Radio frequency signal high-precision distribution method based on FPGA

Abstract

The invention discloses a radio frequency signal high-precision distribution method based on an FPGA, and relates to the field of signal processing, and the method comprises the steps: S1, enabling the working composition of a target FPGA platform to be divided into a preprocessing module, a distribution monitoring module and an error code correction module which are in mutual signal connection; s2, using a mean drift optimization method in the distribution monitoring module to judge whether signal crosstalk exists in the distribution radio frequency signals on each branch; s3, inputting the distribution radio frequency signal of the target branch to an error code correction module for bit correction; and S4, parameter configuration of signal distribution is carried out again, and the corrected distribution configuration is maintained to be updated. Compared with the prior art, the signal crosstalk in the distributed radio-frequency signals is detected in real time through the distribution monitoring module, bit correction is carried out, the advantages of high-precision signal distribution and processing are achieved, bit errors are recognized and corrected in time, and the bit error rate is reduced.

Description

Technical Field

[0001] The present invention relates to the field of signal processing, and in particular to a high-precision distribution method of radio frequency signals based on FPGA. Background Art

[0002] In modern communication systems, FPGA's high concurrent processing capability and flexible logic configuration make it an ideal choice for RF signal processing. However, with the increase in signal frequency and data transmission speed, signal crosstalk has become a major technical problem. Signal crosstalk refers to the interference of a signal channel on the adjacent channel signal, which usually originates from electromagnetic interference and electrical coupling effects. When multiple signal channels in an FPGA platform work in parallel at high frequency or high speed, the electromagnetic field interaction between signal lines may cause crosstalk.

[0003] In the process of distributing RF signals, signal crosstalk may cause data bit errors. The signal distortion caused by crosstalk will affect the accuracy of data decoding at the receiving end, increase the bit error rate, and even increase the bit BER value. This error makes the received data inconsistent with the sent data, affecting the integrity of the data. In the process of using the FPGA platform for RF signal distribution, the signal crosstalk during signal distribution is mainly manifested as rapidly changing transient interference. Due to the real-time data sampling rate and data bandwidth resource limitations of the FPGA platform itself, and the rapid occurrence of signal crosstalk, the FPGA distribution system in the existing technology that increases the bit error rate due to transient signal crosstalk lacks a real-time and fast bit correction mechanism, resulting in the difficulty in timely control of the number of bit errors in the distribution branch after the FPGA platform distributes the signal. Due to the continuity of the signal distribution process, there is a problem of insufficient timely correction of bit errors during signal distribution, resulting in data loss or information distortion in the distributed RF signal, resulting in the risk of reduced RF signal data integrity. Summary of the invention

[0004] The present invention provides a high-precision radio frequency signal distribution method based on FPGA, which solves the problem in the prior art that after using an FPGA platform for signal distribution, a bit correction process is not timely when the number of bit errors of the radio frequency signal increases.

[0005] The present invention is achieved through the following technical solutions: A high-precision radio frequency signal distribution method based on FPGA, the method comprising: Step S1: presetting the original RF signal for signal distribution and setting the parameters as initial signal parameters, setting the workgroup components of the target FPGA platform to the preprocessing module, the distribution monitoring module and the error correction module which are all signal-connected to each other, applying the initial signal parameters to the distribution monitoring module to perform parameter configuration of signal distribution and marking it as initial distribution configuration; Step S2: After the original RF signal is input into the preprocessing module and preprocessed, the original RF signal is transferred to the distribution monitoring module for signal distribution and output as distributed RF signals respectively, and in the distribution monitoring module, a mean shift optimization method is used to determine whether there is signal crosstalk in the distributed RF signals on each branch; Step S3: When the distribution monitoring module determines that there is no signal crosstalk in the branch, the distribution RF signal of the branch completes normal signal output; when the distribution monitoring module determines that there is signal crosstalk in the branch, the distribution RF signal of the target branch is input into the error code correction module for bit correction, and the parameters of the corrected distribution RF signal are set as the correction signal parameters; Step S4: Set a weighted allocation strategy, merge all corrected signal parameters into corrected total parameters through the weighted allocation strategy, apply the corrected total parameters to the distribution monitoring module to reconfigure the signal distribution parameters and mark them as corrected distribution configurations, return to step S2 to perform bit correction on the newly generated distribution RF signal again to keep the corrected distribution configuration updated until the original RF signal completes the distribution process or there is no new signal crosstalk in the distribution process of the target FPGA platform.

