A Design Method and Application of Digital Predistortion Model

Abstract

The present invention discloses a design method and application of a digital pre-distortion model, which relates to the field of wireless communication technology. By introducing a maximum amplitude segmentation value and a minimum amplitude segmentation value, a signal amplitude decomposition method in a decomposition vector rotation digital pre-distortion model is improved to obtain an initial digital pre-distortion model. The optimal values ​​of the maximum amplitude segmentation value and the minimum amplitude segmentation value are further determined based on an operator matrix and model parameters to obtain a final digital pre-distortion model. This method can reduce the complexity of model parameter extraction while accurately representing the nonlinear characteristics of a power amplifier and improving the linearity of a wireless transmitter.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication technology, and in particular to a design method and application of a digital predistortion model with low model parameter extraction complexity. Background Art

[0002] To fully utilize limited spectrum resources, modulation techniques with high spectrum efficiency are gaining increasing popularity. However, the complex modulated signals generated by these techniques, characterized by non-constant envelopes, wide bandwidths, and high peak-to-average ratios, pose significant challenges to the linear performance of wireless transmitters. Therefore, linearization techniques are required to correct the nonlinear distortion of RF power amplifiers in wireless transmitters.

[0003] Digital pre-distortion (DPD) is the most widely used power amplifier linearization technology. Its basic principle is to insert an equivalent inverse model of the power amplifier's nonlinear distortion characteristics into the digital baseband signal, making the wireless transmitter after cascaded DPD equivalent to a linear system. This is used to improve the nonlinearity of the RF power amplifier, optimize in-band signal quality and out-of-band intermodulation indicators, and improve the efficiency of the RF power amplifier.

[0004] With the development of digital pre-distortion (DPD) technology, various behavioral models have emerged that can accurately model the nonlinear characteristics of power amplifiers. Among them, the Decomposed Vector Rotation (DVR) DPD model has been widely used to improve PA nonlinearity due to its easy-to-implement hardware structure and good linearization performance. Although the DVR DPD model can well characterize the nonlinear characteristics of power amplifiers, the complexity and overhead of extracting the model parameters of the DVR DPD model increase dramatically with the number of operator matrix items and data length, consuming a large amount of computing resources and increasing hardware implementation costs.

[0005] To reduce the complexity of extracting model parameters for DVR digital pre-distortion (DVR) models, the simplified regularized piecewise linear model (SCPWL) reduces the number of multiplication operations and complexity by increasing the number of zero values ​​in the operator matrix. However, this results in a loss of linearization, which can affect practical applications. Therefore, it is crucial to develop a DPD model that can effectively characterize the nonlinear characteristics of power amplifiers and reduce the complexity of model parameter extraction. Summary of the Invention

[0006] The purpose of the present invention is to provide a design method and application of a digital predistortion model, which can accurately represent the nonlinear characteristics of a power amplifier while reducing the complexity of extracting model parameters, thereby improving the linearity of a wireless transmitter.

[0007] To achieve the above object, the present invention provides the following solutions:

[0008] A design method for a digital predistortion model, comprising:

[0009] Constructing an initial digital predistortion model; the initial digital predistortion model is obtained by improving the signal amplitude decomposition method in the decomposition vector rotation digital predistortion model by introducing a maximum amplitude segment value and a minimum amplitude segment value;

[0010] Acquire a first baseband signal and a second baseband signal; the first baseband signal is a baseband signal input to a radio frequency power amplifier; the second baseband signal is a baseband signal generated by down-converting an output signal of the radio frequency power amplifier;

[0011] Randomly setting multiple groups of initial values, the initial values ​​including the values ​​of the maximum amplitude segment value and the minimum amplitude segment value; substituting each group of the initial values ​​into the initial digital pre-distortion model to obtain an intermediate digital pre-distortion model; using the first baseband signal and the second baseband signal as input, and calculating an operator matrix and model parameters of the intermediate digital pre-distortion model using the intermediate digital pre-distortion model; and determining an optimal initial value based on the operator matrices and model parameters corresponding to all the initial values;

[0012] The optimal initial value and the model parameters corresponding to the optimal initial value are substituted into the initial digital predistortion model to obtain a final digital predistortion model.

