CN109347452B - Double-frequency power amplifier digital predistortion device and method based on piecewise linear function - Google Patents
Double-frequency power amplifier digital predistortion device and method based on piecewise linear function Download PDFInfo
- Publication number
- CN109347452B CN109347452B CN201811208279.7A CN201811208279A CN109347452B CN 109347452 B CN109347452 B CN 109347452B CN 201811208279 A CN201811208279 A CN 201811208279A CN 109347452 B CN109347452 B CN 109347452B
- Authority
- CN
- China
- Prior art keywords
- digital
- predistorter
- signal
- pass filter
- band
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03F—AMPLIFIERS
- H03F3/00—Amplifiers with only discharge tubes or only semiconductor devices as amplifying elements
- H03F3/20—Power amplifiers, e.g. Class B amplifiers, Class C amplifiers
- H03F3/24—Power amplifiers, e.g. Class B amplifiers, Class C amplifiers of transmitter output stages
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/11—Complex mathematical operations for solving equations, e.g. nonlinear equations, general mathematical optimization problems
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03F—AMPLIFIERS
- H03F1/00—Details of amplifiers with only discharge tubes, only semiconductor devices or only unspecified devices as amplifying elements
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- General Physics & Mathematics (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Power Engineering (AREA)
- Operations Research (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Amplifiers (AREA)
Abstract
The invention discloses a double-frequency power amplifier digital predistortion device based on a piecewise linear function, which comprises a first digital predistorter, a second digital predistorter, a first digital-to-analog converter, a second digital-to-analog converter, a first low-pass filter, a second low-pass filter, a first modulator, a second modulator, a first power divider, a power amplifier, an attenuation coupler, a second power divider, a first band-pass filter, a second band-pass filter, a first demodulator, a second demodulator, a third band-pass filter, a fourth band-pass filter, a first analog-to-digital converter, a second analog-to-digital converter and a predistortion training module. The invention also discloses a double-frequency power amplifier digital predistortion method based on the piecewise linear function. The invention reduces the number of parameters in the existing 2D-DPD dual-band power amplifier predistortion model, reduces the realization difficulty and consumed operation resources of the model on the FPGA, and accelerates the calculation time of the optimal model parameters.
Description
Technical Field
The invention relates to the field of digital predistortion of a multi-band power amplifier, in particular to a digital predistortion device and a digital predistortion method of a dual-band power amplifier based on a piecewise linear function.
Background
In recent years, mobile communication technology has been rapidly developed, and fifth-generation mobile communication (5G) is coming. In the 5G era, a communication system supporting multiple standards and multiple frequency bands simultaneously is one of the hot spots of research. A power amplifier is the most important component of a communication system, and research on a power amplifier supporting a dual band or a multiband is one of important issues in the field of communication technology. In 5G, power amplifiers also face power efficiency issues due to data transmission rate requirements up to the order of Gbps. In order to improve the transmission efficiency of the power amplifier, the power amplifier needs to operate in a saturation region. However, the efficiency and linearity of conventional power amplifiers are contradictory, and a power amplifier operating in the saturation region may generate severe nonlinear distortion, which is expressed as in-band distortion, out-of-band spectral leakage, and memory effect. For dual or multi-band power amplifiers, this non-linearity also manifests as intermodulation distortion in the different bands. In modern communication systems, digital predistortion techniques are often employed to resolve the conflict between the efficiency and linearity of the power amplifier.
An efficient digital predistortion scheme requires a suitable power amplifier model. Most of the existing power amplifier models adopt Volterra series with memory. For a dual-band or multi-band power amplifier, because of intermodulation distortion of different frequency bands, an original single-band power amplifier model needs to be improved, and a cross term capable of describing the intermodulation distortion is introduced to accurately describe the behavior of the power amplifier. The existing dual-band power amplifier model is mostly based on a simplified Volterra polynomial-Memory Polynomial (MP), and the model has reliable modeling precision and can improve adjacent channel power leakage of each frequency band of the dual-band power amplifier. However, when the model describes the intermodulation distortion, a high-order power term is introduced, and the computational complexity of the model is increased, so that the model occupies a large amount of computational resources when being implemented on an FPGA. Therefore, how to simplify the existing dual-band power amplifier model is an important matter of current research.
Disclosure of Invention
The purpose of the invention is as follows: the invention aims to provide a double-frequency power amplifier digital predistortion device and a method based on a piecewise linear function, which can greatly reduce the operation complexity of a behavior model of the existing double-frequency power amplifier and reduce the realization difficulty and consumed operation resources of a double-frequency power amplifier digital predistortion system.
