Digital pre-distortion method and system for spectrum derivation and harmonic proliferation of double-frequency concurrent signals

Abstract

The invention belongs to the field of signal linearization in a parallel double-frequency transmitter, and particularly relates to a digital pre-distortion method and system for frequency spectrum derivation and harmonic proliferation of double-frequency concurrent signals, and the method comprises the steps: synthesizing the double-frequency signals, and obtaining the double-frequency concurrent signals; inputting the dual-frequency concurrent signal into a power amplifier to obtain an output signal; performing modeling on spectrum derivation and harmonic proliferation of the double-frequency concurrent signals to construct a power amplifier model; a forward signal and a feedback signal are collected, and the forward signal and the feedback signal are aligned through a signal frequency domain cross-correlation algorithm; estimating a coefficient of an inverse model of the power amplifier model according to the aligned forward signal and feedback signal, and updating the estimated coefficient to a digital predistorter; according to the method, the power amplifier model is constructed by performing accurate modeling on the higher harmonics and the intermodulation distortion of the double-frequency concurrent signals, the coefficient of the inverse model of the power amplifier model is updated to the digital predistorter in combination with a feedback mechanism, and the updated digital predistorter can effectively suppress harmonic derivation and spectrum pollution.

Description

Technical Field

[0001] The invention belongs to the field of signal linearization in a parallel dual-frequency transmitter, and in particular relates to a digital predistortion method and system for spectrum derivation and harmonic proliferation of dual-frequency concurrent signals. Background Art

[0002] In modern wireless communication systems, as communication frequency band resources become increasingly scarce, how to improve spectrum utilization has become one of the key issues. Concurrent transmission of multi-band signals has gradually become an important means to improve system capacity and efficiency. However, in this transmission mode, the nonlinear effects of the power amplifier (PA) will cause complex distortion phenomena, especially harmonic derivatives and cross-modulation distortion problems. These nonlinear effects will cause the performance of the communication system to degrade, affect the signal transmission quality, and even cause interference to adjacent frequency bands.

[0003] Harmonic problem is one of the typical manifestations of nonlinear effects of power amplifiers. When driven by high-power signals, power amplifiers will generate multi-order harmonics. These harmonic signals not only fall within the original signal frequency band, but also extend to higher frequency ranges, causing spectrum pollution. This spectrum expansion phenomenon is particularly significant in the concurrent transmission of dual-frequency signals. In particular, the presence of high-order harmonics will cause serious interference to other communication signals, especially affecting the normal operation of adjacent channel communication systems.

[0004] Traditional single-frequency digital pre-distortion (DPD) technology aims to reduce nonlinear distortion by modeling the nonlinear characteristics of the power amplifier and correcting the input signal through pre-distortion technology. However, single-frequency DPD technology is difficult to effectively deal with multi-frequency concurrent signals, especially the harmonics and cross-modulation distortion of dual-frequency signals. This is because the mutual influence of multi-frequency signals will not only aggravate the nonlinear effect of the power amplifier, but also cause intermodulation distortion and harmonic proliferation problems, which brings great challenges to the stability of the communication system.

[0005] In order to deal with the nonlinear distortion problem in multi-frequency concurrent signals, a technology based on a two-dimensional predistortion (2-D DPD) model has been proposed in recent years. This technology can more effectively correct the nonlinear behavior of the power amplifier by simultaneously considering the nonlinear distortion and intermodulation effect of the dual-frequency signals.

[0006] However, although the two-dimensional pre-distortion model performs well in suppressing intermodulation distortion, it is not very effective in suppressing higher harmonics. The complexity and spectrum diffusion effect of higher harmonics pose great challenges to the pre-distortion model, especially in communication systems with a wider frequency range, where the presence of higher harmonics has a significant impact on the overall performance of the system. Summary of the invention

[0007] In order to solve the above-mentioned problems in the prior art, the present invention adopts a digital predistortion method for spectrum derivation and harmonic proliferation of dual-frequency concurrent signals, including: obtaining a dual-frequency signal, inputting the dual-frequency signal into a trained digital predistorter to obtain a predistorted dual-frequency signal, synthesizing the predistorted dual-frequency signal to obtain a predistorted dual-frequency concurrent signal, and inputting the predistorted dual-frequency concurrent signal into a power amplifier PA to obtain an output signal; the training process of the digital predistorter includes:

[0008] S1. Acquire a dual-frequency signal, synthesize the dual-frequency signal, and obtain a dual-frequency concurrent signal; input the dual-frequency concurrent signal into a power amplifier PA to obtain an output signal;

[0009] S2, constructing a power amplifier model by modeling the spectrum derivative and harmonic proliferation of the dual-frequency concurrent signal;

[0010] S3, collecting the forward signal in the dual-frequency signal, collecting the feedback signal in the output signal, and aligning the forward signal and the feedback signal by using a signal frequency domain cross-correlation algorithm;

[0011] S4. Estimate coefficients of an inverse model of the power amplifier model according to the aligned forward signal and feedback signal, and update the estimated coefficients to the digital predistorter to obtain a trained digital predistorter.

