high performance digital pre-distortion for wideband systems€¦ · connect the mxg/awg directly...

Welcome

Jake SandersonApplication Engineer Modular Product Operation

Daren McClearnonProduct Planning ManagerElectronic System-Level EDA

Wideband DPD webcastSept 1, 2011

Copyright 2011 Agilent Technologies1

Agenda

About Digital Pre-Distortion (DPD)

DPD challenges for 4G/wideband systems

Wideband modeling approaches

Hardware modeling demonstration

Beyond Hardware

Q&A



Conflicting requirements

Problem Statement

3

How to handle signals with high Crest Factor, while driving the PA to operate with high PAE, while also having low signal distortion?

High DC-RF Efficiency

HighCrest Factor

HighSpectral

Efficiency

IncreaseDrive levels

Causes high distortion

levels

“Back off” the drive

levels


Copyright 2011 Agilent Technologies

Solution Approach

4

High DC-RF Efficiency High

Crest Factor

HighSpectral

Efficiency

IncreaseDrive levels

Causes high distortion

levels

Higher Throughput

rates for subscribers

CFR

DPD

CFRDPD



5

Psat

Pin

LINEAR GAIN

INPUT POWER

OUTPUTPOWER

Pdesired

Pactual

Pin neededto achieve Pdesired

PA, WITH GAINCOMPRESSION

Digital Pre-distortion principles – compressing PA



6

Digital Pre-distortion principles – pre-expansion

Maximum correctablepower

Psat

LINEAR GAIN

INPUT POWER

OUTPUTPOWER

LINEAR REGION

DPDREGION


DPD GAINEXPANSION

+



7

Maximum correctablepower

Psat

INPUT POWER

OUTPUTPOWER

LINEAR REGION

DPDREGION

LINEARIZEDDPD + PA

Digital Pre-distortion principles – linearized result


DPD GAINEXPANSION

+ =



Linear Operation with time-varying envelope

8

Psat

LINEAR GAIN

INPUT POWER

OUTPUTPOWER

Peak-to-AvgPower Ratio COMPLEX ENVELOPE

time

PeakAverage



Nonlinear Operation – peaks are compressed

9

Psat

LINEAR GAIN

INPUT POWER

OUTPUTPOWER

(compressed peaks)

CCDF (LTE)



+19 dBm output, for a Handset PALTE signal, PA operating at ~1.5 dB overall gain compression

10



DPD Pre-Expansion – peaks are exaggerated

11

Psat

LINEAR GAIN

INPUT POWER

OUTPUTPOWER

Further Improvements:• Compensate for artificially higher avg. signal power • Condition signal w/Crest Factor Reduction (CFR)

(expanded peaks)



DPD Net Result: Linear gain of complex-valued RF carrier envelope over a specific range of power levels

12

LINEAR

INPUT POWER

OUTPUTPOWER

DPD pre-expanded peaks

INPUT POWER

PA compresses peaksLINEAR

Baseband Digital Pre-Distortion

RF Power Amplification



AM-AM Effects (Change in Gain vs. Power level)

13

Pre-Distorter AM-to-AM

Power Amp AM-to-AM

DPD + PAAM-to-AM

Linearized PA

Definitions• AM-AM : Change in Gain vs. Power level, compared to small-signal (dB(S21))• AM-PM : Change in Transmission Phase, compared to small-signal (phase(S21))• CCDF: Percentage of time a particular amplitude level spends above avg power



Additional issues: Memory EffectsOutput is dependent on previous history; “path dependence”

14

PA with memoryAM-to-AM

PA without memoryAM-to-AM

Output waveform has an instantaneous 1:1 correspondence in time with input waveform

Output waveform depends on previous values



What does a DPD look like? (Volterra Model)

15

∑=

=K

kk nznz

1)()( ∑ ∑ ∏

= = =

−=Q

m

Q

m

k

llkkk

k

mnymmhnz0 0 1

11

)(),,()(

∑∑∑= ==

+−−+−+=Q

m

Q

m

Q

mmnymnymmhmnymhhnz

0 021212

01110

1 21

)()(),()()()(

Volterra series pre-distorter can be described by

where

Which is a 2-dimensional summation of power series & past time envelope responses

A full Volterra produces a huge computational load. People usually simplify it into• Wiener model • Hammerstein model • Wiener-Hammerstein model• Memory polynomial model



DPD principles – Memory Polynomial Model

If only diagonal terms are kept, Volterra reduces to “Memory polynomial” model.

Agilent uses an indirect learning algorithm to extract MP coefficients.

As of SystemVue 2011.10, you can now add your own model, extraction algorithm, and even create your own GUI.

