System Simulation of Wireless Communication Systems using ... · System Simulation of Wireless...

System Simulation of Wireless Communication Systems

using HP-ADS

Sung,Youngchul Lee, Ju-Yeol

LG Information & Communications, Ltd.Digital CommunicationsR & D CENTER(0343)450-7233sargon@lgic.co.kr jylee@lgic.co.kr

1999 HP EEsof

User’s Workshop

AGENDA

I. Introduction

II. Simulation Techniques for Communication System

- Circuit Simulation (DC,TRAN,HB,CE)

- Digital Algorithm Simulation(Ptolemy)

- How does the Co-simulation work?

III. System Simulation of W-CDMA

- Base Station Waveform Generation

- Cosimulation of Reverse Access Channel

- CDMA Receiver Example

IV. Conclusion

1. Physical Layer of Wireless Digital Communications

I. Introduction

Source

Format

Channel

CodingRF/DemodChannelMod/RF

Source

Deformat

Channel

Decoding

.Source

Coding

.Block Code

(R-S code)

.Convolutional

Code

.Moduation on

Real Sinusoidal

Waveform

-Amplitude

-Phase

-Frequency

.Amplifying

.Mixing

.Oscillation

.Large Scale Loss

-Frii’s Formula

-Hata Model

.Small Scale Fading

-Amplitude/Phase

-Multipath Effect

-Time Dispersive Linear Channel

-Time Varing Property due to Mobility

.Noise and Jamming

.Block Decoder

.Viterbi Decoder

(Hard/Soft)

.Synchronization

-Timing RC

-Freq. RC/Phase RC

.Channel Estimation

-ADE, AGC/LIMITING

-Matched Filtering

.Diversity

Discrete-Time Continuous-Time Discrete-Time

2. Consideration Factors for Wireless Digital Communication System Design

Main Categories of Concurrent Wireless Digital System

. Digital Radio Link

. Digital Multiple Access Network

General Items for Selecting Modulation and Coding scheme

.Power Efficiency/ Power Margin-> Link Availability

.Spectral Efficiency

.Implementation Complexity

Specific Measurement Items for Given System Architecture

.Bit Error Rate(AWGN/Fading Channel, Phase Jittering, Timing Error)

- Long term/Short Term

.Frame Error Rate

.Immunity to Other Jamming Sources(ACPR, CW Jamming, Wideband Jamming)

1. Circuit Simulation

-DC, HF SPICE(TRAN), Hamonic Balance(HB), Circuit Envelope(CE)

DC Analysis

- Eleminate all energy storing elements (capcitor, inductor)

Circuit is reduced to general resistive one.

- Iteratively find solution such that sum of all DC currents into each

circuit node is zero

- Newton-Raphson algorithm

General nonlinear tableau or MNA eq.

II. Simulation Techniques for Communication Systems

i

ijijjj

jjjj

jjj

xxxx

xfxJxx

tohxxxJxfxf

2

,,11

1

1

)(

)()]([

...))(()()(

Time-Domain Analysis (TRAN, CONV)

- DC analysis is performed to find the initial condition

- Solution to general dynamic circuits

.Circuit description: tableau, MNA(Modified Node Analysis)

.A small time step h is selected(several options) so that

derivative of x(t) is assumed to be constant over this time interval h

.Integration method

Backward Euler or trapezoidal

.Convolution Integration : includes frequency-domain described components

Backward Euler Approx.

Trapezoidal Approx.

Initial Condition

)))1(,(),((5.0

))1(,(

))1(,(

)))1(,(),((5.0)(

))1(,()(

11

11

1

1

hkxfkhxfhxx

hkxfhxx

hkkhtfor

hkxfkhxftx

hkxftx

kkkk

kkk

kk

k

Typical SPICE model

DCxtx

ttxftx

)(

)),(()(

0

t

dxthtxftx0

)()(),()(

Hamonic Balance Analysis

-Steady-state solution to multitone excitation(time-invariant frequency sources)

including hamonics and intermodulation terms

-Frequency nonlinear simulation technique

Current flowing from nodes into linear elements, including all distributed element,

are calculated by straightforward Laplace domain linear analysis

Current flowing from nodes into nonlivear elements are calculated by SPICE model

FFT-1

Start with

an initial

spectrum

SPICE ANAL.

