Principles of Communications Lecture 8: Baseband...
Transcript of Principles of Communications Lecture 8: Baseband...
-
Principles of CommunicationsLecture 8: Baseband Communication Systems
Chih-Wei Liu 劉志尉National Chiao Tung [email protected]
-
Commun.-Lec8 [email protected] 2
Outlines
Introduction
Line codes
Effects of filtering
Pulse shaping toward zero ISI
Zero-forcing equalization
Eye diagrams
Synchronization
-
Commun.-Lec8 [email protected] 3
Introduction
Digital vs. analog signals
Analog sampling quantization digital
Baseband vs. passband
Channel distortion – ISI
(Channel) bandwidth limitation
-
Commun.-Lec8 [email protected] 4
Line Codes
Baseband data format used to represent digital data (for transmission purpose).
Examples are given on the next page
Operation: time or frequency shaping
Purposes: usually to cope with the channel limitations (or provide extra function such as synchronization)
Needed for certain applications
-
Commun.-Lec8 [email protected] 5
Non-return-to-zero (NRZ) change
NRZ mark (data1-> change in level; data0 -> no change)
Unipolar return-to-zero (URZ)
Polar RZ
Bipolar RZ (“0”-> 0 level; “1”-> alternate sign
Split phase (Manchester) (“1”-> level A to level –A at 1/2 interval; “0” -> level –A to level A at 1/2 interval)
-
Commun.-Lec8 [email protected] 6
Purposes of Line Codes
Self synchronizationProper power spectrumTransmission bandwidthTransparency
Error detection capability
Good error probability performance
-
Commun.-Lec8 [email protected] 7
Power Spectra (I)The transmitted signal is a pulse train:
( ) ( ).
The amplitudes can be viewed as random variables with 0, 1, 2,
The autocorrelation function of the waveform is
( ) (
kk
m k k m
X m
X t a p t kT
R a a m
R R rτ
∞
=−∞
+
= − − Δ
= = ± ±
=
∑
K
),
1in which ( ) ( ) ( ) .
m
mT
r p t p t dtT
τ
τ τ
∞
=−∞
∞
−∞
−
= +
∑
∫
-
Commun.-Lec8 [email protected] 8
Power Spectra (II)
2
2
The power spectral density is the Fourier transofrm of ( ):
( ) [ ( )] [ ( )]
[ ( )] ( )
( ) .
1Note that ( ) [ ( )] [ ( ) (
X
X X mm
j mTfm m r
m m
j mTfr m
m
r
R
S f R R r mT
R r mT R S f e
S f R e
S f r p t p tT
π
π
τ
τ τ
τ
τ
∞
=−∞
∞ ∞−
=−∞ =−∞
∞−
=−∞
= = −
= − =
=
= = − ∗
∑
∑ ∑
∑
FT FT
FT
FT FT2| ( ) |)] .P f
T=
-
Commun.-Lec8 [email protected] 9
Example 1: NRZ
Assume the message m[n] is random (white noise) with equally “0”and “1” values.
Step 1: Compute Rm based on the above assumption and “format”
Step 2: Compute r(t) based on the pulse shape.
2 2 21 12 2
2 21 1 1 14 4 4 2
2 2 2
1. NRZ, 50% "0" and 50% "1".= ( ) , 0;
= ( ) ( ) ( ) 0 , 0. ( ) ( / ) ( ) sinc( )
Therefore, ( ) ( ) sinc ( ).
m
m
NRZ r
R A A A m
R A A A A A A mp t t T P f T Tf
S f A S f A T Tf
+ − = =
+ − + − + − = ≠
= Π → =
= =
-
Commun.-Lec8 [email protected] 10
Example 2: Unipolar RZ
2 2 2
2
2 2
2 2 2
2. Unipolar RZ, 50% 1-level and 50% 0-level. 1 1 1(0) , 02 2 21 1 1 1 10 0 0 0 , 04 4 4 4 4
( ) (2 / ) ( ) sinc( )
1 1( ) sinc ( )4 2 2 4
m
T T
URZ
A A mR
A A A A A m
p t t T P f f
T TS f f A A
⎧ + = =⎪⎪= ⎨⎪ ⋅ + ⋅ + ⋅ + ⋅ = ≠⎪⎩= Π → =
= + 2, 0
2 2 2 2
22 2 2
1 1sinc ( )4 2 4 4
1 1 1sinc ( ) ( ) . ( )4 2 4 4
mj mTf
m m
mj mTf
m
n m nj mTf
n m n
e
T T f A A e
T T A n nf A f e fT T T T
π
π
πδ δ
=∞−
=−∞ ≠
=∞−
=−∞
=∞ =∞ =∞−
=−∞ =−∞ =−∞
⎡ ⎤⎢ ⎥⎣ ⎦
⎡ ⎤= +⎢ ⎥⎣ ⎦⎡ ⎤
= + − = −⎢ ⎥⎣ ⎦
∑
∑
∑ ∑ ∑Q
-
Commun.-Lec8 [email protected] 11
-
Commun.-Lec8 [email protected] 12
Inter-symbol Interference (ISI)Intersymbol interference refers to one specific type of distortion. It is typically due to insufficient bandwidth of the channel. Note: Narrow BW Long time-duration
Example: Rectangular waveforms through a lowpass RC filter. The neighboring pulses smear out and interfere with each other.
