CN102624338A - Volterra-filtering-based power amplifier pre-distortion method for double-loop feedback model - Google Patents
Volterra-filtering-based power amplifier pre-distortion method for double-loop feedback model Download PDFInfo
- Publication number
- CN102624338A CN102624338A CN2012100712492A CN201210071249A CN102624338A CN 102624338 A CN102624338 A CN 102624338A CN 2012100712492 A CN2012100712492 A CN 2012100712492A CN 201210071249 A CN201210071249 A CN 201210071249A CN 102624338 A CN102624338 A CN 102624338A
- Authority
- CN
- China
- Prior art keywords
- vector
- error
- predistorter
- parameter vector
- distortion
- 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.)
- Granted
Links
Images
Landscapes
- Amplifiers (AREA)
Abstract
The invention discloses a Volterra-filtering-based power amplifier pre-distortion method for a double-loop feedback model. The method mainly comprises the three parts of pre-distortion crude parameter vector extraction, error regulation and pre-distorter fine parameter vector extraction, and specifically comprises the following steps of: performing down-conversion output and pre-distorter output by adopting a first-order dynamic truncation Volterra filtering structure to obtain a pre-distortion crude parameter vector in the pre-distortion crude parameter vector extraction; dynamically regulating an error vector caused by the pre-distortion crude parameter vector according to an average criterion; combining a regulated error vector which serves as an expected signal and a base-band input signal, and obtaining a pre-distorter fine parameter vector by adopting the first-order dynamic truncation Volterra filtering structure; and correcting the crude parameter vector by utilizing the fine parameter vector to obtain a final pre-distorter parameter vector. The problems of slow adaptive convergence, great pre-distortion parameter calculation amount, complexity in implementation, incapability of effectively compensating the complex memory effect of a power amplifier in high-speed communication, and the like of the conventional pre-distortion method are solved.
Description
Technical field
The present invention relates to digital communication technology field, is a kind of pre-distortion method that is used for power amplifier linearization, is specially a kind of power amplifier predistortion method of the two circulation feedback models based on Volterra filtering.
Background technology
Along with the development of digital communication technology and the use in civil area of communication technology of satellite; It is more and more precious that band resource seems, in order to improve the utilance of frequency spectrum, proposed various linearity modulation techniques such as QPSK, 16-QAM and OFDM etc.; These digital modulation modes all belong to non-permanent envelope modulation; Therefore, high peak-to-average force ratio is the common trait of above-mentioned modulation, and the peak is all compared the linearisation of power amplifier and required higher; And after introducing satellite communication,,, cause spaceborne power amplifier all to be operated in saturation region even the cut-off region in order to pursue higher transmitting power owing to receive the restriction of satellite volume; Therefore the linearisation of power amplifier directly affects the fine or not degree that transmits and receives signal.
For the conventional power amplifier, in order to guarantee to be operated in the non-saturated region, generally adopt the back-off method to reach linearizing requirement, this just makes high power device can only export very little power, has caused the efficient of power amplifier lower.In recent years, the research of various power amplifier models and linearization technique more and more widely.The linearization technique of common power amplifier mainly is divided into linear without memory method and linear with memory method two big classes.Memoryless power amplifier linearization research mainly concentrates on narrow band communication; When the bandwidth of input signal is ignored memory effect during much smaller than the bandwidth of power amplifier self; But development along with modern digital communication; The bandwidth of signal to improve the utilance of entire spectrum, causes memoryless power amplifier linearization to study more and more incompatibility modern digital communication more and more near transmitted bandwidth.Elimination has the pre-distortion method of memory power amplifier to mainly contain two box Hammerstein pre-distortion methods, two box Wiener pre-distortion method, Volterra filtering pre-distortion method and Artificial Neural Network etc.Predistorter mainly adopts Hammerstein pre-distortion method and Wiener pre-distortion method at present; These two kinds of methods are divided into memoryless nonlinear block with predistorter and the linear with memory module solves nonlinear distortion and Memorability distortion respectively in order to have solved the memory nonlinear distortion; Through feedovering or feeding back the coefficient of regulating predistorter, these two kinds of methods are easy to realize.But along with signal of communication develops towards the broadband high-speed communication direction, this subsystem is regulated because renewal speed is inconsistent can produce bigger error, and when the transmission speed of signal is big, thereby this error can be amplified fast and caused the inefficacy of pre-distortion system really.Function neural network predistortion method has mode of learning flexibly, and main models has time-delay three layers of feed-forward network model, radial primary function network model, feedback-type network model and dynamic error network model etc.Neural network model is minimum to have the three-layer network structure; The network configuration of this multilayer can not effectively extract the coefficient of predistorter; Secondly neural network model in output error hour; Neuron can be selected coefficient near null value, thereby produces a continuous fluctuating error, and transmission produces considerable influence to this error to high-speed digital communication.Existing Volterra filtering pre-distortion method can approach continuous nonlinear function arbitrarily, but along with the increase of system's order and memory span, the amount of calculation of pre-distortion parameters is increase sharply, and this has just limited the realization and the application of Volterra filtering predistortion.
