Weighted average digital pre-distortion device and method for orthogonal optimization of primary function

Abstract

The invention belongs to the technical field of electromagnetic compatibility, and particularly discloses a weighted average digital pre-distortion device and method for primary function orthogonal optimization, and the method comprises the steps: processing a memory polynomial model through employing a primary function orthogonal optimization method and historical data, and obtaining a broadband signal amplification model of a power amplifier module; performing inversion on the broadband signal amplification model to obtain a digital pre-distortion model; determining an optimal coefficient set of the digital pre-distortion model, and optimizing the optimal coefficient set based on the historical data so as to perform pre-distortion processing on an input signal based on the optimized digital pre-distortion model; and performing power amplification processing on the pre-distorted signal to obtain a power amplifier output signal. According to the invention, the precision of the digital pre-distortion model can be improved.

Description

Technical Field

[0001] The present application belongs to the field of electromagnetic compatibility technology, and more specifically, relates to a weighted average digital predistortion device and method for orthogonal optimization of basis functions. Background Art

[0002] In modern communication systems, the important role of broadband signal amplifiers cannot be ignored. Because of its nonlinear, harmonic and other non-ideal characteristics that seriously affect the communication signal, the design and research of broadband signal amplifiers has always been an important topic in the development of wireless communications.

[0003] In order to solve the above problems, pre-distortion technology has received attention and has been widely used. Many research teams abroad have conducted research on this technology. The research hotspots in the field of digital pre-distortion mainly focus on the selection of digital pre-distortion models, model parameter identification algorithms and parameter learning structures, as well as corresponding improvements and optimizations. Common digital pre-distortion models include the earlier proposed lookup table model. This model has a simple structure and its accuracy is related to the size of the lookup table. It is generally used to model memoryless amplifiers. If you want to describe a memory amplifier, you need to establish a multidimensional lookup table, but this will increase resource consumption. The second is the most complete Volterrra series model. Although this model has high accuracy, the parameter solution is very complicated. Therefore, a series of simplified polynomial models are derived, such as: memory polynomial (MP) model, generalized memory polynomial (GMP) model, Volterrra series (DDR) model based on dynamic deviation reduction, etc. The parameter identification algorithms of these models include least squares (LS), least mean square error (LMS), recursive least squares (RLS), etc. In broadband communication systems, the performance of digital pre-distortion technology is still limited by limited sampling rate and analysis bandwidth. In broadband scenarios, the model accuracy and linearization effect of traditional digital pre-distortion DPD technology are relatively poor. For wireless communication systems with large bandwidth and high peak-to-average ratio, although there are many types of DPD models, there are still some relatively difficult problems to be overcome. For example, the band-limited DPD model generally requires the relevant parameters of the high-order FIR filter, which increases the complexity of model training and also takes up a lot of hardware resources.

[0004] In summary, the existing digital pre-distortion DPD technology has the problems of poor accuracy of the digital pre-distortion model, high training complexity and high resource occupancy rate, and still needs to be continuously improved and optimized to adapt to the rapidly developing modern communication systems. Summary of the invention

[0005] In view of the defects of the prior art, the purpose of the present application is to provide a weighted average digital predistortion device and method for orthogonal optimization of basis functions, aiming to solve the problem of poor accuracy of the digital predistortion model.

[0006] In a first aspect, the present application provides a weighted average digital predistortion device for orthogonal optimization of basis functions, comprising:

[0007] Model processing module, inversion module, optimization module and power amplifier module;

[0008] The model processing module is used to process the memory polynomial model using a basis function orthogonal optimization method and historical data to obtain a broadband signal amplification model of the power amplifier module;

[0009] The inversion module is used to invert the broadband signal amplification model to obtain a digital predistortion model;

[0010] The optimization module is used to determine an optimal coefficient set of the digital predistortion model, and optimize the optimal coefficient set based on the historical data, so as to perform predistortion processing on the input signal based on the optimized digital predistortion model;

[0011] The power amplifier module is used to perform power amplification processing on the predistorted signal to obtain a power amplifier output signal.

[0012] Optionally, the process of solving the optimal coefficient set includes:

[0013] Determine a basis function matrix including each basis function vector, determine a transposed matrix of the basis function matrix, and construct a power amplifier output vector according to an output signal of the power amplifier module;

[0014] Determine the objective function of the error signal between the digital predistortion model and the power amplifier inverse model according to the power amplifier output signal, the basis function vector and the optimal coefficient set;

[0015] The objective function is solved by using the least square method, and when the objective function takes a minimum value, the optimal coefficient set is determined based on the basis function matrix, the transposed matrix and the power amplifier output vector.

