CN102939716B - Multi-band wideband power amplifier digital predistortion system and method - Google Patents

Abstract

Disclose a kind of for multi-frequency band broadband communication system applications, the high-performance with high linearity and high efficiency RF digital hybrid mode power amplifier system and cost effective method.Present disclosure can make the on-the-spot restructural of power amplifier system and in the bandwidth of non-constant width, support the multiple operating frequency band in same PA system.In addition, the present invention supports many modulation schemes (modulation is unknown), multicarrier and multichannel.

Description

Multi-band wideband power amplifier digital predistortion system and method

Related application

This application claims the rights and interests of following application:

On December 21st, 2009 that submit, application number be 61/288,838, denomination of invention is " MULTI-BANDWIDEBANDPOWERAMPLIFIERDIGITALPREDISTORTIONSYST EMANDMETHOD " and invention people is the U.S. Patent application of Wan-JongKim, Kyoung-JoonCho and ShawnPatrickStapleton.

By reference this application is incorporated in herein for all objects.

Technical field

Present invention relates in general to the wireless communication system using complex modulation technique.More specifically, the present invention relates to the power amplifier system for radio communication.

Background technology

Use the wide-band mobile communication system of the complex modulation technique of such as Wideband Code Division Multiple Access (WCDMA) (WCDMA) and OFDM (OFDM) to have large papr (PAPR) specification, thus need the High Linear power amplifier transmitted for its radio frequency (RF).Conventional digital pre-distortion (DPD) technology has bandwidth of operation restriction.

The conventional DPD scheme based on DSP uses FPGA, DSP or microprocessor to estimate, calculates and the nonlinearity of masking amplifier (PA): they are followed the tracks of fast the signal in PA system and regulate.But the DPD scheme based on DSP of change to routine outputing signal caused by asymmetric distortion, that power amplifier is on wide bandwidth linearity performance due to the environmental change of such as temperature and the PA that caused by memory effect proposes challenge.Conventional DPD algorithm is based on broadband feedback signal, and these algorithms need high-speed AD converter (ADC) to catch the information of needs.Multi-frequency band application (or referred to as multiband application) can make its operating frequency significantly spaced apart.Conventional DPD framework uses the ADC sample rate larger than the twice of the nonlinear distortion bandwidth of input signal.This sample rate is usually larger than the twice of 5 times of factors of the bandwidth of operation of multiplex modulated signal.5 times of factors cause the spectral re-growth owing to the nonlinear distortion produced by power amplifier.The feasibility that the conventional predistortion architecture of this constrained to sample rate is applied single band.Sample rate ADC is higher, then resolution lower, consume more power and more expensive.

Summary of the invention

Therefore, consider that above problem completes the present invention, the object of this invention is to provide a kind of for multi-frequency band broadband communication system applications, the high-performance with high linearity and high efficiency power amplifier system and cost effective method.Present disclosure can make the on-the-spot restructural of power amplifier system and in the bandwidth of non-constant width, support the many operating frequency band in same PA system.In addition, the present invention supports many modulation schemes (modulation unknown (agnostic)), multicarrier and multichannel.

In order to realize above object, according to the present invention, this technology is based on the method for adaptive digital pre-distortion making RF power amplifier linearization.The present invention is based on the unlike signal (multi-band signal) using different frequency.The nonlinear distortion that these multi-band signals will also be produced by the distortion of power amplifier experience centered by each carrier wave of 5 times of each bandwidth being approximately multi-band signal.From power amplifier feedback signal by under convert intermediate frequency (IF) to, intermediate frequency (IF) guarantee to sample in the adc after first-harmonic bandwidth can not aliasing mutually.The present invention can adapt to the aliasing of the nonlinear distortion of each carrier wave.

Disclose various execution mode of the present invention.In one embodiment, in PA system, employ the combination of peak factor reduction (CFR), DPD, power efficiency lift technique and coefficient adaptive algorithm.In another embodiment, in order to strengthen the property, also use analogue quadrature moducator (AQM) collocation structure.

Some embodiments of the present invention can be monitored the fluctuation of power amplifier characteristics and can be carried out self-regulation by adaptive algorithm.Disclosed a kind of such adaptive algorithm is referred to as self adaptation DPD algorithm at present, and it realizes in the digital domain and is being incorporated in by reference herein and is giving teaching in the application of enclosing as annex.

Application of the present invention is suitable for using together with other wireless communication system of satellite communication with all wireless base stations, access point, Mobile Equipment wireless terminal, portable radio machine and such as microwave.