[0013] In some embodiments, the initial digital predistortion model includes:

[0014]

[0015] in, is the sampling value of the nth sampling point of the second baseband signal; Q is the memory depth; a q are model parameters; is the sampling value of the nqth sampling point of the first baseband signal; K is the nonlinear order; b kq,1 is the model parameter; F is the signal amplitude decomposition expression; k represents the kth order; θ max is the maximum amplitude segment value; θ min is the minimum amplitude segment value; j is the imaginary unit; θ is the signal phase; c kq,21 are model parameters; is the sampling value of the nth sampling point of the first baseband signal; d kq,22 is the model parameter; e kq,23 are model parameters;

[0016] The signal amplitude decomposition expression is:

[0017]

[0018] Among them, β k is the threshold corresponding to the k-th order nonlinearity; p is a constant.

[0019] In some embodiments, before using the first baseband signal and the second baseband signal as input and calculating the operator matrix and the model parameters of the intermediate digital predistortion model using the intermediate digital predistortion model, the method further includes:

[0020] Normalizing the first baseband signal and the second baseband signal respectively to obtain a first normalized signal and a second normalized signal, using the first normalized signal as a new first baseband signal and using the second normalized signal as a new second baseband signal.

[0021] In some embodiments, before using the first baseband signal and the second baseband signal as input and calculating the operator matrix and the model parameters of the intermediate digital predistortion model using the intermediate digital predistortion model, the method further includes:

[0022] The first baseband signal is sampled using a uniform distribution histogram sampling selection algorithm to obtain a first sampling signal; the second baseband signal is sampled based on the first sampling signal to obtain a second sampling signal; the first sampling signal is used as a new first baseband signal, and the second sampling signal is used as a new second baseband signal.

[0023] In some embodiments, determining the optimal initial value based on the operator matrices and model parameters corresponding to all the initial values ​​specifically includes:

[0024] Taking the first baseband signal and the second baseband signal as input, calculating and obtaining true model parameters of a decomposition vector rotation digital predistortion model;

[0025] For each set of the initial values, calculating the proportion of zero values ​​in the operator matrix; calculating the mean square error of the model parameters and the true model parameters to obtain the mean square error of the model parameters;

[0026] The optimal initial value is determined based on the proportion of zero values ​​in the operator matrix corresponding to all the initial values ​​and the mean square error of the model parameters.

[0027] A computer device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-mentioned method for designing a digital predistortion model.

[0028] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned method for designing a digital predistortion model.

[0029] A computer program product includes a computer program, which implements the steps of the above-mentioned method for designing a digital predistortion model when executed by a processor.

[0030] A wireless transmitter comprises: a signal source, a predistorter, an upconverter and a radio frequency power amplifier connected in sequence; the predistorter is provided with a digital predistortion model, and the digital predistortion model is generated by using the above-mentioned design method of a digital predistortion model.

[0031] A wireless communication system includes: a wireless transmitter and a wireless receiver in communication connection; the wireless transmitter adopts the above-mentioned wireless transmitter.

[0032] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0033] The present invention provides a design method and application of a digital pre-distortion model. By introducing maximum amplitude segmentation values ​​and minimum amplitude segmentation values, the signal amplitude decomposition method in the decomposition vector rotation digital pre-distortion model is improved to obtain an initial digital pre-distortion model. The optimal values ​​of the maximum amplitude segmentation value and the minimum amplitude segmentation value are further determined based on the operator matrix and model parameters to obtain a final digital pre-distortion model. This method can reduce the complexity of model parameter extraction while accurately representing the nonlinear characteristics of the power amplifier and improving the linearity of the wireless transmitter. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0035] Figure 1 A schematic flow chart of a method for designing a digital predistortion model provided in embodiment 1 of the present invention.

[0036] Figure 2 This is a schematic diagram of the proportion of zero values ​​in the operator matrix under different amplitude segmentation values ​​provided in Example 1 of the present invention.

[0037] Figure 3 This is a principle block diagram of the digital predistortion system provided in Example 1 of the present invention.