[0012] The advantages and beneficial effects of the present invention are as follows:

[0013] Compared with the traditional pre-distortion method, the present invention constructs a power amplifier model by accurately modeling the high-order harmonics and intermodulation distortion of the dual-frequency concurrent signal, estimates the coefficients of the inverse model of the power amplifier model, and updates the coefficients of the inverse model of the power amplifier model to the digital pre-distorter in combination with the feedback mechanism. The updated digital pre-distorter can effectively suppress harmonic derivatives and spectrum pollution, so that the output signal after passing through the PA is closer to the ideal state, significantly improving the transmission performance of the multi-band communication system, and meeting the needs of modern wireless communications for efficient and reliable signal transmission. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 is a flow chart of a digital predistortion method for dual-frequency concurrent signal spectrum derivation and harmonic proliferation provided by an embodiment of the present invention;

[0015] Figure 2 is a block diagram of a digital predistortion system for dual-frequency concurrent signal spectrum derivation and harmonic generation provided by an embodiment of the present invention;

[0016] Figure 3 is a flow chart of aligning a forward signal and a feedback signal by using a signal frequency domain cross-correlation algorithm provided by an embodiment of the present invention;

[0017] Figure 4It is a comparison diagram of upper frequency band power density spectra after three digital predistortion methods provided by an embodiment of the present invention;

[0018] Figure 5 It is a comparison diagram of the power density spectrum of the lower frequency band after three digital pre-distortion methods provided by the embodiment of the present invention;

[0019] Figure 6 It is a schematic diagram of the harmonic and out-of-band intermodulation suppression provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] like Figure 1 As shown, the present invention adopts a digital predistortion method for dual-frequency concurrent signal power amplifier spectrum derivation and harmonic proliferation, including: obtaining a dual-frequency signal, inputting the dual-frequency signal into a trained digital predistorter to obtain a predistorted dual-frequency signal, synthesizing the predistorted dual-frequency signal to obtain a predistorted dual-frequency concurrent signal, and inputting the predistorted dual-frequency concurrent signal into a power amplifier PA to obtain an output signal; the training process of the digital predistorter includes:

[0022] S1. Acquire a dual-frequency signal, synthesize the dual-frequency signal, and obtain a dual-frequency concurrent signal; input the dual-frequency concurrent signal into a power amplifier PA to obtain an output signal;

[0023] Obtaining dual-frequency signals includes:

[0024] The initial upper frequency band signal s1[n] and the initial lower frequency band signal s2[n] are obtained by using DDS (direct digital synthesis) technology, and the signal s1[n] and the signal s2[n] are frequency-converted to obtain the upper frequency band signal s out_1 [n] and the lower band signal s out_2 [n];

[0025] Acquiring an initial upper frequency band signal and an initial lower frequency band signal includes:

[0026] DDS includes: a phase accumulator and a triangular ROM. DDS generates a control frequency word CF and uses the control frequency word CF as the address value of the triangular ROM. The control frequency word CF carries the frequency information of the signal, the resolution of the frequency adjustment, and the phase increment. The triangular ROM output frequency is determined by the following relationship:

[0027]

[0028] Among them, f cos is the output frequency signal, f clk is the clock frequency of the DDS, and N is the bit width of the phase accumulator.

[0029] Performing signal frequency conversion on the initial upper frequency band signal and the initial lower frequency band signal respectively includes:

[0030] The center frequencies of the frequency conversion are f1 and f2 respectively; the initial upper frequency band signal and the initial lower frequency band signal are multiplied by the sine and cosine signal sequences with the center frequencies f1 and f2 respectively to obtain the upper frequency band signal s out_1 [n] and the lower band signal s out_2 [n]:

[0031] s out_1 [n] = s1[n] × cos(2πf c1 nT s )+j×s1[n]×sin(2πf c1 nT s )

[0032] s out_2 [n] = s2[n] × cos(2πf c2 nT s )+j×s2[n]×sin(2πf c2 nT s )

[0033] Among them, T s is the sampling time interval between the initial upper-band signal and the initial lower-band signal, n is the time variable of the signal, and j is the imaginary unit.