Accompanying drawing explanation

Other object of the present invention and advantage can be understood more fully according to the following detailed description of carrying out by reference to the accompanying drawings, in the accompanying drawings:

Fig. 1 is the block diagram of the citation form that digital pre-distorting power amplifier system is shown.

Fig. 2 illustrates according to an embodiment of the invention, block diagram for the simple digital predistortion block diagram of power amplifier system.

Fig. 3 illustrates the block diagram based on polynomial predistorter of the present invention.

Fig. 4 is the block diagram of the direct learning algorithm of adaptive digital pre-distortion be applied in digital pre-distorting power amplifier system of the present invention.

Fig. 5 is the block diagram of the adaptive digital pre-distortion indirect learning algorithm be applied in digital pre-distorting power amplifier system of the present invention.

Fig. 6 is the diagram of the frequency-region signal existed in aliasing digital pre-distortion system.

Fig. 7 is the execution mode of quadrature modulator compensation block framework.

Nomenclature

ACLR adjacent channel leakage ratio

ACPR adjacent channel power ratio

ADC analog to digital converter

AQDM analog quadrature demodulator

AQM analogue quadrature moducator

AQDMC analog quadrature demodulation adjuster

AQMC analog quadrature modulation adjuster

BPF band pass filter

CDMA code division multiple access

CFR peak factor reduces

DAC digital to analog converter

DET detector

DHMPA digital hybrid mode power amplifier

DDC digital down converter

DNC low-converter

DPADoherty power amplifier

DPD digital pre-distortion

DQDM digital orthogonal demodulator

DQM digital quadrature modulators

DSP Digital Signal Processing

DUC digital up converter

EER envelope is eliminated and is recovered

EF envelope following

ET envelope-tracking

EVM Error Vector Magnitude

FFLPA feedforward linear power amplifier

FIR finite impulse response (FIR)

FPGA field programmable gate array

GSM global system for mobile communications

I-Q In-Phase-Quadrature

IF intermediate frequency

LlNC uses the Linear Amplifer of non-linear components

LO local oscillator

LPF low pass filter

MCPA Multi-Carrier Power Amplifier

The multi-direction search of MDS

OFDM OFDM

PA power amplifier

PAPR papr

PD predistortion

PLL phase-locked loop

QAM quadrature amplitude modulation

QPSK Quadrature Phase Shift Keying

RF radio frequency

SAW surface acoustic wave filter

UMTS universal mobile telecommunications system

UPC upconverter

WCDMA Wideband Code Division Multiple Access (WCDMA)

WLAN WLAN (wireless local area network)

Embodiment

The present invention is the new multiband pre-distortion system using adaptive digital pre-distortion algorithm.The present invention is the hybrid system of digital module and analog module.The digital module of hybrid system and the interaction of analog module make spectral re-growth linearisation and strengthen the power efficiency of PA, keep simultaneously or increase wide bandwidth.Therefore, present invention achieves the higher efficiency of broadband complex modulated carrier wave simultaneously worked on multiple different frequency band and the linearity of Geng Gao.

Fig. 1 is the high-level block diagram that basic system architecture is shown, at least for some execution modes, can think that this basic system architecture comprises digital module and analog module and feedback path.Digital module is the digital pre-distortion controller 101 comprising DPD algorithm, other auxiliary DSP algorithm and associated digital electronics.As discussed in addition in merged by reference application, analog module is other secondary analog circuit of main power amplifier 102, such as DPA and the related peripheral analog circuit of whole system.The present invention is " flight data recorder ", plug and play type system, this is because its signal 100 accepting RF modulation is as its input, and provides substantially identical but the RF signal 103 amplified exports as it, therefore, its for RF enter/RF goes out.According to an embodiment of the invention, baseband modulation signal directly can be applied to digital predistorter controller.According to an embodiment of the invention, base-band input signal directly can be put on digital predistorter controller.According to an embodiment of the invention, optical input signals directly can be put on digital predistorter controller.Feedback path provides in fact the expression of output signal to predistortion controller 101.

Under arbitrary input pattern, compensate the memory effect owing to producing from the frequency dependence of heating, biasing networks and active equipment by the adaptive algorithm 204 in the DPD201 of Fig. 2.In fig. 2, multi-band RF is inputted x [n] and be supplied to DPD201 and DPD algorithm logic 204.The output z [n] of DPD201 is supplied to DAC202 and logic 204.The output of DAC202 provides input to power amplifier 203.Sensed the distorted characteristic of PA by feedback samples y (t), feedback samples y (t) is converted into data ya [n] and is provided to alignment logic 205 in ADC206.In the registered, data ya [n] provides feedback data to algorithm logic 204.

Synchronous with reference to multi-band signal x [n] (input signal) by making the broadband capture aliasing from feedback path export multi-band signal ya [n] (sampling feedback aliasing signal), regulate the coefficient of DPD.DPD algorithm performs synchronous and compensates.This synchronously makes reference signal aim at the aliasing feedback signal in alignment block.In an execution mode of DPD algorithm, the sampling feedback aliasing signal ya [n] of Reference Signal and aliasing is used in and directly learns in adaptive algorithm.In another execution mode of DPD algorithm, aliasing pre-distorted signals za [n] (predistortion output aliasing signal) and sampling feedback aliasing signal ya [n] is used in indirect learning adaptive algorithm.

In some embodiments, utilize the adaptive algorithm in a digital processing unit carry out DPD before apply peak factor reduction (CFR), to reduce PAPR, EVM and ACPR and compensate for memory effects and the change of the linearity that causes due to the variations in temperature of PA.Digital processing unit can take almost arbitrary form; Conveniently, usually use FPGA to realize, but general processor is also acceptable in many embodiments.The CFR realized in the digital module of execution mode eliminates (scalediterativepulsecancellation) based on the proportional iterative pulse proposed in following patent application: this patent application be on March 31st, 2008 that submit, denomination of invention be the US61/041 of " AnEfficientPeakCancellationMethodForReducingThePeak-To-A veragePowerRatioInWidebandCommunicationSystems ", 164, it is incorporated in herein by reference.Comprise CFR to strengthen the property, and thus CFR is optional.Can from execution mode, delete CFR and not affect overall function.

In all embodiments, the memory effect owing to producing from the frequency dependence of heating, biasing networks and active equipment is compensated by the adaptive algorithm in DPD.By the coefficient making broadband capture output signal and reference signal from feedback path synchronously regulate DPD.Digital pre-distortion algorithm performs synchronous and compensates.Pre-distorted signals to generate real signal, is then converted into IF analog signal via DAC by DQM.In all embodiments, do not need or do not need to implement DQM in FPGA.If do not use DQM in FPGA, then two DAC can be utilized to implement AQM and to realize to generate real signal and empty signal respectively.

Fig. 3 is the block diagram of predistortion (PD) part illustrated in DPD system of the present invention.PD in the present invention generally uses based on polynomial adaptive digital pre-distortion system.Another execution mode of PD uses the digital pre-distortion system based on look-up table (LUT).More specifically, PD shown in Fig. 3 is processed by following adaptive algorithm in digital processing unit: this adaptive algorithm be application number be 11/961,969, denomination of invention is propose in the U.S. Patent application of " AMethodforBasebandPredistortionLinearizationinMulti-Chan nelWidebandCommunicationSystems ".The PD of the PD system in Fig. 3 has multiple finite impulse response (FIR) (FIR) filter, that is, FIR1301, FIR2303, FIR3305 and FIR4307.PD also comprises three factorials and generates block 302, five factorial generation block 304 and seven factorials generation block 306.Output signal from FIR filter is merged in sum block 308.The coefficient of multiple FIR filter is upgraded by digital pre-distortion algorithm.

Digital predistorter algorithm

Digital pre-distortion (DPD) is the technology for linear power amplifier (PA).Fig. 2 illustrates the block diagram of digital pre-distortion PA system.In DPD block, use memory polynomial model as predistortion function (Fig. 3).

$z\left(n\right)=\underset{i=0}{\overset{n-1}{\mathrm{\Σ}}}{x}_i(n-i)\left(\underset{j=0}{\overset{k-1}{\mathrm{\Σ}}}{a}_{\mathrm{ij}}{\left|x_i(n-i)\right|}^j\right)$

Wherein, a _ijfor DPD coefficient.

In DPD estimator block, least square method is used to ask DPD coefficient a _ij, then by DPD coefficient a _ijbe sent to DPD block.Fig. 4 and Fig. 5 illustrates detailed DPD algorithm.Estimate to use QRRLS adaptive algorithm to obtain these coefficients in block at DPD.

Fig. 4 illustrates an execution mode of multiband digital predistorter.Direct study adaptive algorithm has two inputs and enters DPD estimator.DPD estimator uses aliasing sampling multi-band signal xa [n] (sampling input aliasing signal) as a reference and uses sampling feedback aliasing signal ya [n] as input.Therefore, x (n) signal is provided to DPD400 and frequency translation/aliasing logic 420.DPD outputs signal z (n), aliasing logic 420 exports xa (n) and xa (n) is supplied to integer delay logic 402 simultaneously, then integer delay logic 402 provides signal to fractional delay logic 403 and multiplexer 414, and the output of receive logic 403 gone back by multiplexer 414.Multiplexer also receives the control signal from delay estimator 406 and provides output to block xa ' 404, and block 404 determines xa ' (n-m).The output of multiplexer is also provided to data buffer 405, and data buffer 405 is exported to be supplied to DPD estimator 412 and control signal to be supplied to phase shift block 410 and gain calibration block 411.Feedback signal y (t) and sample frequency F _sbe provided to ADC421, wherein, F _sbe selected to suitable Nyquist (Nyquist) district producing aliasing signal.ADC exports and is provided to feedback data buffer 408, and ya (n) data from block 407 are also provided to feedback data buffer 408.Data buffer 408 provides data to delay estimator 406, also provides this data to phase shifter 410 to gain calibration block 411.Gain calibration is provided to DPD estimator, and this gain calibration is multiplexed and returns DPD400 together with the pre-distortion coefficients in internal memory shown in 401.

Fig. 5 illustrates another execution mode of multiband digital predistorter.The indirect learning adaptive algorithm of Fig. 5 has two inputs and enters DPD estimator.DPD estimator use predistortion export aliasing signal za [n] as a reference and use sampling feedback aliasing signal ya [n] as input, and other side and Fig. 4 closely similar, be not thus described in addition.

Fig. 6 illustrates the diagram of spectrum domain figure.Reference-input signal x [n] illustrates that centre frequency is F _aand F _btwo different frequency bands.The bandwidth of operation of each carrier wave is very little compared with the frequency interval between carrier wave.Nonlinearity in power amplifier will cause as the spectral re-growth shown in analog feedback multi-band signal y (t) (analog feedback signal).Analog feedback signal by under convert intermediate frequency F to _i, intermediate frequency F _ibe selected to and be in the first Nyquist district as shown in fig. 6.Frequency F _iselective dependency in use ADC sample rate or F _s, frequency interval (F between the bandwidth of operation of carrier wave and carrier wave _a-F _b).ADC sample rate F _sbe generally F _irestriction factor during selection.For two carrier systems, at least some execution mode, F _abe positioned at the first Nyquist district and F _bbe positioned at the second Nyquist district, but depend on concrete realization, signal F _aor F _bthe the 3rd, the 4th or n-th Nyquist district can be positioned at.Basic demand is that these two signals are positioned at different Nyquist districts.F _iselectional restriction F _aand F _bhow far can be separated from each other.

Image as shown in the frequency spectrum of sampling feedback aliasing signal ya [n] is generated to the sampling of analog feedback signal y (t).The aliased portion of signal allows the mutual aliasing of nonlinear distortion from each carrier wave, as long as can not adversely affect original multi-band signal.Direct learning algorithm uses the difference between xa [n] and ya [n] that obtained error signal is minimized.In DPD estimator, QRRLS algorithm uses this error to regulate predistorter coefficient.Indirect learning algorithm first uses predistortion to export aliasing signal za [n] and sampling feedback aliasing signal ya [n] comes power amplifier modeling.Then, the power amplifier coefficient of modeling is used to calculate predistorter coefficient.

Fig. 7 is the block diagram that analogue quadrature moducator collocation structure is shown.The baseband signal exported from DAC is transformed into RF frequency by analogue quadrature moducator.Input signal is divided into in-phase component X _iwith quadrature component X _q.Analogue quadrature moducator collocation structure comprises four real filters { g11, g12, g21, g22} and two DC bias compensation parameter c1, c2.DC in AQM is biased will be compensated by parameter c1, c2.AQM frequency dependence will { g11, g12, g21, g22} compensate by filter.The exponent number of real filter depends on required compensation level.Output signal Y _iand Y _qto be to the homophase port of AQM and orthogonal port.

In a word, multi-band wideband power amplifier pre-distortion system of the present invention significantly can reduce feedback ADC sample rate requirement.This can realize multiband broadband application and reduce power consumption and cost.This system is also restructural and field-programmable, this is because can regulate algorithm and power efficiency Enhanced feature at any time as software in digital processing unit, this point has been done to discuss in more detail in application that is merged by reference and that enclose as annex.

In addition, multiband broadband DPD system does not need the modulation scheme of QPSK, QAM, the OFDM etc. known such as in CDMA, GSM, WCDMA, CDMA2000 and Wireless LAN system.This means that DPD system can support many modulation schemes, multicarrier and multichannel.Other benefit of DPD system comprises: the correction to PA nonlinearity in the repeater without the necessary base band signal information easily obtained or indoor covering system.

Although describe the present invention with reference to preferred implementation, be to be understood that the details that the invention is not restricted to described in preferred implementation.In description before, suggested various substitutions and modifications, and those of ordinary skill in the art can expect other substitutions and modifications.Therefore, all such substitutions and modifications are intended to be included in the scope of the present invention as limited in the following claims.

Claims (20)

1. a digital pre-distortion system, comprising:

Radio-frequency input signals;

Power amplifier, exports for providing the amplification comprising distorted characteristic;

Amplify from described the feedback signal exporting and derive, described feedback signal comprises the expression at least partially of described distorted characteristic; And

Described feedback signal and described input signal are made to the predistortion logic of response, described predistortion logic represents for the aliasing generating described input signal and described aliasing based on described input signal represents derivation pre-distortion coefficients, and the sample rate that the described aliasing of wherein said input signal represents is less than the twice of the maximum bandwidth of described input signal.

2. digital pre-distortion system according to claim 1, also comprises the digital predistorter receiving described radio-frequency input signals, the described pre-distortion coefficients that described digital predistorter impact is derived by described predistortion logic.

3. digital pre-distortion system according to claim 2, wherein, described digital predistorter utilize adaptive, based on polynomial digital pre-distortion system, describedly utilize the described pre-distortion coefficients derived by described predistortion logic based on polynomial digital pre-distortion system.

4. digital pre-distortion system according to claim 2, wherein, described digital predistorter utilize have finite impulse response filter, based on the digital pre-distortion system of look-up table, the coefficient of wherein said finite impulse response filter comprises the described pre-distortion coefficients derived by described predistortion logic.

5. digital pre-distortion system according to claim 2, also comprises:

Digital to analog converter, it is coupled to described digital predistorter and described power amplifier; And

Analog to digital converter, described feedback signal is converted to digital form from analog form by it.

6. digital pre-distortion system according to claim 1, also comprises alignment logic, and described feedback signal is aimed at relative to described radio-frequency input signals by described alignment logic.

7. digital pre-distortion system according to claim 1, the frequency that the described aliasing of wherein said input signal represents is in Nyquist district.

8. digital pre-distortion system according to claim 1, wherein, described pre-distortion coefficients uses directly study adaptive algorithm to derive.

9. digital pre-distortion system according to claim 1, wherein, described pre-distortion coefficients uses indirect learning adaptive algorithm to derive.

10. digital pre-distortion system according to claim 1, wherein, described pre-distortion coefficients is exported as the memory effect for compensating due to one or more generation in the frequency dependence of the active equipment in heating, biasing networks and described digital pre-distortion system.

11. 1 kinds of methods for amplifying signal, described method comprises:

Received RF input signal;

The amplification detected from power amplifier exports, and described output is made response to described radio-frequency input signals and comprised distorted characteristic;

Export from described amplification and derive feedback signal, described feedback signal comprises the expression at least partially of described distorted characteristic; And

The aliasing generating described input signal via the predistortion logic described feedback signal and described input signal being made to response represents, and represent derivation pre-distortion coefficients based on the described aliasing of described input signal, wherein, the sample rate that the described aliasing of described input signal represents is less than the twice of the maximum bandwidth of described input signal.

12. methods according to claim 11, also comprise:

Receive the second radio-frequency input signals; And

Utilize digital predistorter, use the described pre-distortion coefficients derived by described predistortion logic to carry out predistortion to described second radio-frequency input signals.

13. methods according to claim 12, wherein, to described second radio-frequency input signals carry out predistortion comprise utilize adaptive, based on polynomial digital pre-distortion system, describedly utilize the described pre-distortion coefficients derived by described predistortion logic based on polynomial digital pre-distortion system.

14. methods according to claim 12, wherein, to described second radio-frequency input signals carry out predistortion comprise utilize have finite impulse response filter, based on the digital pre-distortion system of look-up table, the coefficient of wherein said finite impulse response filter comprises the described pre-distortion coefficients derived by described predistortion logic.

15. methods according to claim 11, also comprise:

Described feedback signal is converted to digital form from analog form; And

The second radio frequency input through predistortion is converted to analog form from digital form.

16. methods according to claim 11, also comprise and described feedback signal being aimed at relative to described radio-frequency input signals.

17. methods according to claim 11, wherein, the frequency that the described aliasing of described input signal represents is in Nyquist district.

18. methods according to claim 11, wherein, described pre-distortion coefficients uses directly study adaptive algorithm to derive.

19. methods according to claim 11, wherein, described pre-distortion coefficients uses indirect learning adaptive algorithm to derive.

20. methods according to claim 11, wherein, described pre-distortion coefficients is exported as the memory effect for compensating due to one or more generation in the frequency dependence of the active equipment in heating, biasing networks and digital pre-distortion system.