数字预失真关键技术（三） .ppt

第二章 数字预失真技术理论,第一节 线性化技术概述,第二节 数字预失真基本原理,第二节 数字预失真线性化系统,第三节 数字预失真器及其参数辨识,第四节 数字预失真器参数辨识算法,第五节 峰均值比及削峰技术简介,第六节 数字预失真器的ADS仿真,线性度和功率效率,2,功率回退,3,高功率射频功放,High Power RF PAs(10W)use multiple driver stages to amplify aninput signal.Different PA architectures(Class A,AB,C,etc)offer variousdegrees of linearity,cost and efficiency.,RF PAs are notoriously inefficient Air is a convenient but poor transmission medium.,RF PAs are designed(tuned)for specific frequency range andbandwidth,MCPA=wideband RF PA,does not have to process multiple carriers,PA Gain is usually fixed so pre-amps may be required to drive the PAinput.Antenna,PA,TX Board,If Gain=30dB,RFout=50dBm(100W),RFin=20dBm 800MHz,DAC,IF-RF,DUC,FromBaseband,50 OhmTypical,APre-Amp,Input3 to 4 gain stages typical4,线性化、峰值因子减小,Linearization techniques allow a PA to be operated at higher powerwith minimal IMD increases,thus greater efficiencyRecent technological advances have made digital pre-distortion thefocus of research effortsCrest factor reduction(CFR)further increases the efficiency of the PAby reducing the peak-to-average ratio(PAR)of the transmitted signalTheoretical Performance of Class AB PA,5,线性化基础,The trade-off betweenefficiency and linearity isthe primary concern for PAdesignersA PA operating at a highpercentage of its powerrating requires externallinearization to maintainlinearityThe linearization of the PAreduces back-off,thusincreasing efficiency,6,线性化技术总览,7,线性化的优势,Technical Comparison of BFLPA vs.LPA,8,线性化的优势With linearizationWithout linearization,Network Planning Tradeoffs,9,预失真、前馈、后馈三种线性化技术比较,Feedforward,Feedback,RF Predistortion/Digital Predistortion,FrequencybandwidthLinearityComplexityPower EfficiencyAdaptation,WideUltraHighLowIntrinsicallynon adaptiveSSPAs forBase Stations,NarrowGoodMediumHighIntrinsicallyadaptivePAsfor MobileStations,Ultra/MediumMedium/UltraMedium/MediumHigh/HighIntrinsically nonadaptiveTWTAs&SSPAs forSatcom applications,10,线性化技术比较,CorrectionCapability*25-35 dB10-20 dB5-10 dB,CorrectionBandwidth100 MHz25MHz,Relative CostHighMedLow,50MHz,Med,LinearizationTechniquesFeedforwardEnvelopeFeedbackAnalog Pre-,Distortion,Adaptive Pre-,10-20 dB,Distortion*IMD Correction based on 8-Tone Continuous Random Phase11,第二章 数字预失真技术理论,第一节 线性化技术概述,第二节 数字预失真基本原理,第二节 数字预失真线性化系统,第三节 数字预失真器及其参数辨识,第四节 数字预失真器参数辨识算法,第五节 峰均值比及削峰技术简介,第六节 数字预失真器的ADS仿真,12,数字预失真原理示意图,Predistorter,RFout,RFin,RFout,RFin,|G|G,|G|G,|G|G,PinPin,PinPin,PinPin13,数字预失真技术数学推导,Pre-distortion effectively performs a mathematical inversion of theVolterra PA modelThe output of the pre-distortion processor is described by the followingThe PA is linearized when,14,数字预失真技术数学推导,Digital pre-distortion(DPD)has become an effective linearizationtechnique due to the renewed possibilities offered by DSPAdaptive PD designs use feedback to compensate for PA variationsLook-up tables are updated to achieve optimal pre-distortion bycomparingg PD input to PA outputThe PD function is expressed as a complex polynomialwhere,15,数字预失真频谱,Digital pre-distortion(DPD)requires feedback for sample-by-sampleadaptation 5 times that of the signal bandwidthMulti-carrier systems use signal bandwidths of up to 20MHz today,thusthe feedback bandwidth must be 100MHz to compensate 3rd and 5thorder IMDLeast-mean-square(LMS)is a popular gradient