DE102016124018B4 - Adaptive digital predistortion system - Google Patents

Abstract

Adaptive digital predistortion system, comprising the following elements:• a radiation module, which serves to receive a first signal and has the following elements:◯ a modulator, which serves to modulate the first signal and to generate a first modulation signal,◯ a predistortion module, which is coupled to the modulator, in which the predistortion works on the basis of the first modulation signal and generates a first predistortion signal, and◯ a radiation circuit, which is coupled to the predistortion module and is used to generate a first radiation signal based on the first predistortion signal, the first radiation signal being signals transmits via at least one antenna unit, and• a receiving module, which is used to receive the first emitted signal and has the following elements:◯ a receiving circuit, which is used first to receive the first emitted signal and then to generate a second received signal is used,◯ a demodulator, which is coupled to the receiving circuit, is used to demodulate the second received signal and to generate a second demodulation signal, and◯ an estimation module, which is coupled to the demodulator and the predistortion module and calculates the predistortion parameters on the basis of a vector error measurement of a vector error between the first signal and the second demodulation signal, wherein the predistortion parameters are sent from the estimation module to the predistortion module, wherein the predistortion module adjusts the predistortion in the predistortion module on the basis of the predistortion parameters, further• the predistortion module during predistortion in a first section on the basis of the first modulation signal the generates the first predistortion signal, which has a linear relationship with the first modulation signal, wherein• the predistortion module during predistortion in a second section on the basis of the first modulation signal generates the first predistortion signal which has a non-linear relationship with the first modulation signal.

Description

The invention relates to a predistortion system, in particular an adaptive digital predistortion system.

Power amplifiers for radio frequency signals are a key element of RF transmitters. However, the non-linear properties of the power amplifier lead to distortion of the radio frequency signals. Therefore, it is important to effectively cancel the nonlinear properties. In 1 an adaptive predistortion (Digital Pre-Distortion in English and abbreviated to DPD) setup is shown, which comprises a modulator (MOD) 110, a predistorter 120, a radiating circuit 130, a receiving circuit 140 and an extraction module 150. In the conventional method, the predistorter 120 is used in conjunction with an appropriate monitoring receiver. In addition, the signal waveform distortion is used as a guide, so that the signal model parameters are extracted by the extraction module 150 . In addition, the response ratio of the input and output is adjusted based on the extracted parameters from the predistorter 120, thereby correcting the distortion.

A typical memory polynomial model and the corresponding formula for parameter extraction follows below. $$\begin{array}{l}{\widehat{\mathrm{right}}}_B\left[n\right]=DPD\left\{{\overrightarrow{\mathrm{right}}}_B\right\}={\displaystyle \sum _{q=0}^Q{\displaystyle \sum _{l=1}^L{a}_{q,l}\cdot {\mathrm{right}}_B\left[n-q\right]}}\cdot |{\mathrm{right}}_B{\left[n-q\right]}^{l-1}\approx {\widehat{s}}_B\left[n\right]\\ \Rightarrow \underset{\overrightarrow{A}}{\text{at least}}{\left|\mathbf{R}\cdot \overrightarrow{A}-{\overrightarrow{\widehat{s}}}_B\right|}^2\\ \overrightarrow{A}=\left[a_{q,l}|q=0\sim Q,l=1\sim L\right]\\ \mathbf{R}=\left[{\mathrm{right}}_B\left[n-q\right]\cdot \left|{\mathrm{right}}_B{\left[n-q\right]}^{l-1}\right|q=0\sim Q,l=1\sim L,n=0\sim N-1\right]\\ \therefore {\overrightarrow{A}}_{SD,Opt}={\left({\mathbf{R}}^H\cdot \mathbf{R}\right)}^{-1}\cdot {\mathbf{R}}^H\cdot {\overrightarrow{\widehat{s}}}_B\end{array}$$

Formula (1) shows that a complicated matrix calculation is required in the parameter extraction. In addition, the setup for digital predistortion includes linear channel distortion in the emission module. Therefore, even if the non-linear amplifier is approximately a non-linear memoryless model, there must be a model of the storage polynomials, so that the complexity of the whole structure is increased.

Furthermore, in the conventional predistortion technique, the polynomial coefficients are mostly optimized using the ratio of input and output in full sections of the digital predistorter. This not only further increases the complexity of the predistortion system, but also occupies a lot of computing resources and lowers the processing efficiency of the predistortion. There is a need for improvement in the predistortion technique in order to be able to reduce the complexity of the predistortion processing.

From the US 8,548,091 B2 a measurement method and an error correction method for data transmission is already known, in which a predistortion module is provided.

Next is from the U.S. 7,346,112 B1 a direct modulation method of a power amplifier for digital predistortion is known.

An object of the invention is to provide a structure of an adaptive digital predistortion system in which the need for computational resources is reduced and the utilization efficiency of the adaptive predistortion can be improved.

To address at least the above object, the invention provides an adaptive digital predistortion system comprising a transmit module and a receive module. The emitter module is for receiving a first signal and includes the following elements: a modulator, a predistortion module, and a emitter circuit. The modulator serves to modulate the first signal and to generate a first modulation signal. The predistortion module is coupled to the modulator, operates in predistortion based on the first modulation signal, and generates a first pre distortion signal. The radiation circuit is coupled to the predistortion module and is used to generate a first radiation signal based on the first predistortion signal. The first radiated signal transmits signals via at least one antenna unit. The receiving module is used to receive the first emission signal and has the following elements: a receiving circuit, a demodulator and an estimation module. The receiving circuit is used first to receive the first emission signal and then to generate a second received signal. The demodulator is coupled to the receiving circuit, serves to demodulate the second received signal and to generate a second demodulation signal. The estimation module is coupled to the demodulator and the predistortion module and generates the predistortion parameters based on a vector error measurement of a vector error between the first signal and the second demodulation signal. In addition, the predistortion parameters are sent to the predistortion module. The predistortion module adjusts the predistortion in the predistortion module on the basis of the predistortion parameters.