Summary of the invention
The technical problem that solves
To the high complexity of existing power amplifier linearization technology and slower convergence rate; Can not satisfy the demand of high-speed radiocommunication of future generation to power amplifier properties, the present invention has designed a kind of power amplifier predistortion method of the two circulation feedback models based on Volterra filtering.
Technical scheme
The present invention comprises that mainly the thick parameter vector of predistortion extracts, error is adjusted and the smart parameter vector of predistorter extracts three parts.What the part of extracting the thick parameter vector of predistortion adopted is that single order dynamically blocks the Volterra filter structure, through the down-conversion output and the predistorter output of whole system, tries to achieve the thick parameter vector of predistortion fast; The error vector that the thick parameter vector of predistortion causes is dynamically adjusted according to average criteria in the error adjustment member; Adjusted error vector as desired signal, is utilized single order dynamically to block the Volterra filter structure in conjunction with base-band input signal in the smart parameter vector extraction of predistortion part and solves the smart parameter vector of predistorter; The smart parameter vector of thick parameter vector of predistortion and predistortion is all confirmed by least square (Least-squares) method, utilizes smart parameter vector that thick parameter vector correction is obtained final predistorter parameter vector.
Technical scheme of the present invention is:
The power amplifier predistortion method of said two circulation feedback models based on Volterra filtering is characterized in that: may further comprise the steps:
Step 1: utilize high-speed AD converter to gather base band input data, the base band dateout of predistorter and the down-conversion base band dateout of power amplifier of predistorter;
Step 2: utilize the base band dateout of the predistorter that step 1 collects and the down-conversion base band dateout of power amplifier, set up the Volterra Filtering Model and adopt the least square learning method under minimum mean square error criterion, to calculate the thick parameter vector of predistorter;
Step 3: the thick parameter vector of predistorter that step 2 is obtained copies in the predistorter, calculates the error information vector of power amplifier this moment; According to average criteria and error empirical coefficient the error information vector that obtains is carried out Error processing; Said error processing method is:
Step a: the difference ε of each element and previous element (i) in the error of calculation data vector=| e (i)-e (i-1) |, i=2,3, Λ, k, wherein k representes the element number in the error information vector, i element in e (i) the expression error information vector;
as error decision threshold th, wherein ε (1)=0;
Step b: i is judged from 1～k successively: when ε (i)＞th; With e (i) * (1+a) as i element in the error information vector; Otherwise as i element in the error information vector, wherein a is the error empirical coefficient, 0.01≤a≤0.1 with e (i) * (1-a);
Step 4: utilize the base band input data of the predistorter that error information vector that step 3 obtains and step 1 collect, set up the Volterra Filtering Model and adopt the least square learning method under minimum mean square error criterion, to calculate the smart parameter vector of predistorter;
Step 5: after each element value in the thick parameter vector deducted each element value in the smart parameter vector correspondingly, the vector that obtains was a predistortion final argument vector.
Beneficial effect
The present invention dynamically blocks the Volterra model through two single orders and replaces existing high-order Volterra model or neural network model; Because the matrix operation in predistorter parametric solution process is more; The exponent number of model can be bigger to the influence of predistorter CALCULATION OF PARAMETERS amount, utilizes two single order Volterra models can effectively reduce the amount of calculation of each iteration.
The present invention utilizes two circulations that are mutually related of dynamically having blocked the Volterra model construction of two single orders; Through the second level smart parameter vector correction first order of the resultant predistorter thick parameter vector of resulting predistorter that circulates that circulates; Be equivalent in the single iteration computing; Carry out twice parameter vector and extracted, can effectively accelerate the convergence rate of predistorter parameter.
