CN111274752B - Power amplifier behavior modeling method based on two-stage open loop and binary function - Google Patents
Power amplifier behavior modeling method based on two-stage open loop and binary function Download PDFInfo
- Publication number
- CN111274752B CN111274752B CN201811478375.3A CN201811478375A CN111274752B CN 111274752 B CN111274752 B CN 111274752B CN 201811478375 A CN201811478375 A CN 201811478375A CN 111274752 B CN111274752 B CN 111274752B
- Authority
- CN
- China
- Prior art keywords
- power amplifier
- model
- sequence
- delay
- tap
- 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
-
- 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
- G06F17/15—Correlation function computation including computation of convolution operations
-
- 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
Abstract
The invention provides a power amplifier behavior modeling method based on two-stage open loop and binary function, which relates to the technical field of nonlinear system modeling and analysis application.
Description
Technical Field
The invention relates to the technical field of nonlinear system modeling and analysis application, in particular to a power amplifier behavior modeling method based on a two-stage open loop structure and a binary function unit.
Background
The power amplifier is an important module of the transmitter and is a complex nonlinear system. In order to make the power amplifier operate with high efficiency, the transistors in the power amplifier are mostly operated in a region close to saturation or even in a cut-off region, so that the power amplifier often generates severe nonlinear distortion, and the power amplifier generates a memory effect due to the equivalent reactance of the device.
The digital predistortion module can effectively remove the nonlinearity of the power amplifier and the system by a low-cost and high-precision method, and therefore becomes an indispensable part in the modern wireless transmitter. In the digital predistortion phase of the signal, the characteristics of the power amplifier need to be modeled. The linearization performance and practical availability of digital predistortion for power amplifiers depends to a large extent on the accuracy and complexity of the relevant non-linear power amplifier behavior model.
The modeling method of the power amplifier can be divided into physical modeling and behavior modeling: physical modeling usually requires knowing the circuit structure of the power amplifier, the characteristics of related components, the basic circuit law, related theoretical rules and the like; and the behavior modeling only needs to utilize the input and output data of the power amplifier, adopts a black box mode to describe the response characteristics of the system, and is simpler and more convenient to analyze.
According to the physical essence reflected by the model, the behavior model is divided into a non-memory model, a memory model and the like. The memoryless power amplifier model is relatively simple and has good effect on the modeling of the narrow-band system. But in practice the power amplifier is a typical non-linear memory system. Memory models typically employ a basic two-block-Hammerstein model, a Time Delay Neural Network (TDNN) model, and a closed-loop nonlinear auto-regressive moving average (NARMA) model, which exhibits relatively higher accuracy in modeling power amplifiers with memory effects than open-loop power amplifier models. However, the closed-loop power amplifier model requires an additional feedback path to return real-time samples of the power amplifier output, which makes implementation of digital predistortion techniques using the closed-loop model inconvenient for mobile terminals and high frequency systems. In addition, the memory model of the power amplifier is generally a unitary function unit, the essence of the unitary function unit is that the input values at different times respectively and independently influence the output value, and the correlation of the input values at different times also has influence on the output value of the power amplifier, which is not considered by the existing power amplifier modeling method.
Disclosure of Invention
In order to overcome the defects in the prior art, the invention aims to provide a power amplifier behavior modeling method based on a two-stage open-loop structure and a binary function unit. The method utilizes a two-stage open loop structure and a two-stage function unit, can complete modeling of the nonlinear power amplifier with the memory effect only by needing fewer parameters and no feedback path, provides good modeling precision for band-internal distortion and spectrum regeneration, and is beneficial to reducing the interference of nonlinear noise in a system containing a power amplifier circuit to the system by improving the modeling precision of the power amplifier.