According to an embodiment of the invention, the estimation module updates the predistortion parameters over time based on the change in vector error between the first signal and the second demodulation signal. If the vector error change is a dip, the predistortion parameters are updated based on a first model. If the vector error change is not a dip, the predistortion parameters are updated based on a second model.

According to an embodiment of the invention, the predistortion module adjusts the response factor when predistorting in the predistortion module based on the predistortion parameters.

According to the invention, during the predistortion in a first section, the predistortion module generates the first predistortion signal, which has a linear relationship with the first modulation signal, on the basis of the first modulation signal. During predistortion, the predistortion module generates the first predistortion signal in a second section on the basis of the first modulation signal, which signal has a non-linear relationship with the first modulation signal.

According to one embodiment of the invention, the non-linear relationship is a polynomial relationship.

• 1 Darstellung des Aufbaus des herkömmlichen Vorverzerrungssystems
• 2 Darstellung des Aufbaus des adaptiven digitalen Vorverzerrungssystems gemäß einer Ausführungsform der Erfindung
• 3A ~ 3B Kennlinie der Vorverzerrung des adaptiven digitalen Vorverzerrungssystems gemäß einer Ausführungsform der Erfindung
• 4 Diagramm des Simulationsergebnisses des adaptiven digitalen Vorverzerrungssystems gemäß einer Ausführungsform der Erfindung
• 5A ~ 5B Diagramm des Simulationsergebnisses des adaptiven digitalen Vorverzerrungssystems gemäß einer Ausführungsform der Erfindung
• 6A ~ 6B Diagramm des Simulationsergebnisses des adaptiven digitalen Vorverzerrungssystems gemäß einer Ausführungsform der Erfindung

Hereby, the invention provides a structure of an adaptive digital predistortion system, in which the need for computational resources is reduced and the utilization efficiency of the adaptive predistortion can be improved. In the case of predistortion, the signal waveform distortion is not used as a guide as in the conventional structure, but instead the deviation value of the modulation vector is used as a guide to measure for adjustment. Thus, the need for computing resources is greatly reduced, which can simplify the hardware or software when used .

• 1 Diagram showing the structure of the conventional predistortion system
• 2 Representation of the structure of the adaptive digital predistortion system according to an embodiment of the invention
• 3A ~ 3B Predistortion characteristic of the adaptive digital predistortion system according to an embodiment of the invention
• 4 Diagram of the simulation result of the adaptive digital predistortion system according to an embodiment of the invention
• 5A ~ 5B Diagram of the simulation result of the adaptive digital predistortion system according to an embodiment of the invention
• 6A ~ 6B Diagram of the simulation result of the adaptive digital predistortion system according to an embodiment of the invention

In order that the objects, features and characteristics of the invention may be fully understood, the invention will be explained in detail by the following specific embodiments in conjunction with the accompanying drawings. Explanations follow below.

The invention provides a structure of an adaptive digital predistortion system in which the need for computational resources is reduced and the utilization efficiency of the adaptive predistortion can be improved. When predistorting, the signal waveform distortion becomes the same as that of the conventional structure is not used as a guide, but instead uses the deviation value of the modulation vector as a guide to measure for the adjustment. Thus, the need for computing resources is greatly reduced, which can simplify the hardware or the software in use. In addition, according to one embodiment, the predistortion can still be produced in two sections by means of the response model of the polynomials. Thus, the need for computing resources in the pre-distortion is further reduced. As a result, the communication system can be implemented cheaply and the entire communication efficiency can be increased.

In 2 shows the structure of the adaptive digital predistortion system according to an embodiment of the invention. This embodiment is a structure of the adaptive digital predistortion system based on the vector error for adjusting the predistortion. As in 2 As shown, the adaptive digital predistortion system 20 according to the embodiment includes the following elements. The emission module and the reception module each have a predistortion module 220 and an estimation module 260 for predistortion. The estimation module 260 determines the tuning method in predistorting the predistortion module 220 based on the vector error. This can be used to reduce the overall complexity of predistortion.

As in 2 1, the emitter module is for receiving a first signal S1 and includes the following elements: the modulator 210, the predistortion module 220, and the emitter circuit 230. The modulator 210 is for modulating the first signal S1 and generating a first modulation signal S2. The predistortion module 220 is coupled to the modulator 210, operates in predistortion based on the first modulation signal S2, and generates a first predistortion signal S3. The radiation circuit 230 is coupled to the predistortion module 220 and is used to receive the first predistortion signal S3 and to generate a first radiation signal S4. For example, the first emission signal S4 sends signals via at least one antenna unit. In addition, the first predistortion signal S3 generates a first radiated signal S4 in the radiating circuit 230 by means of at least one non-linear radio frequency element, such as a power amplifier. Therefore, the first radiated signal serves to compensate for the signal distortion of the radio frequency non-linear element such as the power amplifier.

As in 2 shown, the reception module serves to receive the first emission signal S4 and has the following elements: the reception circuit 240, the demodulator 250 and the estimation module 260. The reception circuit 240 serves to receive the first emission signal S4 and to generate a second reception signal S5. The demodulator 250 is coupled to the receiving circuit 240 and is used to demodulate the second received signal S5 and to generate a second demodulation signal S6. The estimation module 260 is coupled to the demodulator 250 and the predistortion module 220 and generates the predistortion parameters based on a vector error measurement of a vector error between the first signal S1 and the second demodulation signal S6. In addition, the estimation module 260 sends the predistortion parameters to the predistortion module 220 .

In addition, the receiving module for the adaptive digital predistortion system 20 serves as a monitoring receiver and for signal sampling of the first emission signal S4 for further processing. Therefore, in one example, the receiving circuit 240 in the receiving module can receive the first emission signal S4 by means of at least one antenna unit. In another example, the receiving circuit 240 can detect the emitted signal S4 by means of a radio frequency coupler (RF coupler) at the output of the emitting circuit 230, i.e. carry out a signal sampling. However, the embodiments of the invention are not limited to these examples.

With the structure of the adaptive digital predistortion system 20 according to FIG 2 the pre-distortion based on the vector error can be adjusted in various ways in some embodiments. For example, according to one embodiment, the estimation module 260 functions as follows. The estimation module updates the predistortion parameters over time based on the change in vector error between the first signal and the second demodulation signal. If the vector error change is a dip, the predistortion parameters are updated based on a first model. For example, the first model generates the updated predistortion parameters to finely modulate the predistortion response in predistortion module 220 . If the change in vector error is non-dip, the predistortion parameters are updated based on a second model. For example, the second model generates the updated predistortion parameters to finely counter-modulate the predistortion response in the predistortion module 220 . This allows the predistortion parameters to be updated by continuously repeating the update or limited in a range by a threshold and other conditions. This allows the predistortion to be adjusted based on the vector error.

In addition, the predistortion in the predistortion module 220 can be carried out with the structure of the adaptive digital predistortion system 20 according to FIG 2 may be further adjusted in various ways in some embodiments, and adaptive predistortion may be performed in conjunction with the estimation module 260 . To this end, an example is given in which the predistortion in the predistortion module 220 based on the piecewise function works as follows. The predistortion response of the predistortion module 220 is based on the polynomials defined in sections. Said predistortion parameters generated by the estimation module 260 are used to adjust the predistortion response.

For example, according to one embodiment, predistortion in the predistortion module 220 to compensate for the nonlinear properties of the nonlinear RF power elements, such as an RF power amplifier, works as follows. The predistortion signal is based on a function in two sections. The two sections may be referred to as a first section and a second section, corresponding to a linear segment and a non-linear segment, respectively. In this way, during the predistortion in a first section, the predistortion module 220 generates the first predistortion signal on the basis of the first modulation signal, which has a linear relationship with the first modulation signal. During predistortion in a second section, the predistortion module 220 generates the first predistortion signal based on the first modulation signal, which has a non-linear relationship with the first modulation signal. As a result, the predistortion module 220 generates the first predistortion signal based on a linear relationship when the first modulation signal is in the first section. The demand for computing resources is lower compared to the second section when the predistortion module 220 is operated in the first section. In a conventional predistorter, the predistortion signal is generated by a same polynomial, independent of the input signals. In contrast to this, the predistortion in the predistortion module 220 according to this embodiment does not only work exclusively by means of a section-wise defined function. In addition, the predistortion is adjusted by the estimation module 260 based on the vector error. Therefore, according to this embodiment, the total amount of computation resources required for the predistortion is further reduced, the communication system can be made cheap, and the overall communication efficiency can be increased.

An example is given where said predistortion module 220 and estimation module 260 can be implemented using hardware and/or software, such as a circuit module or the combination of a microprocessor, a digital signal processor, a microcontroller, an FPGA element (Field Programmable Gate Array) or an application specific integrated circuit (ASIC). Furthermore, the predistortion module 220 and the estimation module 260 can each be implemented as two separate circuit modules in any of a named example circuit module. The predistortion module 220 and the estimation module 260 can also be embodied together as one and the same circuit module, any of a named example circuit module. However, the embodiments of the invention are not limited to these examples. The predistortion module 220 and/or the estimation module 260 can also be implemented as a specific circuit module.

(durch eine Hardware- oder Software-Berechnung) das abzustrahlende Basisbandsignal ändern. Der Abstrahlschaltkreis 230 kann die folgenden Elemente umfassen: einen Digital-Analog-Wandler (auf Englisch DAC abgekürzt), einen RF-Aufwärtswandler (RF Up-Conversion auf Englisch) und einen RF-Leistungsverstärker. Das vom Vorverzerrungsmodul 220 erzeugte erste Vorverzerrungssignal S3 wird zuerst durch den Digital-Analog-Wandler (auf Englisch DAC abgekürzt) und den RF-Aufwärtswandler (RF Up-Converter auf Englisch) in ein Radiofrequenz-Signal umgewandelt, danach durch den Radiofrequenz-(nichtlinearen)-Verstärker verstärkt und anschließend von den Antennen abgestrahlt.Furthermore, the embodiment of the receiving module and the emitting module is explained using an example. According to an embodiment of the adaptive digital predistortion system, the predistortion module 220 in the emission module can be started by means of a start digital predistortion function ${ƒ}_{DPD}^{\left(0\right)}\left\{x\right\}$

change (by a hardware or software calculation) the baseband signal to be radiated. The radiating circuit 230 may include the following elements: a digital-to-analog converter (abbreviated DAC), an RF up-converter, and an RF power amplifier. The first predistortion signal S3 generated by the predistortion module 220 is first converted into a radio frequency signal by the digital-to-analog converter (abbreviated as DAC in English) and the RF up-converter (RF up-converter in English), then by the radio frequency (nonlinear ) amplifier and then radiated by the antennas.

