CN110276100B - Behavior model modeling and implementation method based on Volterra series - Google Patents
Behavior model modeling and implementation method based on Volterra series Download PDFInfo
- Publication number
- CN110276100B CN110276100B CN201910410450.0A CN201910410450A CN110276100B CN 110276100 B CN110276100 B CN 110276100B CN 201910410450 A CN201910410450 A CN 201910410450A CN 110276100 B CN110276100 B CN 110276100B
- Authority
- CN
- China
- Prior art keywords
- model
- input signal
- mag
- parameters
- parameter
- 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
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
Abstract
The invention discloses a behavior model modeling and realizing method based on Volterra series, which firstly reduces redundant parameters for initializing a Volterra series behavior model, then identifies and extracts input signal amplitude sensitive parameters in the model, and realizes the behavior model in a microwave circuit and system simulation software. Compared with the prior art, the method has the advantages of low complexity, convenience and quickness in parameter extraction, high precision, wide applicable frequency band and the like, greatly reduces the realization difficulty of a Volterra series behavior model, and provides an efficient means for analysis, design and optimization of a radio frequency front terminal system in systems such as wireless communication and radar.
Description
Technical Field
The invention belongs to the field of behavior model modeling, and particularly relates to a behavior model modeling and implementation method based on Volterra series.
Background
The behavioral model of the microwave device, the circuit and the system is based on the measured data of the measured piece, and the input and output characteristics of the measured piece are accurately modeled. In the modeling process of the behavior model, the measured piece is regarded as a black box, the internal structure of the measured piece does not need to be determined, the time domain and frequency domain characteristics of the measured piece under the actual measurement condition are directly simulated based on a mathematical method, and compared with a physical basic model, the behavior model has the advantages of high precision, less occupied computing resources, high simulation speed and the like. Therefore, the behavioral model plays an important role in the analysis, design and optimization of the radio frequency front end subsystem in the wireless communication, radar and other systems.
Aiming at the research of the behavior model, various behavior models such as a Volterra series model, a memory polynomial model, a Saleh model, a Wiener model, a Hammerstein model and the like are proposed at present. The Volterra series model is most comprehensive and universal, and can simulate the input and output characteristics of a dynamic nonlinear system in any order and memory length. The traditional Volterra series model usually has higher order and memory length for improving the precision, however, the complexity and the parameter number of the Volterra series model increase exponentially along with the increase of the order and the memory length of the model, so that the difficulty of parameter extraction of the Volterra series model is greatly increased, and the calculation efficiency of the model is reduced. In order to reduce the complexity of a high-order Volterra series model, related researches in academic communities provide a plurality of methods for reducing model parameters (such as a V-vector algebra method, a dynamic deviation reduction method and the like) and a dynamic Volterra series model based on first-order truncation. However, these parameter reduction methods use more complex algorithms or need to know the physical characteristics of the measured object in advance; the dynamic Volterra series model based on first-order truncation has reduced modeling precision for a tested piece with strong nonlinearity and memory effect, so the method has limitation in practical application. In addition, the reported Volterra series models are realized by being programmed based on Matlab and other numerical simulation software, and are difficult to be compatible with circuits and System-level simulation software (such as an Advanced Design System, a System Vue and the like), so that the application of behavior models in radio frequency front-end terminal System simulation is limited.
In view of the above existing problems, a Volterra series behavior model which has low complexity and is convenient and fast to extract parameters and can be integrated in a circuit and system-level simulation software is needed, so as to provide an efficient means for analysis, design and optimization of a radio frequency front-end terminal system.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides a behavior model modeling and realizing method based on a Volterra series, so that the behavior model has the advantages of low complexity, convenient parameter extraction, high precision, capability of being integrated in circuits and system-level simulation software and the like, and efficient means is provided for analysis, design and optimization of radio frequency front terminal systems in systems such as wireless communication and radar.
The invention adopts the following technical scheme for realizing the aim of the invention:
a behavior model modeling method based on Volterra series comprises the following steps:
and 3, identifying and extracting amplitude sensitive parameters of the input signals.
When initializing in the step 1, setting the order N and the memory length M of the Volterra series model, and obtaining an initialization model equation according to the Volterra series theory as follows:
where t is time, x (t) is the input signal, y (t) is the output signal, a 1m 、
……
、
Are all model parameters, T s To memorize the interval time, m 1 、m 2 、……、m N The interval number is memorized.
