points to obtain the DFT result of a single frequency point. Perform K N-point complex multiplication and accumulation operations to obtain the DFT results of K frequency points. The mathematical formula for DFT calculation is Where S(k) is the DFT calculation result of the kth frequency point; s′(n) is the signal after spectrum shifting; Step 36: The real part and the imaginary part of the DFT calculation result are stored in RAM respectively, and the DFT calculation result buffer RAM and the segmented accumulation result buffer RAM are time-shared multiplexed; Step 37: After the DFT calculation of a single frequency point is completed, determine whether the DFT calculation of K frequency points is completed. If completed, go to step 4; otherwise, switch to the next frequency point and re-execute step 3.

5. The FPGA-based efficient method for extracting amplitude and phase of comb spectrum signals according to claim 4, characterized in that: The amplitude and phase calculation in step 4 is as follows: Step 41: After completing the DFT calculation of K frequency points, read the DFT calculation results from the RAM; Step 42: Call the Cordic kernel to perform an inverse tangent operation on the DFT calculation result to obtain phase information, and write the phase result to RAM; Step 43: Perform modulus and LOG operations on the DFT calculation results to obtain amplitude information, and write the amplitude results to RAM. To save FPGA resources, the squaring operation of the amplitude calculation continues to reuse the complex multiplier, and the LOG operation is implemented using a lookup table. The amplitude and phase result buffer RAM and the spectrum shift buffer RAM are time-shared. Step 44: Control the read enable and read address of the amplitude and phase result cache RAM through the read enable and read address input signals of the module input port, thereby reading the amplitude and phase data of all frequency points and outputting them to the module port.

Images