von einem ersten Modell fein moduliert wird. Daraufhin vergleicht das Schätzmodul 260 kontinuierlich zur Kontrolle die Änderungen des Vektorfehlers. Wenn der Vektorfehler im Vergleich zum früheren Vektorfehler sinkt, wird die digitale Vorverzerrungsfunktion weiter vom ersten Modell fein moduliert Wenn der Vektorfehler im Gegenteil verglichen zum früheren Vektorfehler nicht sinkt, wird die digitale Vorverzerrungsfunktion von einem zweiten Modell (zum Beispiel entgegengesetzt zur Modulation durch das erste Modell) fein moduliert. Mit dieser Anpassungsart kann der Vektorfehler als Richtwert verwendet werden bzw. kann ein adaptiver digitaler Vorverzerrungseffekt erreicht werden.The first emission signal S4 emitted by the emission circuit 230 is converted into a digital baseband signal by the receiving circuit 240 serving as a monitoring receiver in the receiving module. For example, the receiving circuit 240 may include the following elements: a low-noise amplifier (LNA), an RF down-converter, and an ANA log-to-digital converter (abbreviated to ADC in English). However, the embodiments of the invention are not limited to these examples. The digital baseband signal generated by the receiving circuit 240 is demodulated by the demodulator 250 . The demodulated signal is then calculated using the estimation module 260 . In addition, the deviation value of the modulation vector between the reception signal code and the original transmission signal code is determined, so that the predistortion parameters arise, thereby the digital predistortion function ${ƒ}_{DPD}^{\left(0\right)}\left\{x\right\}$

is finely modulated by a first model. The estimation module 260 then continuously compares the changes in the vector error as a check. If the vector error decreases compared to the previous vector error, the digital predistortion function is further finely modulated by the first model If, on the contrary, the vector error does not decrease compared to the previous vector error, the digital predistortion function is finely modulated by a second model (e.g. opposite to the modulation by the first model ) finely modulated. With this type of adaptation, the vector error can be used as a guide value or an adaptive digital predistortion effect can be achieved.

For example, modulator 210 refers to a module that converts the signal code to be transmitted (such as quadrature amplitude modulation symbol, abbreviated as QAM) into a signal waveform, such as orthogonal frequency division multiplexing (OFDM abbreviated). It looks like this. $$\begin{array}{l}{s}_B\left[n\right]={\displaystyle \sum _k{S}_l\left[k\right]}\cdot {\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="DE102016124018B4\_0020.png,DE102016124018B4\_0021.png"\; state="\{\{state\}\}">e</figure-callout>}}^j,n=0\sim N-1\\ s_B\left[n\right]:{\text{signal waveform; S}}_l\left[k\right]:\text{QAM}-\text{abbreviated}\end{array}$$

The predistortion module 220 predistorts the signal waveform to cancel out the non-linear characteristics of the non-linear amplifier, thus achieving the near-linear characteristics of an ideal amplifier. Indeed, when digital pre-distortion is used in conjunction with a non-linear amplifier, they can be viewed as a whole as a linear amplifier. The formula looks like this. $$\begin{array}{l}DPD:{\widehat{s}}_B\left[n\right]=DPD\left\{s_B\left[n\right]\right\}={ƒ}_{DPD}\left\{\left|s_B\left[n\right]\right|\right\}\cdot e^{j\angle s_B\left[n\right]}\\ DAC+QM+NPA:{\widehat{s}}_B\left[n\right]\to {\widehat{s}}_B\left(t\right)={ƒ}_{DPD}\left\{\left|s_B\left(t\right)\right|\right\}\cdot e^{j\angle s_B\left(t\right)}\\ x_{Rf}\left(t\right)=Re\left\{{ƒ}_{NPA}\left\{\left|{\widehat{s}}_B\left(t\right)\right|\right\}\cdot e^{j\angle s_B\left(t\right)}\cdot {\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="DE102016124018B4\_0020.png,DE102016124018B4\_0021.png"\; state="\{\{state\}\}">e</figure-callout>}}^j\right\}\approx \text{re}\left\{s_B\left(t\right)\cdot {\mathrm{<figure-callout\; id="2"\; label="e\; j"\; filenames="DE102016124018B4\_0020.png,DE102016124018B4\_0021.png"\; state="\{\{state\}\}">e</figure-callout>}}^j\right\}\\ \because {ƒ}_{NPA}\left\{\left|{\widehat{s}}_B\left(t\right)\right|\right\}=\left\{{ƒ}_{NPA}\left\{\left|s_B\left(t\right)\right|\right\}\right\}\approx \left|s_B\left(t\right)\right|\\ {ƒ}_{DPD}\left\{x\right\}:\text{predistortion function};\\ {ƒ}_{NPA}\left\{x\right\}:\text{Properties of a nonlinear amplifier}\end{array}$$

$$\begin{array}{l}{\tilde{R}}_l\left[k\right]={\displaystyle \sum _n{\widehat{r}}_B}\left[n\right]\cdot e^{-j\frac{2\pi }{N}k\cdot n}\approx S_l\left[k\right]\cdot H_B\left[k\right]\\ {\tilde{R}}_l\left[k\right]={\tilde{R}}_l\left[k\right]/H_B\left[k\right]\approx S_l\left[k\right]\end{array}$$

Furthermore, the demodulator 250 refers to, for example, a module which detects from the reception signal waveform the signal code (QAM symbol) transmitted from a transmitter, such as orthogonal frequency division multiplexing (abbreviated OFDM), and functions as follows. $${\widehat{\mathrm{right}}}_B\left[n\right]={\widehat{s}}_B\left[n\right]\ast H_B\left[n\right],\text{* : convolution - convolution}$$

$$\begin{array}{l}{\tilde{R}}_l\left[k\right]={\displaystyle \sum _n{\widehat{\mathrm{right}}}_B}\left[n\right]\cdot e^{-\mathrm{<figure-callout\; id="2"\; label="j"\; filenames="DE102016124018B4\_0020.png,DE102016124018B4\_0021.png"\; state="\{\{state\}\}">j</figure-callout>}\frac{2\pi }{N}k\cdot n}\approx S_l\left[k\right]\cdot H_B\left[k\right]\\ {\tilde{R}}_l\left[k\right]={\tilde{R}}_l\left[k\right]/H_B\left[k\right]\approx S_l\left[k\right]\end{array}$$

In addition, the estimation module 260 can be used, for example, to calculate the error vector measurement of the modulation vector (abbreviated EVM). It is defined as follows. $$EVM=E\left\{{\left|{\widehat{R}}_l\left[k\right]-S_l\left[k\right]\right|}^2\right\},E\left\{\right\}:S\text{statistical mean}$$