[0038] Figure 4 This is a schematic diagram of the structure of the test platform provided in Example 1 of the present invention.

[0039] Figure 5 This is a schematic diagram of the model parameter extraction process provided in Example 1 of the present invention. DETAILED DESCRIPTION

[0040] 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 making creative efforts are within the scope of protection of the present invention.

[0041] The purpose of the present invention is to provide a design method and application of a digital predistortion model, which can accurately represent the nonlinear characteristics of a power amplifier while reducing the complexity of extracting model parameters, thereby improving the linearity of a wireless transmitter.

[0042] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0043] Example 1

[0044] like Figure 1 As shown, a design method of a digital predistortion model in this embodiment includes:

[0045] S1: constructing an initial digital predistortion model; the initial digital predistortion model is obtained by introducing a maximum amplitude segment value and a minimum amplitude segment value to improve the signal amplitude decomposition method in the decomposition vector rotation digital predistortion model.

[0046] S2: Obtain a first baseband signal and a second baseband signal; the first baseband signal is a baseband signal input to the RF power amplifier; the second baseband signal is a baseband signal generated by down-converting the output signal of the RF power amplifier.

[0047] S3: Randomly set multiple groups of initial values, where the initial values ​​include the values ​​of the maximum amplitude segment value and the values ​​of the minimum amplitude segment value; for each group of the initial values, substitute the initial values ​​into the initial digital pre-distortion model to obtain an intermediate digital pre-distortion model; use the first baseband signal and the second baseband signal as input, and use the intermediate digital pre-distortion model to calculate the operator matrix and model parameters of the intermediate digital pre-distortion model; determine the optimal initial value based on the operator matrices and model parameters corresponding to all the initial values.

[0048] S4: Substituting the optimal initial value and the model parameters corresponding to the optimal initial value into the initial digital predistortion model to obtain a final digital predistortion model.

[0049] The general expression of the decomposition vector rotation digital predistortion model is shown in the following equation (1):

[0050]

[0051] In formula (1), is the sampling value of the nth sampling point of the second baseband signal; Q is the memory depth; a q is the model parameter, which changes with q, and there are Q+1 of them; is the sampling value of the nqth sampling point of the first baseband signal, q can be understood as the delay; K is the nonlinear order; b kq,1 is a model parameter, which changes with the difference of k and q, and there are K(Q+1) of them in total; |·| is the absolute value operation of a complex number, which determines the signal amplitude (also known as the signal amplitude) of the signal within |·|; β k is the threshold corresponding to the kth order nonlinearity, which is determined by the nonlinearity order k; j is the imaginary unit; θ is the signal phase; c kq,21 is the model parameter, which changes with the difference of k and q, and there are K(Q+1) of them; is the sampling value of the nth sampling point of the first baseband signal; d kq,22 is the model parameter, which changes with the difference of k and q, and there are KQ of them in total; e kq,23 is a model parameter, which changes with the difference of k and q, and there are KQ of them in total.

[0052] The signal amplitude decomposition method in the decomposition vector rotation digital predistortion model is shown in the following equation (2). The signal amplitude is decomposed into multiple linear regions through different thresholds, which simplifies the high-order operations of the polynomial model.

[0053]

[0054] In order to reduce the complexity and computational overhead of extracting model parameters of a decomposed vector rotation digital pre-distortion model, this embodiment provides a design method for a digital pre-distortion model for nonlinear distortion modeling with low-complexity model parameter extraction. This design method aims to reduce the complexity of model parameter extraction while accurately representing the nonlinear characteristics of the power amplifier by increasing the number of 0 values ​​in the operator matrix and introducing the mean square error of the model parameters. Compared with the simplified regular piecewise linear model, it can better improve the linearity of the wireless transmitter when the complexity of model parameter extraction is comparable.