The specific steps of the step 2 are as follows:
(2-1) applying an excitation signal x (t) with the amplitude being a typical value when the tested piece works at the input end of the tested piece, and measuring to obtain the waveforms of an input signal x (t) and an output signal y (t);
(2-2) inputting one period N of the signal x (t) para Are equally divided to obtain N para -1 bisector point; measuring the input signal x (t) and the output signal y (t) at each of the time of the division point and the time of the end of the cycle) Substituting the measured value into the initialized model equation to obtain N para A system of linear equations of order one, where N para The number of model parameters after the initialization model equation is expanded;
obtained N para The system of first order linear equations is as follows:
wherein T is the period of the input signal x (T); solving the linear equation set can obtain each model parameter a in the initialized model equation 1m 、
……、
Is started.
(2-3) calculating the Mean value Mean of the initial values of the model parameters in the initialized model equation obtained in the step (2-2), namely
A redundancy parameter decision Threshold is set. The initialized model equation is a complete mathematical form, however, the influence of part of high-order term parameters on the model prediction result is negligible, so that the negligible parameters can be screened out according to the set Threshold value Threshold. For each model parameter in the initialized model equation, if its initial value is less than Mean/Threshold, then the model parameter may be clipped (i.e., its value set to 0) in the initialized model equation. The model equation after the parameter reduction is recorded as y 1 (t), wherein the number of model parameters contained in the model is recorded as N' para 。
The specific steps of the step 3 are as follows:
(3-1) when the input and output signal waveforms are measured in the step (2-1), only the excitation signal x (t) with the amplitude which is typical when the tested piece works is applied to the input end of the tested piece, and in practical application, the amplitude of the input signal when the tested piece works is in oneAnd (c) within a range. Therefore, to improve the versatility of the model, the interval [ Mag ] is swept over the range x (t) L ,Mag H ]Amplitude of the internal scanning input signal x (t), where the number of scanning points is N Mag Simultaneously measuring the waveforms of the input signal x (t) and the output signal y (t);
(3-2) performing N 'for one period of the input signal x (t) for each input signal x (t) corresponding to the amplitude scan value' para Are equally divided to obtain N' para -1 bisector point; substituting the measured value of the input signal x (t) and the measured value of the output signal y (t) at each of the division point time and the period end time into the model equation y with the parameters reduced obtained in the step (2-3) 1 (t) to obtain N' para A system of linear equations of primitive first order; solving the linear equation set to obtain the model equation y under the input signal amplitude 1 Values of each model parameter in (t);
(3-3) for model equation y 1 (t) for each model parameter, calculated in step (3-2) at N Mag Values at the amplitude of the input signal x (t), from which N Mag Finding the Max of the maximum values Mag Minimum value Min Mag And Mean Mag Setting a Threshold for identifying amplitude sensitive parameters Mag . For y 1 (t) each model parameter of (t), if (Max) of the model parameter Mag -Min Mag )/Mean Mag Greater than Threshold Mag Identifying the model parameter as an input signal amplitude sensitive parameter; if the model parameter is (Max) Mag -Min Mag )/Mean Mag Not more than Threshold Mag Then the model parameters are identified as input signal amplitude insensitive parameters and Mean is used Mag As the final extracted value of the model parameter;
(3-4) substituting the final extracted value of the input signal amplitude insensitive parameter identified in the step (3-3) into a model equation y 1 (t), setting optimization targets (such as output signal waveform, output signal amplitude and the like) under each input signal x (t) amplitude scanning value, and calling corresponding algorithms (such as a gradient descent method, a conjugate gradient method, a Newton method or a genetic method) according to specific situationsAlgorithm, etc.) to obtain the input signal amplitude sensitive parameters identified in the step (3-3) and obtain discrete values of the amplitude sensitive parameters under each input signal x (t) amplitude scanning value;
(3-5) selecting a proper fitting function (such as a polynomial function, an exponential function and the like) for each input signal amplitude sensitive parameter, and fitting the discrete value of the amplitude sensitive parameter obtained in the step (3-4) under each input signal x (t) amplitude scanning value by adopting a least square method, so as to extract the model parameter of the fitting function. So far, the extraction of all parameters of the model is completed.