The technical scheme is as follows: in order to achieve the purpose, the invention adopts the following technical scheme:
the invention relates to a double-frequency power amplifier digital predistortion device based on a piecewise linear function, which comprises:
a first digital predistorter: the first digital predistorter comprises a predistortion model based on a piecewise linear function, a digital baseband input signal x of a first frequency band1(n) inputting a first digital predistorter, the first digital predistorter using its own model parametersAnd a digital baseband input signal x of a first frequency band1(n) input into its own predistortion model, which outputs a first predistortion output signal x1-pre(n), the first predistorted output signal x1-pre(n) also as the output signal of the first digital predistorter;
a second digital predistorter: the second digital predistorter comprises a predistortion model based on a piecewise linear function, and a digital baseband input signal x of a second frequency band2(n) inputting a second digital predistorter, wherein the second digital predistorter uses model parameters of the second digital predistorter and a digital baseband input signal x of a second frequency band2(n) input into its own predistortion model, which outputs a second predistortion output signal x2-pre(n), second predistorted output signal x2-pre(n) also as the output signal of the second digital predistorter;
a first digital-to-analog converter: first predistortion output signal x1-pre(n) performing digital-to-analog conversion through a first digital-to-analog converter, and inputting the converted signal into a first low-pass filter for filtering;
a second digital-to-analog converter: second predistorted output signal x2-pre(n) performing digital-to-analog conversion through a second digital-to-analog converter, and inputting the converted signal into a second low-pass filter for filtering;
a first low-pass filter: filtering the signal output by the first digital-to-analog converter;
a second low-pass filter: filtering the signal output by the second digital-to-analog converter;
a first modulator: the signal output by the first low-pass filter is modulated by a first modulator;
a second modulator: the signal output by the second low-pass filter is modulated by the first modulator;
a first power divider: the signal output by the first modulator and the signal output by the second modulator are combined into a signal through the first power divider;
a power amplifier: amplifying the signal combined by the first power divider through a power amplifier;
an attenuation coupler: the signal amplified by the power amplifier is input into an attenuation coupler;
the second power divider: the output signal of the attenuation coupler is divided into two paths of signals through a second power divider, and the two paths of signals are respectively transmitted to a first band-pass filter and a second band-pass filter;
a first band-pass filter: filtering the first path of signals output by the second power divider through a first band-pass filter;
a second band-pass filter: filtering a second path of signals output by the second power divider through a second band-pass filter;
a first demodulator: the signal output by the first band-pass filter is demodulated by a first demodulator;
a second demodulator: the signal output by the second band-pass filter is demodulated by a second demodulator;
a third band-pass filter: filtering the signal output by the first demodulator through a third band-pass filter;
fourth bandpass filter: the signal output by the second demodulator is filtered by a fourth band-pass filter;
a first analog-to-digital converter: the signal output by the third band-pass filter is subjected to analog-to-digital conversion by the first analog-to-digital converter to obtain a first path of digital baseband output signal y1(n)；
A second analog-to-digital converter: the signal output by the fourth band-pass filter is subjected to analog-to-digital conversion by the second analog-to-digital converter to obtain a second path of digital baseband output signal y2(n)；
A predistortion training module: digital baseband input signal x of a first frequency band1(n) digital baseband input signal x of second frequency band2(n) the first path of digital baseband output signal y1(n) and a second digital baseband output signal y2(n) the model parameters of the first digital predistorter and the model parameters of the second predistorter are input into a predistortion training module, the predistortion training module trains the model parameters of the first digital predistorter and the model parameters of the second predistorter, the trained model parameters of the first digital predistorter are copied to the first digital predistorter, and the trained model parameters of the second predistorter are copied to the second digital predistorterA distorter.
The invention relates to a double-frequency power amplifier digital predistortion method based on a piecewise linear function, which comprises the following steps:
s1: the first digital predistorter comprises a predistortion model based on a piecewise linear function, a digital baseband input signal x of a first frequency band1(n) inputting a first digital predistorter, wherein the first digital predistorter inputs model parameters of the first digital predistorter and a digital baseband input signal x of a first frequency band1(n) input into its own predistortion model, which outputs a first predistortion output signal x1-pre(n), the first predistorted output signal x1-pre(n) also as the output signal of the first digital predistorter; the second digital predistorter comprises a predistortion model based on a piecewise linear function, and a digital baseband input signal x of a second frequency band2(n) inputting a second digital predistorter, wherein the second digital predistorter uses model parameters of the second digital predistorter and a digital baseband input signal x of a second frequency band2(n) input into its own predistortion model, which outputs a second predistortion output signal x2-pre(n), second predistorted output signal x2-pre(n) also as the output signal of the second digital predistorter;
s2: first predistortion output signal x1-pre(n) performing digital-to-analog conversion by using a first digital-to-analog converter, wherein the converted signals respectively pass through a first low-pass filter and a first modulator; second predistorted output signal x2-pre(n) performing digital-to-analog conversion by a second digital-to-analog converter, wherein the converted signals respectively pass through a second low-pass filter and a second modulator;
s3: the signal output by the first modulator and the signal output by the second modulator are combined into a signal through the first power divider;
s4: amplifying the signal combined by the first power divider through a power amplifier;
s5: the signal amplified by the power amplifier is input into an attenuation coupler, and the attenuation coupler outputs two paths of signals to a first band-pass filter and a second band-pass filter respectively through a second power divider;
s6: filtering the first path of signals output by the second power divider through a first band-pass filter; filtering a second path of signals output by the second power divider through a second band-pass filter;
s7: the signal output by the first band-pass filter is demodulated by a first demodulator; the signal output by the second band-pass filter is demodulated by a second demodulator;
s8: filtering the signal output by the first demodulator through a third band-pass filter; the signal output by the second demodulator is filtered by a fourth band-pass filter;
s9: the signal output by the third band-pass filter is subjected to analog-to-digital conversion by the first analog-to-digital converter to obtain a first path of digital baseband output signal y1(n); the signal output by the fourth band-pass filter is subjected to analog-to-digital conversion by the second analog-to-digital converter to obtain a second path of digital baseband output signal y2(n)；
S10: digital baseband input signal x of a first frequency band1(n) digital baseband input signal x of second frequency band2(n) the first path of digital baseband output signal y1(n) and a second digital baseband output signal y2And (n) the model parameters of the first digital predistorter and the model parameters of the second predistorter are input into a predistortion training module, the predistortion training module trains the model parameters of the first digital predistorter and the model parameters of the second predistorter, the trained model parameters of the first digital predistorter are copied to the first digital predistorter, and the trained model parameters of the second predistorter are copied to the second digital predistorter.