[0006] In the process of using the FPGA platform to distribute radio frequency signals, the signal crosstalk during signal distribution is mainly manifested as rapidly changing transient interference. Due to the real-time data sampling rate and data bandwidth resource limitations of the FPGA platform itself, and the rapid occurrence of signal crosstalk, the FPGA distribution system part of the prior art that causes the bit error rate to increase due to transient signal crosstalk lacks a real-time and fast bit correction mechanism, resulting in the difficulty in timely control of the number of bit errors in the distribution branch after the FPGA platform distributes the signal, and due to the continuity of the signal distribution process, the problem of insufficient timely correction of bit errors during signal distribution exists, resulting in the possibility of data loss or information distortion in the distributed radio frequency signal, resulting in the risk of reduced data integrity of the radio frequency signal. Based on this, the present invention provides a high-precision distribution method for radio frequency signals based on FPGA, which solves the problem of untimely bit correction process when the bit errors of the radio frequency signal increase after the FPGA platform is used for signal distribution in the prior art.

[0007] Furthermore, the specific process of the mean shift optimization method includes: Step A1: collecting initial signal data of the original radio frequency signal from the distribution monitoring module as an input value, presetting a Gaussian kernel function for calculating the mean value of the initial signal data, and setting a smoothing function for the Gaussian kernel function to limit the range of the Gaussian kernel function; Step A2: All feature values ​​in the original data are set as data points, one group of data points is set as the initial drift point, the number of iterations is preset for the drift point, and the weighted mean around each drift point is calculated using the Gaussian kernel function in each iteration; Step A3: iterate the initial drift point, and the process of moving the initial drift point to the calculated weighted mean position represents iterative update. When the number of iterations is reached or the initial drift point reaches a convergence state, the iteration is stopped and the final drift point is output; Step A4: outputting the initial signal data including all final drift points as distribution signal data representing distribution radio frequency signal parameters, and setting a crosstalk threshold for each identical signal parameter of the distribution signal data and the initial signal data; Step A5: When there is at least one signal parameter difference exceeding the crosstalk threshold, it is determined that signal crosstalk exists; when there is no signal parameter difference exceeding the crosstalk threshold, it is determined that the distributed RF signal has no signal crosstalk.

[0008] Furthermore, a mean change criterion is set to represent a drift point convergence condition used when the initial drift point reaches a convergence state; the content of the mean change criterion includes: Step B1: Mark the initial position for each initial drift point, mark the distance between the initial drift point position and the initial position after each iteration, and calculate the position weight value of the initial drift point of the current iteration to the initial position based on the Gaussian kernel function; Step B2: after each iteration, the weighted mean of the initial drift point is calculated using the corresponding position weight value, and the weighted mean is used to represent the adjusted position of the initial drift point at the end of the current iteration and is marked as the updated position; Step B3: preset a convergence judgment value, calculate the position change between the initial position and the updated position, compare the position change with the convergence judgment value, and when the position change is less than the convergence judgment value, it is determined that the initial drift point has converged; when the position change is greater than the convergence judgment value, it is determined that the initial drift point has not converged.

[0009] Furthermore, the weighted mean calculation process includes: assuming that the data point of the current initial drift point is represented by x, the position weight value is represented by ω, and the weighted mean is represented by x a , Then the weighted mean x a The calculation formula is: , The weighted mean x a represents the weighted average of all data points x in the initial drift point, and the position weight value ω is calculated by the Gaussian kernel function; Represents the sum of all data points x in the current initial drift point multiplied by the corresponding position weight value ω, The sum of all position weights ω.

[0010] Furthermore, the process of calculating the position weight value based on the Gaussian kernel function includes: assuming that the initial position and the updated position are represented by i and j respectively, the data point x at the initial position is further defined as the initial data point x i , the updated position of the data point x is further defined as the updated data point x j ; Let the distance between the initial drift point position and the initial position after each iteration be d ij , let the smoothing function of the Gaussian kernel function be represented by h, and the position weight value be represented by ω ij , Then the position weight value ω ij The calculation formula is expressed as: , Wherein the position weight value ω ij Indicates updating data point x j For the initial data point x i The smoothing function h is used to control the width of the Gaussian kernel function.