Aiming at the concrete realization of the behavior model modeling method: the method adopts a Symbol Definition Device (SDD) function module in a commercial microwave/radio frequency circuit and System simulation software Advanced Design System (ADS) to realize a Volterra series behavior model, and comprises the following specific steps:
firstly, a Volterra series behavior model with the memory length of M is realized in ADS software by adopting an SDD (service description device) with an M +1 port. In which the 1 st port is used for connecting an input signal x (t) and is modeled by the equation y 1 (t) calculating output signal, i +1 (i is more than or equal to 1 and less than or equal to M) th port for simulating memory length iT s Signal x (t-iT) of s ). Signal x (t-iT) s ) The method is realized by connecting a time Delay (TimeDelay) component in ADS in series after x (t), wherein the Delay time Delay in the TimeDelay component is set as iT s ；
Then, an input excitation signal x (t) is applied at the 1 st port of the SDD; applying an excitation signal x (t) which is the same as that of the 1 st port to the far end of the TimeDelay component of the (i + 1) th (i is more than or equal to 1 and less than or equal to M) ports for simulating the memory length iT s Input signal x (t-iT) s )；
And finally, simulating the established SDD model by adopting a Harmonic Balance (Harmonic Balance) simulator in the ADS, and calculating an output signal y (t).
The invention has the following beneficial effects:
firstly, the behavioral model based on the Volterra series reduces redundant parameters in the traditional Volterra series, sets the input signal amplitude sensitive parameters to be related to the input signal amplitude, and further obtains a high-precision result by using a small amount of easily-extracted parameters.
Secondly, compared with the traditional programming implementation mode based on numerical simulation software, the behavioral model based on the Volterra series is implemented in the commercial microwave/radio frequency circuit and system simulation software, so that the model implementation difficulty is greatly reduced, the model can be directly used for the analysis, design and optimization of the circuit and the system, and the efficiency of circuit-level and system-level modeling and design is improved.
Thirdly, the behavior model based on the Volterra series has no requirement on the frequency of the excitation signal, has broadband characteristics, and is suitable for microwave, millimeter wave and terahertz frequency bands.
Drawings
FIG. 1 is a flow chart of the modeling and implementation of the present invention.
FIG. 2 is an input signal amplitude sensitive parameter a in the embodiment 12 ，a 200 ，a 3222 Discrete value and fitting effect graph.
Fig. 3 is a schematic diagram of a behavior model based on a Volterra series implemented in ADS simulation software in a specific embodiment.
FIG. 4 is a graph showing the effect of fitting the Volterra series-based behavior model to the waveform of the output signal y (t).
FIG. 5 is a graph showing the effect of fitting the fundamental wave, the second harmonic wave and the third harmonic wave output power by the Volterra series-based behavior model.
Detailed Description
The following description of the embodiments of the invention is provided to further explain the embodiments of the invention by describing an implementation example in detail with reference to the accompanying drawings.
The flow of the behavioral model modeling and implementation method based on the Volterra series is shown in figure 1.
The behavior model modeling method based on the Volterra series comprises the following specific steps:
step 1: and initializing a model equation.
Setting the order N =3 and the memory length M =2 of the Volterra series model, and obtaining an initialization model equation according to the basic theory of the Volterra series as follows:
where t is time, x (t) is the input signal, y (t) is the output signal,
(i is more than or equal to 1 and less than or equal to 3) as model parameters, m and m 1 、m 2 、m 3 To memorize the number of intervals, T s Is the interval time of memory; t is a unit of s Without loss of generality, this embodiment can be set to one third of the period of the input signal x (t), i.e. 3.03ps.
And 2, step: model parameter initialization and redundant parameter reduction.
(2-1), without loss of generality, this embodiment uses a gallium arsenide hemt as the device under test. An excitation signal x (t) with the amplitude which is the typical value (such as 0 dBm) of the tested piece in operation is applied to the input end of the tested piece, and the waveforms of the input signal x (t) and the output signal y (t) are obtained by measuring with an oscilloscope. Without loss of generality, when hardware conditions required by measurement are not available, waveforms of the input signal x (t) and the output signal y (t) under the same measurement conditions can be obtained based on an equivalent circuit model or physical base model simulation of the measured piece and used as measurement data for behavioral model modeling.
(2-2) performing N for one cycle of the input signal x (t) para Is equally divided to obtain N para -1 bisector point. Substituting the measured value of the input signal x (t) and the measured value of the output signal y (t) for each of the division point time and the cycle end time into the initialization model equation (1) yields N shown below para A system of linear equations of first order:
where T =9.09ps is the period of the input signal x (T). According to the number N of model parameters contained in the initialized model equation (1) para ＝1And 9, solving the linear equation set to obtain initial values of the model parameters in the initialized model equation (1).
And (2-3) calculating a Mean value of initial values of the model parameters in the initialization model equation (1) obtained in the step (2-2), and setting a redundancy parameter judgment Threshold value Threshold =100. The initialized model equation (1) is a complete form in mathematics, however, the influence of part of high-order term parameters on the model prediction result is negligible, so that the negligible parameters can be screened out according to the set Threshold value Threshold. For each model parameter in the initialized model equation (1), if its initial value is less than Mean/Threshold (i.e., the value of the parameter is less than 1% of the Mean of all parameters), then the model parameter may be clipped (i.e., its value set to 0) in the initialized model equation (1). The model equation after the parameter reduction is recorded as y 1 (t), wherein the number of model parameters contained in the model is recorded as N' para In the present embodiment, the model equation after reduction includes N 'model parameters' para =7,7 model parameters each being a 10 ，a 11 ，a 12 ，a 200 ，a 211 ，a 3000 And a 3222 Reduced model equation y 1 (t) is as follows:
and step 3: and identifying and extracting the input signal amplitude sensitive parameters.
(3-1) when the input and output signal waveforms are measured in the step (2-1), only the excitation signal x (t) with the amplitude being the typical value (0 dBm) when the tested piece works is applied to the input end of the tested piece, and the amplitude of the input signal can change within a range when the tested piece works in practical application. Therefore, to improve the versatility of the model, the interval [ Mag ] is scanned in the range of x (t) L ,Mag H ]Amplitude (number of scanning points N) of internal scanning input signal x (t) Mag ) The waveforms of the input signal x (t) and the output signal y (t) are measured simultaneously. In this embodiment, mag L ＝-30dBm，Mag H ＝5dBm，N Mag ＝8。
For each input signal x (t) corresponding to an amplitude sweep value, N 'is carried out for one period of the input signal x (t)' para Are equally divided to obtain N' para -1 bisector point. Substituting the measured value of the input signal x (t) and the measured value of the output signal y (t) at each of the halving point time and the cycle end time into the model equation (3) obtained in the step (2-3) after the parameter reduction to obtain N' para A system of linear equations of first order. And solving the linear equation set to obtain values of each model parameter in the model equation (3) under the input signal amplitude.
(3-3) for each model parameter in the model equation (3), the calculation of step (3-2) yields its value at N Mag Value at the amplitude of the input signal x (t), from which N Mag Finding the Max of the maximum values Mag Minimum value Min Mag And Mean Mag Setting a Threshold for identifying amplitude sensitive parameters Mag =0.05. For each model parameter in model equation (3), (Max) if the model parameter is Mag -Min Mag )/Mean Mag Greater than Threshold Mag (i.e. the relative variation of the model parameter under different input signal amplitudes is greater than 5%), identifying the model parameter as an input signal amplitude sensitive parameter; if the model parameter is (Max) Mag -Min Mag )/Mean Mag Not more than Threshold Mag Then the model parameters are identified as input signal amplitude insensitive parameters and Mean is used Mag As the final extracted value of the model parameter. In the present embodiment, the model parameter a 12 、a 200 And a 3222 Is identified as an input signal amplitude sensitive parameter.
(3-4) substituting the final extracted value of the input signal amplitude insensitive parameter identified in the step (3-3) into the model equation (3), setting an optimization target (in the embodiment, the optimization target is set as an output signal waveform and an output signal power) under the amplitude scanning value of each input signal x (t), calculating the input signal amplitude sensitive parameter identified in the step (3-3) through an optimization algorithm, and obtaining an amplitude sensitive parameter a 12 、a 200 And a 3222 At each input signal x: (t) discrete values under amplitude sweep values.
And 3-5, selecting a fitting function shown as follows for each input signal amplitude sensitive parameter:
and (5) fitting the discrete values of the amplitude sensitive parameters calculated in the step (3-4) under the amplitude scanning value of each input signal x (t) by adopting a least square method, thereby extracting and obtaining the parameters of the fitting function. a is 12 、a 200 、a 3222 The discrete values at each input signal x (t) amplitude and the fitting effect of equations (4) to (6) on the discrete values are shown in fig. 2. At this point, the extraction of all the parameters of the model is completed.
Aiming at the concrete realization of the behavior model modeling method: the Volterra series behavior model provided by the invention is realized by adopting a Symbol Defined Device (SDD) functional module in a commercial microwave/radio frequency circuit and System simulation software Advanced Design System (ADS). The model implemented in ADS is shown in fig. 3. The method comprises the following specific steps:
firstly, a Volterra series behavior model with the memory length of 2 is realized in ADS software by adopting a SDD with 3 ports. The 1 st port is used to connect the input signal x (t) and calculate the output signal according to the model equation (3). The (i + 1) th port (i is more than or equal to 1 and less than or equal to 2) is used for simulating the memory length iT s Signal x (t-iT) of s ). Signal x (t-iT) s ) The method is realized by connecting a time Delay (TimeDelay) component in ADS in series after x (t), wherein the Delay time Delay in the TimeDelay component is set as iT s 。
Then, an input excitation signal x (t) is applied to the 1 st port of the SDD; the excitation signal x (t) which is the same as that of the 1 st port is applied to the far end of the TimeDelay component of the (i + 1) (i is more than or equal to 1 and less than or equal to 2) th port and is used for simulating the memory length iT s Input signal x (t-iT) s )。
And finally, simulating the established SDD model by adopting a Harmonic Balance (Harmonic Balance) simulator in the ADS, and calculating an output signal y (t). The results of comparing the waveforms of the output signal y (t) simulated and measured by the behavior model and the fundamental wave, the second harmonic and the third harmonic output power thereof are shown in fig. 4 and 5 respectively.
The foregoing detailed description of the embodiments of the invention has been presented with reference to the accompanying drawings and examples, but the invention should not be construed as being limited thereto. Variations can be made within the knowledge of those skilled in the art without departing from the spirit of the invention.
Claims (7)
1. The behavior model modeling method based on the Volterra series is characterized by comprising the following steps:
step 1, initializing a model, setting the order N and the memory length M of a Volterra series model, and obtaining an initialized model equation according to a Volterra series theory;
step 2, initializing model parameters and reducing redundant parameters according to an initialized model equation, specifically:
(2-1), applying an excitation signal x (t) with the amplitude of a typical value of the tested piece during working at the input end of the tested piece, and measuring to obtain the waveforms of an input signal x (t) and an output signal y (t);
(2-2) performing N for one cycle of the input signal x (t) para Are equally divided to obtain N para -1 bisector point; substituting the measured value of the input signal x (t) and the measured value of the output signal y (t) of each equal division point time and period end time into an initialization model equation to obtain N para Solving the linear equation set to obtain initial values of the model parameters in the initialized model equation; said N is para The number of model parameters after the initialization model equation is expanded;
(2-3) determining a Mean value Mean of initial values of each model parameter according to the initialized model equation obtained in the step (2-2), setting a redundant parameter judgment Threshold value Threshold, and screening out model parameters needing to be ignored according to the set Threshold value Threshold; for each model parameter in the initialized model equation, if the initial value of the model parameter is less than Mean/Threshold, the model parameter is reduced in the initialized model equation, namely the value of the model parameter is set to 0; the model equation after the parameter reduction is recorded as y 1 (t), wherein the number of model parameters contained in the model is recorded as N' para ；
Step 3, identifying and extracting amplitude sensitive parameters of the input signals, specifically:
(3-1) scanning the amplitude of the input signal x (t) within the range enabling the tested piece to work normally, and setting the amplitude scanning interval of x (t) as [ Mag ] L ,Mag H ]Where the number of scanning points is N Mag Simultaneously measuring the waveforms of the input signal x (t) and the output signal y (t);
(3-2), for each input signal x (t) corresponding to the amplitude sweep value, performing N 'for one period of the input signal x (t)' para Are equally divided to obtain N' para -1 bisector point; substituting the measured value of the input signal x (t) and the measured value of the output signal y (t) at each of the division point time and the period end time into the model equation y with the parameters reduced obtained in the step (2-3) 1 (t) to give N' para Solving the linear equation system to obtain the model equation y under the input signal amplitude 1 Values of each model parameter in (t);
(3-3) for model equation y 1 (t) each model parameter obtained in step (3-2) at N Mag Values at the amplitude of the input signal x (t), from which N Mag Finding the Max of the maximum values Mag Minimum Min Mag And Mean Mag Setting a Threshold for identifying amplitude sensitive parameters Mag (ii) a For the model equation y 1 (t) each model parameter if (Max) of that model parameter Mag -Min Mag )/Mean Mag Greater than Threshold Mag Then will beThe model parameters are identified as input signal amplitude sensitive parameters; if the model parameter is (Max) Mag -Min Mag )/Mean Mag Not more than Threshold Mag Then the model parameters are identified as input signal amplitude insensitive parameters and Mean is used Mag As the final extracted value of the model parameter;
(3-4) substituting the final extracted value of the input signal amplitude insensitive parameter identified in the step (3-3) into a model equation y 1 (t), setting an optimization target under each input signal x (t) amplitude scanning value, calling a corresponding algorithm according to specific conditions to obtain the input signal amplitude sensitive parameters identified in the step (3-3), and obtaining discrete values of the amplitude sensitive parameters under each input signal x (t) amplitude scanning value;
(3-5) for each input signal amplitude sensitive parameter, fitting the discrete value of the amplitude sensitive parameter obtained in the step (3-4) under each input signal x (t) amplitude scanning value, and extracting a model parameter of a fitting function; so far, the extraction of all parameters of the model is completed;
realizing a Volterra series behavior model according to a symbol definition Device Symbolic Defined Device functional module in a commercial microwave/radio frequency circuit and System simulation software Advanced Design System; wherein, the Advanced Design System is abbreviated as ADS, and the Symbolic Defined Device is abbreviated as SDD.
2. The behavioral model modeling method based on Volterra series according to claim 1, characterized in that the initialization model equation in step 1 is as follows:
where t is time, x (t) is the input signal, y (t) is the output signal, a 1m 、
……、
Are all model parameters, T s To memorize the interval time, m 1 、m 2 、……、m N The number of intervals is memorized.
3. The Volterra series-based behavior model modeling method according to claim 2, wherein N in step 2 is para The system of linear equations of first order is:
wherein T is the period of the input signal x (T), and each model parameter a in the initialized model equation can be obtained by solving the linear equation set 1m 、
……、
Of (4) is calculated.
4. The behavioral model modeling method based on Volterra series according to claim 2, characterized in that the method for determining the Mean of the initial values of each model parameter in step (2-3) is:
5. the Volterra series-based behavior model modeling method according to claim 1, wherein the corresponding algorithm invoked on a case-by-case basis in step (3-4) for the optimization objective comprises a gradient descent method, a conjugate gradient method, a Newton method, or a genetic algorithm.
6. The Volterra series-based behavioral model modeling method according to claim 1, characterized in that in step (3-5), for each input signal amplitude sensitive parameter, an appropriate fitting function is selected and least square method is adopted for fitting.
7. The Volterra series-based behavior model modeling method according to claim 1, characterized by comprising the following concrete implementation steps:
firstly, realizing a Volterra series behavior model with a memory length of M in ADS software by adopting an SDD (service description device) with an M +1 port; in which the 1 st port is used for connecting an input signal x (t) and is modeled by the equation y 1 (t) calculating the output signal, the (i + 1) th port being used for simulating the memory length iT s Signal x (t-iT) of s ) (ii) a Signal x (t-iT) s ) The method is realized by serially connecting a Delay TimeDelay component in the ADS after an input signal x (t), wherein the Delay TimeDelay in the TimeDelay component is set as iT s ；
Then, an input excitation signal x (t) is applied at the 1 st port of the SDD; the same excitation signal x (t) as that applied to the 1 st port is applied to the far end of the TimeDelay component of the (i + 1) th port for simulating the memory length iT s Input signal x (t-iT) s )；
Finally, simulating the established SDD model by adopting a harmonic balance simulator in the ADS, and calculating an output signal y (t);
wherein i is more than or equal to 1 and less than or equal to M.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910410450.0A CN110276100B (en) | 2019-05-16 | 2019-05-16 | Behavior model modeling and implementation method based on Volterra series |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910410450.0A CN110276100B (en) | 2019-05-16 | 2019-05-16 | Behavior model modeling and implementation method based on Volterra series |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110276100A CN110276100A (en) | 2019-09-24 |