[0034] The synthesis of dual-frequency signals includes:

[0035] The signal s out_1 [n] and signal s out_2 [n] is converted into an analog signal to obtain an analog signal s out_1 (n) and the analog signal s out_2 (n), the analog signal s out_1 (n) and the analog signal s out_2 (n) is synthesized to obtain the dual-frequency concurrent signal s(n)=s out_1 (n)+s out_2 (n); where n is the time variable of the signal.

[0036] S2, constructing a power amplifier model by modeling the spectrum derivative and harmonic proliferation of the dual-frequency concurrent signal;

[0037] The two-dimensional predistortion (2-D-DPD) model is improved by using the Volterra series model that retains the characteristics of the dual-frequency predistortion signal, and the high-order harmonics and intermodulation distortion of the dual-frequency signal are accurately modeled using the improved two-dimensional predistortion (2-D-DPD) model to obtain a power amplifier model, including:

[0038]

[0039] Among them, a dk 、b dk 、c dk is the coefficient of the power amplifier model, K is the nonlinear order of the power amplifier model, k is the index of the nonlinear order, l is the ergodic value from 0 to k, Q is the memory depth of the power amplifier model, d is the index of the memory depth Q, indicating the number of sampling periods of signal delay, Q1 and Q2 respectively represent the memory depth of the frequency band fitting corresponding to signals s1(n) and s2(n), d1 is the index of the memory depth Q1, d2 is the index of the memory depth Q2, Q=Q1=Q2, M is the delay characterization term, I represents the harmonic order, i represents the index of the harmonic order, s1(n) and s2(n) respectively represent the input signals of the power amplifier model, y1(n) and y2(n) respectively represent the output signals of the power amplifier model.

[0040] Preferably, I=3 (indicating that three orders are taken into consideration, including first-order, second-order and third-order harmonics).

[0041] Preferably, K=5, Q=4, M=4.

[0042] S3, collecting the forward signal in the dual-frequency signal, collecting the feedback signal in the output signal, and aligning the forward signal and the feedback signal by using a signal frequency domain cross-correlation algorithm;

[0043] In the signal simulation signal s out_1 (n) and the analog signal s out_2 (n), respectively collecting the forward signal x1[n] of the upper frequency band and the forward signal x2[n] of the lower frequency band; filtering the output signal y(n) to obtain the output signals of the two frequency bands, collecting the feedback signal y1[n] of the upper frequency band and the feedback signal y2[n] of the lower frequency band from the output signals of the two frequency bands, Figure 3 The method for calculating the signal delay parameter of the present invention specifically comprises:

[0044] S31, perform fast Fourier transform (FFT) on the forward signal x1[n] and the feedback signal y1[n] according to the Blackman window function ω[n], and obtain the frequency domain representation x of the forward signal x1[n] and the feedback signal y1[n] 1,w [k] and y 1,w [k];

[0045] x1,w [k] = FFT(x 1,w [n]) = FFT(x1[n]*ω[n])

[0046] y 1,w [k] = FFT(y 1,w [n]) = FFT(y1[n]*ω[n])

[0047]

[0048]

[0049] Where N is the discrete time length of the signal x1[n] and k is the index in the frequency domain.

[0050] S32, according to the frequency domain representation y 1,w The complex conjugate and frequency domain representation of [k]x 1,w [k] Calculate the frequency domain cross-correlation R of the forward signal x1[n] and the feedback signal y1[n] 1,xy [k]:

[0051] R 1,xy [k] = x 1,w [k]×y 1,w * [k]

[0052] Among them, y 1,w * [k] is the frequency domain representation of y 1,w The complex conjugate of [k].

[0053] S33, frequency domain cross-correlation R 1,xy [k] Perform inverse Fourier transform (IFFT) to obtain the discrete time signal r 1,xy [n], according to the discrete time signal r 1,xy [n] gets the signal delay Δn;

[0054]

[0055] in, Indicates that r 1,xy The time variable n whose absolute value of [n] is the largest.

[0056] S34, adjusting y1[n] according to the signal delay Δn to obtain a signal y′1[n] aligned with x1[n] in the time domain; the alignment method is as follows;

[0057] y′1[n]=y1[n+Δn]

[0058] The alignment of the forward signal x2[n] and the feedback signal y2[n] is similar.

[0059] S4. Estimate coefficients of an inverse model of the power amplifier model according to the aligned forward signal and feedback signal, and update the estimated coefficients to the digital predistorter to obtain a trained digital predistorter.

[0060] The coefficients of the inverse model of the power amplifier model estimated by the least squares estimation algorithm include:

[0061] According to the forward signal x1[n] and the forward signal x2[n], the output matrix X of the inverse model of the power amplifier model is constructed respectively. (1) =[x1[1],...,x1[N]] and the output matrix X (2) =[x2[1],...,x2[N]];

[0062] The input matrix Y of the inverse model of the power amplifier model is constructed based on the aligned feedback signals y′1[n], y′2[n] and the power amplifier model (1) , Y (2) :

[0063]

[0064] in, is estimated by least squares according to the model The input matrix constructed by the feedback signals y′1[n] and y′2[n] is similar. is estimated by least squares according to the model The input matrix constructed by the feedback signal y′1[n], is estimated by least squares according to the model The input matrix constructed by the feedback signal y′1[n], s1(n) and s2(n) correspond to the feedback signals y′1[n] and y′2[n] respectively; similarly, Y (2) .

[0065] According to the output matrix X (1) and the input matrix Y (1) Estimate the coefficients U of the inverse model of the power amplifier model (1) , according to the output matrix P (2) and the input matrix Y (2) Estimate the coefficients U of the inverse model of the power amplifier model (2) :

[0066] U (1) =[A (1) ,B (1) ,C (1) ]=[(Y (1) ) H Y (1) ] -1 (Y (1) )H X (1)

[0067] U (2) =[A (2) ,B (2) ,C (2) ]=[(Y (2) ) H Y (2) ] -1 (Y (2) ) H X (2)

[0068]

[0069] in, is the coefficient of the inverse model of the power amplifier model, A (1) , B (1) , C (1) , A (2) , B (2) , C (2) The coefficients are The coefficient matrix of , H represents conjugate transpose, and -1 represents inversion.

[0070] like Figure 2 As shown, a digital predistortion elimination system for spectrum derivation and harmonic generation of a dual-frequency concurrent signal power amplifier includes:

[0071] DDS, an upper-band signal processing module, a lower-band signal processing module, a synthesizer, a power amplifier PA and a DPD coefficient training module; the upper-band signal processing module includes: a first signal frequency conversion module, a first digital predistorter DPD1, a first switch, a second switch and a first digital-to-analog converter DAC; the lower-band signal processing module includes: a second signal frequency conversion module, a second digital predistorter DPD2, a third switch, a fourth switch and a second digital-to-analog converter DAC;

[0072] The DDS generates an upper-band signal and a lower-band signal, and the input ends of the upper-band signal processing module and the lower-band signal processing module are respectively connected to the DDS to receive the upper-band signal and the lower-band signal; the input end of the synthesizer is respectively connected to the output ends of the upper-band signal processing module and the lower-band signal processing module, and is used to synthesize the signals output by the upper-band signal processing module and the lower-band signal processing module; the input end of the power amplifier PA is connected to the output end of the synthesizer, and is used to amplify the power of the synthesizer output signal; the input end of the DPD coefficient training module is respectively connected to the output ends of the power amplifier PA, the upper-band signal processing module and the lower-band signal processing module; the output end of the DPD coefficient training module is respectively connected to DPD1 of the upper-band signal processing module and DPD2 of the lower-band signal processing module.

[0073] The first switch includes an input terminal, an output terminal 1 and an output terminal 2, and the second switch includes an input terminal 1, an input terminal 2 and an output terminal; the output terminal of the first signal frequency conversion module is connected to the input terminal of the first switch, the output terminal 2 of the first switch is connected to the input terminal of the first digital predistorter DPD1, the output terminal of DPD1 is connected to the input terminal 2 of the second switch, the output terminal of the second switch is connected to the input terminal of the first digital-to-analog converter DAC, and the output terminal 1 of the first switch is connected to the input terminal 1 of the second switch; the output terminal of the first digital-to-analog converter DAC is connected to the input terminal of the synthesizer;

[0074] The third switch includes an input terminal, an output terminal 1 and an output terminal 2, and the fourth switch includes an input terminal 1, an input terminal 2 and an output terminal; the output terminal of the second signal frequency conversion module is connected to the input terminal of the third switch; the output terminal 2 of the third switch is connected to the input terminal of the second digital predistorter DPD2, the output terminal of DPD2 is connected to the input terminal 2 of the fourth switch, the output terminal of the fourth switch is connected to the input terminal of the second digital-to-analog converter DAC, the output terminal 1 of the third switch is connected to the input terminal 1 of the fourth switch; the output terminal of the second digital-to-analog converter DAC is connected to the input terminal of the synthesizer;

[0075] Before the digital predistorter is updated, the output end 1 of the first switch and the input end 1 of the second switch in the upper-band signal processing module are connected, and the output end 1 of the third switch and the input end 1 of the fourth switch of the lower-band signal processing module are connected; after the digital predistorter is updated, the output end 2 of the first switch and the input end 2 of the second switch in the upper-band signal processing module are connected, and the output end 2 of the third switch and the input end 2 of the fourth switch of the lower-band signal processing module are connected.

[0076] like Figure 4 , Figure 5 , Figure 6 As shown, the effect of the digital predistorter (HC-2-D-DPD) of the present invention is better than other methods; wherein, No-DPD means not using a digital predistorter, and 2-D-DPD means a two-dimensional predistortion model.

[0077] The above embodiments further illustrate the purpose, technical solutions and advantages of the present invention in detail. It should be understood that the above embodiments are only preferred implementation modes of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made to the present invention within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A digital predistortion method for dual-frequency concurrent signal spectrum derivation and harmonic proliferation, including: A dual-frequency signal is obtained, the dual-frequency signal is input into a trained digital predistorter to obtain a predistorted dual-frequency signal, the predistorted dual-frequency signal is synthesized to obtain a predistorted dual-frequency concurrent signal, and the predistorted dual-frequency concurrent signal is input into a power amplifier PA to obtain an output signal, wherein the training process of the digital predistorter includes: S1. Acquire a dual-frequency signal, synthesize the dual-frequency signal, and obtain a dual-frequency concurrent signal; input the dual-frequency concurrent signal into a power amplifier PA to obtain an output signal; S2, constructing a power amplifier model by modeling the spectrum derivative and harmonic proliferation of the dual-frequency concurrent signal; S3, collecting the forward signal in the dual-frequency signal, collecting the feedback signal in the output signal, and aligning the forward signal and the feedback signal by using a signal frequency domain cross-correlation algorithm; S4. Estimate coefficients of an inverse model of the power amplifier model according to the aligned forward signal and feedback signal, and update the estimated coefficients to the digital predistorter to obtain a trained digital predistorter.

2. The digital predistortion method for dual-frequency concurrent signal spectrum derivation and harmonic generation according to claim 1, characterized in that: Acquiring the dual-frequency signal includes: acquiring an initial upper frequency band signal s1[n] and an initial lower frequency band signal s2[n], performing signal frequency conversion on the signal s1[n] and the signal s2[n] respectively, and obtaining an upper frequency band signal s out_1 [n] and the lower band signal s out_2 [n]; where n is the time variable of the signal.

3. The digital predistortion method for dual-frequency concurrent signal spectrum derivation and harmonic generation according to claim 2, characterized in that: Performing signal frequency conversion on the initial upper frequency band signal and the initial lower frequency band signal respectively includes: s out_1 [n]=s1[n]×cos(2πf c1 nT s )+j×s1[n]×sin(2πf c1 nT s ) s out_2 [n]=s2[n]×cos(2πf c2 nT s )+j×s2[n]×sin(2πf c2 nT s ) Among them, T s is the sampling time interval between the initial upper-band signal and the initial lower-band signal, f c1 、f c2 is the center frequency of the frequency conversion.

4. The digital predistortion method for dual-frequency concurrent signal spectrum derivation and harmonic generation according to claim 1, characterized in that: The amplifier model is: Among them, a dk (1) , b dk (1) 、c dk (1) 、a dk (2) , b dk (2) 、c dk (2) is the coefficient of the power amplifier model, K is the nonlinear order of the power amplifier model, k is the index of the nonlinear order, l is the ergodic value from 0 to k, Q is the memory depth of the power amplifier model, d is the index of the memory depth Q, Q1 and Q2 represent the memory depths of the frequency band fitting corresponding to signals s1(n) and s2(n) respectively, d1 is the index of the memory depth Q1, d2 is the index of the memory depth Q2, Q=Q1=Q2, M is the delay characterization term, I represents the harmonic order, i represents the index of the harmonic order, s1(n) and s2(n) represent the input signals of the power amplifier model, y1(n) and y2(n) represent the output signals of the power amplifier model.