16

L. Ding, G. T. Zhou, D. R. Morgan, Z. Ma, J. S. Kenney, J. Kim, and C. R. Giardina, “Memory polynomial predistorter based on the indirect learning architecture,” in Proc. of GLOBECOM, Taipei, Taiwan, 2002, vol. 1, pp. 967–971.

∑∑= =

−−−=K

k

Q

q

kkq qnyqnyanz

1 0

1)()()(

Where• K is Nonlinearity order • Q is Memory length



MORE INFO: “4G for Everyone” webcast

Agenda





Beyond Hardware

Q&A

17




18

• LTE-Advanced (100MHz) and 802.11ac (160MHz) are physically 5x-8x wider than previous generation

• Oversampling increases this bandwidth an additional 3x-5x

• Drives wider ADC/DAC, data rates, test equipment, & more

• Requires powerful embedded processors : DSP/FPGA/ASIC

Wider Bandwidth

Higher Crest Factor

Rapidly changing environment



Oversampling increases the Measurement BW

19

1x

3x-5x

oversampling

oversampling

Measures: In-band EVM, Throughput

Measures:Out-of-band Spectral Masks

Most Effective DPD Region




20

• Carrier aggregation increases PAPR (drives Efficiency down)

• Highly-configurable signals (time-varying RBs) can lead to worst-case RF scenarios

• People apply Crest Factor Reduction differently….….how to estimate the effect of CFR on your PA if someone else is doing the DSP?

Wider Bandwidth

Higher Crest Factor




CCDF

The Effect of Carrier Aggregation on PAPR

21

CarrierAgg.

Scenario

LinkType Configuration PAPR of single CC,

before aggregationPAPR with CCs, after aggregation

Scenario 1 FDD DL 4x20 MHz CCs 8.45 dB 9.98 dB

Scenario 2 TDD DL 5x20 MHz CCs 9.17 dB 11.71 dB

Scenario 4 FDD DL 2x20+2x20MHz 8.38 dB 9.58 dB

FDD UL 20 + 20MHz 5.79 dB 6.86 dB

CA Scenario 1 FDD DL

CA Scenario 2 TDD DL

CA Scenario 4 FDD DL CA Scenario 4 FDD UL




22

• LTE-Advanced, 802.11ac, and other Standards still changing

• IP issues: interoperability of signals, algorithms, channels, coded performance

• Closed DPD IP (no control)• Availability of commercial DPD

solutions • Ecosystem & vendor

re-alignments • BB/RF hardware platform

neutrality for local spectral variations, vendors, standards

Wider Bandwidth

Higher Crest Factor




Agenda





Beyond Hardware

Q&A

23



Wideband modeling platform

24

N5182A MXG, orE8267D PSG(as external modulator)

W1461 SystemVueW1716 DPD89600B VSA

M9330A AWG

M9392A PXI VSA

Device Under Test

Baseband I,Q



How is this wideband modeling platform used?

25

WIDEBAND MODELING PLATFORM

WIDEBANDRF SOURCE

WIDEBANDRF ANALYZER

.m, C++, VHDL

Test Vectors

Standards reference

DPD, CFR

Early R&DScenarios

ArchitecturesAlgorithms

System-level ValidationVendor QualificationComponent Design

DeploymentScenarios

COMMS PHY …..BASEBAND

IF/RFUPCONVERSIONDPD

COMMS PHYBASEBAND

IF/RFUPCONVERSION

FPGA / DSPREALIZATION

Wireless SoC DPD



Wideband modeling – software

Flexible modeling environment, integrates .m, C++, VHDL, along with instrument drivers, simulators, scripting

DPD modeling & extraction algorithms

Instrument control

Wireless Standards libraries, for test vector generation and system-level tests

Connection to VSA software, RF EDA design flows, MATLAB, and other tools

26

W1461 SystemVueW1716 DPD89600B VSA



SystemVue for unified architecture, verification

PHY system integrationand verification

Cross-domain PHY modeling framework, for Model-Based Design

Complete a working PHY using combinations of Software, RF/BB Hardware, Simulation, and Measurements

Baseband AlgorithmsDataflow Simulation

RF Sys ArchitectureRF Simulators

Agilent SystemVue

PHY IP

TESTRF Hardware FlowsRFIC / MMIC

Hardware SiP / BoardHardware

Baseband Hardware Flows

GPP/ARMSoftware

DSP/ASSPSoftware

FPGA/ASIC/SoCHardware



Wideband platform – PXI modular instruments

Stimulus– User-defined (or locally generated) test vectors– Wideband, calibrated AWG– RF/MW signal generator (modulate & upconvert)– Driver pre-amp (optional)

Response– RF/MW downconverter, attenuator, signal

conditioning– Wideband baseband digitizer

Integration pieces– DC bias source, cables, connectors, PA output

attenuator, etc.