For example, the digitized operation of the adaptive digital predistortion system can be explained using the following equation. $$\begin{array}{l}{\overrightarrow{s}}_B=MOD\left\{{\overrightarrow{S}}_l\right\}\\ {\overrightarrow{S}}_l={\left[S_l\left[k\right]|k=1\sim K\right]}^T{\text{, p}}_l\left[k\right]\in QAM\\ {\overrightarrow{\text{right}}}_B=NL\left\{DPD\left\{{\overrightarrow{s}}_B\right\}\right\}\ast {\overrightarrow{H}}_B\\ {\overrightarrow{\text{R}}}_l=\left[R_l\left[k\right]|k=1\sim K\right]=DEMOD\left\{{\overrightarrow{\mathrm{right}}}_B\right\}=DEMO{D}_{EQ}\left\{NL\left\{DPD\left\{{\overrightarrow{s}}_B\right\}\right\}\right\}\\ \therefore DP{D}_{EVM,Opt}=a\mathrm{right}G\underset{DPD}{\text{at least}}E\left\{{\left|{\overrightarrow{R}}_l-{\overrightarrow{S}}_l\right|}^2\right\}=\underset{DPD}{\text{at least}}E\left\{{\left|DEMO{D}_{EQ}\left\{NL\left\{DPD\left\{{\overrightarrow{s}}_B\right\}\right\}\right\}-{\overrightarrow{S}}_l\right|}^2\right\}\end{array}$$

DEMOD _EQ { } denotes a demodulation function with which the linear channel distortion was equalized. From formula (2) consider that the non-linear properties are a pure AM-AM non-linearity and DPD){ } is its corresponding linearization function when the radio frequency non-linear power amplifier is memoryless.

Said modulator or demodulator can be implemented in hardware and/or software, such as by means of a circuit module or a combination of a microprocessor, a digital signal processor, a microcontroller, a field programmable gate array (FPGA) element or an application specific integrated circuit Circuit (application specific integrated circuit in English, abbreviated ASIC).

In addition, the modules or the elements in said emitter module or in the receiver module pertaining to digital processing can each be embodied as separate circuit modules from any of said example digital circuit modules or together as one and the same circuit module from any of said example digital circuit modules. However, the embodiments of the invention are not limited to these examples. The modules or the elements can also be implemented as a specific circuit module.

According to one embodiment, the predistortion in the predistortion module 220 works based on the architecture of the adaptive digital predistortion system 20 in FIG 2 as follows. The predistortion signal is based on a function in two sections. The two sections can be referred to as a first section and a second section, which correspond to a linear segment and a non-linear segment, respectively. During predistortion in a first section, the predistortion module 220 generates the first predistortion signal on the basis of the first modulation signal and a linear relationship. During predistortion in a second section, the predistortion module 220 generates the first predistortion signal based on the first modulation signal and a polynomial ratio. In 3A and 3B 1 shows the predistortion characteristic of the adaptive digital predistortion system according to an embodiment of the invention. In 3A shows the AM-AM characteristics of the performance of a typical non-linear RF amplifier, where, outside of the saturation region, the characteristics of a linear segment with a low signal level and a non-linear segment near the saturation region can be distinguished. According to this embodiment, the predistortion works in the predistortion module 220 based on the structure of the adaptive digital predistortion system 20 in FIG 2 as follows. The predistortion signal is based on a function in two sections. The two sections can be referred to as a first section and a second section, which correspond to a linear segment and a non-linear segment, respectively. During predistortion in a first section, the predistortion module 220 generates the first predistortion signal, which has a linear relationship with the first modulation signal, on the basis of the first modulation signal. During predistortion in the second section, the predistortion module 220 generates the first predistortion signal based on the first modulation signal, which has a non-linear relationship with the first modulation signal. It is assumed that the boundary value at the input in the linear segment and the boundary value at the input and output in the saturation region are known parameters. If the characteristics of the non-linear power amplifier (Power Amplifier in English, abbreviated PA) have normalized (namely, the margin value at the input has been divided by the margin value at the input in the saturation region, as well as the gain at the input and output has normalized to 1), the equation sees for example like this. $$\begin{array}{l}y=NL\left\{x\right\}=\{\begin{array}{c}x,\text{x}\le {\text{I}}_L\\ G\left(x\right),{\text{I}}_L<x\le 1\end{array},{\text{y}}_{\text{Max}}=G\left(1\right)\\ x=DP{D}_{\stackrel{\text{right}}{a}}\left\{s\right\}=\{\begin{array}{c}{y}_{\text{Max}}\cdot s,s\le I_L\\ \mathrm{i.e}p{\mathrm{i.e}}_{\stackrel{\text{right}}{a}}\left(s\right),I_L<s\le 1\end{array},\stackrel{\text{right}}{a}=\left[a_p|p=2\sim P\right]\\ \mathrm{i.e}p{\mathrm{i.e}}_{\stackrel{\text{right}}{a}}\left(s\right)=y_{\text{Max}}\cdot I_L+y_{\text{Max}}\cdot \left(s-I_L\right)+{\displaystyle \sum _{p=2}^P{a}_p\cdot {\left(s-I_L\right)}^2}\text{\&}\mathrm{i.e}p{\mathrm{i.e}}_{\stackrel{\text{right}}{a}}\left(1\right)=\text{1}\\ \Rightarrow \underset{\stackrel{\text{right}}{a}}{\text{at least}}E\left\{{\left|DEMO{D}_{EQ}\left\{NL\left\{DP{D}_{\stackrel{\text{right}}{a}}\left\{{\stackrel{\text{right}}{s}}_B\right\}\right\}\right\}-{\stackrel{\text{right}}{S}}_l\right|}^2\right\}\end{array}$$

With 3A the characteristic curve of a setup of digital predistortion outside the saturation region is simply and clearly schematically represented The horizontal axis refers to the AM characteristics of the nonlinear amplifier at the input (abbreviated NPA_in in English), while the vertical axis refers to the AM characteristics of the nonlinear amplifier at the output (abbreviated to NPA_out in English). With 3B clarifies that the characteristics are normalized and further differentiated between a linear segment with a low signal level and a non-linear segment that is near the saturation region. The characteristic 301 represents the properties of the non-linear power amplifier. The characteristic 302 represents the properties of the non-linear power amplifier with the ideal predistortion structure. The characteristic 303 represents the properties of the ideal linear power amplifier. The amplification in the linear segment of the characteristic 302 is suppressed.