Two circulation feedback pre-distortion methods that the present invention designed are because the first thick parameter vector that calculates predistorter through the base band output and the output after the power amplifier down-conversion of predistorter; Guarantee the renewal speed when high speed transmission of signals; And the input of the base band through predistorter and error revise thick parameter vector, guaranteed the accuracy of two circulation feedback pre-distortion methods.Thereby when making analog signal have the fast state variation, also can realize the output signal line propertyization exactly.
The power amplifier pre-distortion method of the two circulation feedback models based on Volterra filtering of the present invention has combined Volterra filtering can approach the advantage and the less advantage of feedback model amount of calculation of any continuous function; In order to guarantee the validity under the high-speed digital communication; The present invention has set up two circulation feedback models and has guaranteed precision; Avoided the shortcoming of Volterra filtering and feedback model simultaneously, the self adaptation convergence rate that has solved existing pre-distortion method is slow, and the pre-distortion parameters amount of calculation is big; Realize more complicated, the effective problems such as memory effect of the complicacy of compensating power amplifier in high-speed communication.When not increasing implementation complexity, can eliminate the nonlinear characteristic and the memory effect of power amplifier fast.Utilize the low-complexity of good the approaching property of Volterra filter structure and two circulation feedback amendment schemes, in the high-speed radio transmission, have accurately, stable and advantage efficiently.
Description of drawings
Fig. 1: the power amplifier predistortion conceptual scheme that the present invention proposes based on two circulation feedback models of Volterra filtering;
Fig. 2: the conceptual scheme of the predistorter coefficient update structure among the present invention;
Fig. 3: the convergence curve figure of different pre-distortion systems;
Fig. 4: the performance test figure of changeable amplitude test signal;
Fig. 5: flow chart of the present invention.
Embodiment
Below in conjunction with specific embodiment the present invention is described:
Embodiment:
The overall structure of the preparatory simulating scheme of power amplifier of present embodiment is as shown in Figure 1, utilizes the two-stage single order dynamically to block the Volterra Filtering Model and makes up the two circulation feedback update system of predistorter coefficient, is used for the nonlinear characteristic and the memory effect of compensating power amplifier.Concrete steps are:
Step 1: utilize base band dateout and power amplifier that high-speed AD converter gathers the base band input data of predistorter, predistorter through the base band dateout after the attenuator down-converted; Wherein, present embodiment is selected the base band input data of broad band multicarrier signal as predistorter.
At n constantly, the base band of predistorter input data are x (n), and the base band dateout of predistorter is u (n), and the base band dateout of power amplifier after through the attenuator down-converted is y (n).
Step 2: utilize the base band dateout u (n) of the predistorter that step 1 collects and the down-conversion base band dateout y (n) of power amplifier; This step of thick parameter vector
of setting up the Volterra Filtering Model and adopting the least square learning method under minimum mean square error criterion, to calculate predistorter is the known method in the present technique field, provides concrete step below:
Structure Volterra Filtering Model:
Input data vector as first order circulation feedback;
is made up of the power amplifier dateout y (n) after the down-conversion; K is a dateout vector length after the power amplifier down-conversion; K=P (M+1), P are non-linear progression, and M is a memory effect length.
Structure first order input data matrix, and obtain after adding forgetting factor:
Wherein, (k * k) is an input data matrix to ψ, and λ is a forgetting factor; This is because in digital communication; Though power amplifier has certain memory effect, the time interval is long more, and is more little to current systematic influence; Therefore adding forgetting factor can be more near actual conditions, and λ gets 0.5 in the present embodiment.
The desired signal vector
of structure predistorter baseband output signal
Wherein, u (n-1), Λ, u (n-k+1) is n-1, Λ, n-k+1 is the base band dateout of predistorter constantly.
The thick parameter vector of predistorter that obtains first order feedback cycle does
Step 3: the thick parameter vector of predistorter
that step 2 is obtained copies in the predistorter; The error information vector
that calculates power amplifier this moment carries out Error processing according to average criteria and error empirical coefficient to the error information vector that obtains, and to reduce that background noise influences be purpose to reach more fast; Said error processing method is:
Step a: the difference ε of each element and previous element (i) in the error of calculation data vector=| e (i)-e (i-1) |, i=2,3, Λ, k, i element in e (i) the expression error information vector; With
as error decision threshold th, ε (1)=0 wherein;
Step b: i is judged from 1～k successively: when ε (i)＞th; With e (i) * (1+a) as i element in the error information vector; Otherwise as i element in the error information vector, wherein a is the error empirical coefficient, 0.01≤a≤0.1 with e (i) * (1-a); The error empirical coefficient gets 0.05 in the present embodiment, obtains adjusted error information vector for
Step 4: the base band input data x (n) that utilize the predistorter that adjusted error information vector
that step 3 obtains and step 1 collect; This step of smart parameter vector
of setting up the Volterra Filtering Model and adopting the least square learning method under minimum mean square error criterion, to calculate predistorter is the known method in the present technique field, provides concrete step below:
Structure Volterra Filtering Model:
Structure second level input data matrix, and obtain after adding forgetting factor:
Wherein, ψ
Fine(k * k) is a second level input data matrix.