In order to achieve the above object, the present invention adopts the following technical solutions:
a power amplifier behavior modeling method based on two-stage open loop and a binary function comprises the following steps:
step 1): collecting input signal data vector x of a power amplifierin=[xin(1),xin(2),xin(3),...,xin(N)]And output signal data vector yout=[yout(1),yout(2),yout(3),...,yout(N)]Wherein each element represents a signal data, N being the data length;
step 2): establishing a first-stage part of a power amplifier behavior model: the forward path μ (k), whose input variables contain the instantaneous and past samples, is expressed as follows:
wherein x (k) and x (k-d)n) Representing instantaneous and past samples, respectively, Fn() Representing a two-dimensional linear function for characterizing the memory effect of a power amplifier, Fn[x(k),x(k-dn)]Is the constructed binary function unit and d represents the delay tap.
Step 3): establishing a second-stage part of the power amplifier behavior model: feedback path λ (k), expression is as follows:
wherein G ism()、Gp()、Gq() And Gr() Representing a two-dimensional linear function, Gm[μ(k),μ(k-dm)]Equal representing binary function unit containing first stage current total output and past total output, d representing delay tap, M, P, Q, R value size measuring the power amplifierThe larger the value is, the larger the memory effect is, and the larger the memory effect is determined according to the actual engineering.
Step 4): the model for power amplifier behavior modeling based on the two-stage open-loop structure and the binary function unit is obtained as follows:
wherein the output y (k) of the power amplifier in the feedback path λ (k) is approximated using the forward path μ (k);
step 5): the selection of the number of terms in the feedback path lambda (k) is determined according to the characteristics of the power amplifier model, and the larger the memory effect of the power amplifier is, the larger the number of terms is, and the general selection isThe feedback path of the power amplifier can be described. Therefore, the parameters of the power amplifier behavior model based on the two-stage open loop structure and the binary function unit are selected, and the number N of the sample delay taps is equal to N0,M=M0,N0、M0The magnitude of (A) measures the memory effect of the power amplifier, N0、M0The larger the memory effect of the power amplifier is, the larger the memory effect is, and the memory effect is determined according to engineering practice; delay tap dnAnd dmRespectively is dn-maxAnd dm-maxActually determining according to engineering; the model estimates the error e ∞, which is reduced by iterative calculations.
Step 6): initializing the behavior model of the power amplifier and selecting a delay tap dnAnd dmThe sequences of (a) and (b) are,
step 7): input signal data vector x of power amplifier to be collectedinAnd output signal data vector youtAnd (4) bringing the behavior model into the power amplifier, and calculating to obtain the behavior model of the power amplifier.
Wherein, youtIs the actual output of the power amplifier and,is the output of a power amplifier behavior model based on a two-level open-loop structure and a binary function unit.
Step 9): update dnAnd dmOf (a) i.e.Anduntil a critical condition is reachedAnduntil now. Is found out so thatMinimized delay tap dnAnd dmThe sequence of (a) is the final output of the power amplifier model. Wherein an optimum tap delay sequence is foundAndincluding but not limited to traversal search, depth first search algorithms, etc.
Preferably, the updating in step 9Andthe algorithm of (a) includes, but is not limited to, the following algorithm steps:
step 91): comparing the value of the previously retained model estimation error e with the value of the current error functionSize of (1), ifThe system updates the delay tap sequence dnAnd dmI.e. by Entering step 7; if it isThe current tap delay sequence value is retained And updating the model estimation errorThe system then updates the delay tap sequence dnAnd dmI.e. by Step 7 is entered and the steps are performed sequentially downwards.
Step 92): when detecting thatAnd isThereafter, the tap delay sequence is updated And entering step 7; when detecting thatAnd isThereafter, step 9 is stopped and the residue is obtainedAnd
the invention adopts a two-stage model, the first stage mu (k) is a parallel memory polynomial, and each branch is represented by a sequence with excellent convergence characteristics and accuracy characteristics. In order to better represent the memory effect of the power amplifier, the input variable of each branch is constructed by a two-stage function unit, the two-stage function unit comprises an instantaneous sample and a past sample, and the influence of the input values at different moments on an output value and the influence of the correlation between the samples at different moments on the output of the power amplifier are considered. The second stage λ (k) of the two-stage model is also a parallel memory polynomial in which the input variables of each branch are not feedback samples from the power amplifier output, but their approximate signals, i.e. the total output of the first stage; meanwhile, the second stage also adopts a two-stage function unit, and the input variables comprise the current total output and the past total output of the first stage.