FFT

KCL Check

Circuit Envelope Analysis

-Mixed time-frequency domain simulator

-Efficiency for complex modulated signal/ RF transient simulation

-Represent the signal as complex envelope with regard to concerned carrier frequency

at each time

-Samples the modulation envelope of signal instead of its RF carrier (Time -Domain Method)

(*Tstep is fixed in CE simulation)

-RF carrier is simultaneously computed in frequency domain for each envelope time sample

(Frequency Domain Method)

-Output is a time-varying spectrum (discrete in frequency domain)

Linear or Nonlinear

Circuit

)(tVenv

where

Constant for Harmonic Balance

Real Time-Varying for Baseband Signal

Complex Time-Varying for Bandpass Signal

})(Re{)( 0

,

tjkw

kenv

k

OUT etVtV })(Re{)( 0tjw

envIN etVtV

2. DSP / Communication System Simulation

PTOLEMY

- Basically Discrete-time Sequence Algebra(Scalar/Matrix,Int,Fixed,Float,Complex,Timed)

- Numeric Domain / Timed Domain

Numeric Domain : Time interval between samples is not defined

Timed Domain : Time interval between samples is defined

- Synchronous Data Flow Control

.Sources to Sinks

.‘Actors’ comsumes a number of ‘tokens’ at each ‘firing’

. Balance Equation

Balancing Producing and Consumption of Tokens

Vector Processing : Multiple Token Consumption on a single firing

ex) FFT

Multirate Processing : Cosumes M samples and Produce N

Multiple Signal Merger: Commutator or Distributor

. Each component is fired the minimum number of times to satisfy the

Balance Eq.

. Iteration of firing until all sinks are filled.

. Inconsistency / Deadlocks

COMMUNICATION SYSTEM SIMULATION

-Basic Goal

. Verifying the DSP Algorithm/ Performance Test

BER,SER,FER versus Channel Effect - Monte Carlo Method

TR,CR Algorithm

ADE Algorithm

.Check the Analog Result

Spectrum(ACPR)

Constellation/ EVM

Eye Diagram

Noise Figure/ Dynamic Range

-Simulation Software Consideration

.Simulation Bandwidth( > (4~10)*Signal BW )

.Baseband versus Bandpass Systems

Baseband : Direct Sampling -> Real Envelope

Bandpass : Complex Envelope(Baseband Equivalent System)

Baseband Equivalent System for Bandpass System

1. Baseband Equivalent Signal

))(2cos()()( tftatx cpass

})(Re{2)( tfjtj ceeta

)()()( tj

LP etatx

)()()( tjxtxtx QILP tfj

LPcetxjtxtx

2)](ˆ)([)(

Real

Imag

frequency

Hermitian Spectrum of Real Bandpass Signal

2. Baseband Equivalent Linear System

tfthtfthth cQcI 2sin)(2cos)()(

)]()([2/1)( tjhthth QILP

3. Whole Behavioral Equivalent System for Quadrature Scheme

including Linear Filtering,Modulation/Demodulation,Up/Down conversion,AWGN,

Rayleigh Fading/Doppler

H_I(t)

H_Q(t)

H_Q(t)

H_I(t)

Rotator

(2*2Matrix)

f_up -f_down

doppler

phase noise

Gaussian

R.P.

Gaussian

R.P.

Fading/

Doppler

(Smith Method)

-

I_IN

Q_IN Q_OUT

I_OUT

3. DSP / Behavioral , Circuit Cosimulation

Full Ptolemy Chain

Ptolemy

DSP

Ptolemy

Down/

Demod

Ptolemy

Channel

Model

Ptolemy

Mod/Up

Ptolemy

DSP

Ptolemy /Circuit Mixed Chain

Ptolemy

DSPCircuit

Ptolemy

Channel

Model

CircuitPtolemy

DSP

Ptolemy /Circuit Mixed Chain

- Circuit takes time sample values as input and dumps calculated

time sample output value.