[0011] Furthermore, the error correction module uses a convolutional code forward error correction method to perform bit correction, and the process includes: presetting the coding rate of the convolutional code and the constraint length of the convolutional code, inputting the initial signal parameters as a bit stream into the convolutional code to generate output bit data, using soft decision decoding to decode and output the output bit data and correct the erroneous data therein, and setting the decoded output signal after the erroneous data is corrected as the corrected distribution RF signal.

[0012] Furthermore, the soft decision decoding uses the Viterbi algorithm to decode the output bit data, and its process includes: defining the initialized encoder parameters, demodulating the received output bit data and calculating the likelihood, using the Viterbi algorithm to update the path metric matrix to select the optimal backtracking path, and finally restoring the bit data to the original bit stream based on the optimal backtracking path.

[0013] Furthermore, the convolutional code has a coding rate limit of generating at least two output bits and at most four output bits for each input bit.

[0014] Furthermore, the path metric matrix is ​​calculated using the Hamming distance; when the path metric matrix is ​​updated, the predecessor state with the smallest path metric is selected.

[0015] Furthermore, the content of the weighted allocation strategy includes: setting a signal performance value to represent the indicator size of the impact of each corrected distributed RF signal on the performance of the target FPGA platform; the signal quality weighting process is to allocate a weight value according to the signal performance value of each corrected distributed RF signal.

[0016] Compared with the prior art, the present invention detects signal crosstalk in distributed radio frequency signals in real time and performs bit correction through a distribution monitoring module, and reconfigures the distribution parameters of the FPGA platform after each correction, thereby ensuring that the bit correction process always keeps up with the actual distribution situation. It has the advantages of high-precision signal distribution and processing and the beneficial effects of timely identifying and correcting bit errors and reducing the bit error rate during the signal distribution process. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings: Figure 1 It is a structural flow chart of the present invention. DETAILED DESCRIPTION

[0018] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention. Example

[0019] like Figure 1 As shown, this embodiment is a high-precision distribution method of radio frequency signals based on FPGA, and the method includes: Step S1: presetting the original RF signal for signal distribution and setting the parameters as initial signal parameters, setting the workgroup components of the target FPGA platform to the preprocessing module, the distribution monitoring module and the error correction module which are all signal-connected to each other, applying the initial signal parameters to the distribution monitoring module to perform parameter configuration of signal distribution and marking it as initial distribution configuration; Step S2: After the original RF signal is input into the preprocessing module and preprocessed, the original RF signal is transferred to the distribution monitoring module for signal distribution and output as distributed RF signals respectively, and in the distribution monitoring module, a mean shift optimization method is used to determine whether there is signal crosstalk in the distributed RF signals on each branch; Step S3: When the distribution monitoring module determines that there is no signal crosstalk in the branch, the distribution RF signal of the branch completes normal signal output; when the distribution monitoring module determines that there is signal crosstalk in the branch, the distribution RF signal of the target branch is input into the error code correction module for bit correction, and the parameters of the corrected distribution RF signal are set as the correction signal parameters; Step S4: Set a weighted allocation strategy, merge all corrected signal parameters into corrected total parameters through the weighted allocation strategy, apply the corrected total parameters to the distribution monitoring module to reconfigure the signal distribution parameters and mark them as corrected distribution configurations, return to step S2 to perform bit correction on the newly generated distribution RF signal again to keep the corrected distribution configuration updated until the original RF signal completes the distribution process or there is no new signal crosstalk in the distribution process of the target FPGA platform.

[0020] The original RF signal used for signal distribution refers to the basic RF signal used as input in the distribution process. These signals are usually generated, modulated or amplified to meet the needs of specific applications. The original RF signal contains the signal information that needs to be distributed. It will serve as the starting point of the system and undergo preprocessing, distribution and correction steps on the FPGA platform to achieve a high-precision signal distribution effect. The initial signal parameters are the basic parameters of the RF signal in the conventional technology contained in the original RF signal, including frequency, amplitude, phase, waveform, modulation mode and other information. The working composition of the target FPGA platform refers to the various functional modules or components that constitute the FPGA platform and support its work. The mutual signal connection of the preprocessing module, the distribution monitoring module and the error code correction module mainly adopts the data bus and interface protocol to cooperate in operation. The parameter configuration of the FPGA platform for signal distribution in the prior art usually involves frequency setting, gain control, phase modulation, distribution path selection, signal synchronization control and other parameters, that is, the initial distribution configuration and the modified distribution configuration are also the same as the options of the existing signal distribution parameter configuration, and the difference lies in the specific setting of the parameter size. The preprocessing module includes signal amplification, filtering and shaping, and sampling and quantization parts, which are used to improve the quality of the signal and provide a good foundation for subsequent signal distribution and correction. The above implementation method can effectively reduce signal crosstalk and improve distribution accuracy through dynamic adjustment and error code correction.