The invention has the beneficial effects that: the invention predicts the behavior model of the nonlinear power amplifier with memory effect by a two-stage open-loop structure model, and simultaneously uses a binary function unit to describe the specific behavior of the model. In this model, the sampled power amplifier output in the feedback path is replaced by an approximation signal. In addition, a sequence having excellent convergence characteristics and accuracy characteristics is adopted in each branch of the parallel memory polynomial used in the model. The power amplifier model is generated by a forward path and a feedback path, which shows that the output of the power amplifier is not only related to the input at the current moment, but also related to the historical input, and the memory effect of the power amplifier is described. Compared with an open-loop time delay neural network, the model requires fewer parameters and no feedback path, and provides good modeling accuracy for in-band distortion and spectrum regeneration. The method has the advantages that the convergence speed of the model is accelerated, the precision of the model is guaranteed, and the behavior modeling of the power amplifier is facilitated. In addition, the influence of the correlation of input values input into the power amplifier at different moments on an output value is considered by introducing the binary function unit, and compared with a power amplifier model of a unitary function, the method has higher modeling precision.
Drawings
Fig. 1 is a graph of test results for a power amplifier.
Fig. 2 is a flow chart of a power amplifier behavior modeling method based on a two-stage open-loop structure and a binary function unit.
Fig. 3A is a graph of the AM-AM distortion characteristics of a measured power amplifier, and fig. 3B is a graph of the AM-AM distortion characteristics of a fitted power amplifier.
Fig. 4 is an AM-AM distortion characteristic error plot of a power amplifier model.
Fig. 5A is a graph of the AM-PM distortion characteristics of a measured power amplifier, and fig. 5B is a graph of the AM-PM distortion characteristics of a fitted power amplifier.
Fig. 6 is an AM-PM distortion characteristic error plot of a power amplifier model.
Detailed Description
In order to make the aforementioned and other features and advantages of the invention more comprehensible, embodiments accompanied with figures are described in detail below.
The embodiment discloses a power amplifier behavior modeling method based on a two-stage open-loop structure and a binary function unit, which is used for constructing a power amplifier model. The power amplifier is tested using a test structure as shown in fig. 1. The data for model extraction and validation was measured from a 2.15GHZ, 170W Lateral Diffused Metal Oxide Semiconductor (LDMOS) power amplifier and the test signal used the 3G wideband code division multiplexing access specification (WCDMA). The modeling method is shown in fig. 2 and is specifically described below.
Step 1): collecting input signal data vector x of a power amplifierin=[xin(1),xin(2),xin(3),...,xin(N)]And output signal data vector yout=[yout(1),yout(2),yout(3),...,yout(N)]Where N is the data length.
Step 2): the first stage part μ (k) of the power amplifier behavior model is built with its input variables containing instantaneous samples and past samples. Fn() The method comprises the steps of representing a two-dimensional linear function, and being capable of representing the memory effect of the power amplifier; fn[x(k),x(k-dn)]Is a constructed binary function unit containing instantaneous samples and past samples. The expression is as follows:
step 3): establishing a second-stage part lambda (k), G of a power amplifier behavior modelm()、Gp()、Gq() And Gr() Representing two-dimensional linesSexual function, Gm[μ(k),μ(k-dm)]Equal represents a binary function unit, containing the first-stage current total output and the past total output. The expression is as follows:
step 4): approximating the output y (k) of the power amplifier in the feedback path λ (k) using the forward path μ (k), the model for power amplifier behavior modeling based on a two-stage open-loop structure and a binary function unit is derived as:
step 5): the choice of the number of terms in the feedback path λ (k) is determined by the characteristics of the power amplifier model, and is generally chosenThe feedback path of the power amplifier can be described. Therefore, the parameters of the power amplifier behavior model based on the two-stage open loop structure are selected, and the number N of the sample delay taps is equal to N0,M=M0Delay tap dnAnd dmRespectively is dn-maxAnd dm-maxThe model estimation error e ═ infinity.
Step 6): initializing the behavior model of the power amplifier and selecting a delay tap dnAnd dmThe sequences of (a) and (b) are,
step 7): input signal data vector x of power amplifier to be collectedinAnd output signal data vector youtAnd (4) bringing the behavior model into the power amplifier, and calculating to obtain the behavior model of the power amplifier.
Wherein, youtIs the actual output of the power amplifier and,is the output of a power amplifier behavior model based on a two-level open-loop structure and a binary function unit.
Step 9): comparing the value of the previously retained model estimation error e with the value of the current error functionSize of (1), ifOrThe system updates the sequence of delay taps, entering step 7; if it isOrThe current tap delay sequence value is retained And updating the model estimation errorThe system then updates the delay tap sequence, i.e. Step 7 is entered and the steps are performed sequentially downwards.
Step 10): when detecting thatAnd isThereafter, the tap delay sequence is updated And entering step 7; when detecting thatAnd isThereafter, step 9 is stopped and the residue is obtainedAnd
the pairs of the results of fitting after model construction and the results of actual measurement are shown in fig. 3 to 6.
As can be seen from fig. 3A-3B, the measured AM-AM distortion characteristic curve of the power amplifier and the AM-AM distortion characteristic curve fitted to the power amplifier model both enter a saturation state when the input signal power is about-14 dBm, which indicates that the power amplifier model better fits the AM-AM distortion characteristic of the power amplifier.
Fig. 4 compares the error of the power actually output by the power amplifier with the error of the power output by the power amplifier model, and it is obvious from the graph that the error of the power of most points is within 1dB except that the error of individual points is more than 1dB, which shows that the modeling accuracy of the power amplifier model is high.
As is apparent from fig. 5A-5B, the measured AM-PM distortion characteristic curve of the power amplifier and the AM-PM distortion characteristic curve fitted to the power amplifier model both enter an apparent phase distortion state when the input signal power is about-10 dBm, which indicates that the power amplifier model better fits the AM-PM distortion characteristic of the power amplifier.
Fig. 6 compares the phase of the signal actually output by the power amplifier with the phase difference of the signal output by the power amplifier model, and it is obvious from the graph that the error of the phase difference of most points is within pi/8 except that the error of individual points is larger than pi/8, which shows that the modeling precision of the power amplifier model is high.
The above embodiments are only intended to illustrate the technical solution of the present invention and not to limit the same, and a person skilled in the art can modify the technical solution of the present invention or substitute the same without departing from the spirit and scope of the present invention, and the scope of the present invention should be determined by the claims.
Claims (9)
1. A power amplifier behavior modeling method based on two-stage open loop and a binary function comprises the following steps:
1) collecting input signal data vector x of a power amplifierin=[xin(1),xin(2),xin(3),…,xin(N′)]And output signal data vector yout=[yout(1),yout(2),yout(3),…,yout(N′)]Wherein each element represents a signal data, N' is a data length;
2) establishing a first stage of a power amplifier behavior model: forward path μ (k), expressed as:
wherein, Fn() Representing a two-dimensional linear function, representing the memory effect of the power amplifier, and constructing a binary function unit; x (k) denotes instantaneous samples, x (k-d)n) Represents past samples, d represents delay taps;
3) establishing a second stage of the power amplifier behavior model: a feedback path λ (k), expressed as:
wherein G ism()、Gp()、Gq() And Gr() Representing a two-dimensional linear function, and constructing a binary function unit, wherein two elements contained in the binary function unit are the current total output and the past total output of the first stage in sequence, and d represents a delay tap;
4) and adding mu (k) and lambda (k) to obtain a power amplifier behavior model y (k), wherein the expression is as follows:
2. the method of claim 1, wherein the term in the feedback path λ (k) is determined based on a power amplifier memory effect magnitude.