When the non-linear characteristics of the non-linear RF power amplifier are close to the memoryless AM-AM non-linear model, the complexity of the matrix-vector calculation can be reduced with the new method according to this embodiment, which uses the lowering of the vector error as a benchmark. Moreover, the parameters for the predistortion are respectively optimized by means of the response model of the polynomials with the two-section predistortion structure according to the embodiment as shown in formula (3). For example, the parameters can be repeatedly searched using the on-line method to find the optimized parameters for predistortion.

$EV{M}_i=E\left\{{\left|DEMO{D}_{EQ}\left\{NL\left\{DP{D}_{{\overrightarrow{a}}^{\left(i\right)}}\left\{{\overrightarrow{s}}_l\right\}\right\}\right\}-{\overrightarrow{S}}_l\right|}^2\right\}$

(S10) i = i+1 $EV{M}_i=E\left\{{\left|DEMO{D}_{EQ}\left\{NL\left\{DP{D}_{{\overrightarrow{a}}^{\left(i\right)}}\left\{{\overrightarrow{s}}_l\right\}\right\}\right\}-{\overrightarrow{S}}_l\right|}^2\right\}$

(S30) Wenn EVM _i < EVM _i-1 dann ${\overrightarrow{a}}^{\left(i\right)}={\overrightarrow{a}}^{\left(i-1\right)}+\Delta {\overrightarrow{a}}^{\left(i\right)},$

gehe zu (S10) (S40) Wenn EVM _i >= EVM _i-1 dann ${\overrightarrow{a}}^{\left(i\right)}={\overrightarrow{a}}^{\left(i-1\right)}-\Delta {\overrightarrow{a}}^{\left(i\right)},$

EVM _i = EVM _i-1 - (EVM _i - EVM _i-1), gehe zu (S10) This method is explained, for example, using the pseudo code and the equations in Table 1. Table 1 Start: i = 0, ${\overrightarrow{a}}^{\left(i\right)}=\left[a_p^{\left(0\right)}|p=2\sim P\right],$

$EV{M}_i=E\left\{{\left|DEMO{D}_{EQ}\left\{NL\left\{DP{D}_{{\overrightarrow{a}}^{\left(i\right)}}\left\{{\overrightarrow{s}}_l\right\}\right\}\right\}-{\overrightarrow{S}}_l\right|}^2\right\}$

(S10) i = i +1 $EV{M}_i=E\left\{{\left|DEMO{D}_{EQ}\left\{NL\left\{DP{D}_{{\overrightarrow{a}}^{\left(i\right)}}\left\{{\overrightarrow{s}}_l\right\}\right\}\right\}-{\overrightarrow{S}}_l\right|}^2\right\}$

(S30) If EVM _i < EVM _{i -1} then ${\overrightarrow{a}}^{\left(i\right)}={\overrightarrow{a}}^{\left(i-1\right)}+\Delta {\overrightarrow{a}}^{\left(i\right)},$

go to (S10) (S40) If EVM _i >= EVM _{i -1} then ${\overrightarrow{a}}^{\left(i\right)}={\overrightarrow{a}}^{\left(i-1\right)}-\Delta {\overrightarrow{a}}^{\left(i\right)},$

EVM _i = EVM _{i -1} - ( EVM _i - EVM _{i -1} ), go to (S10)

welches a_p und p = 2~P umfasst) bei der Vorverzerrungsreaktion im Vorverzerrungsmodul 220 fein zu modulieren. Wenn es sich bei der Änderung des Vektorfehlers um eine Nichtsenkung handelt, werden die Vorverzerrungsparameter (wie im Ablauf S40 gezeigt) auf Basis eines zweiten Modells aktualisiert. Beispielsweise erzeugt das zweite Modell die aktualisierten Vorverzerrungsparameter, um die Koeffizienten der Polynome bei der Vorverzerrungsreaktion im Vorverzerrungsmodul 220 (wie im Ablauf S30 gezeigt) fein gegenzumodulieren. Hiermit kann die Aktualisierung der Vorverzerrungsparameter durch eine sich kontinuierlich wiederholende Aktualisierung der Vorverzerrungsparameter oder mittels des Schwellenwerts und anderer Bedingungen in einem Bereich begrenzt erfolgen. Dadurch kann die Vorverzerrung auf Basis des Vektorfehlers angepasst werden. Darüber hinaus sind die Ausführungsformen der Erfindung nicht auf das genannte Beispiel beschränkt. Die Verfahren des Vorverzerrens oder der Koeffizienten der Polynome bei der Vorverzerrungsreaktion können natürlich gemäß diesem Beispiel oder anderen Verfahren bestimmt bzw. angepasst werden. According to the embodiment based on the pseudocodes in said Table 1, the estimation module 260 works as follows. Over time, the deviations of the modulation vector between the first signal and the second demodulation signal (as shown in the start flow and the flow S20) are calculated. In addition, the predistortion parameters (as shown in process S30 and S40) are updated based on the change in the deviations of the modulation vector. If the change in vector error is a dip, the predistortion parameters are updated (as shown in flow S30) based on a first model. For example, the first model generates the updated predistortion parameters to reflect the coefficients of the polynomials (such as $\overrightarrow{a}$

which a _p and p = 2˜P) in the predistortion response in the predistortion module 220 to be finely modulated. If the change in vector error is non-decrease, the predistortion parameters are updated (as shown in flow S40) based on a second model. For example, the second model generates the updated predistortion parameters to finely countermodulate the coefficients of the polynomials in the predistortion response in the predistortion module 220 (as shown in flow S30). With this, the update of the predistortion parameters can be limited by a continuously repeated update of the predistortion parameters or by means of the threshold value and other conditions in a range. This allows the predistortion to be adjusted based on the vector error. Furthermore, the embodiments of the invention are not limited to the example mentioned. The methods of predistortion or the coefficients of the polynomials in the predistortion response can of course be determined according to this example or other methods.