[0055] To achieve the above-mentioned object, this embodiment provides an initial digital predistortion model for extracting low-complexity model parameters based on a decomposed vector rotation digital predistortion model. The initial digital predistortion model is obtained by improving the signal amplitude decomposition method in the decomposed vector rotation digital predistortion model by introducing maximum amplitude segment values ​​and minimum amplitude segment values. The general expression of the initial digital predistortion model is shown in the following equation (3):

[0056]

[0057] In formula (3), is the sampling value of the nth sampling point of the second baseband signal; Q is the memory depth; a q are model parameters; is the sampling value of the nqth sampling point of the first baseband signal; K is the nonlinear order; b kq,1 is the model parameter; F is the signal amplitude decomposition expression; is the amplitude value of the sampling value of the nqth sampling point of the first baseband signal; k represents the kth order; θ max is the maximum amplitude segment value; θ min is the minimum amplitude segment value; j is the imaginary unit; θ is the signal phase; c kq,21 are model parameters; is the sampling value of the nth sampling point of the first baseband signal; d kq,22 is the model parameter; e kq,23 Wherein, the first baseband signal and the second baseband signal are both complex forms of I / Q signals, and the I / Q signal refers to a signal expressed using amplitude and phase.

[0058] The signal amplitude decomposition expression is shown in the following formula (4):

[0059]

[0060] In formula (4), β k is the threshold corresponding to the kth order nonlinearity; p is a constant. It should be noted that the minimum amplitude segment value is smaller than the maximum amplitude segment value.

[0061] The principle of the above formula (4) is: when the signal amplitude is greater than or equal to θ max When the signal amplitude is greater than the threshold β k Relatively large, in order to make the operator matrix have more 0 values, so when the signal amplitude is decomposed, if the signal amplitude is greater than the threshold β k , the value of this item in the operator matrix is ​​0; similarly, when the signal amplitude is less than or equal to θ min When the signal amplitude is greater than the threshold β kRelatively small, in order to make the operator matrix have more 0 values, so when the signal amplitude is decomposed, if the signal amplitude is less than the threshold β k , the value of this item in the operator matrix is ​​0; when the signal amplitude is in the interval (θ min ,θ max ), the signal amplitude is relatively moderate, so when the signal amplitude is decomposed, when the threshold β k When the nonlinear order in is an odd number, if the signal amplitude is greater than the threshold β k , the value of this item in the operator matrix is ​​0; when the threshold β k When the nonlinear order in is even, if the signal amplitude is less than the threshold β k , the value of this item in the operator matrix is ​​0.

[0062] In the initial digital pre-distortion model, the maximum amplitude segment value, the minimum amplitude segment value and the model parameters are all unknown and need to be solved to obtain the final digital pre-distortion model. Figure 2 As shown, when the input signal is a 16QAM signal with a bandwidth of 20MHz, the proportion of 0 values ​​in the operator matrix under different amplitude segment values ​​is obtained after the uniform distribution histogram sampling selection algorithm is adopted. When the signal amplitude is uniformly distributed, as θ max The decrease of θ min The increase is located at (θ min ,θ max ) interval, the fewer signals there are, the more zero values ​​there are in the operator matrix. Based on this, the maximum and minimum amplitude segment values ​​affect the performance of the digital predistortion model, and their optimal values ​​need to be determined.

[0063] like Figure 3 As shown, it is a principle block diagram of the digital predistortion system. The first baseband signal sent by the signal source is After pre-distortion processing by the pre-distorter, we get After up-conversion, the signal is input to the RF power amplifier, which performs power amplification to obtain the power amplifier output signal. In order to extract the pre-distortion model parameters (i.e., determine the maximum amplitude segment value, minimum amplitude segment value, and model parameters of the initial digital pre-distortion model), this embodiment further couples the power amplifier output signal. At this time, the first baseband signal is input to the RF power amplifier after up-conversion, and the output signal of the RF power amplifier is down-converted to obtain the feedback signal (i.e., the second baseband signal). The first baseband signal and the second baseband signal are used to perform the predistortion model parameter extraction operation, specifically, the operation can be performed using S2-S4. After obtaining the final digital predistortion model, the subsequent operation is performed according to Figure 3 The first baseband signal sent by the source is transmitted After pre-distortion processing by the pre-distorter, we get After up-conversion, it is input into the RF power amplifier, which amplifies the power and outputs it to the wireless receiver.