Further, in step S10, the predistortion training module trains the model parameters of the first digital predistorter and the model parameters of the second predistorter by using a least square method, calculates a normalized mean square error between the digital baseband input signal and the digital baseband output signal under different model parameters, and obtains a cross modulation coefficient with the minimum normalized mean square error by using a binary search method, thereby obtaining the model parameters of the trained first digital predistorter and the model parameters of the trained second predistorter.
Further, in step S10, the predistortion training module includes a dual-frequency predistortion model based on a piecewise linear function, where the dual-frequency predistortion model based on the piecewise linear function is represented by equation (1):
in the formula (1), the reaction mixture is,
representing the model parameters of the kth first digital predistorter when the memory depth of the power amplifier is m;
representing model parameters of a kth second digital predistorter when the memory depth is m; beta is ak (1)A kth segmentation point representing a threshold for the first frequency band; beta is ak (2)A kth segmentation point representing a threshold for a second frequency band;
representing a first cross modulation coefficient between a (k +1) th first frequency band and a (k +1) th second frequency band;
representing a second cross modulation coefficient between the (k +1) th first frequency band and the (k +1) th second frequency band; m represents the total memory depth of the power amplifier, K represents the total threshold number of the piecewise linear function, y1(n-m) represents the first digital baseband output signal at time n-m, y2And (n-m) represents the second path of digital baseband output signal at the moment of n-m.
Further, the value range of the first cross modulation coefficient and the second cross modulation coefficient is [0.8,2 ].
Further, the process for establishing the dual-frequency predistortion model based on the piecewise linear function comprises the following steps:
1) obtaining a digital baseband input signal x of a first frequency band1(n) digital baseband input signal x of second frequency band2(n) the first path of digital baseband output signal y1(n) and secondDigital baseband output signal y2(n) carrying out normalization treatment;
2) selecting a memory depth M and a threshold number K according to the following ranges: m is less than or equal to 5, and K is less than or equal to 10;
3) a kth segmentation point beta according to a threshold of the first frequency bandk (1)The kth segmentation point beta of the threshold of the second frequency bandk (2)Model memory depth M and threshold number K, and the value ranges of the first cross modulation coefficient and the second cross modulation coefficient are [0.8,2]]Searching by an internal binary search method, wherein each search is carried out according to the digital baseband input signal x of the first frequency band by using a least square method1(n) digital baseband input signal x of second frequency band2(n) the first path of digital baseband output signal y1(n) and a second digital baseband output signal y2(n) training the model parameters of the first digital predistorter and the model parameters of the second predistorter, calculating the normalized mean square error of the digital baseband input signal and the digital baseband output signal by using the model parameters of the first digital predistorter and the model parameters of the second predistorter, and taking the corresponding model parameters of the first digital predistorter and the corresponding model parameters of the second predistorter as the trained model parameters of the first digital predistorter and the trained model parameters of the second predistorter when the normalized mean square error is minimum.
Has the advantages that: the invention discloses a double-frequency power amplifier digital predistortion device and method based on a piecewise linear function, which have the following beneficial effects compared with the prior art:
1) the number of parameters in the existing 2D-DPD dual-band power amplifier predistortion model is reduced;
2) absolute value operation is used for replacing high-order power item operation in the existing double-frequency model, so that the difficulty in realizing the model on the FPGA and consumed operation resources are reduced;
3) the model adopts a least square and binary search method, so that the calculation time of the optimal model parameter is accelerated.
Drawings
FIG. 1 is a diagram of a digital predistortion apparatus in an embodiment of the present invention;
FIG. 2 is a comparison graph of power spectra of two frequency bands before and after pre-distortion according to an embodiment of the present invention;
FIG. 2(a) is a comparison graph of the power spectrum at 1.5 GHz;
FIG. 2(b) is a graph comparing the power spectrum at 1.7 GHz.
Detailed Description
The technical solution of the present invention will be further described with reference to the following embodiments.
The specific embodiment of this patent discloses a dual-frequency power amplifier digital predistortion device based on piecewise linear function, as shown in fig. 1, includes:
a first digital predistorter: the first digital predistorter comprises a predistortion model based on a piecewise linear function, a digital baseband input signal x of a first frequency band1(n) inputting a first digital predistorter, wherein the first digital predistorter inputs model parameters of the first digital predistorter and a digital baseband input signal x of a first frequency band1(n) input into its own predistortion model, which outputs a first predistortion output signal x1-pre(n), the first predistorted output signal x1-pre(n) also as the output signal of the first digital predistorter;
a second digital predistorter: the second digital predistorter comprises a predistortion model based on a piecewise linear function, and a digital baseband input signal x of a second frequency band2(n) inputting a second digital predistorter, wherein the second digital predistorter uses model parameters of the second digital predistorter and a digital baseband input signal x of a second frequency band2(n) input into its own predistortion model, which outputs a second predistortion output signal x2-pre(n), second predistorted output signal x2-pre(n) also as the output signal of the second digital predistorter;
a first digital-to-analog converter: first predistortion output signal x1-pre(n) performing digital-to-analog conversion through a first digital-to-analog converter, and inputting the converted signal into a first low-pass filter for filtering;
a second digital-to-analog converter: second predistorted output signal x2-pre(n) performing digital-to-analog conversion by a second digital-to-analog converter, after the conversionThe signal is input into a second low-pass filter for filtering processing;
a first low-pass filter: filtering the signal output by the first digital-to-analog converter;
a second low-pass filter: filtering the signal output by the second digital-to-analog converter;