- Circuit Engine : transient, circuit envelope

Timed

Numeric Timed

TSDFHF SPICE

CIRCUIT

ENVELOPE

Real Time

Sample,

Tstep

Complex

Time Sample,

Tstep,

F_carrier

F_carrier

Selector

TSDF

Sink

III. System Simulation of W-CDMA

BASE STATION WAVEFORM GENERATION

Channel Code Channel Linear Power Power ,dB Sym,bol Rate PN ChipRatePilot 0 0.2 -7 None 8.192 McpsSync 1 0.03 -15.2 8kb p s8.192 McpsPaging 2~5 0.07 -11.5 32kbps 8.192 McpsTr a ff ic / P CS 16~63 0.116 -9.4 32k/4kbps 8.192 McpsTr a ff ic / P CS 26~63 0.116 -9.4 32k/4kbps 8.192 McpsTr a ff ic / P CS 36~63 0.116 -9.4 32k/4kbps 8.192 McpsTr a ff ic / P CS 46~63 0.116 -9.4 32k/4kbps 8.192 McpsTr a ff ic / P CS 56~63 0.116 -9.4 32k/4kbps 8.192 McpsTr a ff ic / P CS 66~63 0.116 -9.4 32k/4kbps 8.192 Mcps

POINTS OF INTEREST

. PEAK TO AVERAGE POWER RATIO

. CONSTELLATION

. SPECTRUM

BandWidth: 10 MHz, TX Center Freq.: 2385 MHz Output Spectrum Shape : defined as follows

D1

D2

Fp Fs

D1D1 <=1.5 dB

D2 >= 40 dB

Fp=3.92 MHz

Fs=4.94 MHz

Whole Base Station Block

Base Station Pilot Channel

Base Station Sync Channel

Base Station Paging Channel

Base Station Traffic/PCS Channel

Results of Simulation

CONSTELLATION SPECTRUM

Power Distribution (Peak to Average Ratio)

I/Q Stream After Band-Limiting Filter

COSIMULATION OF REVERSE ACCESS CHANNEL

POINTS OF INTEREST

. PEAK TO AVERAGE POWER RATIO- Battery Time

. CONSTELLATION DISPERSION

. SPECTRUM- Power Regrowth,ACPR

CHANNEL PARAMETER

Code Channel Symbol Rate PN Rate Power LinearH0/H1 None 8.192 Mcps 0.2H2/H3 8 kbps 8.192 Mcps 0.8

ACCESS CHANNEL IMPLEMENTATION

ANALOG/ RF SUBCIRCIUT

SIMULATION RESULTS

ENVELOPE DISTRIBUTION OF MODULATED

SIGNALCONSTELLATION AT DIGITAL SUMMER

I SIGNAL AFTER ROOT RAISED COSINE

FILTERING

Q SIGNAL AFTER ROOT RAISED COSINE

FILTERING

CONSTELLATION OF BASEBAND

SPECTRUM OF IF SIGNAL SPECTRUM OF RF SIGNAL

CONSTELLATION OF RF SIGNAL

BASIC RECEIVER FUNCTIONS IN CDMA DETECTION

RF INTERFACE

RX AGC, Offset Correction, Freq. Acquistion

SEARCH OPERATION

PN Generation

Correlation Calculator/Selector

FINGER OPERATION

Time Tracking

Frequency/Phase Tracking

Signal Level Estimation

RAKE COMBING

Symbol Combing

Power Control Unit

VITERBI DECODING UNIT

.Pilot Aided Coherent Quadrature Modulation Scheme

.Consider Time-Varying Fading Effect

.Add a Additional Frequency Offset

Channel Rate PowerPilot 1.2288 Mcps 0.5Data 19.2 ksps 0.5

Frequency Tracking Algorithm Simulation

using CDMA Design Library

SIMULATION RESULT

Estimated Frequency Error

Binary Symbols Sent

Binary Symbols Received

Symbol Error Rate

IV. Conclusion

.Simulation is very useful in predicting system’s behavior , analyzing

algorithm’s performance , and designing circuits.

. No Single simulation method can be sufficient to resolve

all the simulation requirements in digital wireless system

. All simulation methods must be combined into a single tool

for top-to-bottom and mixed domain simulation of complete system

. Hardware implementation can be easier with predicted simulation

results

. Connecting simulation and hardware implementation automatically

is a challenging work

Math.

Physics

Communication

Theory

Component Circuit

Signal

Processing

Subsystem

Algorithm

COMMUNICATION SYSTEM