In the feedback cycle of the second level; The adjusted error information vector
that employing step 3 obtains is as the desired signal vector in the feedback cycle of the second level, and the smart parameter vector of predistorter that obtains second level feedback cycle does
Step 5: because thick parameter vector and smart parameter vector same order; After each element value in the thick parameter vector deducted each element value in the smart parameter vector correspondingly, the vector that obtains was predistortion final argument vector
Through the smart parameter vector of predistorter is eliminated from the thick parameter vector of predistorter; Just can well eliminate broadband signal parameter in the digital communication and change the error that circulation is found the solution with the first order the thick parameter vector of predistorter is caused owing to the precision reason, finally the parameter vector of predistorter is:
Copy to the resulting predistorter final argument of two circulation feedback models
in the predistorter based on Volterra filtering; Pass through predistorter to the predistorter base-band input signal; Through upconverter and power amplifier, be sent to the transmission link after the completion predistortion again.
In the present embodiment this method has been carried out the correlated performance analysis:
1, analysis of convergence speed
Power amplifier adopts the Saleh model, and 3 road QPSK mixed signals are adopted in system's input, and carry out waveform processing through the raised cosine roll off filter, and its parameter is respectively: rolloff-factor is 0.5, and delay is 3, and rising sample rate is 8.Through the actual measurement experiment, forgetting factor adopts 0.5, and the error dynamics step-length adopts 0.05.In order to compare, traditional two box Hammerstein structure pre-distortion methods, three rank Volterra structure pre-distortion method and neural network structure pre-distortion methods have also been calculated respectively with of the present invention pair of circulation feedback pre-distortion method.In order to reduce initial influence to pre-distortion system, the predistorter initial coefficients is full zero vector.Its result is as shown in Figure 3.
When the mean square error fluctuation range within 5dB, can regard algorithmic statement as.From Fig. 3, can see; The convergence rate of two circulation feedback pre-distortion methods proposed by the invention is the fastest in four kinds of pre-distortion methods; And mean square error can reach-75dB about; More traditional two box Hammerstein structure pre-distortion methods reduce about 2ldB approximately, compare with the neural network structure pre-distortion method with three rank Volterra structure pre-distortion methods; Also have the reduction about 3-4dB, and the convergence rate of two circulation feedback pre-distortion methods also has very big raising compared to additive method.This is because the present invention dynamically blocks the amount of calculation that model can effectively reduce each iteration through two single order Volterra, and, come the further accuracy of revising whole pre-distortion system through the base band input of system and the error of first order circulation.Because iteration speed faster, in limited time of satellite communication, can better reduce the error rate and settling time.
2, pre-distortion system error rate analyzer
In order to show the actual performance difference of each pre-distortion system, in the predistorter base-band input signal, added the performance that stochastic regime conversion fast is beneficial to distinguish each pre-distortion method.Because power amplifier output signal envelope after the down-conversion and ideal signal envelope are very approaching; So in emulation, added the error curve of each pre-distortion method; And, having added the output signal envelope that does not pass through pre-distortion in order to contrast, its simulation result is as shown in Figure 4.
In Fig. 4; Can find out two circulation feedback pre-distortion methods that the present invention designs because the thick parameter vector that base band output and the output after the power amplifier down-conversion through predistorter earlier calculates predistorter; Guarantee the renewal speed when high speed transmission of signals; And the input of the base band through predistorter and error revise thick parameter vector, guaranteed the accuracy of two circulation feedback pre-distortion methods.Thereby when making the predistorter base-band input signal have the fast state variation, also can realize the output signal line propertyization exactly.