a first modulator: the signal output by the first low-pass filter is modulated by a first modulator;
a second modulator: the signal output by the second low-pass filter is modulated by the first modulator;
a first power divider: the signal output by the first modulator and the signal output by the second modulator are combined into a signal through the first power divider;
a power amplifier: amplifying the signal combined by the first power divider through a power amplifier;
an attenuation coupler: the signal amplified by the power amplifier is input into an attenuation coupler;
the second power divider: the output signal of the attenuation coupler is divided into two paths of signals through a second power divider, and the two paths of signals are respectively transmitted to a first band-pass filter and a second band-pass filter;
a first band-pass filter: filtering the first path of signals output by the second power divider through a first band-pass filter;
a second band-pass filter: filtering a second path of signals output by the second power divider through a second band-pass filter;
a first demodulator: the signal output by the first band-pass filter is demodulated by a first demodulator;
a second demodulator: the signal output by the second band-pass filter is demodulated by a second demodulator;
a third band-pass filter: filtering the signal output by the first demodulator through a third band-pass filter;
fourth bandpass filter: the signal output by the second demodulator is filtered by a fourth band-pass filter;
a first analog-to-digital converter: the signal output by the third band-pass filter is processed by the first analog-to-digital converterObtaining a first path of digital baseband output signal y after analog-to-digital conversion1(n)；
A second analog-to-digital converter: the signal output by the fourth band-pass filter is subjected to analog-to-digital conversion by the second analog-to-digital converter to obtain a second path of digital baseband output signal y2(n)；
A predistortion training module: digital baseband input signal x of a first frequency band1(n) digital baseband input signal x of second frequency band2(n) the first path of digital baseband output signal y1(n) and a second digital baseband output signal y2And (n) the model parameters of the first digital predistorter and the model parameters of the second predistorter are input into a predistortion training module, the predistortion training module trains the model parameters of the first digital predistorter and the model parameters of the second predistorter, the trained model parameters of the first digital predistorter are copied to the first digital predistorter, and the trained model parameters of the second predistorter are copied to the second digital predistorter.
The double-frequency power amplifier digital predistortion method based on the piecewise linear function comprises the following steps:
s1: the first digital predistorter comprises a predistortion model based on a piecewise linear function, a digital baseband input signal x of a first frequency band1(n) inputting a first digital predistorter, wherein the first digital predistorter inputs model parameters of the first digital predistorter and a digital baseband input signal x of a first frequency band1(n) input into its own predistortion model, which outputs a first predistortion output signal x1-pre(n), the first predistorted output signal x1-pre(n) also as the output signal of the first digital predistorter; the second digital predistorter comprises a predistortion model based on a piecewise linear function, and a digital baseband input signal x of a second frequency band2(n) inputting a second digital predistorter, wherein the second digital predistorter uses model parameters of the second digital predistorter and a digital baseband input signal x of a second frequency band2(n) input into its own predistortion model, which outputs a second predistortion output signal x2-pre(n), second predistorted output signal x2-pre(n) also as the output signal of the second digital predistorter;
s2: first predistortion outputSignal x1-pre(n) performing digital-to-analog conversion by using a first digital-to-analog converter, wherein the converted signals respectively pass through a first low-pass filter and a first modulator; second predistorted output signal x2-pre(n) performing digital-to-analog conversion by a second digital-to-analog converter, wherein the converted signals respectively pass through a second low-pass filter and a second modulator;
s3: the signal output by the first modulator and the signal output by the second modulator are combined into a signal through the first power divider;
s4: amplifying the signal combined by the first power divider through a power amplifier;
s5: the signal amplified by the power amplifier is input into an attenuation coupler, and the attenuation coupler outputs two paths of signals to a first band-pass filter and a second band-pass filter respectively through a second power divider;
s6: filtering the first path of signals output by the second power divider through a first band-pass filter; filtering a second path of signals output by the second power divider through a second band-pass filter;
s7: the signal output by the first band-pass filter is demodulated by a first demodulator; the signal output by the second band-pass filter is demodulated by a second demodulator;
s8: filtering the signal output by the first demodulator through a third band-pass filter; the signal output by the second demodulator is filtered by a fourth band-pass filter;
s9: the signal output by the third band-pass filter is subjected to analog-to-digital conversion by the first analog-to-digital converter to obtain a first path of digital baseband output signal y1(n); the signal output by the fourth band-pass filter is subjected to analog-to-digital conversion by the second analog-to-digital converter to obtain a second path of digital baseband output signal y2(n)；
S10: digital baseband input signal x of a first frequency band1(n) digital baseband input signal x of second frequency band2(n) the first path of digital baseband output signal y1(n) and a second digital baseband output signal y2(n) are input into a predistortion training module, and the predistortion training module is used for carrying out model parameters and second predistortion on the first digital predistorterAnd training the model parameters of the true machine, copying the trained model parameters of the first digital predistorter to the first digital predistorter, and copying the trained model parameters of the second predistorter to the second digital predistorter.
In step S10, the predistortion training module trains the model parameters of the first digital predistorter and the model parameters of the second predistorter by using a least square method, calculates a normalized mean square error between the digital baseband input signal and the digital baseband output signal under different model parameters, and obtains a cross modulation coefficient with the minimum normalized mean square error by using a binary search method, thereby obtaining the model parameters of the trained first digital predistorter and the model parameters of the trained second predistorter.