28

M9330A AWG

M9392A PXI VSA

N5182A MXG, orE8267D PSG(as external modulator)



Agenda





Beyond Hardware

Q&A

29



Wideband platform – TODAY’S DEMONSTRATION

30

Jake SandersonApplication Engineer,Agilent Modular Products



Wideband platform – TODAY’S DEMONSTRATIONSystemVue with modular PXI instruments (bandwidth ~250MHz)

~10 dB improvement

in spectral leakage

(1st iteration)



1. Instrument setup to capture PA input signal

2. Instrument setup to capture PA output signal

Wideband platform – TODAY’S DEMONSTRATIONSystemVue with modular PXI instruments (bandwidth ~250MHz)

External Trigger

M9330A AWG

MXG as external modulator M9392A PXI VSA

M9392A PXI VSA

External Trigger

Attenuator

M9330A AWG

MXG as external modulator



SystemVue DPD Hardware Flow for LTE/LTE-A

The download power and length of the waveform can also be set.

33

Step 1. Create DPD stimulus waveform• LTE parameters such as bandwidth, Resource Block allocation and others

can be set.• Switch LTE or LTE-Advanced waveform



Connect the MXG/AWG directly to the PXA/M9392A and click the “Capture Waveform” button. The captured signal is the input of the PA.

Connect the MXG to the PA, connect the PA to the PXA/M9392A, and click the “Capture Waveform” button. The captured signal is the output of the PA DUT.

The measured I/Q files are stored and used in following steps.

34

Step 2. Capture PA response• SystemVue interfaces directly to the MXG or M9330A AWG (source) and PXA

or M9392A(analyzer).• Instrument parameters such as number of signal, trace assignment and file

name can be set.




DPD AM-to-AM Characteristic

35


Step 3. DPD Model Extraction• DPD model parameters such as number

of training samples, memory order and nonlinear order can be set.

PA AM-to-AM Characteristic



36


Step 3. DPD Model Extraction (Custom IP)• Provide UI to allow customers to export their own

Matlab code of DPD algorithm (IP) into MathLang to verify their DPD algorithm performance.

Custom DPD Model Extraction

Custom Digital Pre-distorter



37

SystemVue DPD Hardware Flow for LTE/LTE-AStep 4. Capture DPD+PA Response• The signal is predistorted by the DPD model

and downloaded into the MXG or M9330A AWG.DPD+PA output (blue) and original raw signal (red) is displayed

Set the RF powerDPD+PA AM-to-AM Characteristic




Step 5. Verify DPD+PA response• LTE performance for the DPD model used with the PA hardware is verified.

Spectrum, EVM andACLR are calculatedand plotted automatically

38



DPD of LTE-Advanced, using M9330A/M9392A 4x20MHz contiguous CA, (80MHz signal BW)

Source = M9330A AWG

Vector Analyzer= M9392A- 12bits ADC- up to 250MHz bandwidth

PA output Spectrum (Blue)PA+DPD Spectrum (Red)PA input Spectrum (Green)

39



DPD of LTE-Advanced, using M9330A/M9392A 2x20MHz + 20MHz non-contiguous CCs, (80MHz signal BW)

40



Source = M9330A AWG



DPD of 802.11ac, using M9330A/M9392A(80MHz Option)

41



Source = M9330A AWG



Agenda





Beyond Hardware

Q&A

42



Simulation vs. Measurement DPD Extraction

External Trigger

Attenuator N5182 MXG

or E8257D PSGas external modulatorM9330A AWG if > 100 MHz

89600VSA

M9392A PXI VSA (>140MHz)or N9030A PXA (<140 MHz)

I,Q RF

RF DUT

SIMULATION-BASED DPD(predictive)

• ADS & GoldenGate Circuits as simulated RF DUTs- Complex loading, memory FX, dynamic behaviors

• NVNA X-parameter measurement model,- Great for smaller solid-state devices

X-parameters

RF DUTN5241,2 PNA-X

MEASUREMENT-BASED DPD

CO-SIM, MODELS

CO-SIM, MODELS

MODEL

ADS

GG



Simulation-based, predictive DPD

ADS circuit-level PA

(circuit envelope simulation)

ADS Ptolemy(circuit-system co-

simulation)

