- 1/40 - Instructor : Jaewook Kang At CSNL-GIST E-mail: [email protected] 2011, Apr. 1st A Simple Communication System Design Lab with MATLAB Simulink - Lab #3: - Phase splitter - up conversion and down conversion
- 1/40 -
Instructor : Jaewook KangAt CSNL-GISTE-mail: [email protected], Apr. 1st
A Simple Communication System Design Lab with MATLAB Simulink
- Lab #3: - Phase splitter- up conversion and down conversion
- 2/40 -
Next time…
v Place: IC203
Weeks Time Hour Instructor
1 weekLab. #0 3.11
(13:00~16:00)
3- Overview of Development with Simulink
- QPSK Model with AWGN Channel/ Rayleigh FadingChannel
- Development Example: Interference Cancellation for Satellite CommuniJunil Ahn
2 weeksLab. #1
3.18(13:00~14:20)
1.5 - Basic OFDM Junil Ahn
3.18(14:30~16:00)
1.5- Introduction - How to use Simulink with interleaver
implementationJaewook Kang
3 weeksLab. #2
3.25(13:00~16:00)
3- How to use S-function builder- PSF and Matched filter design- Concept of upsampling and downsampling
Jaewook Kang
4 weeksLab. #3
4.1(13:00~16:00)
3 - Phase splitter - Up conversion and down conversion Jaewook Kang
5 weeksLab. #4
4.8(13:00~16:00)
3 - How to make subsystem- Channel implementation using Jaewook Kang
- 3/40 -
Reviewv Q1) Why ?
v Q2) What is happen in freq. domain if sampling rate is up ?
v Q3) What about down case?
v Q4)Why PSF ?
v Q5) calculate bandwidth when Rsym=1 sym/sec and L=4.
dw p£
- 4/40 -
Today’s main pointsv Understand up/down sampling
v Understand characteristic of complex exponential signal
v Learn about Hilbert transform
v Implement Phase splitter using FIR filter
v Understand up/down conversion
- 5/40 -
Our target systemvTx part
vRx part
Tx Source Interleaver QAMMapper PSF X
NCO
↑4
PhaseSplitter
Matchedfilter
QAMDemapperX
NCO
↓4 De-Interleaver Rx Source
:Real:Complex
- 6/40 -
up conversionvWhy upconvert to IF ?
§ 1) To reduce the work of D/A and A/D convertor.
§ 2) To bring the signal only using Inphase channel.
- 7/40 -
up conversion§ As a result, rotational transform of basis in s plane
- 8/40 -
up conversion
Re[ ( )] cos(2 ) sin(2 )
Im [ ( )] sin(2 ) cos(2 )shift shift
shift shift
s t x f t y f tag s t x f t y f t
p p
p p
= -
= +
- 9/40 -
up conversion§ When baseband is up-converted to Fs/2
Re[ ( )] cos(2 )shifts t x f tp=
cos(2 ) sin(2 ) cos( )2 2s s
s sf fx nT y nT x np p p= - =
- 10/40 -
up conversion§ NCO (Numerically Controlled Oscillator ) implementation
§ - Carrier frequency is determined by adjusting ‘k’§ In general k=4à Then, IF is pi/4§ - Thus, higher Fs should be required in order to up-convert
baseband into higher band
02s
shiftff< £ (s
shiftff kk
= ³ is positive number and k 2.0)
2
( ) ( ) ( )
( ) cos(2 ) sin(2 )
1cos(2 ) cos(2 ) cos(2 )
shift s
s s sj f nT
s shift s shift s
sshift s
s
s nT g nT c nT
c nT e f nT j f nTf nf nT nk f k
p p p
p p p
=
= = +
= =
- 11/40 -
up conversion
§ The relationship between up-sampling rate and up-conversion
§ In this case, Fs is sampling rate of baseband signal, g(t)§ Usually g(t) is the output of PSF, so Fs is up-sampling rate
- 12/40 -
Phase splitterv Purpose of the phase splitter
§ Since Tx only transmit real part of signals, we have to recover the full complex signal from real part of the received signal.
§ When using FIR filter to implement Phase splitter, we have to consider the delay of the filter (N-1)/2.
(N- 1)/2 sampledelay
N tap Hilbert Transform FIR filter
Xr(t) Xr(t)
Xi(t)
Xc(t)
- 13/40 -
Phase splitterv Basic of complex exponential
§ What is shape of ?§ Can you draw Xc(t) in 3D domain ?
Ø (Hint: real imag, and time axis)
02c
0
0 fourier transform
x ( )( ) cos(2 )( ) sin(2 )
( ) ( ) ( ) ( ) ( ) ( )
j f t
r
i
c r i c r i
Let t Aex t A f tx t A f t
x t x t jx t X f X f jX f
p
pp
===
= + ¾¾¾¾¾® = +
02cx ( )
j f tt Ae p=
- 14/40 -
Phase splitterv Basic of complex exponential
§ As t increase, (+) freq. component rotates with CCW and (-) freq. component rotates with CW.
§ Sine and cosine is the same signal without phase. § Real part : even symmetric§ Imaginary part: odd symmetric
Imag
- f
Re
Acos(2* pi* f0* t)
- f0
f0
Imag
- f
Re
Asin(2* pi* f0* t)
- f0
f0
0
( 0)cos(2 0)rx tA f t Ap
= == = 0
( 0)sin(2 0) 0ix tA f tp
= =
= =
Conjugate symmetric
- 15/40 -
Phase splitterv Basic of complex exponential
§ Why cosine and sine are real ?Ø Imaginary component of (-) and (+) freq. are cancelled each other.