8. The method of claim 6, wherein d is updatednAnd dmThe method of (a) comprises the steps of:
1) comparing the value of the previously retained model estimation error e with the value of the current error functionSize of (1), ifThe delay tap sequence d is updatednAnd dm: If it isThe current tap delay sequence value is retained: and updating the model estimation errorAfter which the delay tap sequence d is updatednAnd dm:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811478375.3A CN111274752B (en) | 2018-12-05 | 2018-12-05 | Power amplifier behavior modeling method based on two-stage open loop and binary function |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811478375.3A CN111274752B (en) | 2018-12-05 | 2018-12-05 | Power amplifier behavior modeling method based on two-stage open loop and binary function |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111274752A CN111274752A (en) | 2020-06-12 |
CN111274752B true CN111274752B (en) | 2022-05-03 |
Family
ID=70998517
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811478375.3A Active CN111274752B (en) | 2018-12-05 | 2018-12-05 | Power amplifier behavior modeling method based on two-stage open loop and binary function |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111274752B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
DE102021126108A1 (en) | 2021-10-08 | 2023-04-13 | IMMS Institut für Mikroelektronik- und Mechatronik-Systeme gemeinnützige GmbH (IMMS GmbH) | Method of extending and using a model to simulate an electronic circuit |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1784879A (en) * | 2003-05-07 | 2006-06-07 | 西门子公司 | Arrangement and method for the digital pre-distortion of a complex baseband input signal |
CN103618521A (en) * | 2013-12-10 | 2014-03-05 | 哈尔滨工业大学 | Radio-frequency binary-pulse generation device and method for dynamic X parameter |
CN105224985A (en) * | 2015-09-28 | 2016-01-06 | 南京航空航天大学 | A kind of power amplifier behavior modeling method based on degree of depth reconstruction model |
CN106411271A (en) * | 2016-08-10 | 2017-02-15 | 成都芯通科技股份有限公司 | Predistortion device for power amplifier and parameter selection method thereof |
CN108153943A (en) * | 2017-12-08 | 2018-06-12 | 南京航空航天大学 | The behavior modeling method of power amplifier based on dock cycles neural network |
CN108256257A (en) * | 2018-01-31 | 2018-07-06 | 南京航空航天大学 | A kind of power amplifier behavior modeling method based on coding-decoding neural network model |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102611661B (en) * | 2012-03-27 | 2015-02-25 | 电子科技大学 | Predistortion device and method based on precise inverse solution memory polynomial model equation |
US9461597B2 (en) * | 2014-11-05 | 2016-10-04 | King Fahd University Of Petroleum And Minerals | Weighted memory polynomial method and system for power amplifiers predistortion |
US9837970B2 (en) * | 2015-03-04 | 2017-12-05 | King Fahd University Of Petroleum And Minerals | Behavioral model and predistorter for modeling and reducing nonlinear effects in power amplifiers |
-
2018
- 2018-12-05 CN CN201811478375.3A patent/CN111274752B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1784879A (en) * | 2003-05-07 | 2006-06-07 | 西门子公司 | Arrangement and method for the digital pre-distortion of a complex baseband input signal |
CN103618521A (en) * | 2013-12-10 | 2014-03-05 | 哈尔滨工业大学 | Radio-frequency binary-pulse generation device and method for dynamic X parameter |
CN105224985A (en) * | 2015-09-28 | 2016-01-06 | 南京航空航天大学 | A kind of power amplifier behavior modeling method based on degree of depth reconstruction model |
CN106411271A (en) * | 2016-08-10 | 2017-02-15 | 成都芯通科技股份有限公司 | Predistortion device for power amplifier and parameter selection method thereof |