-
Commun.-Lec8 [email protected] 13
BW= 1bit/sec
-
Commun.-Lec8 [email protected] 14
BW= 0.5bit/sec
-
Commun.-Lec8 [email protected] 15
ISI Reduction
Pulse shaping: Assume the channel is ideal (flat) with the minimum BW (W). What shape of “pulse” warrants ISI-free transmission?
Equalization: If the channel is non-ideal (not ideally flat LPF), an equalizer helps in reducing ISI.
-
Commun.-Lec8 [email protected] 16
Pulse Shaping
Consider 2W independent samples per second are transmitted through a channel with bandwidth W Hz. The output is as follows.
( ) ( ) sinc 2 .2
The samples at / 2 are ISI-free, sinc( ) 0.
n nn n
m
ny t y t a W tW
t m W m n
∞ ∞
=−∞ =−∞
⎡ ⎤⎛ ⎞= = −⎜ ⎟⎢ ⎥⎝ ⎠⎣ ⎦= − =
∑ ∑Q
Are there other pulse shapes warrant ISI-free? -T 0 T 2T t
t
Sampling of p(t-2T)
)2(2 Ttpa −
)3(3 Ttpa −
)(1 Ttpa −
)(tp
-
Commun.-Lec8 [email protected] 17
Raised Cosine FamilyRaised Cosine: a example of ISI-free pulses – raised cosine spectra at BW edges.
Roll-off factor β: β=0 rectangular pulse (sinc)β=1 “cosine” in freq; time pulse has narrow main lobe with very low sidelobes
1, | |2
1 1 1( ) 1 cos | | | |2 2 2 2
10 | |2
RC
T fT
T TP f f fT T T
fT
β
π β β ββ
β
−⎧ ≤⎪⎪
⎧ ⎫⎡ ⎤− − +⎪ ⎛ ⎞= + − < ≤⎨ ⎨ ⎬⎜ ⎟⎢ ⎥⎝ ⎠⎣ ⎦⎩ ⎭⎪⎪ +
>⎪⎩
-
Commun.-Lec8 [email protected] 18
2
Time pulse of the raised cosine:cos( / )( ) sinc( ), is called the roll-off factor.
1 (2 / )RCt T tp tt T T
πβ ββ
=−
cosine curves
-
Commun.-Lec8 [email protected] 19
Nyquist’s Pulse Shaping CriterionA pulse shape ( ) with a Fourier transform ( ) satisfying
1( ) , | |2
1, 0then its sampled values ( ) .
0, 0.
k
p t P fkP f T fT T
np nT
n
∞
=−∞
+ = ≤
=⎧= ⎨ ≠⎩
∑
-
Commun.-Lec8 [email protected] 20
( )
2
2
2 12 2
2 12
122
12
2
Proof of Nyquist's Pulse Shaping Criterion
( ) ( )
Sampling at , ( ) ( )
( ) ( )
( ) ;
( )
j ft
j fnT
kT j fnT
kkT
j f unT knTk
T
j funT
k
p t P f e df
nT p nT P f e df
p nT P f e df
k kP u e du u fT T
kP u eT
π
π
π
π
π
∞
−∞
∞
−∞
+∞
−=−∞
∞ +
−=−∞
=−
=
=
=
= + = −
= +
∫∫
∑ ∫
∑ ∫1
21
2
12 2
12
1, 00, 0
T
T
T j funT
T
du
nTe du
nπ
∞
−∞
−
=⎧= = ⎨ ≠⎩
∑∫
∫
-
Commun.-Lec8 [email protected] 21
Pulse Transmission System
( ) ( ).