[0021] Furthermore, as a feasible implementation method, the specific process of the mean shift optimization method includes: Step A1: collecting initial signal data of the original radio frequency signal from the distribution monitoring module as an input value, presetting a Gaussian kernel function for calculating the mean value of the initial signal data, and setting a smoothing function for the Gaussian kernel function to limit the range of the Gaussian kernel function; Step A2: All feature values ​​in the original data are set as data points, one group of data points is set as the initial drift point, the number of iterations is preset for the drift point, and the weighted mean around each drift point is calculated using the Gaussian kernel function in each iteration; Step A3: iterate the initial drift point, and the process of moving the initial drift point to the calculated weighted mean position represents iterative update. When the number of iterations is reached or the initial drift point reaches a convergence state, the iteration is stopped and the final drift point is output; Step A4: outputting the initial signal data including all final drift points as distribution signal data representing distribution radio frequency signal parameters, and setting a crosstalk threshold for each identical signal parameter of the distribution signal data and the initial signal data; Step A5: When there is at least one signal parameter difference exceeding the crosstalk threshold, it is determined that signal crosstalk exists; when there is no signal parameter difference exceeding the crosstalk threshold, it is determined that the distributed RF signal has no signal crosstalk.

[0022] The Gaussian kernel function is a mathematical function used for data smoothing and feature weighting. It is widely used in kernel methods in machine learning and signal processing. Its core function is to improve the effect of data processing and analysis through smoothing and weighting. The smoothing function is usually related to the standard deviation in the kernel function, and its function is to limit the influence range of the kernel function. The eigenvalue in the original data refers to the key parameters or measurement results in the original RF signal data, which are used for mean shift optimization. In the mean shift algorithm, a data point refers to an observation value or eigenvalue in the data space, usually represented by a vector. Each data point consists of one or more eigenvalues, representing the values ​​of the data in different dimensions. In the mean calculation process, the drift direction of the data point is determined by calculating the weighted mean of the data point, and its position is gradually adjusted to find the high-density area of ​​the data. The weighted mean refers to the different contribution of different data points when calculating the mean around each data point, and this contribution is determined by the weight; the Gaussian kernel function is used to assign weights to each data point, and the weight is related to the distance from the data point to the target drift point. The crosstalk threshold represents the critical value of the difference between the distribution signal data and the initial signal data. The mean shift optimization reduces signal crosstalk, improves the accuracy of signal distribution, and ensures the timeliness and effectiveness of bit correction.

[0023] Further, as a feasible implementation manner, a mean change criterion is set to represent the drift point convergence condition used when the initial drift point reaches a convergence state; the content of the mean change criterion includes: Step B1: Mark the initial position for each initial drift point, mark the distance between the initial drift point position and the initial position after each iteration, and calculate the position weight value of the initial drift point of the current iteration to the initial position based on the Gaussian kernel function; Step B2: after each iteration, the weighted mean of the initial drift point is calculated using the corresponding position weight value, and the weighted mean is used to represent the adjusted position of the initial drift point at the end of the current iteration and is marked as the updated position; Step B3: preset a convergence judgment value, calculate the position change between the initial position and the updated position, compare the position change with the convergence judgment value, and when the position change is less than the convergence judgment value, it is determined that the initial drift point has converged; when the position change is greater than the convergence judgment value, it is determined that the initial drift point has not converged.

[0024] In the working principle of using the mean change criterion to indicate that the initial drift point reaches a convergence state, the mean change criterion is used to determine whether the iterative process of the drift point reaches a stable state. The position weight value refers to the degree of influence of the initial drift point on the initial position when calculating the weighted mean of the drift point. The Gaussian kernel function reflects the contribution of the data point at the current position through a given weight value. The Gaussian kernel function is used for smoothing and weighted processing. It assigns a weight to each data point according to the distance between the data point and the target drift point. Data points with closer distances will have larger weights, while data points with farther distances will have smaller weights. The weighted mean is used to smooth the data, and the position of the drift point is adjusted by the weight value determined by the Gaussian kernel function to reduce the error and noise generated during the iteration process. There are many ways to set the convergence judgment value, which mainly depends on the specific requirements of the optimization task and the system characteristics. In the specific implementation, as a preferred embodiment, the size of the convergence judgment value can be set by a fixed threshold, a first change rate or an empirical rule. The mean change criterion ensures the convergence and stability of the optimization process by dynamically tracking the initial drift point position and calculating the weighted mean.