[0016] Optionally, the pre-distortion compensation module solves the broadband signal amplification model in the following manner:

[0017]

[0018] Among them, φ 2k+1 (z(nl)) is the polynomial basis function of the memory polynomial model, and the polynomial basis function is used to describe the nonlinear characteristics of the power amplifier module. 2k+1,lare the coefficients of the polynomial basis function, K is the highest nonlinear order of the broadband signal amplification model, L is the memory depth of the memory polynomial model, y(n) is the signal after predistortion processing, z(n) is the power amplifier output signal, l is the number of memory layers of the power amplifier memory polynomial model, and k is the nonlinear order of the broadband signal amplification model.

[0019] Optionally, the objective function of the error signal between the digital predistortion model and the power amplifier inverse model is determined by the following formula:

[0020]

[0021] in, is the output signal of the power amplifier inverse model, β n is the predistortion coefficient of the memory polynomial model, B k (n) is the kth basis function of the digital predistortion model, B(n) is the basis function vector, and β is the optimal coefficient set;

[0022] The kth basis function of the digital predistortion model is defined as:

[0023]

[0024] The basis function vector is defined as:

[0025] B(n)=[B1(n),B3(n),...,B k (n)]

[0026] Among them, a kq is the coefficient, x(nq) is the value of the input signal at time nq, q is the number of memory layers of the predistortion memory polynomial model, and Q represents the memory depth of the predistortion memory polynomial model;

[0027] The objective function of the error signal between the digital pre-distortion model and the power amplifier inverse model is expressed as:

[0028]

[0029] The objective function is expressed as follows by the least squares method:

[0030]

[0031] Where N is the number of data samples for model training;

[0032] Then the optimal coefficient set to be solved is:

[0033] β=(B H B) -1 B H z

[0034] z=[z(1),z(2),...,z(N)] T

[0035]

[0036] Among them, z is the output vector of the power amplifier inverse model, and B is the basis function matrix.

[0037] Optionally, the pre-distortion compensation module obtains optimized digital pre-distortion model coefficients in the following manner:

[0038] Based on the normalized RMS value of the historical data, the weights of the digital predistortion model coefficients are calculated as shown in the following formula:

[0039]

[0040] Among them, ω i is the weight of the digital pre-distortion model coefficient, NRMS i is the normalized RMS value of i training samples, G is the number of training sample sequences;

[0041] By weighting the coefficients of the same order, the final weight expression is:

[0042]

[0043] Where k is the nonlinear order of the model, is the optimized k-th order digital predistortion model coefficient, α i,k is the k-th order model coefficient corresponding to the i-th training sample.

[0044] Optionally, the predistortion compensation module is further used for:

[0045] Determine the root mean square of each training data, and sort the root mean square of each training data;

[0046] A small weight factor is assigned to training data with a large RMS value, and a large weight factor is assigned to training data with a small RMS value.

[0047] Optionally, the pre-distortion compensation module includes:

[0048] A coefficient determination unit, used to determine an optimal coefficient set of the digital predistortion model using a least squares method;

[0049] The weighted averaging unit is used to obtain the normalized root mean square value of the historical data, and optimize the optimal coefficient set based on the normalized root mean square value by using the weighted averaging method to obtain the optimized digital pre-distortion model coefficients.

[0050] The present application also provides a weighted average digital predistortion method for orthogonal optimization of basis functions, characterized by comprising:

[0051] The memory polynomial model is processed by using a basis function orthogonal optimization method and historical data to obtain a broadband signal amplification model of the power amplifier module;

[0052] Inverting the broadband signal amplification model to obtain a digital predistortion model;

[0053] Determining an optimal coefficient set of the digital predistortion model, and optimizing the optimal coefficient set based on the historical data, so as to perform predistortion processing on the input signal based on the optimized digital predistortion model;

[0054] The predistorted signal is power amplified to obtain a power amplifier output signal.

[0055] Optionally, determining an optimal coefficient set of the digital pre-distortion model includes:

[0056] Determine a basis function matrix including each basis function vector, determine a transposed matrix of the basis function matrix, and construct a power amplifier output vector according to an output signal of the power amplifier module;

[0057] Determine the objective function of the error signal between the digital predistortion model and the power amplifier inverse model according to the power amplifier output signal, the basis function vector and the optimal coefficient set;

[0058] The objective function is solved by using the least square method, and when the objective function takes a minimum value, the optimal coefficient set is determined based on the basis function matrix, the transposed matrix and the power amplifier output vector.

[0059] In a third aspect, the present application provides an electronic device comprising: at least one memory for storing programs; and at least one processor for executing the programs stored in the memory. When the program stored in the memory is executed, the processor is used to execute the method described in the first aspect or any possible implementation of the first aspect.

[0060] In a fourth aspect, the present application provides a computer-readable storage medium, which stores a computer program. When the computer program runs on a processor, the processor executes the method described in the first aspect or any possible implementation of the first aspect.

[0061] In a fifth aspect, the present application provides a computer program product. When the computer program product runs on a processor, the processor executes the method described in the first aspect or any possible implementation of the first aspect.

[0062] It can be understood that the beneficial effects of the second to fifth aspects mentioned above can be found in the relevant description of the first aspect mentioned above, and will not be repeated here.

[0063] In general, the above technical solutions conceived by this application have the following beneficial effects compared with the prior art:

[0064] (1) The present application can improve the flexibility and expressiveness of the model by using orthogonal basis functions for optimization. The orthogonal basis functions can reduce the correlation problem of the model and help to more accurately capture the nonlinear characteristics of the power amplifier module. In addition, the present application can more effectively capture the characteristics of the system in different states during the model optimization process by introducing historical data as a reference, thereby improving the accuracy of the digital pre-distortion model.

[0065] (2) The memory polynomial model can take into account the nonlinearity of the power amplifier module and the influence of its historical signals. The model describes the dynamic characteristics of the power amplifier module by including memory terms, so it can more comprehensively capture the nonlinear behavior of the power amplifier and improve the accuracy of the model.

[0066] (3) By introducing the basis function orthogonal optimization method and weighted average technology, it is possible to improve model performance while reducing computational complexity, avoiding the huge computational burden brought by high-order parameters and complex model training.

[0067] (4) After the pre-distortion signal is amplified by the power amplifier module, it can significantly reduce the nonlinear distortion introduced by the power amplifier, effectively improving the gain performance and overall efficiency of the power amplifier. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 This is one of the structural schematic diagrams of the weighted average digital predistortion device for orthogonal optimization of basis functions provided in the embodiments of the present application;

[0069] Figure 2 This is the second structural schematic diagram of the weighted average digital predistortion device for orthogonal optimization of basis functions provided in the embodiment of the present application;

[0070] Figure 3 Schematic diagram of a digital pre-distortion test platform provided in an embodiment of the present application;

[0071] Figure 4 1 is a characteristic diagram of a broadband signal amplifying device before and after compensation according to an embodiment of the present application; wherein, Figure (a) is a schematic diagram of normalized output amplitude; Figure (b) is a relationship diagram of normalized input phase shift and output phase shift; Figures (c) and (d) are frequency mixing output power relationship diagrams;

[0072] Figure 5 This is a comparison chart of the modeling accuracy of this application and other models;

[0073] Figure 6 The figures are comparison diagrams of modeling results of different methods; wherein, Figure (a) is the output power spectrum modeling accuracy curve using the present application; Figures (b) and (c) are the local enlarged modeling curves of the upper and lower sidebands respectively; and Figure (d) is the detailed modeling error curve;

[0074] Figure 7 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0075] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0076] The term "and / or" in this application is a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. The symbol " / " in this application indicates that the associated objects are in an or relationship, for example, A / B means A or B.

[0077] The terms "first" and "second" in the specification and claims of this application are used to distinguish different objects rather than to describe a specific order of objects. For example, a first response message and a second response message are used to distinguish different response messages rather than to describe a specific order of response messages.

[0078] In the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.

[0079] In the description of the embodiments of the present application, unless otherwise specified, "multiple" means two or more than two. For example, multiple processing units refer to two or more processing units, etc.; multiple elements refer to two or more elements, etc.

[0080] Next, the technical solutions provided in the embodiments of the present application are introduced.

[0081] Reference Figure 1 , the present application provides a weighted average digital predistortion device for orthogonal optimization of basis functions, comprising: a model processing module, an inversion module, an optimization module and a power amplifier module;

[0082] The model processing module is used to process the memory polynomial model using a basis function orthogonal optimization method and historical data to obtain a broadband signal amplification model of the power amplifier module;

[0083] The inversion module is used to invert the broadband signal amplification model to obtain a digital predistortion model;

[0084] The optimization module is used to determine an optimal coefficient set of the digital predistortion model, and optimize the optimal coefficient set based on the historical data, so as to perform predistortion processing on the input signal based on the optimized digital predistortion model;

[0085] The power amplifier module is used to perform power amplification processing on the predistorted signal to obtain a power amplifier output signal.

[0086] Further, refer to Figure 2 , a predistortion compensation module, including a signal processing unit, a signal predistortion distortion compensation unit, an adaptive coefficient extraction unit and a storage unit;

[0087] A coefficient determination unit 21, configured to determine an optimal coefficient set of the digital predistortion model using a least squares method;

[0088] The weighted averaging unit 22 is used to obtain the normalized root mean square value of the historical data, and optimize the optimal coefficient set based on the normalized root mean square value by using a weighted averaging method to obtain optimized digital pre-distortion model coefficients.

[0089] Among them, the input signal of the signal processing unit is x(n), the output of the signal predistortion and distortion compensation unit is z(n), and the output of the power amplifier module is y(n).

[0090] The functions of the above-mentioned modules are described in detail below to further understand the working principle and process of the entire digital predistortion (DPD) system:

[0091] 1. Model processing module:

[0092] This module is responsible for processing the input signal and establishing a broadband signal amplification model of the power amplifier module using the basis function orthogonal optimization method and historical data.

[0093] By selecting a set of orthogonal basis functions (such as polynomial basis functions or Fourier basis functions), a broadband signal amplification model of the power amplifier module is constructed. The use of orthogonal basis functions can ensure that different basis functions do not interfere with each other, thereby improving the accuracy and stability of the model.

[0094] By using historical data collected by the power amplifier module under different operating conditions, the actual behavior of the power amplifier can be modeled more accurately. This data helps capture the nonlinear characteristics of the power amplifier and its response to different signals, ensuring that the model reflects the true system characteristics. The model obtained based on historical data and basis function orthogonal optimization is usually a memory polynomial model, which considers not only the current input signal, but also the influence of past signals, thereby more comprehensively describing the dynamic characteristics of the power amplifier.

[0095] 2. Inversion module:

[0096] This module is mainly responsible for extracting and optimizing the coefficients of the digital predistortion model based on the established broadband signal amplification model.

[0097] In the inversion step of the broadband signal amplification model, the adaptive coefficient extraction unit performs a mathematical inversion operation on the broadband signal amplification model to obtain a digital predistortion model. This process is to design a predistortion model that can compensate for the nonlinear distortion of the power amplifier module.

[0098] 3. Optimization module: In order to determine the best set of coefficients for the digital pre-distortion model, the adaptive coefficient extraction unit uses the least squares method for optimization. The least squares method aims to determine the optimal values ​​of the model parameters by minimizing the sum of squares of the errors between the model output and the actual data.

[0099] The optimal coefficient set obtained by the least squares method is further optimized, and the normalized root mean square value (RMS value) of the historical data is used for weighted averaging, which can balance the impact of different data samples and reduce the interference of noise on coefficient optimization. Through the optimization process, the model can maintain good performance under various working conditions.

[0100] Furthermore, the optimization module is also responsible for performing pre-distortion processing on the input signal using the optimized digital pre-distortion model, thereby compensating for the nonlinear distortion in the power amplifier module.

[0101] Before the signal enters the power amplifier module, the signal predistortion compensation unit preprocesses the input signal. Specifically, the input signal is transformed based on the optimized digital predistortion model so that the signal output by the power amplifier module can be as close to the original ideal signal as possible after nonlinear amplification.

[0102] Distortion compensation can effectively compensate for the nonlinear distortion introduced by the power amplifier module in actual work, thereby improving the linearity and signal quality of the system.

[0103] In this embodiment, a broadband signal amplification model is established by a signal processing unit using a basis function orthogonal optimization method and historical data to ensure that the model can accurately reflect the characteristics of the power amplifier module. The adaptive coefficient extraction unit inverts the broadband signal amplification model to obtain a digital pre-distortion model; then the coefficient set is optimized by the least squares method and weighted average to improve the accuracy of the model. The signal pre-distortion distortion compensation unit performs pre-distortion processing on the input signal based on the optimized digital pre-distortion model to compensate for the nonlinear distortion of the power amplifier module, thereby improving the quality of the output signal. The overall system ensures the efficiency and accuracy of the digital pre-distortion model in practical applications, and helps to improve the overall performance of the power amplifier system.

[0104] Specifically, the weighted average digital predistortion device for orthogonal optimization of basis functions provided in this embodiment is implemented based on the following method flow:

[0105] S1. Send signal data x(n) to the hardware communication system, and obtain the output Z(n) of the digital predistorter compensation module and the output signal y(n) of the RF broadband signal amplification device through the hardware feedback channel, and then enter step S2;

[0106] S2. Perform synchronization alignment on all input and output signals, and then proceed to step S3;

[0107] S3. Perform uniform normalization on all data sets by using the maximum amplitude value of the complete long sequence, and calculate the NRMS value of all data, and then proceed to step S4;

[0108] S4. Scanning based on the entire sequence, using the basis function orthogonal optimization method to process the memory polynomial in the MP model, obtain the parameters of the broadband signal amplification device model, and proceed to step S5;

[0109] S5. Using the inverse method to obtain the inverse model of the power amplifier, that is, the model of the power amplifier digital predistorter, and then proceeding to step S6;

[0110] S6. Using the LS algorithm to determine the effect of the predistorter, the coefficient data set with the best effect of the predistorter is obtained, which is included in step S7;

[0111] S7. Calculate the weights of the predistortion coefficients obtained from each data set, and use a weighted average digital predistortion algorithm to optimize the coefficients to obtain an optimized coefficient set, and then proceed to step S8;

[0112] S8: Send the finally obtained optimization coefficient to the digital pre-distortion unit to perform pre-distortion processing on the input signal.

[0113] S9: Perform weighted average digital predistortion experimental test with orthogonal optimization of basis functions.

[0114] Among them, the above S1-S3 are the specific processes of hardware system deployment.

[0115] Reference Figure 3 The system of this embodiment includes a broadband signal amplifying device, an FPGA module, a spectrum analyzer (VSA) for observation, up and down frequency conversion joints, a digital power supply, an attenuator, a coupler, etc. The power amplifier used in the implementation is a 30MHz-512MHz broadband 100W broadband signal amplifying device designed for research and development.

[0116] A dual-carrier WCDMA signal with a bandwidth of 10MHz was used for testing. A vector signal generator was connected to a broadband signal amplification device to form the transmitter of the digital pre-distortion system. The output signal generated by the power amplifier was connected to a spectrum analyzer through an attenuator to form the feedback channel of the digital pre-distortion.

[0117] The vector signal generator and spectrum analyzer are connected to the FPGA module via a router. The FPGA module is used to generate the input signal of the broadband signal amplifier. The original input signal is then passed through the signal source and the broadband signal amplifier to obtain the output signal of the broadband signal amplifier. The signal is then read back through the FPGA module, and the collected signal is synchronously processed with the input signal to obtain the input and output signals of the broadband signal amplifier. The broadband signal amplifier model is then established using the measured experimental signal.

[0118] Using the digital predistortion of the indirect learning structure, the power amplifier digital predistortion model is obtained according to the weighted average digital predistortion algorithm of the orthogonal optimization of the basis functions from S4 to S7 in the above process, and the MP model, the orthogonalized MP model, the AWMP model and the orthogonalized AWMP model are used as controls;

[0119] Specifically, in the experiment, the model selects the first 47,000 data of the test signal for modeling the broadband signal amplification device. In order to verify the modeling accuracy of different models, 8,000 data of the sample signal are selected for modeling.

[0120] The first analog-to-digital converter collects the signal sent by the previous intermediate frequency signal through low-noise broadcasting as a digital signal and encodes it to generate a digital coded signal. The number of sampling bits used is 16 bits, and the sampling frequency is mainly determined by the intermediate frequency signal carrier frequency f0 and bandwidth B0 of the actual input signal. The sampling frequency f>2f0 and the sampling bandwidth B>5B0 are required.

[0121] Using the inverse model parameters of the broadband signal amplifying device, a pre-distortion operation is performed on the verification signal in the signal generated by the signal generator to obtain a compensation signal, and finally obtain a compensated output signal of the broadband signal amplifying device;

[0122] Reference Figure 4 and Figure 5 , Figure 4 This is a characteristic diagram of the broadband signal amplification device before and after compensation in this application; Figure 5 This is a comparison chart of the modeling accuracy of this application and other models;

[0123] from Figure 4 and Figure 5 It can be seen that with the improvement of the training algorithm, the restored power amplifier signal power spectrum is closer to the original power amplifier signal power spectrum. By comparison, it can be seen that the power amplifier signal restored by the weighted average digital predistortion method with orthogonal optimization of basis functions is closer to the original power amplifier signal power spectrum.

[0124] Further, refer to Figure 6 As shown in Figure (a), this is the output power spectrum modeling accuracy curve of the present application. For ease of observation, the local amplified modeling curves of the upper and lower sidebands are given in Figures (b) and (c) respectively. It can be clearly seen that the present application maintains a high modeling accuracy, especially the modeling accuracy of the out-of-band extension distortion. The main reason for the deterioration of the modeling accuracy of the upper and lower sidebands by the MP model is that the cross-memory effect parameters are ignored in the model. A detailed modeling error curve is given in Figure (d). From the error curve in Figure (d), it can be seen that the traditional MP model also has distortion errors in the modeling within the band, which is mainly caused by the correlation of the basis functions. It can be seen that the weighted average digital pre-distortion method with orthogonal optimization of the basis functions proposed in the present application has significantly reduced the problem of modeling distortion within the band.

[0125] The parameter values ​​are N=8000 and the short sequence length L=500. Under different values ​​of the number of model coefficients K, the coefficient identification time of different methods is analyzed and compared. As can be seen from Table 4.2, when the coefficient K=40, the embodiment of the present application saves 63% of the identification time compared with the LS algorithm of the original overall sample; when the coefficient K=107, the embodiment of the present application saves 77.48% of the identification time compared with the traditional LS algorithm of the original overall sample, and the more coefficients, the faster the running time of the present application. The main reason is that the present application effectively reduces the operation time of matrix multiplication by splitting the large matrix, which is of great significance and advantage for the realization of digital hardware circuits.

[0126] After DPD compensation, NRMSE is reduced from 14.15% to 0.49%, and the adjacent channel ACPR is improved from -36.67 / -33.68dBc to -65 / -62.78dBc when the carrier is 10MHz away, and from -47.44 / -48.87dBc to -68.04 / -66.6dBc when the carrier is 0MHz away, and the out-of-band interference is improved by 20dB. It can be seen that the present application effectively overcomes the bias convergence problem of the polynomial model, and is an excellent modeling and training method.

[0127] Optionally, the process of solving the optimal coefficient set includes:

[0128] Determine a basis function matrix including each basis function vector, determine a transposed matrix of the basis function matrix, and construct a power amplifier output vector according to an output signal of the power amplifier module;

[0129] Determine the objective function of the error signal between the digital predistortion model and the power amplifier inverse model according to the power amplifier output signal, the basis function vector and the optimal coefficient set;

[0130] The objective function is solved by using the least square method, and when the objective function takes a minimum value, the optimal coefficient set is determined based on the basis function matrix, the transposed matrix and the power amplifier output vector.

[0131] Specifically, the present application first needs to determine the basis function matrix, denoted as B. The basis function vector is a mathematical function used to describe the characteristics of the power amplifier module, usually including polynomial basis functions, Fourier basis functions or other basis functions suitable for modeling nonlinearity. All basis function vectors are organized into a matrix. Each column of the matrix represents a basis function vector, and each row represents the basis function value of a sample point.

[0132] By solving the least squares problem by the least squares method, the optimal coefficient set can be obtained, which makes the pre-distortion model have the best effect on the nonlinear distortion compensation of the power amplifier module. When the objective function takes the minimum value, the optimal coefficient set is determined based on the basis function matrix, the transposed matrix and the power amplifier output vector. This optimal coefficient set can be used for actual pre-distortion processing to improve the performance of the power amplifier module.

[0133] This application uses the basis function matrix and the least squares method to optimize the coefficients, which can accurately model the nonlinear characteristics of the power amplifier module, and help to more effectively perform signal pre-distortion compensation. The least squares method can minimize the error between the model output and the actual power amplifier output, thereby improving the accuracy of the digital pre-distortion model and reducing the impact of nonlinear distortion on the signal. The optimized coefficient set can significantly improve the performance of the power amplifier module, improve the linearity and quality of the signal, reduce signal distortion and interference, and thus improve the reliability and efficiency of the entire system.

[0134] Optionally, the pre-distortion compensation module solves the broadband signal amplification model in the following manner:

[0135] That is, given the input signal x(n), the impact of the input signal at the historical moment on the output of the power amplifier is introduced, and the signal is processed using the MP model to obtain the output y(n) expressed as:

[0136]

[0137] Among them, φ 2k+1 (z(nl)) is the polynomial basis function of the memory polynomial model, and the polynomial basis function is used to describe the nonlinear characteristics of the power amplifier module. 2k+1,l are the coefficients of the polynomial basis function, K is the highest nonlinear order of the broadband signal amplification model, L is the memory depth of the memory polynomial model, y(n) is the signal after predistortion processing, z(n) is the power amplifier output signal, l is the number of memory layers of the power amplifier memory polynomial model, and k is the nonlinear order of the broadband signal amplification model.

[0138] In this embodiment, the linear component and nonlinear component of y(n) can be expressed as follows using the Gram-Schmidt method:

[0139]

[0140] Optionally, the objective function of the error signal between the digital predistortion model and the power amplifier inverse model is determined by the following formula:

[0141]

[0142] in, is the output signal of the power amplifier inverse model, β n is the predistortion coefficient of the memory polynomial model, B k (n) is the kth basis function of the digital predistortion model, B(n) is the basis function vector, and β is the optimal coefficient set;

[0143] The kth basis function of the digital predistortion model is defined as:

[0144]

[0145] The basis function vector is defined as:

[0146] B(n)=[B1(n),B3(n),...,B k (n)]

[0147] Among them, a kq is the coefficient, x(nq) is the value of the input signal at time nq, q is the number of memory layers of the predistortion memory polynomial model, and Q represents the memory depth of the predistortion memory polynomial model;

[0148] The objective function of the error signal between the digital pre-distortion model and the power amplifier inverse model is expressed as:

[0149]

[0150] The objective function is expressed as follows by the least squares method:

[0151]

[0152] Where N is the number of data samples for model training;

[0153] Then the optimal coefficient set to be solved is:

[0154] β=(B H B) -1 B H z

[0155] z=[z(1),z(2),...,z(N)] T

[0156]

[0157] Among them, z is the output vector of the power amplifier inverse model, and B is the basis function matrix.

[0158] This application defines the basis function B k (n) is the sum of the basis function vectors B(n). The digital pre-distortion model can effectively capture the nonlinear characteristics and enhance the adaptability under different signal conditions. The application of the least squares method not only simplifies the parameter optimization process, but also improves the pre-distortion effect, significantly reducing the error between the output signal and the power amplifier model, thereby improving the system performance.

[0159] The error signal objective function of the digital pre-distortion model and the power amplifier inverse model in the embodiment of the present application provides an effective technical solution for improving the nonlinear distortion compensation of modern broadband communication systems, which not only optimizes the signal quality, but also improves the overall performance and reliability of the system.

[0160] Optionally, the pre-distortion compensation module obtains optimized digital pre-distortion model coefficients in the following manner:

[0161] Based on the normalized RMS value of the historical data, the weights of the digital predistortion model coefficients are calculated as shown in the following formula:

[0162]

[0163] Among them, σ i is the weight of the digital pre-distortion model coefficient, NRMS i is the normalized RMS value of i training samples, G is the number of training sample sequences;

[0164] By weighting the coefficients of the same order, the final weight expression is:

[0165]

[0166] Where k is the nonlinear order of the model, is the optimized k-th order digital predistortion model coefficient, α i,k is the k-th order model coefficient corresponding to the i-th training sample.

[0167] Optionally, the predistortion compensation module is further used for:

[0168] Determine the root mean square of each training data, and sort the root mean square of each training data;

[0169] A small weight factor is assigned to training data with a large RMS value, and a large weight factor is assigned to training data with a small RMS value.

[0170] Specifically, in this embodiment, the weights of the digital pre-distortion model coefficients are calculated based on the normalized root mean square value (NRMS) of historical data, aiming to optimize the model coefficients of each order by weighted average, thereby improving the accuracy and adaptability of the model.

[0171] The normalized root mean square value of historical data is used to define the weights, so that the model can be more in line with the actual situation, thereby improving data adaptability. The influence of different samples is adjusted dynamically, which is conducive to training sets with rich data characteristics.

[0172] The introduction of weights in this embodiment reduces the negative impact of outliers or noise on the overall model, ensuring the stability and consistency of the model output. Through the weighted average method, in-depth analysis of historical data can be performed during the model design process, thereby promoting the continuous improvement of the pre-distortion model performance.

[0173] The present application also provides a weighted average digital predistortion method for orthogonal optimization of basis functions, characterized by comprising:

[0174] The memory polynomial model is processed by using a basis function orthogonal optimization method and historical data to obtain a broadband signal amplification model of the power amplifier module;

[0175] Inverting the broadband signal amplification model to obtain a digital predistortion model;

[0176] Determining an optimal coefficient set of the digital predistortion model, and optimizing the optimal coefficient set based on the historical data, so as to perform predistortion processing on the input signal based on the optimized digital predistortion model;

[0177] The predistorted signal is power amplified to obtain a power amplifier output signal.

[0178] Optionally, determining an optimal coefficient set of the digital pre-distortion model includes:

[0179] Determine a basis function matrix including each basis function vector, determine a transposed matrix of the basis function matrix, and construct a power amplifier output vector according to an output signal of the power amplifier module;

[0180] Determine the objective function of the error signal between the digital predistortion model and the power amplifier inverse model according to the power amplifier output signal, the basis function vector and the optimal coefficient set;

[0181] The objective function is solved by using the least square method, and when the objective function takes a minimum value, the optimal coefficient set is determined based on the basis function matrix, the transposed matrix and the power amplifier output vector.

[0182] Reference Figure 7 Based on the method in the above embodiment, the embodiment of the present application provides an electronic device, which may include: a processor (processor) 710, a communication interface (Communications Interface) 720, a memory (memory) 730 and a communication bus 740, wherein the processor 710, the communication interface 720, and the memory 730 communicate with each other through the communication bus 740. The processor 710 can call the logic instructions in the memory 730 to execute the method in the above embodiment.

[0183] In addition, the logic instructions in the above-mentioned memory 730 can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application.

[0184] Based on the method in the above embodiment, an embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program runs on a processor, the processor executes the method in the above embodiment.

[0185] Based on the method in the above embodiment, an embodiment of the present application provides a computer program product. When the computer program product runs on a processor, the processor executes the method in the above embodiment.

[0186] It is understandable that the processor in the embodiment of the present application can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. The general-purpose processor can be a microprocessor or any conventional processor.

[0187] The method steps in the embodiments of the present application can be implemented by hardware or by a processor executing software instructions. The software instructions can be composed of corresponding software modules, and the software modules can be stored in random access memory (RAM), flash memory, read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), registers, hard disks, mobile hard disks, CD-ROMs, or any other form of storage medium known in the art. An exemplary storage medium is coupled to a processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can be located in an ASIC.

[0188] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted through the computer-readable storage medium. The computer instructions may be transmitted from a website site, a computer, a server or a data center to another website site, a computer, a server or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or a data center that includes one or more available media integrated. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid state disk (SSD)), etc.

[0189] It should be understood that the various numerical numbers involved in the embodiments of the present application are only used for the convenience of description and are not used to limit the scope of the embodiments of the present application.

[0190] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A weighted average digital predistortion device with orthogonal optimization of basis functions, characterized in that: include: Model processing module, inversion module, optimization module and power amplifier module; The model processing module is used to process the memory polynomial model using a basis function orthogonal optimization method and historical data to obtain a broadband signal amplification model of the power amplifier module; The inversion module is used to invert the broadband signal amplification model to obtain a digital predistortion model; The optimization module is used to determine an optimal coefficient set of the digital predistortion model, and optimize the optimal coefficient set based on the historical data, so as to perform predistortion processing on the input signal based on the optimized digital predistortion model; The power amplifier module is used to perform power amplification processing on the predistorted signal to obtain a power amplifier output signal.