In step S10, the predistortion training module includes a dual-frequency predistortion model based on a piecewise linear function, where the dual-frequency predistortion model based on the piecewise linear function is represented by equation (1):
in the formula (1), the reaction mixture is,
representing the model parameters of the kth first digital predistorter when the memory depth of the power amplifier is m;
representing model parameters of a kth second digital predistorter when the memory depth is m; beta is ak (1)A kth segmentation point representing a threshold for the first frequency band; beta is ak (2)A kth segmentation point representing a threshold for a second frequency band;
representing a first cross modulation coefficient between a (k +1) th first frequency band and a (k +1) th second frequency band;
representing a second cross modulation coefficient between the (k +1) th first frequency band and the (k +1) th second frequency band; m represents the total memory depth of the power amplifier, K represents the total threshold number of the piecewise linear function, y1(n-m) represents the first digital baseband output signal at time n-m, y2And (n-m) represents the second path of digital baseband output signal at the moment of n-m. Beta is ak (1)And betak (2)Range of values of (1) from 0 to
Are uniformly distributed.
And
using binary search method to find the value in [0.8,2]]Find the optimal intermodulation coefficient in the interval of (2). The specific method comprises the following steps:
1) first, several parameters of a binary search method are defined: the left and right boundaries and the midpoint of the search are respectively cleft，cright，cmiddle＝(cright+cleft)/2. Step Δ c of search ═ c (c)right-cleft)/2。
2) Initialization:
cleft＝0.8
cright＝2.0
cmiddle＝(cright+cleft)/2＝1.4
when c is going to(i)＝cmiddleCalculating the unknown parameters of the model by means of least squares
And calculating the minimum normalized mean square error of the model under the intermodulation coefficient, and recording the minimum normalized mean square error as
3) Starting iteration: each iteration begins, updates first
Δc＝Δc/2
cleft＝cmiddle-Δc/2
cright＝cmiddle+Δc/2
Then, the method uses the least square method to calculate c(i)＝cleft,c(i)＝crightUnknown parameters of the model
And calculating the minimum normalized mean square error under the intermodulation coefficient and recording the minimum normalized mean square error as
Find out
The smallest of them is recorded as
C corresponding to the minimum normalized error(i)Record as
4) Judging whether Delta c is less than or equal to Delta cstop，(Δcstop0.1), if not, continue with execution 3). Otherwise, the iteration is exited and the coefficients c of the model are returned(i)And
the establishment process of the dual-frequency predistortion model based on the piecewise linear function comprises the following steps:
1) obtaining a digital baseband input signal x of a first frequency band1(n) digital baseband input signal x of second frequency band2(n) the first path of digital baseband output signal y1(n) and a second digital baseband output signal y2(n) carrying out normalization treatment;
2) selecting a memory depth M and a threshold number K according to the following ranges: m is less than or equal to 5, and K is less than or equal to 10;
3) a kth segmentation point beta according to a threshold of the first frequency bandk (1)The kth segmentation point beta of the threshold of the second frequency bandk (2)Model memory depth M and threshold number K, and the value ranges of the first cross modulation coefficient and the second cross modulation coefficient are [0.8,2]]Searching by an internal binary search method, wherein each search is carried out according to the digital baseband input signal x of the first frequency band by using a least square method1(n) digital baseband input signal x of second frequency band2(n) the first path of digital baseband output signal y1(n) and a second digital baseband output signal y2(n) training the model parameters of the first digital predistorter and the model parameters of the second predistorter, calculating the normalized mean square error of the digital baseband input signal and the digital baseband output signal by using the model parameters of the first digital predistorter and the model parameters of the second predistorter, and taking the corresponding model parameters of the first digital predistorter and the corresponding model parameters of the second predistorter as the trained model parameters of the first digital predistorter and the trained model parameters of the second predistorter when the normalized mean square error is minimum.
Take the example of a 4-carrier LTE-A signal input frequency band 1.5 and 1.7GHz radio frequency power amplifier with 40MHz bandwidth. And carrying out normalization processing after the input and output data of the power amplifier are synchronously acquired. As shown in FIGS. 2(a) and 2(b)
Before predistortion, the output of the power amplifier is subjected to a dual-band power amplifier model based on a piecewise linear function, and ACPR of left and right sidebands of 1500MHz and 1700MHz is compared as follows:
TABLE 1
Looking at table 1, it can be seen that greater than 12dB of ACPR improvement can be achieved when modeling the dual band power amplifier using the present model. This effect is similar to the conventional 2D-DPD model, but for the 2D-DPD model used in the current dual-frequency predistortion:
there is a great difference in performance from the model in this patent. Firstly, in terms of the number of parameters: when a 2D-DPD model is selected for modeling, when the memory depth is selected to be M, the nonlinear order of the model is K, and the number of model parameters at the moment is (M +1) × (K +1) × (K + 2); when the dual-frequency power amplifier model based on the piecewise linear function is selected, the memory depth M and the threshold number K are selected, and the number of the parameters of the model at the moment is (M +1) multiplied by K. The complexity of the two models is O (n), respectively3) And O (n)2) (ii) a Secondly, on the realization difficulty of the model: . For the 2D-DPD model, a high-order power item exists, and a large amount of multiplier resources are consumed when FPGA is realized; and the high-order power term in the piecewise linear function model is replaced by absolute value operation, so that the method is easier to realize and saves computing resources.