[0025] Further, as a feasible implementation method, the weighted mean calculation process includes: assuming that the data point of the current initial drift point is represented by x, the position weight value is represented by ω, and the weighted mean is represented by x a , Then the weighted mean x a The calculation formula is: , The weighted mean x a represents the weighted average of all data points x in the initial drift point, and the position weight value ω is calculated by the Gaussian kernel function; Represents the sum of all data points x in the current initial drift point multiplied by the corresponding position weight value ω, The sum of all position weights ω.

[0026] The weight value ω is calculated by the Gaussian kernel function. This function takes into account the distance from the data point to the initial drift point. The closer the distance, the greater the weight. The weight product represents the contribution of each data point in the weighted calculation, and the weighted sum represents the sum of the weighted products of all data points to obtain the total weighted sum. The weighted mean calculation process uses the weight value calculated by the Gaussian kernel function and combines it with the value of each data point to obtain a mean that comprehensively reflects the distribution of the data points. This method can effectively smooth data, reduce noise, and improve the accuracy of the drift point position, thereby improving the effect and stability of the optimization process.

[0027] Further, as a feasible implementation, the process of calculating the position weight value based on the Gaussian kernel function includes: assuming that the initial position and the updated position are represented by i and j respectively, the data point x at the initial position is further defined as the initial data point x i , the updated position of the data point x is further defined as the updated data point x j ; Let the distance between the initial drift point position and the initial position after each iteration be d ij , let the smoothing function of the Gaussian kernel function be represented by h, and the position weight value be represented by ω ij , Then the position weight value ω ij The calculation formula is expressed as: , Wherein the position weight value ω ij Indicates updating data point x j For the initial data point x i The smoothing function h is used to control the width of the Gaussian kernel function.

[0028] The smoothing function h determines the width of the Gaussian kernel function and affects the distribution of the position weight value. A larger h value will cause the weight value to decay more slowly, and more neighboring data points will be given higher weights; a smaller h value will cause the weight value to decay faster, and the weights of neighboring data points will be higher, while the influence of distant data points will be smaller. The calculation formula of the Gaussian kernel function is used to determine the weight of each data point. The weight value reflects the degree of influence of the data point on the initial drift point. The weight value is an exponential function, which increases with the distance d. ij It decreases with the increase of , which reflects the influence of the distance of the data point from the initial drift point on the weight.

[0029] Furthermore, the error correction module uses a convolutional code forward error correction method to perform bit correction, and the process includes: presetting the coding rate of the convolutional code and the constraint length of the convolutional code, inputting the initial signal parameters as a bit stream into the convolutional code to generate output bit data, using soft decision decoding to decode and output the output bit data and correct the error data therein, and setting the decoded output signal after the error data is corrected as the corrected distribution RF signal. The process of calculating the position weight value based on the Gaussian kernel function is performed by setting the distance d ij And the smoothing function h effectively measures and adjusts the impact of the data point on the initial drift point. ij It is calculated through the Gaussian kernel function, which reflects the influence of the distance of the data point from the initial drift point on the weight, thereby providing an accurate degree of influence in the weighted mean calculation.

[0030] Furthermore, as a feasible implementation method, the soft decision decoding uses the Viterbi algorithm to decode the output bit data, and its process includes: defining the initialized encoder parameters, demodulating the received output bit data and calculating the likelihood, using the Viterbi algorithm to update the path metric matrix to select the optimal backtracking path, and finally restoring the bit data to the original bit stream based on the optimal backtracking path; the convolutional code coding rate is limited to generate at least two output bits and at most four output bits for each input bit; the path metric matrix is ​​calculated using the Hamming distance; when updating the path metric matrix, the predecessor state with the smallest path metric is selected.