Claims (1)
1. power amplifier predistortion method based on two circulation feedback models of Volterra filtering is characterized in that: comprise with
Following step:
Step 1: utilize high-speed AD converter to gather base band input data, the base band dateout of predistorter and the down-conversion base band dateout of power amplifier of predistorter;
Step 2: utilize the base band dateout of the predistorter that step 1 collects and the down-conversion base band dateout of power amplifier, set up the Volterra Filtering Model and adopt the least square learning method under minimum mean square error criterion, to calculate the thick parameter vector of predistorter;
Step 3: the thick parameter vector of predistorter that step 2 is obtained copies in the predistorter, calculates the error information vector of power amplifier this moment; According to average criteria and error empirical coefficient the error information vector that obtains is carried out Error processing; Said error processing method is:
Step a: the difference ε of each element and previous element (i) in the error of calculation data vector=| e (i)-e (i-1) |, i=2,3, Λ, k, wherein k representes the element number in the error information vector, i element in e (i) the expression error information vector; With
as error decision threshold th, ε (1)=0 wherein;
Step b: i is judged from 1～k successively: when ε (i)＞th; With e (i) * (1+a) as i element in the error information vector; Otherwise as i element in the error information vector, wherein a is the error empirical coefficient, 0.01≤a≤0.1 with e (i) * (1-a);
Step 4: utilize the base band input data of the predistorter that error information vector that step 3 obtains and step 1 collect, set up the Volterra Filtering Model and adopt the least square learning method under minimum mean square error criterion, to calculate the smart parameter vector of predistorter;
Step 5: after each element value in the thick parameter vector deducted each element value in the smart parameter vector correspondingly, the vector that obtains was a predistortion final argument vector.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210071249.2A CN102624338B (en) | 2012-03-18 | 2012-03-18 | Volterra-filtering-based power amplifier pre-distortion method for double-loop feedback model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210071249.2A CN102624338B (en) | 2012-03-18 | 2012-03-18 | Volterra-filtering-based power amplifier pre-distortion method for double-loop feedback model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102624338A true CN102624338A (en) | 2012-08-01 |
| CN102624338B CN102624338B (en) | 2014-09-03 |
Family
ID=46564046
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210071249.2A Expired - Fee Related CN102624338B (en) | 2012-03-18 | 2012-03-18 | Volterra-filtering-based power amplifier pre-distortion method for double-loop feedback model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102624338B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103036514A (en) * | 2012-11-21 | 2013-04-10 | 南京航空航天大学 | Method for calculating output quantity of power amplifier by using Volterra correction model |
| CN104219182A (en) * | 2014-08-12 | 2014-12-17 | 上海航天电子通讯设备研究所 | Radio-frequency power amplifier linearization system based on broadband modulator |
| CN104410593A (en) * | 2014-12-04 | 2015-03-11 | 西北工业大学 | Decision feedback model-based digital symbol nonlinear error correction equalization method |
| CN105610408A (en) * | 2015-12-17 | 2016-05-25 | 西安空间无线电技术研究所 | Non-linear Volterra filtering optimization method based on contribution factor |
| CN113114125A (en) * | 2021-04-20 | 2021-07-13 | 上海矽昌微电子有限公司 | Digital predistortion correction method and system for double-loop resolving |
| US11901951B2 (en) | 2021-07-07 | 2024-02-13 | Fujitsu Limited | Distorter coefficient updating apparatus, method and digital predistortion apparatus |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7782974B2 (en) * | 2006-03-24 | 2010-08-24 | Texas Instuments Incorporated | Apparatus and method providing non-linear adaptive signal tracking |
| CN101860326A (en) * | 2010-06-29 | 2010-10-13 | 奥维通信股份有限公司 | Indirect learning predistortion linearized system based on Volterra series |
-
2012
- 2012-03-18 CN CN201210071249.2A patent/CN102624338B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7782974B2 (en) * | 2006-03-24 | 2010-08-24 | Texas Instuments Incorporated | Apparatus and method providing non-linear adaptive signal tracking |
| CN101860326A (en) * | 2010-06-29 | 2010-10-13 | 奥维通信股份有限公司 | Indirect learning predistortion linearized system based on Volterra series |
Non-Patent Citations (4)
| Title |