Claims (5)
1. A double-frequency power amplifier digital predistortion device based on a piecewise linear function is characterized in that: the method comprises the following steps:
a first digital predistorter: the first digital predistorter comprises a predistortion model based on a piecewise linear function, a digital baseband input signal x of a first frequency band1(n) inputting a first digital predistorter, wherein the first digital predistorter inputs model parameters of the first digital predistorter and a digital baseband input signal x of a first frequency band1(n) input into its own predistortion model, which outputs a first predistortion output signal x1-pre(n), the first predistorted output signal x1-pre(n) also as the output signal of the first digital predistorter;
a second digital predistorter: the second digital predistorter includes piecewise linear function-based predistortionModel, digital baseband input signal x in the second frequency band2(n) inputting a second digital predistorter, wherein the second digital predistorter uses model parameters of the second digital predistorter and a digital baseband input signal x of a second frequency band2(n) input into its own predistortion model, which outputs a second predistortion output signal x2-pre(n), second predistorted output signal x2-pre(n) also as the output signal of the second digital predistorter;
a first digital-to-analog converter: first predistortion output signal x1-pre(n) performing digital-to-analog conversion through a first digital-to-analog converter, and inputting the converted signal into a first low-pass filter for filtering;
a second digital-to-analog converter: second predistorted output signal x2-pre(n) performing digital-to-analog conversion through a second digital-to-analog converter, and inputting the converted signal into a second low-pass filter for filtering;
a first low-pass filter: filtering the signal output by the first digital-to-analog converter;
a second low-pass filter: filtering the signal output by the second digital-to-analog converter;
a first modulator: the signal output by the first low-pass filter is modulated by a first modulator;
a second modulator: the signal output by the second low-pass filter is modulated by the first modulator;
a first power divider: the signal output by the first modulator and the signal output by the second modulator are combined into a signal through the first power divider;
a power amplifier: amplifying the signal combined by the first power divider through a power amplifier;
an attenuation coupler: the signal amplified by the power amplifier is input into an attenuation coupler;
the second power divider: the output signal of the attenuation coupler is divided into two paths of signals through a second power divider, and the two paths of signals are respectively transmitted to a first band-pass filter and a second band-pass filter;
a first band-pass filter: filtering the first path of signals output by the second power divider through a first band-pass filter;
a second band-pass filter: filtering a second path of signals output by the second power divider through a second band-pass filter;
a first demodulator: the signal output by the first band-pass filter is demodulated by a first demodulator;
a second demodulator: the signal output by the second band-pass filter is demodulated by a second demodulator;
a third band-pass filter: filtering the signal output by the first demodulator through a third band-pass filter;
fourth bandpass filter: the signal output by the second demodulator is filtered by a fourth band-pass filter;
a first analog-to-digital converter: the signal output by the third band-pass filter is subjected to analog-to-digital conversion by the first analog-to-digital converter to obtain a first path of digital baseband output signal y1(n)；
A second analog-to-digital converter: the signal output by the fourth band-pass filter is subjected to analog-to-digital conversion by the second analog-to-digital converter to obtain a second path of digital baseband output signal y2(n)；
A predistortion training module: digital baseband input signal x of a first frequency band1(n) digital baseband input signal x of second frequency band2(n) the first path of digital baseband output signal y1(n) and a second digital baseband output signal y2(n) the model parameters of the first digital predistorter and the model parameters of the second predistorter are input into a predistortion training module, the predistortion training module trains the model parameters of the first digital predistorter and the model parameters of the second predistorter, the trained model parameters of the first digital predistorter are copied to the first digital predistorter, and the trained model parameters of the second predistorter are copied to the second digital predistorter;
the predistortion training module comprises a dual-frequency predistortion model based on a piecewise linear function, and the dual-frequency predistortion model based on the piecewise linear function is as follows:
in the formula (I), the compound is shown in the specification,
representing the model parameters of the kth first digital predistorter when the memory depth of the power amplifier is m;
representing model parameters of a kth second digital predistorter when the memory depth is m; beta is ak (1)A kth segmentation point representing a threshold for the first frequency band; beta is ak (2)A kth segmentation point representing a threshold for a second frequency band;
representing a first cross modulation coefficient between a (k +1) th first frequency band and a (k +1) th second frequency band;
representing a second cross modulation coefficient between the (k +1) th first frequency band and the (k +1) th second frequency band; m represents the total memory depth of the power amplifier, K represents the total threshold number of the piecewise linear function, y1(n-m) represents the first digital baseband output signal at time n-m, y2And (n-m) represents the second path of digital baseband output signal at the moment of n-m.
2. The double-frequency power amplifier digital predistortion method based on the piecewise linear function is characterized in that: the method comprises the following steps:
s1: the first digital predistorter comprises a predistortion model based on a piecewise linear function, a digital baseband input signal x of a first frequency band1(n) inputting a first digital predistorter, wherein the first digital predistorter inputs model parameters of the first digital predistorter and a digital baseband input signal x of a first frequency band1(n) input into its own predistortion model, which outputs a first predistortion output signal x1-pre(n), the first predistorted output signal x1-pre(n) also as the output signal of the first digital predistorter; the second digital predistorter includes a segmentation-based basisPredistortion model of linear function, digital baseband input signal x of second frequency band2(n) inputting a second digital predistorter, wherein the second digital predistorter uses model parameters of the second digital predistorter and a digital baseband input signal x of a second frequency band2(n) input into its own predistortion model, which outputs a second predistortion output signal x2-pre(n), second predistorted output signal x2-pre(n) also as the output signal of the second digital predistorter;