( ) ( ) ( ) [ ( ) ( )] ( ).Consider the combined effect of three filters, we would liketo have ( ) to be ISI-free.
k Tk
R C R
x t a h t kT
v t y t h t x t h t h t
v t
∞
=−∞
= −
= ∗ = ∗ ∗
∑
-
Commun.-Lec8 [email protected] 22
Transmitter and Receiver Filters
Practically, Hc(f) is unknown of time-varying
2
Let ( ) ( )
where represents a scale factor and a possible delay.( ) ( ) ( ) ( ), or in frequency domain,
( ) ( ) ( ) ( ).If ( ) is given, then
d
k RC dk
d
RC d T C Rj ft
RC T C R
C T
v t A a p t kT t
A tAp t t h t h t h t
AP f e H f H f H fH f H
π
∞
=−∞
−
= − −
− = ∗ ∗
=
∑
( ) ( ) 1 ( )The overall spectrum should satisfy the Nyquist criterion to avoid ISI. A filter design problem.
R Cf H f H f=
⇒
-
Commun.-Lec8 [email protected] 23
1/2 Let ( ) ( ) 1 ( )T R CH f H f H f= =
-
Commun.-Lec8 [email protected] 24
Zero-Forcing Equalization
Now, assume digital processing at the receiver.Purpose: Design an FIR filter that compensates for the channel distortion zero-ISI.Show an example of zero-ISI equalizer design.
-
Commun.-Lec8 [email protected] 25
Let the combined impulse response after the channel be ( ),Pass such a pulse through our finite-length equalizer, then
( ) ( ).
If the resulting pulse satisfies the zero-ISI condition
c
n
eq n cn N
p t
p t p t nα=−
= − Δ∑, ISI-free
transmission is achieved. Assume that , the condistion is1, 0
( ) [( ) ] 0, 1, , .0, 0
Notice that we only enforce the zero-ISI condition on the 2 1 samples
n
eq n cn N
Tm
p mT p m n T m Nm
N
α=−
Δ =
=⎧= − = = ± ±⎨ ≠⎩
+
∑ K
. Why? We only have 2 1 coefficients (variables)and can only achieve this much. Solve the 2 1 equations andyou obtain the zero-forcing equalizer.
NN
++
-
Commun.-Lec8 [email protected] 26
1
0
(0) ( ) ( 2 )0
( ) (0) (( 2 1) )10
(2 ) (0)
0
c c c N
c c c N
c c N
p p T p NT ap T p p N T a
p NT p a
−
− +
⎡ ⎤⎢ ⎥⎢ ⎥ − −⎡ ⎤ ⎡ ⎤⎢ ⎥ ⎢ ⎥ ⎢ ⎥− +⎢ ⎥ ⎢ ⎥ ⎢ ⎥= ⋅⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥ ⎢ ⎥⎣ ⎦ ⎣ ⎦⎢ ⎥⎢ ⎥⎣ ⎦
ML
L
M M M
M
Example: A 3-tap equalizer The equalizer cannot
eliminate ISI beyond its span.
-
Commun.-Lec8 [email protected] 27
Eye Diagrams
Eye diagram: Constructed by overlapping a number ofsegments of the base-band signals
A qualitative measure of the system performance.
(Lee et al., Digital Comm., 1994)
-
Commun.-Lec8 [email protected] 28
The best place tosample the signal.
Amplitude jitter:ICI causes the amplitudeof each symbol fluctuate.
Timing jitter:ICI causes the timing ofeach symbol fluctuate.
Increasing bandwidth may mitigate ICI, butwill also allows more noise to enter. Anotherexample of trade-offs in commu systems design
eye eye
-
Commun.-Lec8 [email protected] 29
Symbol, Bit, Word, and Frame
Bits Symbol (a single pulse in transmission) ---carrier sync., symbol timing
Bits Word --- word sync.
Words Frame ---frame sync.
(Benedetto et al., Digital TansmissionTheory, 1987)
-
Commun.-Lec8 [email protected] 30
Synchronization
Methods : (1) external sync signals;
(2) self-synchronization: derivation from the modulated signals
Example 1: squaring the received NRZ signals and PLL
Sync signal
-
Commun.-Lec8 [email protected] 31
Example 2
4
Sync signal
-
Commun.-Lec8 [email protected] 32
Carrier Modulation
Simple modulation applied to baseband digital signals
Example: Let ( ) be the NZR waveform. Amplitude Shift-Keying (ASK): ( ) [1 ( )]cos(2 )
Phase Shift-Keying (PSK): ( ) cos[2 ( )]2
Frequency Shift-Keying (FSK): ( ) cos[2
ASK c c
PSK c c
FSK c c
d tx t A d t f t
x t A f t d t
x t A f t
πππ
π
= +
= +
= + ( ) ]t
fk d dα α∫
-
Commun.-Lec8 [email protected] 33