|---|
| 吕海飞等: "基于简化VOLTERRA级数的数字预失真系统设计方法", 《制导与引信》 * |
| 徐文波等: "一种新型数字预失真算法的设计", 《电讯技术》 * |
| 房保卫等: "基于Volterra级数间接学习型预失真器的FPGA实现", 《中国集成电路》 * |
| 王勇等: "宽带正交频分复用系统的记忆型预失真器", 《系统工程与电子技术》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103036514A (en) * | 2012-11-21 | 2013-04-10 | 南京航空航天大学 | Method for calculating output quantity of power amplifier by using Volterra correction model |
| CN103036514B (en) * | 2012-11-21 | 2015-09-09 | 南京航空航天大学 | Utilize the method for Volterra correction model rated output amplifier output variable |
| CN104219182A (en) * | 2014-08-12 | 2014-12-17 | 上海航天电子通讯设备研究所 | Radio-frequency power amplifier linearization system based on broadband modulator |
| CN104219182B (en) * | 2014-08-12 | 2017-11-10 | 上海航天电子通讯设备研究所 | RF power amplification linearized system based on wideband modulator |
| CN104410593A (en) * | 2014-12-04 | 2015-03-11 | 西北工业大学 | Decision feedback model-based digital symbol nonlinear error correction equalization method |
| CN104410593B (en) * | 2014-12-04 | 2017-06-30 | 西北工业大学 | Numerical chracter nonlinearity erron amendment equalization methods based on decision-feedback model |
| CN105610408A (en) * | 2015-12-17 | 2016-05-25 | 西安空间无线电技术研究所 | Non-linear Volterra filtering optimization method based on contribution factor |
| CN105610408B (en) * | 2015-12-17 | 2018-04-10 | 西安空间无线电技术研究所 | A kind of Nonlinear Volterra filtering optimization method based on contribution factor |
| CN113114125A (en) * | 2021-04-20 | 2021-07-13 | 上海矽昌微电子有限公司 | Digital predistortion correction method and system for double-loop resolving |
| US11901951B2 (en) | 2021-07-07 | 2024-02-13 | Fujitsu Limited | Distorter coefficient updating apparatus, method and digital predistortion apparatus |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102624338B (en) | 2014-09-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102624338B (en) | Volterra-filtering-based power amplifier pre-distortion method for double-loop feedback model | |
| CN100571023C (en) | A kind of adaptive pre-distortion method of broadband linear power amplifier and system | |
| EP3089414B1 (en) | Digital pre-distortion parameter obtaining method and pre-distortion system | |
| CN103715992B (en) | Based on the power amplifier predistortion device and method simplifying Volterra progression | |
| EP3267579A1 (en) | Linearizing power amplifiers outputs in multi-antenna system | |
| CN111245375A (en) | Power amplifier digital predistortion method of complex value full-connection recurrent neural network model | |
| AU2010364181A1 (en) | Joint process estimator with variable tap delay line for use in power amplifier digital predistortion | |
| CN102969987A (en) | Undersampling-based broadband power-amplifier pre-distortion method | |
| CN110765720A (en) | Power amplifier predistortion method of complex value assembly line recurrent neural network model | |
| CN101127534B (en) | Simulated pre-distortion circuit and method | |
| CN101072220A (en) | Radical basis function neural network predistortion method for adaptive power amplifier | |
| CN101860326A (en) | Indirect learning predistortion linearized system based on Volterra series | |
| CN101741787A (en) | Predistortion fast-convergence method and system for collecting training data | |
| CN103929136A (en) | Broadband mobile communication LTE system transmitter power amplifier predistortion processing method | |
| CN201127032Y (en) | Digital predistortion apparatus based on TD-SCDMA signal | |
| CN201409180Y (en) | Self-adapting baseband linearization device of digital television transmitter | |
| CN103179074A (en) | Self-adapting predistortion system and method based on orthogonal polynomials | |
| CN107786174B (en) | Circuit and method for predistortion | |
| CN102081751A (en) | Method for modeling synchronous double-frequency power amplifier based on real number time delay neural network | |
| CN108449294B (en) | Pre-distortion processing method and device and pre-distortion processing system | |
| CN114565077B (en) | Deep neural network generalization modeling method for power amplifier | |
| CN103888395B (en) | Digital pre-distortion method and digital pre-distortion device | |
| CN101001327A (en) | Excitation apparatus of digital TV transmitter based on nerval net predistortion technology | |
| CN105680919A (en) | Crossover MIMO (Multiple Input Multiple Output) system predistortion method capable of compensating IQ (In-phase and Quadrature-phase) imbalance effect | |
| CN104009717A (en) | Self-adaptive pre-distortion processing method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20140903 Termination date: 20150318 |
|
| EXPY | Termination of patent right or utility model |