[0064] In S3, multiple groups of initial values ​​are randomly set, and the initial values ​​include the values ​​of the maximum amplitude segment value and the minimum amplitude segment value. For each group of initial values, the initial values ​​are substituted into the initial digital pre-distortion model to obtain an intermediate digital pre-distortion model. At this time, only the model parameters in the intermediate digital pre-distortion model are unknown values. The first baseband signal is used as input, and the operator matrix is ​​calculated using formula (4). The operator matrix is ​​a matrix with I rows and J columns, where I is the total number of sampling points in the first baseband signal, and J is K(Q+1). The value of the element in the i-th row and j-th column of the operator matrix is ​​determined as follows: the values ​​of k and q are determined according to the value of j, and the iq-th sampling point in the first baseband signal is substituted into formula (4) (at this time, the threshold β in formula (4) is k The signal amplitude is calculated as the value of the element in the i-th row and j-th column. Substituting the first and second baseband signals into equation (3) yields the model parameters of the intermediate digital predistortion model. Thus, the operator matrix and the model parameters of the intermediate digital predistortion model are calculated using the first and second baseband signals as inputs using the intermediate digital predistortion model.

[0065] It should be noted that, in this embodiment, the expressions of the initial digital pre-distortion model, the intermediate digital pre-distortion model and the final digital pre-distortion model are all equation (3). The only difference between the three is that the maximum amplitude segment value, the minimum amplitude segment value and the model parameters of the initial digital pre-distortion model are all unknown values, the maximum amplitude segment value and the minimum amplitude segment value of the intermediate digital pre-distortion model are known values, and the model parameters are unknown values, and the maximum amplitude segment value, the minimum amplitude segment value and the model parameters of the final digital pre-distortion model are all known values.

[0066] In S3, after obtaining the operator matrix and model parameters corresponding to each group of initial values, the optimal initial values ​​are further determined based on the operator matrix and model parameters corresponding to all the initial values. Specifically, the method includes: using the first baseband signal and the second baseband signal as input, using formula (1) to calculate the true model parameters of the decomposition vector rotation digital predistortion model; for each group of initial values, calculating the proportion of zero values ​​in the operator matrix, calculating the mean square error between the model parameters and the true model parameters, and obtaining the mean square error of the model parameters; and determining the optimal initial values ​​based on the proportion of zero values ​​in the operator matrix and the mean square error of the model parameters corresponding to all the initial values. The larger the proportion of zero values ​​in the operator matrix, the better, and the smaller the mean square error of the model parameters, the better. According to this principle, the optimal initial values ​​are determined according to user needs, and the optimal initial values ​​of the combined optimal proportion of zero values ​​in the operator matrix and the mean square error of the model parameters can be selected as the optimal initial values.

[0067] Preferably, before using the first baseband signal and the second baseband signal as input and calculating the operator matrix and model parameters of the intermediate digital pre-distortion model using the intermediate digital pre-distortion model, the design method of this embodiment further includes: normalizing the first baseband signal and the second baseband signal to obtain a first normalized signal and a second normalized signal, using the first normalized signal as a new first baseband signal, and using the second normalized signal as a new second baseband signal. In this case, the maximum amplitude segment value and the minimum amplitude segment value have a value range of 0-1.

[0068] Preferably, before using the first baseband signal and the second baseband signal as input and calculating the operator matrix and the model parameters of the intermediate digital pre-distortion model using the intermediate digital pre-distortion model, the design method of this embodiment also includes: sampling the first baseband signal using a uniform distribution histogram sampling selection algorithm to obtain a first sampled signal; sampling the second baseband signal based on the first sampled signal to obtain a second sampled signal; using the first sampled signal as a new first baseband signal and using the second sampled signal as a new second baseband signal, thereby reducing the number of sampling points of the first baseband signal and the second baseband signal and improving calculation efficiency.

[0069] Among them, a uniform distribution histogram sampling selection algorithm is used to sample the first baseband signal to obtain a first sampling signal, specifically including: taking the minimum value of the signal amplitude of the first baseband signal as the starting value, taking the maximum value of the signal amplitude of the first baseband signal as the ending value, uniformly dividing the interval formed by the starting value and the ending value to obtain multiple amplitude distribution intervals, all sampling points of the first baseband signal belong to the multiple amplitude distribution intervals obtained by division, determining the amplitude distribution interval with the least number of sampling points, determining a sampling value based on the number of sampling points in the amplitude distribution interval, the sampling value is less than the number of sampling points in the amplitude distribution interval, sampling the sampling value sampling points in each amplitude distribution interval to obtain the first sampling signal.

[0070] The method of sampling the second baseband signal based on the first sampling signal to obtain the second sampling signal specifically includes: since the sampling points of the first baseband signal and the sampling points in the second baseband signal correspond to each other, after determining the first sampling signal, determining in the second baseband signal the sampling points corresponding to each sampling point in the first sampling signal to sample the second baseband signal to obtain the second sampling signal.

[0071] In this embodiment, the first baseband signal and the second baseband signal may be normalized first, and then the first normalized signal and the second normalized signal may be sampled.

[0072] In order to complete the acquisition process of the first baseband signal and the second baseband signal, this embodiment sets up a test platform, such as Figure 4 As shown, the test platform consists of a PC, a digital RF experimental platform, an RF power amplifier, a power splitter, an attenuator and a spectrum analyzer. The PC is bidirectionally connected to the digital RF experimental platform, the digital RF experimental platform is connected to the RF power amplifier, and the RF power amplifier is connected to two attenuators, one attenuator is connected to the spectrum analyzer, and the other attenuator is connected to the digital RF experimental platform. Among them, the digital RF experimental platform includes an FPGA and an RF board, and the RF board can adopt an ADRV RF board. The required test signal (i.e., the first baseband signal) is generated by the PC and downloaded to the digital RF experimental platform. The digital RF experimental platform up-converts the test signal, and the power amplifier output signal generated by the power amplifier is down-converted by the digital RF experimental platform to obtain the second baseband signal, which is then sent to the PC for digital pre-distortion model parameter extraction.

[0073] Based on the above-mentioned test platform, this embodiment provides a method for reducing the computational complexity of extracting model parameters of a decomposed vector rotation digital predistortion model. The predistortion model parameter extraction operation specifically includes: establishing a test platform, recording a first baseband signal at a transmitting end and a second baseband signal at a receiving end; normalizing the recorded first and second baseband signals; using a uniform distribution histogram sampling selection algorithm to select sampling signals (i.e., sampling points) based on the signal amplitude distribution characteristics of the normalized first baseband signal, so that the number of sampling signals (i.e., sampling points) within different amplitude distribution intervals after selection is consistent, thereby reducing the number of sampling signals; analyzing the proportion of zero values ​​in the operator matrix and the mean square error of the extracted model parameters under different amplitude segmentation values ​​(i.e., the maximum amplitude segmentation value and the minimum amplitude segmentation value). When the proportion of zero values ​​and the mean square error are both moderate, the amplitude segmentation value is the required value; solving the model parameters based on the selected amplitude segmentation value to obtain a final digital predistortion model, and copying the model parameters to the predistorter to achieve nonlinear modeling of the power amplifier.

[0074] like Figure 5 As shown in FIG, the above-mentioned predistortion model parameter extraction operation process is mainly divided into three parts: (1) recording the first baseband signal input to the power amplifier and the second baseband signal output from the power amplifier to complete the nonlinear characteristic acquisition of the RF power amplifier; (2) analyzing the influence of different amplitude segmentation values ​​on the proportion of zero values ​​in the operator matrix in the digital predistortion model and the mean square error of the extracted model parameters, and selecting the amplitude segmentation value to be used; (3) comparing the adjacent channel power ratio and error vector amplitude of the power amplifier output signal before and after predistortion based on the selected amplitude segmentation value to determine whether its linearization performance meets the requirements of practical applications.

[0075] The test process of the test platform is:

[0076] Step 1: Generate a test signal on the PC through MATLAB. When generating the test signal, set the signal modulation mode, symbol rate, peak-to-average ratio, sampling rate, and data length.

[0077] Step 2: Download the test signal generated by MATLAB to the digital RF experiment platform.

[0078] Step 3: The digital RF experimental platform up-converts the test signal and sends it to the RF power amplifier.

[0079] Step 4: The up-converted test signal is output as a power amplified signal after passing through the RF power amplifier.

[0080] Step 5: After passing through the power splitter, the power amplifier output signal (i.e., the power amplified signal output by the RF power amplifier after power amplifying the up-converted test signal) is transmitted to a spectrum analyzer via an attenuator, where the spectrum diagram and constellation diagram of the signal are observed; and to a digital RF experimental platform via another attenuator.

[0081] Step 6: The digital RF experimental platform down-converts the received power amplifier output signal to obtain a second baseband signal, and sends it to the PC end, which extracts the model parameters of the digital pre-distortion model.

[0082] Step 7: The PC performs digital predistortion processing on the test signal according to the extracted model parameters to obtain a predistorted signal.

[0083] Step 8: Download the predistorted signal back to the digital RF experimental platform. The digital RF experimental platform up-converts the predistorted signal and sends it to the RF power amplifier. The RF power amplifier then outputs the signal as a power amplified signal.

[0084] Step 9: The pre-distorted power amplifier output signal (i.e., the power amplified signal output by the RF power amplifier after power amplification of the up-converted pre-distorted signal) passes through a power splitter and then an attenuator to a spectrum analyzer. The spectrum analyzer is used to observe the signal's spectrum diagram and constellation diagram to determine the spectral characteristics of the pre-distorted power amplifier output signal.

[0085] Step 10: Analyze the linearization performance of the power amplifier output signal after predistortion to see if it meets actual application requirements.

[0086] Among them, the model parameter extraction process on the PC side is:

[0087] Step 1: The PC side normalizes the received power amplifier input and output baseband signals (ie, the first baseband signal and the second baseband signal).

[0088] Step 2: Use the uniform distribution histogram sampling selection algorithm to select the signal based on the amplitude distribution characteristics of the normalized baseband signal. The selected signal has the same number of sampling points within each amplitude distribution interval. By using the uniform distribution histogram sampling selection algorithm to select the signal, the number of sampled signals can be reduced.

[0089] Step 3: Calculate the proportion of 0 values ​​in the operator matrix under different amplitude segmentation values.

[0090] Step 4: Calculate the mean square error of the extracted model parameters under different amplitude segmentation values.

[0091] Step 5: When the proportion of zero values ​​in the operator matrix and the mean square error can be compromised, the amplitude segment value is selected, and the selected amplitude segment value is subsequently used for parameter extraction and power amplifier nonlinear improvement.

[0092] Step 6: Extract the model parameters of the digital predistortion model according to the selected amplitude segment values ​​and copy them to the predistorter to achieve nonlinear modeling of the power amplifier.

[0093] This embodiment relates to a digital predistortion technology for correcting nonlinear distortion of a power amplifier in a wireless communication system, and more particularly to a method for reducing the computational complexity of extracting model parameters from a decomposition vector rotation digital predistortion model. This method can reduce the computational complexity of extracting model parameters from the decomposition vector rotation digital predistortion model. By designing a new digital predistortion model, increasing the number of zero values ​​in an operator matrix, and introducing a mean square error (MSE) of the model parameters, the computational complexity of parameter extraction can be significantly reduced while maintaining linearization accuracy. This method has strong applicability and feasibility.

[0094] Example 2

[0095] A computer device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method for designing a digital predistortion model described in Example 1.

[0096] Example 3

[0097] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for designing a digital predistortion model described in embodiment 1.

[0098] Example 4

[0099] A computer program product includes a computer program, which implements the steps of the method for designing a digital predistortion model described in embodiment 1 when executed by a processor.

[0100] Example 5

[0101] A wireless transmitter includes: a signal source, a predistorter, an upconverter, and a radio frequency power amplifier connected in sequence; the predistorter is provided with a digital predistortion model, and the digital predistortion model is generated using the design method of a digital predistortion model described in Example 1.

[0102] Example 6

[0103] A wireless communication system includes: a wireless transmitter and a wireless receiver in communication connection; the wireless transmitter adopts a wireless transmitter described in Example 5.