based optimizationalgorithm that requires wideband feedback16,16,数字预失真频谱,17,第二章 数字预失真技术理论,第一节 线性化技术概述,第二节 数字预失真基本原理,第二节 数字预失真线性化系统,第三节 数字预失真器及其参数辨识,第四节 数字预失真器参数辨识算法,第五节 峰均值比及削峰技术简介,第六节 数字预失真器的ADS仿真,Predistorter,D/AConverter,QuadratureModulator,PowerAmplifier,自适应数字预失真系统框图Digital Analog,Iin,Ipd,Qpd,QinLOPredistortion,Estimator,A/D Converter,QuadratureDemodulator,Iout,Qout19,自适应数字预失真关键事项ADCs sampling speed must be at least 5 times the signal bandwidth andhave 12 bits resolution.Modulation can be performed RF or base band using FPGAprocessor.Analogue implementation of the MOD/DEMOD introduces amplitude and,phase unbalances,digital MOD/DEMOD is preferable the cost of,increasing the processing power/time needed.,Limited ACPR improvement due to PA memory effects,compensate for,memory effects at base band processing.The PA monitoring/modelling loop must be independent from modulationloop.For high volume production,a tailor made ASIC can be a cost effectivesolution for the DPD Technology.20,数字预失真的优缺点,21,数字预失真系统,22,审视数字预失真,23,数字预失真IMD交调修正,24,数字预失真的附带修正功能,25,数字预失真的附带修正功能,26,实验验证系统,27,实验验证系统框图,28,实验验证系统,X,PA,Down-Converter,Hardware Implementation of Wideband Pre-DistortionWaveform,Generator,Attenuator,20dB,Agilent4432B,LO,AnalogRF,DUT,Pre-DistortedInput Signal,Tektronics TDS224Oscilloscope,PC,AnalogIF,29,第二章 数字预失真技术理论,第一节 线性化技术概述,第二节 数字预失真基本原理,第二节 数字预失真线性化系统,第三节 数字预失真器及其参数辨识,第四节 数字预失真器参数辨识算法,第五节 峰均值比及削峰技术简介,第六节 数字预失真器的ADS仿真,30,Transmitter,zTX(n),yTX(n),离线训练原理Hammerstein Predistorter(Copy of A),uTX(n),xTX(n),MemorylessLUT,LinearFilter,1,x(n),Memoryless,LUT,u(n),Linear,y(n),Glinear,Filter,Hammerstein Predistorter Training(A)31,功放的自适应记忆预失真,The term memory effects refer to thebandwidth-dependant nonlineareffects often present in PAs.These encompass envelope memoryeffects and frequency responsememory effects.Envelope memory effects areprimarily a result of thermalhysteresis and electrical propertiesinherent to PAs.Frequency memory effects are due tothe variations in the frequencyspacing of the transmitted signal andare characterized by shorter timeconstants.,32,记忆多项式数字预失真器,Memory Polynomial Pre-Distortion Implementation,Where(K=7)And(D=2)33,自适应记忆多项式数字预失真器,Simulated Performance of Wideband Pre-DistortionThis traditional approach uses and LMS algorithm to adapt the PDcoefficients on a sample-by-sample basis.The memory PA model has D=1(delay)and K=5(order).,34,记忆多项式数字预失真器输出频谱,Simulated Performance of Wideband Pre-Distortion,DPD=0:the LMS algorithm indicates an ACPL improvement of-3dB and anACPH improvement of 3dB.DPD=1:the LMS algorithm indicates an ACPL improvement of-15dB and anACPH improvement of-11dB.35,35,记忆多项式数字预失真器输出频谱,Simulated Performance of Wideband Pre-Distortion,DPD=2:the LMS algorithm indicates an ACPL improvement of-24dB and anACPH improvement of-23dB.DPD=3:the LMS algorithm indicates an ACPL improvement of-24dB and anACPH improvement of-20dB.36,36,Hammerstein 预失真器,u(n),u(n),x(n),y(n),AM/AM andAM/PM LUT,Complexmultiplier,u(n),Memory effect,LinearFilter,GiGq,subsystemMemoryless static nonlinear subsystemx(n)=Gi(u(n)+jGq(u(n)u(n)=G(u(n)u(n)M 1j=037,预失真器的AM/AM 和AM/PM 曲线,（三载波WCDMA信号）,38,滤波器的动态AM/AM和AM/PM 特性,39,PSD(dB/Hz),PSD(dB/Hz),A,ACPR(dBc),-10.0,Hammerstein预失真器频谱比较20.0-15.010.0-20.000.00,(b),-25.0-30.0,(a),(b),(c),-40.0-50.0,-20.0-30.0,(a),(c),(e),(d),-40.0,-35.0,(d),1.95,1.97,1.965,1.96,1.955,Frequency(GHz),-45.01.9625,1.963,1.9635,(e)1.964,1.9645,Frequency(GHz),-35,(a),(a)Without predistorter.,-45,(b),(c),(d),(b)With memoryless predistorter.