[0031] During the decoding process of the Viterbi algorithm, the state and path metric matrix of the encoder need to be initialized. Encoder parameters include state transition diagram, codeword dictionary, and state transition probability. During demodulation, the received bit data is processed by the demodulator to obtain the bit value or probability of the received signal; the likelihood calculation calculates the likelihood between the received bit data and each possible coded bit sequence, which usually involves calculating the matching degree of each received bit with the corresponding coded bit. The path metric matrix is ​​used to record the minimum path metric to reach each state, and each matrix element represents the cost of the path to a certain state; when updating, the predecessor state with the minimum path metric is selected to update the path metric of the current state. This helps to find the minimum cost path to each state. The Hamming distance is used to measure the different number of bits between two bit sequences and is suitable for error detection and correction. At the end of bit recovery, the coded bit stream is converted back to the original bit stream according to the selected optimal backtracking path. The Viterbi algorithm optimizes the path metric through dynamic programming to ensure the accuracy and efficiency of decoding. In this way, the original bit stream can be effectively recovered from the received bits affected by noise.

[0032] Furthermore, as a feasible implementation method, the content of the weighted allocation strategy includes: setting a signal performance value to represent the indicator size of the impact of each corrected distributed RF signal on the performance of the target FPGA platform; the signal quality weighting process is to allocate a weight value according to the signal performance value of each corrected distributed RF signal.

[0033] The signal performance value is an indicator that indicates the impact of each corrected distributed RF signal on the performance of the target FPGA platform. This can be an indicator of signal quality, stability, bit error rate (BER), power efficiency, etc., which is used to evaluate the impact of each distributed RF signal on the system by quantifying its performance, thereby providing a basis for weighted allocation. The weight allocation of the signal performance value refers to the allocation of corresponding weight values ​​according to the signal performance value of each corrected distributed RF signal. The higher the signal performance value, the greater the impact of the signal on the system, and the corresponding weight value will also increase accordingly. The weight value allocation method can be designed according to specific indicators and requirements. For example, a linear ratio, a logarithmic ratio or other customized allocation strategies can be adopted. Through this weighted allocation strategy, signals with better performance can be effectively allocated first, thereby improving the overall performance and stability of the target FPGA platform.

[0034] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A high-precision distribution method of radio frequency signals based on FPGA, characterized in that: The method includes: Step S1: presetting the original RF signal for signal distribution and setting the parameters as initial signal parameters, setting the workgroup components of the target FPGA platform to the preprocessing module, the distribution monitoring module and the error correction module which are all signal-connected to each other, applying the initial signal parameters to the distribution monitoring module to perform parameter configuration of signal distribution and marking it as initial distribution configuration; Step S2: After the original RF signal is input into the preprocessing module and preprocessed, the original RF signal is transferred to the distribution monitoring module for signal distribution and output as distributed RF signals respectively, and in the distribution monitoring module, a mean shift optimization method is used to determine whether there is signal crosstalk in the distributed RF signals on each branch; Step S3: When the distribution monitoring module determines that there is no signal crosstalk in the branch, the distribution RF signal of the branch completes normal signal output; when the distribution monitoring module determines that there is signal crosstalk in the branch, the distribution RF signal of the target branch is input into the error code correction module for bit correction, and the parameters of the corrected distribution RF signal are set as the correction signal parameters; Step S4: Set a weighted allocation strategy, merge all corrected signal parameters into corrected total parameters through the weighted allocation strategy, apply the corrected total parameters to the distribution monitoring module to reconfigure the signal distribution parameters and mark them as corrected distribution configurations, return to step S2 to perform bit correction on the newly generated distribution RF signal again to keep the corrected distribution configuration updated until the original RF signal completes the distribution process or there is no new signal crosstalk in the distribution process of the target FPGA platform.

2. The high-precision distribution method of radio frequency signals based on FPGA according to claim 1, characterized in that: The specific process of the mean shift optimization method includes: Step A1: collecting initial signal data of the original radio frequency signal from the distribution monitoring module as an input value, presetting a Gaussian kernel function for calculating the mean value of the initial signal data, and setting a smoothing function for the Gaussian kernel function to limit the range of the Gaussian kernel function; Step A2: All feature values ​​in the original data are set as data points, one group of data points is set as the initial drift point, the number of iterations is preset for the drift point, and the weighted mean around each drift point is calculated using the Gaussian kernel function in each iteration; Step A3: iterate the initial drift point, and the process of moving the initial drift point to the calculated weighted mean position represents iterative update. When the number of iterations is reached or the initial drift point reaches a convergence state, the iteration is stopped and the final drift point is output; Step A4: outputting the initial signal data including all final drift points as distribution signal data representing distribution radio frequency signal parameters, and setting a crosstalk threshold for each identical signal parameter of the distribution signal data and the initial signal data; Step A5: When there is at least one signal parameter difference exceeding the crosstalk threshold, it is determined that signal crosstalk exists; when there is no signal parameter difference exceeding the crosstalk threshold, it is determined that the distributed RF signal has no signal crosstalk.

