System-Level Time-Domain Behavioral Modeling for

a Mobile WiMax Transceiver

2006. 9. 15

Brian Y. Kim, Ph. D.

Telecomm. R&D Centre

Samsung Electronics

System-Level Time-Domain Behavioral Modeling for

a Mobile WiMax Transceiver

2006. 9. 15

Brian Y. Kim, Ph. D.

Telecomm. R&D Centre

Samsung Electronics

© SAMSUNG Electronics Co., Ltd. 2/19

• Background• RF Time-domain Behavioral Modeling• Modeling Methods• Simulation Results• Conclusion

AgendaAgenda

3/19

Industry Trend - Highly Integrated SoC/SiPRF Ckt, ADC/DAC & Digital baseband

Design MethodologyTop-down vs. Bottom-upTop-down methodology becomes necessary for SoC/SiP

Simulation & Verification System-level vs. Transistor-levelSystem-level simulation & verification play a key role in Top-down methodology

Background

4/19

Time-domain vs. Frequency-domain System-level Specification <-> Time-domain

Constellation, EVM, BER, FER,…

RF Performance <-> Frequency-domainNonlinearity, Noise Figure, Phase noise, …

System-level simulationFull signal path and time-consuming simulationTime-domain behavioral models are required Frequency-domain model -> Time-domain mode

Application ExampleMobile WiMax IEEE 802.16e (WiBro) in Simulink

Background

5/19

RF TimeRF Time--domain Behavioral Modelingdomain Behavioral Modeling

Passband vs. Baseband Real signal processing -> passband modelComplex signal processing -> baseband equivalent model

( ) ( ) cos( ) ( )sin( )1 {[ ( ) ( )] [ ( ) ( )] } 2

c c

c c

j t j t

rf t I t t Q t t

I t jQ t e I t jQ t eω ω

ω ω−

= +

= + + − 2

{[ ( ) ( )] [ ( ) ( )] }

[ ( ) ( )] [ ( ) ( )]

c c c

c

j t j t j t

j t

I t jQ t e I t jQ t e e

I t jQ t I t jQ t e

ω ω ω

ω

−+ + − ×

= + + −

6/19

RF TimeRF Time--domain Behavioral Modelingdomain Behavioral Modeling

Nonlinearity Polynomial memoryless modelPassband expression

x(t) is the passband signal, and able to expressed as x(t) = C(t)e-jωct + C*(t)ejωct,

C(t) is the complex baseband equivalent signal.

Baseband equivalent expression

the polynomial coefficients changed in baseband model

2 3 51 2 3 5( ) ( ) ( ) ( ) ( )y t k x t k x t k x t k x t= + + +

2 41 3 5

3 5( ) [ ( ) ( ) ] ( )4 8BBy t k k C t k C t C t= + + ⋅

7/19

RF TimeRF Time--domain Behavioral Modelingdomain Behavioral Modeling

NoiseNoise figure (NF) – Frequency-domainRMS noise power (Pn) – Time-domain

Where k is the Boltzmann constant, T is the noise temperature,

R is the source impedance and fs is the sampling frequency.

Other Noise sourceFlicker noisePhase noiseQuantization noise

2 /104 (10 1)NFn n sP kRT fσ= = ⋅ ⋅ +

8/19

Modeling Methods

AmplifierLNA

Polynomial nonlinearity model with hard saturationNoise model

Power AmplifierPolynomial nonlinearity model with hard saturationPhase distortion model by AM-PM conversion

3 51 3 5

3 54 8

+ +k x k x k x3 5

1 3 53 54 8

+ +k x k x k x

LNA PA

9/19

Modeling Methods

Amplifier Model Verification

LNA PA

More Nonlinearity is in PA

10/19

Modeling Methods

MixerSame nonlinearity and noise model as LNANonlinearity

Equivalent to the RF input

Noise Equivalent to the output port

11/19

Modeling Methods

LOPhase noise

1/f3, 1/f2, and Noise floorNoise floor is modeled as AWGN1/f2 is modeled as AWGN with transfer function

1/f3 is modeled as AWGN with transfer function

PLL loop Second -order high pass filter

21/ 11( )

1fH zz−=

−

31/ 1 3/ 21( )

(1 )fH zz−=

−31/ 1 2

0 1 2

1( )...fH z

a a z a z− −=+ + +

IIR

12/19

Modeling Methods

LO Model VerificationPhase noise: -50dBc/Hz @10kHz, -80dBc/Hz @100kHz ,&

-120 dBc/Hz @Noise Floor.Loop bandwidth: 100kHz

PLL Disabled PLL Enabled

13/19

Modeling Methods

ADC/DACNonlinearity:

INL and DNL: Modeled as non-uniform quantization levelDNL is generated by Gaussian random noise INL is the integration of DNL

, ,1

k

INL k DNL ii

v v=

=∑ , , 2q k INL k ref offsetnkv v v v= + +

Non-uniform Quantization level

14/19

System Simulation in MATLAB SimulinkSystem-level Diagram

Simulation Results

15/19

Simulation Results

Mobile WiMax Signal SourceSignal Source generation by m-codingDownlink modulation :

64QAM/ 16QAM/ QPSKUplink modulation : 16QAM/ QPSKCell ID : 2Segment number : 0DL/ UL ratio : 27/15 symbols

Modulation Scrambling RearrangementPilot Insertion

IFFTCyclic Prefix InsertionRF Input

16/19

System Performance Constellation and Spectrum for 16QAM

Simulation Results

A B C

ACPR SNR for 1e-6 BER

Signal source TX RX

17/19

Simulation Results

80dB55dB

45dB

35dB

TxSource

Rx

Constellation and Spectrum for 64 QAM

18/19

Simulation Results

System Performance for Various Cases

Simulation EfficiencyThe whole simulation for one frame OFDM signals takes about 4 minutes

on a Pentium 4 3.0GHz PC with 1GB memory.

19/19

Conclusion

Behavioral Mobile WiMax Transceiver design has been implemented on Matlab/Simulink platform.

Overall Functionality and block specification could be validated based on 802.11e standard.

The impact of nonlinearity and noise from RF/Mixed-signal building blocks is addressed by the time-domain system simulation.

The proposed baseband equivalent model technique is able to be applied by wide range of the system-level simulation covering RF nonlinearity and noise effects

© SAMSUNG Electronics Co., Ltd.