5. The digital predistortion method for dual-frequency concurrent signal spectrum derivation and harmonic generation according to claim 4, characterized in that: The forward signal includes: a forward signal x1[n] of an upper frequency band and a forward signal x2[n] of a lower frequency band, and the feedback signal includes: a feedback signal y1[n] of an upper frequency band and a feedback signal y2[n] of a lower frequency band; aligning the forward signal x1[n] and the feedback signal y1[n] includes: S31, perform fast Fourier transform on the forward signal x1[n] and the feedback signal y1[n] according to the Blackman window function ω[n], and obtain the frequency domain representation x of the forward signal x1[n] and the feedback signal y1[n] 1,w [k] and y 1,w [k]; where n is the time variable of the signal and k is the index in the frequency domain; S32, according to the frequency domain representation y 1,w The complex conjugate and frequency domain representation of [k]x 1,w [k] Calculate the frequency domain cross-correlation R of the forward signal x1[n] and the feedback signal y1[n] 1,xy [k]; S33, frequency domain cross-correlation R 1,xy [k] performs inverse Fourier transform to obtain the discrete time signal r 1,xy [n], according to the discrete time signal r 1,xy [n] gets the signal delay Δn; S34, adjusting y1[n] according to the signal delay Δn to obtain a feedback signal y′1[n] aligned with the forward signal x1[n]; The way the forward signal x2[n] and the feedback signal y2[n] are aligned is the same as the way the forward signal x1[n] and the feedback signal y1[n] are aligned, and the feedback signal y2′[n] aligned with the forward signal x2[n] is obtained.

6. The digital predistortion method for dual-frequency concurrent signal spectrum derivation and harmonic generation according to claim 5, characterized in that: According to the discrete time signal r xy [n] The signal delay Δn includes: in, Indicates that r 1,xy The time variable n whose absolute value of [n] is the largest.

7. The digital predistortion method for dual-frequency concurrent signal spectrum derivation and harmonic generation according to claim 6, characterized in that: Estimating the coefficients of the inverse model of the power amplifier model includes: using a least squares estimation algorithm to estimate the coefficients of the inverse model of the power amplifier model based on the forward signal x1[n], the forward signal x2[n], the aligned feedback signal y1′[n] and the aligned feedback signal y2′[n].

8. A digital predistortion elimination system using the digital predistortion method according to claims 1 to 7, characterized in that: include: DDS, upper band signal processing module, lower band signal processing module, synthesizer, power amplifier PA and DPD coefficient training module; The upper-band signal processing module and the lower-band signal processing module both include a digital predistorter; wherein the DDS is a direct digital synthesizer; The DDS generates an initial upper-band signal and an initial lower-band signal, and the input ends of the upper-band signal processing module and the lower-band signal processing module are respectively connected to the DDS; the input end of the synthesizer is respectively connected to the output ends of the upper-band signal processing module and the lower-band signal processing module; the input end of the power amplifier PA is connected to the output end of the synthesizer; the input end of the DPD coefficient training module is respectively connected to the output ends of the power amplifier PA, the upper-band signal processing module and the lower-band signal processing module; the output end of the DPD coefficient training module is respectively connected to the digital predistorter of the upper-band signal processing module and the digital predistorter of the lower-band signal processing module.

9. A digital pre-distortion elimination system according to claim 8, characterized in that: The upper-band signal processing module also includes: a signal frequency conversion module, a first switch, a second switch and a digital-to-analog converter DAC; the first switch includes an input end, an output end 1 and an output end 2, and the second switch includes an input end 1, an input end 2 and an output end; the input end of the signal frequency conversion module is connected to the input end of the first switch, the output end 2 of the first switch is connected to the input end of the digital predistorter DPD, the output end of the DPD is connected to the input end 2 of the second switch, the output end of the second switch is connected to the input end of the digital-to-analog converter DAC, and the output end 1 of the first switch is connected to the input end 1 of the second switch; the output end of the digital-to-analog converter DAC is connected to the input end of the synthesizer; the structure of the lower-band signal processing module is consistent with that of the upper-band signal processing module.

10. A digital pre-distortion elimination system according to claim 9, characterized in that: Before the digital predistorter is updated, the output end 1 and the input end 1 of the switch in the upper-band signal processing module and the lower-band signal processing module are connected; after the digital predistorter is updated, the output end 2 and the input end 2 of the switch in the upper-band signal processing module and the lower-band signal processing module are connected.

Images