[0104] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0105] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the method and core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A method for designing a digital predistortion model, characterized in that: include: Building an initial digital predistortion model; The initial digital pre-distortion model is obtained by improving the signal amplitude decomposition method in the decomposition vector rotation digital pre-distortion model by introducing the maximum amplitude segment value and the minimum amplitude segment value; Acquire a first baseband signal and a second baseband signal; the first baseband signal is a baseband signal input to a radio frequency power amplifier; The second baseband signal is a baseband signal generated by down-converting the output signal of the radio frequency power amplifier; Randomly setting multiple groups of initial values, the initial values ​​including the value of the maximum amplitude segment value and the value of the minimum amplitude segment value; for each group of the initial values, substituting the initial values ​​into the initial digital predistortion model to obtain an intermediate digital predistortion model; Taking the first baseband signal and the second baseband signal as input, and calculating an operator matrix and model parameters of the intermediate digital predistortion model using the intermediate digital predistortion model; determining an optimal initial value based on the operator matrix and model parameters corresponding to all the initial values; Substituting the optimal initial value and the model parameters corresponding to the optimal initial value into the initial digital predistortion model to obtain a final digital predistortion model; The initial digital predistortion model includes: in, is the sampling value of the nth sampling point of the second baseband signal; Q is the memory depth; a q are model parameters; is the sampling value of the nqth sampling point of the first baseband signal; K is the nonlinear order; b kq,1 is the model parameter; F is the signal amplitude decomposition expression; k represents the kth order; θ max is the maximum amplitude segment value; θ min is the minimum amplitude segment value; j is the imaginary unit; θ is the signal phase; c kq,21 are model parameters; is the sampling value of the nth sampling point of the first baseband signal; d kq,22 is the model parameter; e kq,23 are model parameters; The signal amplitude decomposition expression is: Among them, β k is the threshold corresponding to the k-th order nonlinearity; p is a constant.

2. The method for designing a digital predistortion model according to claim 1, wherein: Before obtaining an operator matrix and model parameters of the intermediate digital pre-distortion model by calculating the operator matrix and the model parameters of the intermediate digital pre-distortion model using the first baseband signal and the second baseband signal as input, the method further includes: Normalizing the first baseband signal and the second baseband signal respectively to obtain a first normalized signal and a second normalized signal, using the first normalized signal as a new first baseband signal and using the second normalized signal as a new second baseband signal.

3. The method for designing a digital predistortion model according to claim 1, wherein: Before obtaining an operator matrix and model parameters of the intermediate digital pre-distortion model by calculating the operator matrix and the model parameters of the intermediate digital pre-distortion model using the first baseband signal and the second baseband signal as input, the method further includes: The first baseband signal is sampled using a uniform distribution histogram sampling selection algorithm to obtain a first sampling signal; the second baseband signal is sampled based on the first sampling signal to obtain a second sampling signal; the first sampling signal is used as a new first baseband signal, and the second sampling signal is used as a new second baseband signal.

4. The method for designing a digital predistortion model according to claim 1, wherein: Determining the optimal initial value based on the operator matrix and model parameters corresponding to all the initial values ​​specifically includes: Taking the first baseband signal and the second baseband signal as input, calculating and obtaining true model parameters of a decomposition vector rotation digital predistortion model; For each set of the initial values, calculating the proportion of zero values ​​in the operator matrix; calculating the mean square error of the model parameters and the true model parameters to obtain the mean square error of the model parameters; The optimal initial value is determined based on the proportion of zero values ​​in the operator matrix corresponding to all the initial values ​​and the mean square error of the model parameters.

5. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the method for designing a digital predistortion model according to any one of claims 1 to 4.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for designing a digital predistortion model according to any one of claims 1 to 4 are implemented.

7. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method for designing a digital predistortion model according to any one of claims 1 to 4 are implemented.

8. A wireless transmitter, characterized in that: include: A signal source, a predistorter, an upconverter and a radio frequency power amplifier are connected in sequence; the predistorter is provided with a digital predistortion model, and the digital predistortion model is generated by using a design method of a digital predistortion model as described in any one of claims 1-4.

9. A wireless communication system, characterized in that: include: A wireless transmitter and a wireless receiver for communication connection; The wireless transmitter adopts the wireless transmitter according to claim 8.

Images