| CN110276100B true CN110276100B (en) | 2022-11-22 |
Family
ID=67959939
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910410450.0A Active CN110276100B (en) | 2019-05-16 | 2019-05-16 | Behavior model modeling and implementation method based on Volterra series |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110276100B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1790896A (en) * | 2005-12-19 | 2006-06-21 | 宁波大学 | Radio-frequency power amplifier modeling method based on memory effect |
| CN102315835A (en) * | 2011-05-06 | 2012-01-11 | 中国工程物理研究院电子工程研究所 | Method for estimating roll-off coefficient of forming filter |
| 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 |
| CN106126804A (en) * | 2016-06-21 | 2016-11-16 | 上海无线电设备研究所 | The behavioral scaling modeling of a kind of power amplifier bottom circuit and verification method |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8855175B2 (en) * | 2012-08-02 | 2014-10-07 | Telefonaktiebolaget L M Ericsson (Publ) | Low complexity all-digital PIM compensator |
| WO2014184774A1 (en) * | 2013-05-16 | 2014-11-20 | Telefonaktiebolaget L M Ericsson (Publ) | Baseband equivalent volterra series for digital predistortion in multi-band power amplifiers |
-
2019
- 2019-05-16 CN CN201910410450.0A patent/CN110276100B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1790896A (en) * | 2005-12-19 | 2006-06-21 | 宁波大学 | Radio-frequency power amplifier modeling method based on memory effect |
| CN102315835A (en) * | 2011-05-06 | 2012-01-11 | 中国工程物理研究院电子工程研究所 | Method for estimating roll-off coefficient of forming filter |
| 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 |
| CN106126804A (en) * | 2016-06-21 | 2016-11-16 | 上海无线电设备研究所 | The behavioral scaling modeling of a kind of power amplifier bottom circuit and verification method |
Non-Patent Citations (2)
| Title |
|---|
| "一种简化 Volterra 级数的记忆功放行为模型";杨新，汪琳娜;《科学技术与工程》;20141231;第1-5页 * |
| Bilel Fehri ; Slim Boumaiza."Baseband Equivalent Volterra Series for Behavioral Modeling and Digital Predistortion of Power Amplifiers Driven With Wideband Carrier Aggregated Signals".《 IEEE Transactions on Microwave Theory and Techniques 》.2014, * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110276100A (en) | 2019-09-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Cai et al. | Padé-approximation-based behavioral modeling | |
| Widanage et al. | Nonlinear system-identification of the filling phase of a wet-clutch system | |
| CN106199185B (en) | A kind of linear impulsive response measurement method and system based on continuous logarithmic frequency sweep | |
| CN103685111A (en) | Calculating method of digital pre-distortion parameters and pre-distortion system | |
| CN104236893A (en) | Performance parameter test system and performance parameter test method of hydraulic damper | |
| CN103064010B (en) | Parameter estimation method for artificial circuit fault component based on Hilbert-Huang transforming (HHT) | |
| CN113392610B (en) | Simulation optimization method and device for transient analysis of large-scale integrated circuit | |
| CN107346017A (en) | The intensive target simulation method of pulse compression radar based on frequency matching filtering | |
| CN105787219A (en) | Method for building conducted interference coupling channel multiple linear regression model by near frequency point sampling | |
| GB2408122A (en) | Generating a circuit model | |
| CN108519436B (en) | Concrete damage assessment method based on acoustic emission signal statistical analysis | |
| CN113095192B (en) | Dynamic load spectrum compiling method based on time domain extrapolation technology | |
| CN101689836B (en) | Method for linearizing non-linear power amplifier and apparatus thereof | |
| CN111967186A (en) | Neural network space mapping method for large signal modeling of power transistor | |
| CN110276100B (en) | Behavior model modeling and implementation method based on Volterra series | |
| CN106405388B (en) | A kind of digit chip function test method and system | |
| CN113189561A (en) | Sea clutter parameter estimation method, system, equipment and storage medium | |
| CN111898320A (en) | Radio frequency power amplifier scattering parameter extraction method and device based on neural network | |
| CN116611378A (en) | Simulation method and device for circuit model, computer equipment and storage medium | |
| CN103197292B (en) | The generation of Multichannel radar analogue echoes signal and verification method | |
| CN107967395A (en) | A kind of nonlinear time_varying system Fast Identification Method based on the expansion of beta wavelet basis functions | |
| Sharma et al. | Automated time domain modeling of linear and nonlinear microwave circuits using recurrent neural networks | |
| CN107872358B (en) | Automatic simulation test method for HDLC protocol | |
| CN104483546A (en) | Spectrum analysis method of FPGA (Field Programmable Gate Array) digital logic signal | |
| King et al. | Time-domain simulation of passband S-parameter networks using complex baseband vector fitting |
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 |