SystemVueSTIMULUS

SystemVueRESPONSE

CO-SIM

• Full, nonlinear RF• Time-varying behaviors

& memory effects• Envelope Tracking• Complex loadings• No H/W limitations

CO-SIM

Applications: Early assessment, early validation, cross-domain troubleshooting



Conclusion

New bandwidth and linearity requirements are driving 4G designers to spec DPD earlier in their system designs

The velocity of the industry is pushing DPD activity in-house, where designers are taking a more active role

The integration of an open Comms EDA environment with versatile wideband instruments enables

– Flexibility for modeling, realization, validation, and troubleshooting– Higher performance– Integration with the baseband DSP you are already doing

45



Further information

About Wideband DPD – Watch a demo:

http://www.youtube.com/watch?v=bocF6P74T9E – Read an app note:

http://cp.literature.agilent.com/litweb/pdf/5990-8883EN.pdf

About Agilent Products – http://www.agilent.com/find/eesof-systemvue-dpd-builder– http://www.agilent.com/find/modular

46



Agenda





Looking ahead

Q&A

47



48

Q&A



1. Lei Ding, Zhou G.T., Morgan D.R., Zhengxiang Ma, Kenney J.S., Jaehyeong Kim, Giardina C.R., “A robust digital baseband predistorter constructed using memory polynomials”, Communications, IEEE Transactions on, Jan. 2004, Volume: 52, Issue:1, page 159-165.

2. Lei Ding, “Digital Predistortionof Power Amplifiers for Wireless Applications”, PhD Thesis, March 2004.3. Roland Sperlich, “Adaptive Power Amplifier Linearization by Digital Pre-Distortion with Narrowband

Feedback using Genetic Algorithms”, PhD Thesis, 2005.4. Helaoui, M. Boumaiza, S. Ghazel, A. Ghannouchi, F.M., “Power and efficiency enhancement of 3G

multicarrier amplifiers using digital signal processing with experimental validation”, Microwave Theory and Techniques, IEEE Transactions on, June 2006, Volume: 54, Issue: 4, Part 1, page 1396-1404.

5. H. A.Suraweera, K. R. Panta, M. Feramez and J. Armstrong, “OFDM peak-to-average power reduction scheme with spectral masking,” Proc. Symp. on Communication Systems, Networks and Digital Signal Processing, pp.164-167, July 2004.

6. Zhao, Chunming; Baxley, Robert J.; Zhou, G. Tong; Boppana, Deepak; Kenney, J. Stevenson, “Constrained Clipping for Crest Factor Reduction in Multiple-user OFDM”, Radio and Wireless Symposium, 2007 IEEE Volume , Issue , 9-11 Jan. 2007 Page(s):341- 344.

7. Olli Vaananen, “Digital Modulators with Crest Factor Reduction Techniques”, PhD Thesis, 2006 8. Boumaiza, et a, “On the RF/DSP Design for Efficiency of OFDM Transmitters” , IEEE Transactions on

Microwave Theory and Techniques, Vol. 53, No. 7, July 2005, pp 2355-2361.9. Boumaiza, Slim, “Advanced Memory Polynomial Linearization Techniques,” IMS2009 Workshop WMC

(Boston, MA), June 2009.

Selected DPD References

49



http://ieeexplore.ieee.org/xpl/RecentIssue.jsp?punumber=26�




Recent DPD resources from Agilent

App Notes http://cp.literature.agilent.com/litweb/pdf/5990-8883EN.pdf (Wideband DPD)

http://cp.literature.agilent.com/litweb/pdf/5990-7818EN.pdf (3G/4G)

http://cp.literature.agilent.com/litweb/pdf/5990-6742EN.pdf

http://cp.literature.agilent.com/litweb/pdf/5990-6534EN.pdf (algorithms used)

Demonstration Videoshttp://www.youtube.com/watch?v=bocF6P74T9E (Wideband DPD)

Previous Webcasts“4G For Everyone: Extended RF Performance with DPD” (June 2010)http://www.home.agilent.com/agilent/eventDetail.jspx?cc=US&lc=eng&ckey=1842093&nid=-11143.0.00&id=1842093

50



You are invited

51

You can find more webcastswww.agilent.com/find/eesof-innovations-in-edawww.agilent.com/find/eesof-webcasts-recorded

Dr. Yi CaoDigital Hardware Designer, RIM (Research In Motion)

Hee-Soo Lee3D EM Applications Specialist, Agilent EEsof EDA



http://www.agilent.com/find/eesof-innovations-in-eda�

http://www.agilent.com/find/eesof-webcasts-recorded�

high performance digital pre-distortion for wideband systems€¦ · connect the mxg/awg directly...

Documents