s2: first predistortion output signal x1-pre(n) performing digital-to-analog conversion by using a first digital-to-analog converter, wherein the converted signals respectively pass through a first low-pass filter and a first modulator; second predistorted output signal x2-pre(n) performing digital-to-analog conversion by a second digital-to-analog converter, wherein the converted signals respectively pass through a second low-pass filter and a second modulator;
s3: the signal output by the first modulator and the signal output by the second modulator are combined into a signal through the first power divider;
s4: amplifying the signal combined by the first power divider through a power amplifier;
s5: the signal amplified by the power amplifier is input into an attenuation coupler, and the attenuation coupler outputs two paths of signals to a first band-pass filter and a second band-pass filter respectively through a second power divider;
s6: filtering the first path of signals output by the second power divider through a first band-pass filter; filtering a second path of signals output by the second power divider through a second band-pass filter;
s7: the signal output by the first band-pass filter is demodulated by a first demodulator; the signal output by the second band-pass filter is demodulated by a second demodulator;
s8: filtering the signal output by the first demodulator through a third band-pass filter; the signal output by the second demodulator is filtered by a fourth band-pass filter;
s9: the signal output by the third band-pass filter is subjected to analog-to-digital conversion by the first analog-to-digital converter to obtain a first path of digital baseband output signal y1(n); of the fourth bandpass filter outputThe signal is subjected to analog-to-digital conversion by a second analog-to-digital converter to obtain a second path of digital baseband output signal y2(n)；
S10: digital baseband input signal x of a first frequency band1(n) digital baseband input signal x of second frequency band2(n) the first path of digital baseband output signal y1(n) and a second digital baseband output signal y2(n) the model parameters of the first digital predistorter and the model parameters of the second predistorter are input into a predistortion training module, the predistortion training module trains the model parameters of the first digital predistorter and the model parameters of the second predistorter, the trained model parameters of the first digital predistorter are copied to the first digital predistorter, and the trained model parameters of the second predistorter are copied to the second digital predistorter;
the predistortion training module comprises a dual-frequency predistortion model based on a piecewise linear function, wherein the dual-frequency predistortion model based on the piecewise linear function is shown as a formula (1):
in the formula (1), the reaction mixture is,
representing the model parameters of the kth first digital predistorter when the memory depth of the power amplifier is m;
representing model parameters of a kth second digital predistorter when the memory depth is m; beta is ak (1)A kth segmentation point representing a threshold for the first frequency band; beta is ak (2)A kth segmentation point representing a threshold for a second frequency band;
representing a first cross modulation coefficient between a (k +1) th first frequency band and a (k +1) th second frequency band;
representing a second cross modulation coefficient between the (k +1) th first frequency band and the (k +1) th second frequency band; m represents the total memory depth of the power amplifier, K represents the total threshold number of the piecewise linear function, y1(n-m) represents the first digital baseband output signal at time n-m, y2And (n-m) represents the second path of digital baseband output signal at the moment of n-m.
3. The piecewise linear function based digital predistortion method for a dual-frequency power amplifier according to claim 2, characterized in that: in step S10, the predistortion training module trains the model parameters of the first digital predistorter and the model parameters of the second predistorter by using a least square method, calculates a normalized mean square error between the digital baseband input signal and the digital baseband output signal under different model parameters, and obtains a cross modulation coefficient with the minimum normalized mean square error by using a binary search method, thereby obtaining the model parameters of the trained first digital predistorter and the model parameters of the trained second predistorter.
4. The piecewise linear function based digital predistortion method for a dual-frequency power amplifier according to claim 2, characterized in that: the value range of the first cross modulation coefficient and the second cross modulation coefficient is [0.8,2 ].
5. The piecewise linear function based digital predistortion method for a dual-frequency power amplifier according to claim 2, characterized in that: the establishment process of the dual-frequency predistortion model based on the piecewise linear function comprises the following steps:
1) obtaining a digital baseband input signal x of a first frequency band1(n) digital baseband input signal x of second frequency band2(n) the first path of digital baseband output signal y1(n) and a second digital baseband output signal y2(n) carrying out normalization treatment;
2) selecting a memory depth M and a threshold number K according to the following ranges: m is less than or equal to 5, and K is less than or equal to 10;
3) second according to the threshold of the first frequency bandk segmentation points betak (1)The kth segmentation point beta of the threshold of the second frequency bandk (2)Model memory depth M and threshold number K, and the value ranges of the first cross modulation coefficient and the second cross modulation coefficient are [0.8,2]]Searching by an internal binary search method, wherein each search is carried out according to the digital baseband input signal x of the first frequency band by using a least square method1(n) digital baseband input signal x of second frequency band2(n) the first path of digital baseband output signal y1(n) and a second digital baseband output signal y2(n) training the model parameters of the first digital predistorter and the model parameters of the second predistorter, calculating the normalized mean square error of the digital baseband input signal and the digital baseband output signal by using the model parameters of the first digital predistorter and the model parameters of the second predistorter, and taking the corresponding model parameters of the first digital predistorter and the corresponding model parameters of the second predistorter as the trained model parameters of the first digital predistorter and the trained model parameters of the second predistorter when the normalized mean square error is minimum.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811208279.7A CN109347452B (en) | 2018-10-17 | 2018-10-17 | Double-frequency power amplifier digital predistortion device and method based on piecewise linear function |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811208279.7A CN109347452B (en) | 2018-10-17 | 2018-10-17 | Double-frequency power amplifier digital predistortion device and method based on piecewise linear function |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109347452A CN109347452A (en) | 2019-02-15 |