CN108153943A (en) * | 2017-12-08 | 2018-06-12 | 南京航空航天大学 | The behavior modeling method of power amplifier based on dock cycles neural network |
CN108256257A (en) * | 2018-01-31 | 2018-07-06 | 南京航空航天大学 | A kind of power amplifier behavior modeling method based on coding-decoding neural network model |
Non-Patent Citations (4)
Title |
---|
Behavioral Modeling of RF Power Amplifiers Using Adaptive Recursive Polynomial Functions;John Dooley 等;《IEEE MMT-S International Microwave Symposium igest》;20061120;第852-855页 * |
Behavioural Modelling of RF Nonlinear Power Amplifiers in Orthogonal Frequency Division Multiplex Transmitter Systems;Yiming LEI;《ulir.ul.ie/handle/10344/1632》;20101231;第1-191页 * |
功放数字预失真模型与处理算法研究;孔潇维;《中国优秀博硕士学位论文全文数据库(博士)信息科技辑》;20140515(第5期);第I135-53页 * |
记忆非线性功率放大器行为建模及其数字预失真;袁晓辉;《中国博士学位论文全文数据库 信息科技辑》;20160815(第8期);第I135-97页 * |
Also Published As
Publication number | Publication date |
---|---|
CN111274752A (en) | 2020-06-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US9225501B2 (en) | Non-linear modeling of a physical system using look-up table with polynomial interpolation | |
Aguirre et al. | On the interpretation and practice of dynamical differences between Hammerstein and Wiener models | |
Vanbeylen et al. | Blind maximum-likelihood identification of Wiener systems | |
CN113037226B (en) | Digital predistortion design method and device based on adaptive step length clipping method | |
CN111274752B (en) | Power amplifier behavior modeling method based on two-stage open loop and binary function | |
Kelly et al. | Low-complexity stochastic optimization-based model extraction for digital predistortion of RF power amplifiers | |
CN111060823A (en) | DP model-based battery SOP online estimation method in low-temperature environment | |
CN112764345B (en) | Strong nonlinear system Kalman filter design method based on target state tracking | |
CN113032988B (en) | Design method of high-order extended Kalman filter based on maximum correlation entropy | |
Gong et al. | A Memorized Recurrent Neural Network Design for Wide Bandwidth PA Linearization | |
CN109995335B (en) | Segmented digital predistortion method capable of realizing hardware efficiently | |
CN111444474A (en) | Multiplicative noise correlation self-adaptive CKF-based target tracking method | |
CN102628897B (en) | Three-order intermodulation test method based on N1dB compression point and N2dB compression point | |
Kantana et al. | A System Approach for Efficiency Enhancement and Linearization Technique of Dual-Input Doherty Power Amplifier | |
Kazlauskas et al. | On intelligent extraction of an internal signal in a Wiener system consisting of a linear block followed by hard-nonlinearity | |
CN111092602A (en) | Modeling method and device of power amplifier, computer equipment and storage medium | |
Yuan et al. | Behavioral modeling of RF power amplifiers with memory effects using orthonormal hermite polynomial basis neural network | |
Averina et al. | Two-block model of a digital corrector for linearization of an analog radio channel | |
Baheti et al. | A new cross correlation algorithm for Volterra kernel estimation of bilinear systems | |
Yu | An experimental evaluation of a digital predistortion system with thermal memory effects modeling | |
Huadong et al. | An Memory polynomial model for power amplifiers | |
CN113472459B (en) | Comprehensive test and modeling platform for nonlinear effect of radio frequency link of receiver | |
Allende-Chávez et al. | Coefficients estimation of MPM through LSE, ORLS and SLS for RF-PA modeling and DPD | |
Gunn et al. | An output-only nonlinear system identification technique suited to integer arithmetic | |
CN110135012B (en) | Regression coefficient determination method of system linear regression prediction model |
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 |