3. The high-precision distribution method of radio frequency signals based on FPGA according to claim 2, characterized in that: Setting a mean change criterion to represent a drift point convergence condition used when the initial drift point reaches a convergence state; The content of the mean change criterion includes: Step B1: Mark the initial position for each initial drift point, mark the distance between the initial drift point position and the initial position after each iteration, and calculate the position weight value of the initial drift point of the current iteration to the initial position based on the Gaussian kernel function; Step B2: after each iteration, the weighted mean of the initial drift point is calculated using the corresponding position weight value, and the weighted mean is used to represent the adjusted position of the initial drift point at the end of the current iteration and is marked as the updated position; Step B3: preset a convergence judgment value, calculate the position change between the initial position and the updated position, compare the position change with the convergence judgment value, and when the position change is less than the convergence judgment value, it is determined that the initial drift point has converged; when the position change is greater than the convergence judgment value, it is determined that the initial drift point has not converged.

4. A high-precision RF signal distribution method based on FPGA according to any one of claims 2 to 3, characterized in that: The weighted mean calculation process includes: assuming that the data point of the current initial drift point is represented by x, the position weight value is represented by ω, and the weighted mean is represented by x a , Then the weighted mean x a The calculation formula is: , The weighted mean x a represents the weighted average of all data points x in the initial drift point, and the position weight value ω is calculated by the Gaussian kernel function; Represents the sum of all data points x in the current initial drift point multiplied by the corresponding position weight value ω, The sum of all position weights ω.

5. The high-precision distribution method of radio frequency signals based on FPGA according to claim 4, characterized in that: The process of calculating the position weight value based on the Gaussian kernel function includes: assuming that the initial position and the updated position are represented by i and j respectively, the data point x at the initial position is further defined as the initial data point x i , the updated position of the data point x is further defined as the updated data point x j ; Let the distance between the initial drift point position and the initial position after each iteration be d ij , let the smoothing function of the Gaussian kernel function be represented by h, and the position weight value be represented by ω ij , Then the position weight value ω ij The calculation formula is expressed as: , Wherein the position weight value ω ij Indicates updating data point x j For the initial data point x i The smoothing function h is used to control the width of the Gaussian kernel function.

6. The high-precision distribution method of radio frequency signals based on FPGA according to claim 1, characterized in that: The error code correction module uses a convolutional code forward error correction method to perform bit correction, and the process includes: presetting the coding rate of the convolutional code and the constraint length of the convolutional code, inputting the initial signal parameters as a bit stream into the convolutional code to generate output bit data, using soft decision decoding to decode and output the output bit data and correct the error data therein, and setting the decoded output signal after the error data is corrected as a corrected distribution radio frequency signal.

7. The high-precision distribution method of radio frequency signals based on FPGA according to claim 6, characterized in that: The soft decision decoding uses the Viterbi algorithm to decode the output bit data, and the process includes: defining the initialized encoder parameters, demodulating the received output bit data and calculating the likelihood, using the Viterbi algorithm to update the path metric matrix to select the optimal backtracking path, and finally restoring the bit data to the original bit stream based on the optimal backtracking path.

8. The high-precision distribution method of radio frequency signals based on FPGA according to claim 6, characterized in that: The convolutional code has a coding rate limit of generating at least two output bits and at most four output bits for each input bit.

9. The high-precision distribution method of radio frequency signals based on FPGA according to claim 7, characterized in that: The path metric matrix is ​​calculated using the Hamming distance; when updating the path metric matrix, the predecessor state with the smallest path metric is selected.

10. The high-precision distribution method of radio frequency signals based on FPGA according to claim 1, characterized in that: The content of the weighted allocation strategy includes: setting a signal performance value to represent the indicator size of the impact of each corrected distributed RF signal on the performance of the target FPGA platform; the signal quality weighting process is to allocate a weight value according to the signal performance value of each corrected distributed RF signal.

Images