| CN109347452B true CN109347452B (en) | 2022-04-26 |
Family
ID=65308942
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811208279.7A Active CN109347452B (en) | 2018-10-17 | 2018-10-17 | Double-frequency power amplifier digital predistortion device and method based on piecewise linear function |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109347452B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111786639B (en) * | 2020-06-23 | 2022-11-01 | 东南大学 | Double-frequency power amplifier digital predistortion device and method |
| CN111988002B (en) * | 2020-08-05 | 2023-09-05 | 东南大学 | Digital predistortion method, device, equipment and storage medium for MIMO power amplifier |
| CN114598578A (en) * | 2020-11-20 | 2022-06-07 | 中国移动通信有限公司研究院 | Digital predistortion training method and device and storage medium |
| CN113162559B (en) * | 2021-03-30 | 2022-07-29 | 西南电子技术研究所（中国电子科技集团公司第十研究所） | Millimeter wave self-adaptive predistortion linearized solid-state power amplifier |
| CN113381705B (en) * | 2021-06-22 | 2022-11-15 | 电子科技大学 | Digital predistortion implementation system and method in hardware scene |
| CN113612454B (en) * | 2021-08-12 | 2023-09-08 | 东南大学 | Power amplifier digital predistortion device and method based on affine function model with amplitude limiting degree selection |
| CN116436461B (en) * | 2023-06-12 | 2023-09-19 | 北京思凌科半导体技术有限公司 | Digital-to-analog converter and electronic device |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102868368A (en) * | 2012-09-14 | 2013-01-09 | 清华大学 | Double-frequency synchronous power amplifier based on T-type network and coupling line and design method thereof |
| CN102969987A (en) * | 2012-12-17 | 2013-03-13 | 东南大学 | Undersampling-based broadband power-amplifier pre-distortion method |
| CN103414671A (en) * | 2013-07-16 | 2013-11-27 | 清华大学 | 2D-DPD iteration reduction algorithm and application based on unit primary function |
| CN104885369A (en) * | 2013-05-22 | 2015-09-02 | 瑞典爱立信有限公司 | Low complexity digital predistortion for concurrent multi-band transmitters |
| CN105320492A (en) * | 2014-08-01 | 2016-02-10 | 英飞凌科技股份有限公司 | Digital pre-distortion and post-distortion based on segmentwise piecewise polynomial approximation |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8611402B2 (en) * | 2011-02-02 | 2013-12-17 | Rf Micro Devices, Inc. | Fast envelope system calibration |
-
2018
- 2018-10-17 CN CN201811208279.7A patent/CN109347452B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102868368A (en) * | 2012-09-14 | 2013-01-09 | 清华大学 | Double-frequency synchronous power amplifier based on T-type network and coupling line and design method thereof |
| CN102969987A (en) * | 2012-12-17 | 2013-03-13 | 东南大学 | Undersampling-based broadband power-amplifier pre-distortion method |
| CN104885369A (en) * | 2013-05-22 | 2015-09-02 | 瑞典爱立信有限公司 | Low complexity digital predistortion for concurrent multi-band transmitters |
| CN103414671A (en) * | 2013-07-16 | 2013-11-27 | 清华大学 | 2D-DPD iteration reduction algorithm and application based on unit primary function |
| CN105320492A (en) * | 2014-08-01 | 2016-02-10 | 英飞凌科技股份有限公司 | Digital pre-distortion and post-distortion based on segmentwise piecewise polynomial approximation |
Non-Patent Citations (2)
| Title |
|---|
| 2-D Digital Predistortion (2-D-DPD) Architecture for Concurrent Dual-Band Transmitters;Seyed Aidin Bassam 等;《IEEE Transactions on Microwave Theory and Techniques》;20110901;第59卷(第10期);第2547-2553页 * |
| Linearization of Concurrent Dual-Band Power Amplifier Based on 2D-DPD Technique;Seyed Aidin Bassam 等;《IEEE Microwave and Wireless Components Letters》;20111018;第21卷(第12期);第685-686页、图3 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109347452A (en) | 2019-02-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109347452B (en) | Double-frequency power amplifier digital predistortion device and method based on piecewise linear function | |
| US20230370098A1 (en) | System and method for increasing bandwidth for digital predistortion in multi-channel wideband communication systems | |
| CN111436225B (en) | Predistortion circuit of wireless transmitter and method for generating predistortion baseband signal | |
| Yu et al. | Band-limited Volterra series-based digital predistortion for wideband RF power amplifiers | |
| CN106506417B (en) | Narrow-band feedback digital predistortion system and method | |
| CN105763495B (en) | digital predistortion method and device | |
| CN104796091A (en) | Polynomial power amplifier modeling and digital pre-distorting method based on segmented memory | |
| US20080218262A1 (en) | Predistortion with asymmetric usage of available bandwidth | |
| CN102969987A (en) | Undersampling-based broadband power-amplifier pre-distortion method | |
| TWI700888B (en) | Digital pre-distortion circuit and digital pre-distortion method | |
| CN110266276B (en) | Low-speed digital predistortion method for 5G ultra-wideband power amplifier | |
| US9595925B2 (en) | Distortion-compensating power amplifier and method for compensating for distortion to amplify power | |
| WO2016101627A1 (en) | Method and apparatus for increasing digital pre-distortion performance of radio frequency power amplifier | |
| CN111147412A (en) | Predistortion processing device, signal transmission system and predistortion processing method | |
| JP2011507444A (en) | Linear power amplifier | |
| Liu et al. | Novel multiband linearization technique for closely-spaced dual-band signals of wide bandwidth | |
| CN111786639B (en) | Double-frequency power amplifier digital predistortion device and method | |
| CN113612453A (en) | Low-sampling-rate feedback digital predistortion correction method and device | |
| Yang et al. | Crest factor reduction for dual-band systems | |
| Yu | Digital predistortion using feedback signal with incomplete spectral information | |
| Xia et al. | Dual-band linear filter assisted envelope memory polynomial for linearizing multi-band power amplifiers | |
| CN111988002B (en) | Digital predistortion method, device, equipment and storage medium for MIMO power amplifier | |
| Velazquez et al. | 1 GHz instantaneous bandwidth digital pre-distortion for multi-concurrent channel wideband power amplifiers | |
| CN113630134A (en) | Digital feedforward-assisted broadband digital predistortion method and device | |
| CN113824446A (en) | High-numerical-stability band-limited digital predistortion solving method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |