Method and system for realizing DPD (Digital Pre-Distortion) with limited transmitter bandwidth
By constructing in-band and out-of-band error weight matrices and optimizing digital predistortion parameters using the Newton-Raphson iteration method, the performance limitations of the DPD system under transmitter bandwidth constraints are solved, achieving excellent linearization of the transmitted signal and meeting in-band and out-of-band performance requirements.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANGHAI JIAOTONG UNIV
- Filing Date
- 2024-11-18
- Publication Date
- 2026-05-19
AI Technical Summary
Existing DPD systems cannot obtain full-band feedback signals or accurate digital predistortion parameters when transmitter bandwidth is limited, resulting in limited power amplifier compensation effects and failing to meet the in-band and out-of-band performance requirements of the transmitted signal.
By constructing in-band and out-of-band error weight matrices, the digital predistortion parameters are optimized using the Newton-Raphson iteration method to meet the in-band and out-of-band performance requirements of the transmitted signal at the antenna end. The specific steps include constructing the objective function for minimizing the full-band error and using the Newton-Raphson iteration method to iteratively update the digital predistortion parameters.
Despite limited transmitter bandwidth, excellent linearization performance of the PA output signal was achieved, meeting the requirements of in-band EVM and out-of-band ACLR, and improving the system's in-band and out-of-band performance balance.
Smart Images
Figure CN122069141A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a technology in the field of wireless communication, specifically a method and system for implementing digital predistortion (DPD) with limited transmitter bandwidth. Background Technology
[0002] Existing DPD systems typically require transmit and feedback links with bandwidths 3-5 times that of the original input signal to operate. The wideband transmit link accommodates the out-of-band spectral spread characteristics of the predistorted signal, while the wideband feedback link captures the main nonlinear behavior of the power amplifier (PA) output. Current predistortion techniques, when transmitter bandwidth is limited, cannot obtain full-band feedback signals or accurate digital predistortion parameters, resulting in limited compensation for the power amplifier and failing to meet the in- and out-of-band performance requirements of the transmitted signal. Summary of the Invention
[0003] To address the aforementioned shortcomings of existing technologies, this invention proposes a digital predistortion implementation method and system with limited transmitter bandwidth, which can meet the in-band and out-of-band performance requirements of the system's transmitted signal even when the transmit low-pass filter (Tx LPF) bandwidth of a broadband nonlinear transmitter is limited.
[0004] This invention is achieved through the following technical solution:
[0005] This invention relates to a method for implementing digital predistortion with limited transmitter bandwidth. Based on the in-band and out-of-band error weight matrix between the baseband input signal and the power amplifier (PA) output signal, an objective function for minimizing the full-band error is constructed. After obtaining the optimal parameters of the DPD model through Newton's iteration method, the PA is linearized to meet the in-band and out-of-band performance requirements of the transmitted signal at the antenna end.
[0006] The in-band and out-of-band error weight matrix A = diag(a1, a2, ..., a... N ), where: diag(.) represents a diagonal matrix, a1, a2, ..., a N The weight values represent the full-band error, and N represents the dimension of the Discrete Fourier Transform (DFT). The specific method for obtaining this in-band and out-of-band error weight matrix is as follows: based on the full-band error between the PA output signal and the baseband input signal, combined with the in-band error vector magnitude (EVM) of the transmitter's transmitted signal and the out-of-band adjacent channel power leakage ratio (ACLR) requirement, appropriate error weight values are set.
[0007] The objective function of the full-band error minimization problem is: Where: b is the digital predistortion parameter vector to be solved, and A is the set diagonal matrix of in-band and out-of-band error weights. and These are the frequency domain representations of the PA output signal and the baseband input signal, respectively. x = W N x, The error represents the full-bandwidth error, where x and y represent the time-domain baseband input signal and the PA output signal, respectively, and W... N For an N-point DFT matrix
[0008] The Newton iteration method, specifically the second-order Newton iteration method, includes:
[0009] Step a: Iteratively update the digital predistortion parameter vector Where: s is the number of iterations, μ is the iteration step size (ranging from 0 to 1), and b0 is the initial value of the digital predistortion parameter vector. K1 and L1 represent the nonlinear order and memory depth of the digital predistortion model, respectively. c1 is the linear gain of PA, and the superscript H indicates the conjugate transpose operation. The intermediate matrix... P and h p These are the order and parameters of Tx LPF, respectively, with the superscript T indicating the transpose operation.
[0010] The initial value of the digital predistortion parameter vector is preferably set to b0 = [1, 0, ..., 0].
[0011] Step b: Convert the iterative update formula for the digital predistortion parameter vector in step a into... Then return to step a for iteration until the preset number of iterations S is reached, where: the intermediate matrix D = AW N G;
[0012] Step c: Substitute the obtained digital predistortion parameter vector b into the predistortion formula to obtain the predistortion signal, and then linearize the PA to meet the in-band and out-of-band performance requirements of the transmitted signal at the antenna end, specifically: Where: u(n) is the predistortion signal, and K1 and L1 are the digital predistortion nonlinearity order and memory depth, respectively. These are the corresponding predistortion parameters. Technical effect
[0013] This invention utilizes a digital predistortion direct learning architecture to minimize the full-band error between the PA output signal and the baseband input signal in the frequency domain. Simultaneously, it incorporates an error weighting diagonal matrix into the objective function to distinguish between in-band and out-of-band errors, thus better balancing the system's in-band and out-of-band performance. Compared to existing technologies, this invention can simultaneously meet the in-band EVM and out-of-band ACLR performance requirements of the antenna-end transmitted signal under hardware limitations such as transmitter bandwidth constraints (e.g., TxLPF bandwidth limited to twice the baseband input signal bandwidth). Attached Figure Description
[0014] Figure 1 This is a flowchart of the present invention;
[0015] Figure 2 This is a schematic diagram of a scenario for an example embodiment;
[0016] Figure 3 The power spectral density plot of the PA output signal in the simulation results with a 100MHz bandwidth input signal and a 200MHz Tx LPF bandwidth;
[0017] Figure 4 The power spectral density plot of the PA output signal is the measured result with a 100MHz bandwidth input signal and a 200MHz Tx LPF bandwidth. Detailed Implementation
[0018] like Figure 2 As shown in the figure, this embodiment relates to a DPD implementation system with limited transmitter bandwidth, including: a digital predistortion module, a digital-to-analog converter module, a low-pass filter module, an RF power amplifier module, a transmit antenna module, a coupler module, an analog-to-digital converter module, and a parameter estimation module. Specifically: the digital predistortion module performs predistortion processing on the baseband input signal; the RF converter module converts the digital signal output by the digital predistortion module into an analog signal; the low-pass filter module filters out out-of-band high-frequency signals; the RF power amplifier module amplifies the energy of the analog signal output by the RF converter module; the transmit antenna module transmits the amplified signal; the coupler module couples the PA output to solve for the digital predistortion parameters; the analog-to-digital converter module converts the analog signal from the coupler module into a digital signal; and the parameter estimation module solves for the digital predistortion scheme parameters by minimizing the full-band error between the RF chain input signal and the output signal of the RF converter module.
[0019] Through specific practical experiments, the baseband input signal was set to a 100MHz bandwidth 64-Quadrature Amplitude Modulation (64-QAM) Orthogonal Frequency Division Multiplexing (OFDM) signal with a peak-to-average power ratio (PA) of 12.4dB. The PA used a three-branch parallel Wiener model, and the Tx LPF was modeled using a 120th-order equiripple finite impulse response filter. Furthermore, the in-band and out-of-band error weighting ratio was set to 1:5. The iteration step size was set to μ = 0.4, and the PA's linearization gain was normalized to c1 = 1. The digital predistortion model used a memory polynomial model with a nonlinear order of 7 and a memory depth of 5. A full-band error objective function was constructed, and the digital predistortion parameters were iteratively solved using the Newton-Raphson method. Upon reaching the set number of iterations, the digital predistortion parameters were output and substituted into the predistortion model to linearize the PA, obtaining a transmit signal that met the in-band and out-of-band performance requirements.
[0020] like Figure 3 The figure shows the normalized power spectral density of the PA output signal when the Tx LPF bandwidth is limited to 200MHz. The corresponding ACLR and EVM values are summarized in Table 1. Figure 3 As shown in Table 1, under a 200MHz Tx LPF bandwidth, the nonlinear transmitter bandwidth-limited digital predistortion (weighted FBEM-DPD) proposed in this invention achieves satisfactory linearization performance both inside and outside the band, reaching an ACLR of -46.91dBc and an EVM of -35.56dB, which meets the performance requirements of the system's transmitted signals inside and outside the band.
[0021] Table 1 shows the ACLR and EVM values of the PA output signal in the simulation results with an input signal of 1100MHz bandwidth and a Tx LPF bandwidth of 200MHz.
[0022] The proposed nonlinear transmitter bandwidth-limited digital predistortion scheme was further validated on a test platform including a PC, signal generator, broadband PA, attenuator, and spectrum analyzer. The PA is a broadband GaN PA operating at 1.6 GHz with a gain of 40 dB and an output power of 30 dBm at a 1 dB compression point. After digital-to-analog conversion, Tx LPF, and digital up-conversion, the baseband input signal is converted to an RF signal, then passes through the PA. The PA output signal is coupled back via a coupler, and then obtained as an equivalent PA output baseband signal through MATLAB digital small-conversion and filtering. The in-band and out-of-band error weighting ratio is set to 1:5. The iteration step size is set to μ = 0.4, and the linearized gain of the PA is normalized to c1 = 1. The digital predistortion model uses a memory polynomial model with a nonlinear order of 9 and a memory depth of 5. A full-band error objective function is constructed, and the digital predistortion parameters are solved using the Newton-Raphson iteration method. After reaching the preset number of iterations, the optimized predistortion parameters are output, substituted into the predistortion model, and the PA is linearized to obtain the transmit signal that meets the in-band and out-of-band performance requirements.
[0023] like Figure 4 As shown, the power spectral density diagram of the PA output signal under a 200MHz Tx LPF bandwidth is presented in this invention. Figure 4 As can be seen, the digital predistortion algorithm proposed in this invention still achieves excellent linearization performance. Furthermore, this invention enjoys an out-of-band spectral suppression advantage of nearly 6dB at a frequency of 100MHz.
[0024] Compared with existing technologies, this invention achieves superior performance under the actual hardware constraints of limited Tx LPF bandwidth, based on the full-band error of the baseband input signal and PA output signal. By introducing an error weight diagonal matrix into the objective function to distinguish between in-band and out-of-band errors, the in-band and out-of-band performance of the system is balanced, so that the transmitted signal simultaneously meets the requirements of in-band EVM and out-of-band ACLR.
[0025] The above-described specific implementations can be partially adjusted by those skilled in the art in different ways without departing from the principles and purpose of the present invention. The scope of protection of the present invention is defined by the claims and is not limited to the above-described specific implementations. All implementation schemes within the scope of the claims are bound by the present invention.
Claims
1. A method for implementing digital predistortion with limited transmitter bandwidth, characterized in that, Based on the in-band and out-of-band error weight matrix between the baseband input signal and the power amplifier (PA) output signal, an objective function for minimizing the full-band error is constructed. After obtaining the optimal parameters of the DPD model through Newton's iteration method, the PA is linearized to meet the in-band and out-of-band performance requirements of the transmitted signal at the antenna end.
2. The transmitter bandwidth-limited digital predistortion implementation method according to claim 1, characterized in that, The objective function of the full-band error minimization problem is: Where: b is the digital predistortion parameter vector to be solved, and A is the set diagonal matrix of in-band and out-of-band error weights. and These are the frequency domain representations of the PA output signal and the baseband input signal, respectively. X = W N x, The error represents the full-bandwidth error, where x and y represent the time-domain baseband input signal and the PA output signal, respectively, and W... N For an N-point DFT matrix, 3. The transmitter bandwidth-limited digital predistortion implementation method according to claim 1 or 2, characterized in that, The in-band and out-of-band error weight matrix A = diag(a1, a2, ..., a N ), where: diag(.) represents a diagonal matrix, a1, a2, ..., a N The weight values represent the full-band error, and N represents the dimension of the Discrete Fourier Transform (DFT). The specific method for obtaining this in-band and out-of-band error weight matrix is as follows: based on the full-band error between the PA output signal and the baseband input signal, combined with the in-band error vector magnitude (EVM) of the transmitter's transmitted signal and the out-of-band adjacent channel power leakage ratio (ACLR) requirement, appropriate error weight values are set.
4. The transmitter bandwidth-limited digital predistortion implementation method according to claim 1, characterized in that, The Newton iteration method, specifically the second-order Newton iteration method, includes: Step a: Iteratively update the digital predistortion parameter vector Where: s is the number of iterations, μ is the iteration step size (ranging from 0 to 1), and b0 is the initial value of the digital predistortion parameter vector. K1 and L1 represent the nonlinear order and memory depth of the digital predistortion model, respectively. c1 represents the linear gain of PA, and the superscript H indicates the conjugate transpose operation. P and h p These are the order and parameters of TxLPF, respectively, with the superscript T indicating the transpose operation; Step b: Convert the iterative update formula for the digital predistortion parameter vector in step a into... Then return to step a for iteration until the preset number of iterations S is reached, where: the intermediate matrix D = AW N G; Step c: Substitute the obtained digital predistortion parameter vector b into the predistortion formula to obtain the predistortion signal, and then linearize the PA to meet the in-band and out-of-band performance requirements of the transmitted signal at the antenna end, specifically: Where: u(n) is the predistortion signal, and K1 and L1 are the digital predistortion nonlinearity order and memory depth, respectively. These are the corresponding predistortion parameters.
5. The transmitter bandwidth-limited digital predistortion implementation method according to claim 4, characterized in that, The initial value of the digital predistortion parameter vector is b0 = [1, 0, ..., 0].
6. A DPD implementation system for implementing any one of the methods described in claims 1-5, characterized in that, include: The system comprises a digital predistortion module, a digital-to-analog converter (DAC) module, a low-pass filter module, an RF power amplifier module, a transmit antenna module, a coupler module, an analog-to-digital converter (ADC) module, and a parameter estimation module. Specifically: the digital predistortion module performs predistortion processing on the baseband input signal; the DAC module converts the digital signal output from the digital predistortion module into an analog signal; the low-pass filter module filters out out-of-band high-frequency signals; the RF power amplifier module amplifies the output analog signal from the DAC module; the transmit antenna module transmits the amplified signal; the coupler module couples the PA output and solves for the digital predistortion parameters; the ADC module converts the analog signal from the coupler module into a digital signal; and the parameter estimation module solves for the digital predistortion scheme parameters by minimizing the full-band error between the RF chain input signal and the ADC module output signal.