2. The device according to claim 1, characterized in that The process of solving the optimal coefficient set includes: Determine a basis function matrix including each basis function vector, determine a transposed matrix of the basis function matrix, and construct a power amplifier output vector according to an output signal of the power amplifier module; Determine the objective function of the error signal between the digital predistortion model and the power amplifier inverse model according to the power amplifier output signal, the basis function vector and the optimal coefficient set; The objective function is solved by using the least square method, and when the objective function takes a minimum value, the optimal coefficient set is determined based on the basis function matrix, the transposed matrix and the power amplifier output vector.

3. The device according to claim 1, characterized in that The pre-distortion compensation module solves the broadband signal amplification model in the following way: Among them, φ 2k+1 (z(nl)) is the polynomial basis function of the memory polynomial model, and the polynomial basis function is used to describe the nonlinear characteristics of the power amplifier module. 2k+1,l are the coefficients of the polynomial basis function, K is the highest nonlinear order of the broadband signal amplification model, L is the memory depth of the memory polynomial model, y(n) is the signal after predistortion processing, z(n) is the power amplifier output signal, l is the number of memory layers of the power amplifier memory polynomial model, and k is the nonlinear order of the broadband signal amplification model.

4. The device according to claim 2, characterized in that The objective function of the error signal between the digital pre-distortion model and the power amplifier inverse model is determined by the following formula: in, is the output signal of the power amplifier inverse model, β n is the predistortion coefficient of the memory polynomial model, B k (n) is the kth basis function of the digital predistortion model, B(n) is the basis function vector, and β is the optimal coefficient set; The kth basis function of the digital predistortion model is defined as: The basis function vector is defined as: B(n)=[B1(n),B3(n),...,B k (n)] Among them, a kq is the coefficient, x(nq) is the value of the input signal at time nq, q is the number of memory layers of the predistortion memory polynomial model, and Q represents the memory depth of the predistortion memory polynomial model; The objective function of the error signal between the digital pre-distortion model and the power amplifier inverse model is expressed as: The objective function is expressed as follows by the least squares method: Where N is the number of data samples for model training; Then the optimal coefficient set to be solved is: β=(B H B) -1 B H z from=[from(1),from(2),...,from(N)] T Among them, z is the output vector of the power amplifier inverse model, and B is the basis function matrix.

5. The device according to claim 1, characterized in that The pre-distortion compensation module obtains the optimized digital pre-distortion model coefficients in the following manner: Based on the normalized RMS value of the historical data, the weights of the digital predistortion model coefficients are calculated as shown in the following formula: Among them, ω i is the weight of the digital pre-distortion model coefficient, NRMS i is the normalized RMS value of i training samples, G is the number of training sample sequences; By weighting the coefficients of the same order, the final weight expression is: Where k is the nonlinear order of the model, is the optimized k-th order digital predistortion model coefficient, α i,k is the k-th order model coefficient corresponding to the i-th training sample.

6. The device according to claim 5, characterized in that The pre-distortion compensation module is also used for: Determine the root mean square of each training data, and sort the root mean square of each training data; A small weight factor is assigned to training data with a large RMS value, and a large weight factor is assigned to training data with a small RMS value.

7. The device according to claim 1, characterized in that The optimization module includes: A coefficient determination unit, used to determine an optimal coefficient set of the digital predistortion model using a least squares method; The weighted averaging unit is used to obtain the normalized root mean square value of the historical data, and optimize the optimal coefficient set based on the normalized root mean square value by using the weighted averaging method to obtain the optimized digital pre-distortion model coefficients.

8. A weighted average digital predistortion method for orthogonal optimization of basis functions, characterized in that: include: The memory polynomial model is processed by using a basis function orthogonal optimization method and historical data to obtain a broadband signal amplification model of the power amplifier module; Inverting the broadband signal amplification model to obtain a digital predistortion model; Determining an optimal coefficient set of the digital predistortion model, and optimizing the optimal coefficient set based on the historical data, so as to perform predistortion processing on the input signal based on the optimized digital predistortion model; The predistorted signal is power amplified to obtain a power amplifier output signal.

9. The method according to claim 8, characterized in that Determining an optimal coefficient set of the digital pre-distortion model includes: Determine a basis function matrix including each basis function vector, determine a transposed matrix of the basis function matrix, and construct a power amplifier output vector according to an output signal of the power amplifier module; Determine the objective function of the error signal between the digital predistortion model and the power amplifier inverse model according to the power amplifier output signal, the basis function vector and the optimal coefficient set; The objective function is solved by using the least square method, and when the objective function takes a minimum value, the optimal coefficient set is determined based on the basis function matrix, the transposed matrix and the power amplifier output vector.

10. An electronic device, characterized in that: include: at least one memory for storing a computer program; At least one processor is used to execute the program stored in the memory. When the program stored in the memory is executed, the processor is used to execute the method according to any one of claims 8 to 9.

Images