§ Such a fact implies that real and imag components co-exists in real periodic signal.
Imag
- f
Re
Acos(2* pi* f0* t)
- f0
f0
Imag
- f
Re
Asin(2* pi* f0* t)
- f0
f0
0
( 0)cos(2 0)rx tA f t Ap
= == = 0
( 0)sin(2 0) 0ix tA f tp
= =
= =
- 16/40 -
Phase splitterv Basic of complex exponential
§ Phase: The angle between real and imag component.§ Magnitude: The sum of real part of two freq component.
Ø According to time, the phase and magnitude is changing.
- 17/40 -
Phase splitterv Basic of complex exponential
§ Complex exp. signal has only one frequency component such that imaginary part exists.
§ How to extract real/imaginary part of Xc(t) ?
02cx ( )
j f tt Ae p=
- 18/40 -
Phase splitterv Hilbert transform
§ Find imaginary part of Xc(t) from real part only when Xc(t) is complex exponential. Imag
- f
Re
- f0
f002j f tAe p
Imag
- f
Re
Acos(2* pi* f0* t)
- f0
f0
Re[]
Imag
- f
Re
Asin(2* pi* f0* t)
- f0
f0
Hilbert TR
+
Imag
- f
Re
- f0
f002j f tAe p
Xc Xr
Xi_hatXc_hat
X
j
- 19/40 -
Phase splitterv Transfer function of Hilbert transform
§ The transfer function have sign shape on imaginary axis.
- 20/40 -
Phase splitterv Impulse response of Hilbert transform
§ The discrete form can be obtained by sampling such that
22
2
0 2 220
2
( ) ( )
1 [1 cos( )]
s
s
s
s
fft
f
fft ft
f
s
h t H f e df
je df je df
f tt
p
p p
pp
-
-
=
= + -
= -
ò
ò ò
:apply iFT to H(f) of HF
1 [1 cos( )]nn
pp
-h(n)=
- 21/40 -
Phase splitterv Implementation of Hilbert transform using FIR filter
§ 1) Determine the number of taps, NØ The larger, the nicer. But it cause longer delay.Ø The delay should be multiple of up sampling rate for synchronization.Ø Let N
§ 2) Generate h(n) : N=25 à delay =12§ 3) put h(n) into FIR filter as coefficient.
- 22/40 -
Phase splitterv allclearv fs=1;%fs musthave'1'togetnomalized magnitude
responsev TapNum=73;%thenumberofFIRfilterv n=0:(TapNum-1);v delay=(TapNum-1)/2%thedelayofFIRfiltervv %generateHilberttransformImpulseresponsev ht=fs./(pi*(n-delay)).*[1-cos(pi*(n-delay))];v %alternateformv %H=2*sin(pi*(n-10)/2).^2./(pi*(n-10))vv ht(delay+1)=0;%//byL'Hopital's rulevv %Impulseresponseofhilbert tansform FIRfilterv figure(1);v stem(ht);v xlabel('Xaxis-Time-index(n)','fontsize',12);v ylabel('Yaxis-Magnitudeofh(n)','fontsize',12);v title('\bf{ImpulseresponseofHilberttansform,
h(n)}','fontsize',12);v %obtainthemagnituderesponseofHilberttransformv Hf=abs(fft(ht));
v %Magnituderesponseofhilbert transformFIRfilter
v figure(2);v stem(Hf);v xlabel('Xaxis-Frequency-index(m)','fontsize',12);v ylabel('Yaxis-Magnitude,|H(m)|','fontsize',12);v title('\bf{MagnituderesponseofHibert
transform,|H(m)|}','fontsize',12);vv Hf(TapNum-delay+1:TapNum)=Hf(TapNum-
delay+1:TapNum).*-1;vv figure(3);v stem(Hf);v xlabel('Xaxis-frequency-index(m)','fontsize',12);v ylabel('Yaxis-imaginaryaxis,jH(m)
','fontsize',12);v title('\bf{Hilberttransforminfrequency
domain}','fontsize',12);
- 23/40 -
Phase splitterv Implementation of Hilbert transform using FIR filter
§ 1) Determine the number of taps, NØ The larger, the nicer. But it cause longer delay.Ø The delay should be multiple of up sampling rate for synchronization.Ø Let N
§ 2) Generate h(n) : N=25 à delay =12§ 3) put h(n) into FIR filter as coefficient.
- 24/40 -
Next time…
v Place: IC203
Weeks Time Hour Instructor
1 weekLab. #0 3.11
(13:00~16:00)
3- Overview of Development with Simulink
- QPSK Model with AWGN Channel/ Rayleigh FadingChannel
- Development Example: Interference Cancellation for Satellite CommuniJunil Ahn
2 weeksLab. #1
3.18(13:00~14:20)
1.5 - Basic OFDM Junil Ahn
3.18(14:30~16:00)
1.5- Introduction - How to use Simulink with interleaver
implementationJaewook Kang
3 weeksLab. #2
3.25(13:00~16:00)
3- How to use S-function builder- PSF and Matched filter design- Concept of upsampling and downsampling
Jaewook Kang
4 weeksLab. #3
4.1(13:00~16:00)
3 - Phase splitter - Up conversion and down conversion Jaewook Kang
5 weeksLab. #4
4.8(13:00~16:00)
3 - How to make subsystem- Channel implementation using Jaewook Kang