Since the rate of change of the non-linear characteristics of the RF non-linear power amplifier is not high (compared to the EVM measurement frequency of the communication signals), the change of the non-linear characteristics can be tracked adaptively by the simple, repetitive parameter search to detect the effect of the adaptive digital predistortion in the Get EVM under relative optimization.

When the non-linear characteristics of the RF non-linear power amplifier are close to the memoryless non-linear AM-AM model and no models and parameters estimate the deviations, the non-linearity of the RF power amplifier can also be corrected by this invention. The reason is explained with the following equations. $$\begin{array}{l}DP{D}_{SD,Opt}=\text{bad}\underset{DPD}{\text{at least}}E\left\{{\left|DPD\left\{{\stackrel{\text{right}}{\mathrm{right}}}_B\right\}-{\stackrel{\text{right}}{\widehat{s}}}_B\right|}^2\right\}\Rightarrow DP{D}_{SD,Opt}=N{L}_{cH}^{-1}\\ \therefore {\stackrel{\text{right}}{\mathrm{right}}}_B=N{L}_{cH}\left\{DP{D}_{SD,Opt}\left\{{\stackrel{\text{right}}{s}}_B\right\}\right\}=N{L}_{cH}\left\{N{L}_{cH}^{-1}\left\{{\stackrel{\text{right}}{s}}_B\right\}\right\}={\stackrel{\text{right}}{s}}_B\\ DP{D}_{EVM,Opt}=\text{bad}\underset{DPD}{\text{at least}}E\left\{\left|{\stackrel{\text{v}}{R}}_l-{\stackrel{\text{right}}{S}}_l\right|\right\}=\underset{DPD}{\text{at least}}E\left\{{\left|DEMO{D}_{EQ}\left\{NL\left\{DPD\left\{{\stackrel{\text{right}}{s}}_B\right\}\right\}\right\}-{\stackrel{\text{right}}{S}}_l\right|}^2\right\}\\ \Rightarrow DP{D}_{EVM,Opt}=N{L}^{-1}\\ \therefore {\stackrel{\text{v}}{R}}_l=DEMO{D}_{EQ}\left\{NL\left\{DP{D}_{EVM,Opt}\left\{{\stackrel{\text{right}}{s}}_B\right\}\right\}\right\}=DEMO{D}_{EQ}\left\{{\stackrel{\text{right}}{s}}_B\right\}={\stackrel{\text{right}}{S}}_l\end{array}$$

In this way, the advantages of the adaptive digital predistortion structure according to the present invention, in which the vector error is used as a guide, lie in the following points. For demodulation signals, only the vector error needs to be calculated and processed to adjust the digital predistortion without using the relatively complicated non-linear model and calculating the corresponding matrix vector to extract the received signal model parameters. Thereby it brings the advantage of lowering the overall complexity of the digital predistortion and the method can be easily applied to various radio communication systems. The invention can apply to various communication systems such as mobile communication, cable communication, satellite communication and radio terminal of digital television communication. However, the invention is not limited to these systems.

sind die benötigten, zu suchenden Parameter eins üblich. Bei dieser Ausführungsform wird der einzige Suchparameter mit a₂ angezeigt. Darüber hinaus ist Δa₂ = 0.01. In 4 wird die Kennlinie von a2 und EVM gezeigt. Dabei werden die spezifischen OFDM-Signale verwendet und das Verhältnis zwischen dem Suchparameter a₂ und EVM(dB) unter OBO (Output Back-Off) = 7dB wird berechnet. Darunter wird EVM als $E\left\{{\left|\overrightarrow{R}-\overrightarrow{S}\right|}^2\right\}/E\left\{{\left|\overrightarrow{S}\right|}^2\right\}$

definiertIn 4 a diagram of the simulation result of the adaptive digital predistortion system according to an embodiment of the invention is shown. In this embodiment according to 3B the sections of the predistortion polynomials in the non-linear domain are assumed to be P=3, subject to the boundary condition $\mathrm{i.e}p{\mathrm{i.e}}_{\overrightarrow{a}}\left(1\right)=1$

the required parameters to look for are one common. In this embodiment, the single search parameter is indicated as a ₂ . In addition, Δa ₂ = 0.01. In 4 the characteristic curve of a2 and EVM is shown. The specific OFDM signals are used and the ratio between the search parameter a ₂ and EVM(dB) under OBO (Output Back-Off) = 7dB is calculated. Below that, EVM is called $E\left\{{\left|\overrightarrow{R}-\overrightarrow{S}\right|}^2\right\}/E\left\{{\left|\overrightarrow{S}\right|}^2\right\}$

Are defined

In the 5A - 5B shows the diagrams of the simulation result of the adaptive digital predistortion system according to an embodiment of the invention. In 5A the polynomial of the digital predistortion function is shown in two sections with the mentioned optimized search parameter. In 5B shows the linearization result when using the predistortion function in conjunction with PA (like the characteristic 503) accordingly. In 5B the characteristic curve 503 is the corresponding linearization result after applying the digital predistortion (DPD) according to the embodiment. This characteristic curve results from precise calculation or simulation software. For practical comparison, in 5B further added the characteristic 501 and the characteristic 502. Characteristic 501 represents the properties of the original PA, with characteristic 502 representing the corresponding linearization result of the optimal DPD as in 5B As shown, the result represented by the characteristic 503 according to the embodiment almost corresponds to the optimal result of the characteristic 502. This means that an effective predistortion effect can be obtained based on the embodiment of the invention.