(c)Hammerstein predistorter with a 10-,tap FIR filter.,(d)Hammerstein predistorter with a 64-,(e),-55-65,-15,-10,-5 5,10,15,Frequency Offset(MHz)40,tap FIR filter.(e)Hammerstein predistorter with a 128-tap FIR filter.,改进型Hammerstein 预失真器FIR1,u(n),x(n),y(n),u(n),FIR2,x(n)x(n),AM/AM andAM/PM LUT,x,+,Memoryless static nonlinear subsystem,Memory effect subsystem,jM1 1 M 2 1y(n)=ai x(n i)+ii=0 i=041,ACPR(dBc),PS,SD(dB/Hz),D(dB/Hz)PS,改进型Hammerstein预失真器频谱比较,-40-45,-20.0,20.010.00.0,-55,(a)(b),(c),-50-10.0(d),-40.0,-30.0,(a),(b),(c),(d),-65,-60,1.955,1.96,1.965,1.97,-50.01.95,Frequency(MHz)-15.0(a),-15,-10,10,15,-5 5Frequency Offset(MHz),-20.0-25.0,(b),(a)Without predistorter.(b)With memoryless predistorter.,(c),(d),-30.0-35.0-40.0-45.0,1.9625,1.963,1.9635,1.964,1.9645,Frequency(MHz)42,(c)Hammerstein predistorter with a 128-tap FIR filter.(d)Augmented Hammerstein predistorterwith two twenty-tap FIR filter.,(dB/Hz)PSD,(,(dBc)ACPR,(,改进型Hammerstein预失真器频谱比较,-30,(a),-60.0,-40.0-50.0,-35-40,(c),(b)(d),(a)(b)(c),-70.0-80.0-90.0-100.0,1.99,2,1.92,1.93,1.94,1.95,1.96,1.97,1.98,Frequency(GHz),(d),-45-50-55,-15,-10,-5 5,10,15,Frequency Offset(MHz),(a)Without predistorter.(b)With memoryless predistorter.(c)Hammerstein predistorter with a 128-tap FIR filter.(d)Augmented Hammerstein predistorterwith two twenty-tap FIR filter.43,PSD(dBm),-50.00,改进型Hammerstein预失真器频谱比较Class B TR-40.00Augmented,HammersteinpredistorterMemoryless,without DPD,Class C TR,-60.00-70.00-80.00-90.00-100.00,1.94E+09,1.94E+09,1.95E+09,1.95E+09,1.96E+09,1.96E+09,1.97E+09,Frequency(Hz)Typical Linearization Results of aDoherty Power Amplifier with a two-carrier 3GPP WCDMA SignalExcitation44,第二章 数字预失真技术理论,第一节 线性化技术概述,第二节 数字预失真基本原理,第二节 数字预失真线性化系统,第三节 数字预失真器及其参数辨识,第四节 数字预失真器参数辨识算法,第五节 峰均值比及削峰技术简介,第六节 数字预失真器的ADS仿真,数字预失真参数辨识算法,LMS,RLS,QRD-RLS,SVD,46,第二章 数字预失真技术理论,第一节 线性化技术概述,第二节 数字预失真基本原理,第二节 数字预失真线性化系统,第三节 数字预失真器及其参数辨识,第四节 数字预失真器参数辨识算法,第五节 峰均值比及削峰技术简介,第六节 数字预失真器的ADS仿真,OFDM信号的峰均值,48,减小峰值因子,The DPD optimal performancedepends greatly on signalcharacteristicsMulti-carrier signals can have aPAR as high as 13dB increasingthe level of back-off to maintainacceptable IMD levelsThe application of CFR allows thePA to operate at higherinput/output power levels whilemaintaining linearity at the outputof the PAAchieved through pulsegeneration and digital clipping,49,减小峰值因子,50,减小峰值因子,51,减小峰值因子,Preferred PA bias point for a typical modulated signal,52,52,减小峰值因子,Preferred PA bias point for a CFR signal,53,减小峰值因子Typical Peak Detection and Cancellation through Pulse InjectionCancellation Signal,Input Signal,Output Signal,-+,54,法CFR,削峰法CFR原理,55,CFR算法汇总比较,56,ACPR改善,ACPR improvement with respect to output power,57,功率和效率改善,The ACPR measurements were recorded according to specifications with a30kHz marker at and offset of 885kHzResults were limited by the performance limitations of the test bedPower and efficiency improvement,58,第二章 数字预失真技术理论,第一节 线性化技术概述,第二节 数字预失真基本原理,第二节 数字预失真线性化系统,第三节 数字预失真器及其参数辨识,第四节 数字预失真器参数辨识算法,第五节 峰均值比及削峰技术简介,第六节 数字预失真器的ADS仿真,复增益LUT,60,ADS数字预失真电路,61,数字预失真查找表,62,优化的预失真查找表相位,63,优化的预失真查找表增益,64,预失真的多音仿真结果,65,谢谢大家！,