In the 6A - 6B shows the diagrams of the simulation result of the adaptive digital predistortion system according to an embodiment of the invention. In 6A and 6B the diagrams of the probability distribution of the EVM to which the randomly generated 10000 OFDM symbol signals correspond by means of the on-line method of adaptive digital predistortion are presented. As in 6A As shown, the characteristic 601 represents the result of the original PA, and the characteristic 602 and the characteristic 603 represent the result of the optimal DPD and the result of the DPD according to the embodiment of the invention, respectively, when OBO is 6 dB. As in 6B As shown, the characteristic 611 represents the result of the original PA, and the characteristic 612 and the characteristic 613 represent the result of the optimal DPD and the result of the DPD according to the embodiment of the invention, respectively, when OBO is 7 dB. As in 6A and 6B shown, the characteristic curve 603 (or the characteristic curve 613) is almost identical to the characteristic curve 602 (or the characteristic curve 612). This means that an efficient predistortion effect can be achieved based on the embodiment of the invention.

With 5A , 5B , 6A and 6B it is proven that the effectiveness of the adaptive predistortion of the invention is almost identical to the effect of the optimal predistortion. In comparison, the optimal predistortion can only correct the characteristics in the non-linear region, and the effect in the saturation region can still delimit the radiation quality of the exemplary OFDM signals, because the OFDM signals have a property that the peak-to-average power ratio (peak-to-average power) -Average Power Ratio in English, abbreviated PAPR) is relatively large. In comparison to the prior art, in the cited embodiment of the invention the vector error for measurement is used as a benchmark together with the simple polynomial responses in two sections, thereby providing a corresponding algorithm for adaptive predistortion. The prior art requires a complicated non-linear model and a corresponding matrix-vector calculation, whereas the complexity according to the embodiment method of the invention is lower. For example, according to the embodiment of the invention, the signals must be demodulated. As with the OFDM signals, the signals must be processed during demodulation using, for example, the Fast Fourier Transform (FFT) algorithm. The complexity is directly proportional to N*log ₂ N (N means the length of normal OFDM symbols). Therefore, it is advantageous to use software and hardware according to the embodiment of the invention.

As mentioned, the invention provides a structure of an adaptive digital predistortion system, in which the need for computational resources is reduced and the utilization efficiency of the adaptive predistortion can be improved. In predistorting, unlike the conventional structure, signal waveform distortion is not used as a guide, but instead the deviation value of the modulation vector is used as a guide to measure for matching. This significantly reduces the need for computing resources, which means that the hardware or software can be simplified when used. In addition, according to one embodiment, the predistortion can still be produced in two sections by means of the response model of the polynomials. Thus, the need for computing resources in the pre-distortion is further reduced. As a result, the communication system can be implemented cheaply and the entire communication efficiency can be increased.

As previously mentioned, the invention is illustrated by the preferred embodiments. Those skilled in the art should take note that the invention is not limited to the embodiments mentioned, but can be varied in many ways within the scope of the disclosure. In this context, all new individual and combination features disclosed in the description and/or drawing are considered essential to the invention. Only the following claims are valid for the scope of protection of the present invention.

Reference List

10

Traditional predistortion system

20

Adaptive digital predistortion system

110

modulator

120

predistortion

130

emission circuit

140

receiving circuit

150

extraction module

210

modulator

220

predistortion module

230

emission circuit

240

receiving circuit

250

demodulator

260

estimation module

S1-S6

signal

301-303

curve

501-503

curve

601-603

curve

S10-S40

process

Claims (4)

Adaptive digital predistortion system, comprising the following elements: • a radiation module, which serves to receive a first signal and has the following elements: ◯ a modulator, which serves to modulate the first signal and to generate a first modulation signal, ◯ a predistortion module, which is coupled to the modulator, in which the predistortion works on the basis of the first modulation signal and generates a first predistortion signal, and ◯ a radiation circuit, which is coupled to the predistortion module and is used to generate a first radiation signal based on the first predistortion signal, the first radiation signal being signals transmits via at least one antenna unit, and • a receiving module, which serves to receive the first emission signal and has the following elements: ◯ a receiving circuit, which firstly receives the first emission signal and then generates a second reception signals, ◯ a demodulator, which is coupled to the receiving circuit, is used to demodulate the second received signal and to generate a second demodulation signal, and ◯ an estimation module, which is coupled to the demodulator and the predistortion module and the predistortion parameters on the basis of a vector error measurement of a vector error generated between the first signal and the second demodulation signal, the predistortion parameters being sent from the estimation module to the predistortion module, the predistortion module adjusting the predistortion in the predistortion module based on the predistortion parameters, further • the predistortion module during predistortion in a first section based on the first modulation signal generates the first predistortion signal, which has a linear relationship with the first modulation signal, wherein • the predistortion module during predistortion in a second section on the basis of the first modula tion signal generates the first predistortion signal, which has a non-linear relationship with the first modulation signal. Adaptive digital predistortion system after claim 1 , characterized in that the estimation module updates the predistortion parameters over time based on the change in the vector error between the first signal and the second demodulation signal, the predistortion parameters being updated based on a first model when the change in the vector error is a When the vector error change is not a dip, the predistortion parameters are updated based on a second model. Adaptive digital predistortion system after claim 2 , characterized in that the predistortion module adjusts the response factor during predistortion in the predistortion module on the basis of the predistortion parameters. Adaptive digital predistortion system after claim 1 , characterized in that the non-linear relationship is a polynomial relationship.

Images