Lecture 6
Lecture 6
ANALYSIS
DR. TRINH VAN CHIEN
SER/BER for (binary antipodal signal)
1E-8
1E-9
1E-10
1E-11
1E-12
1E-13
1E-14
-2 0 2 4 6 8 10 12 14 16
Eb/N0 [dB]
2
BER COMPARISON
Error probability of antipodal signals and orthogonal signals:
1 Eb 1 1 Eb
Pb (e) |antipodal = erfc Pb (e) orthogonal = erfc
2 N 2 N
0 2 0
3
BER COMPARISON
1 Eb 1 1 Eb
Pb (e) |antipodal = erfc = erfc
2 N Pb (e) orthogonal
2 N
0 2 0
1
ortogonale
0.1 antipodale
0.01
1E-3
1E-4
1E-5
1E-6
1E-7
BER
1E-8
1E-9
1E-10
1E-11
1E-12
1E-13
1E-14
-2 0 2 4 6 8 10 12 14 16 18 20
Eb/N0 [dB]
4
BER COMPARISON
With Eb/N0 =12 dB:
• Antipodal signals give Pb(e) =1e-8
• Orthogonal signals give Pb(e) =5e-5 (Higher error probability → Lower
system performance)
To achieve Pb(e)=1e-6:
• Antipodal signals require: Eb/N0 = 10.6 dB;
• Orthogonal signals require: Eb/N0 =13.6 dB
5
BER COMPARISON
Example: For a communication environment with
GT GR
PR = PT
4 d
2
The system with antipodal signals has Pb(e)=1e-6 ➔ The received
signals are ½ that of orthogonal signals
With the same transmit power, the propagation distance corresponding the
antipodal signals are 2 longer than that of the orthogonal signals
For a given propagation distance, the network can reduce ½ the transmit
power if the network uses antipodal signals instead of orthogonal
signals
6
BER COMPARISON
For two constellation sets M1 and M2 with
Eb E
Pb (e) |1 erfc y1 Pb (e) |2 erfc y2 b
N N0
0
7
BER COMPARISON
1
ortogonale
0.1 antipodale
0.01
1E-3
1E-4
1E-5
1E-6
1E-7
BER
1E-8
1E-9
1E-10
1E-11
1E-12
1E-13
Eb 1E-14
-2 0 2 4 6 8 10 12 14 16 18 20
N 0 2 Eb/N0 [dB]
G = 10 log10 [dB]
Eb
0 1
N
G
8
ASYMPTOTIC PERFORMANCE
1 d2
PS (e) Amin erfc min
2 4 N0
d min = min d E ( s1 , s j )
s1 s j M
1 wmin d2
Pb (e) erfc min
2 k 4 N0
Where:
➔ All these formulations are upper bounds and the approximations for high
SNR values
10
EXAMPLE
For 4-PSK
01/ 00/s0
s1 a
d min = 2a Amin = 2 wmin = 2
a
11/s2 10/s3
d2 Eb
PS (e) erfc min
= erfc
4 N0
N0
1 d2 1 Eb
Pb (e) erfc min
= erfc
2
4 N0 2 N0
11
GRAY LABELLING
Asymptotic BER performance
1 wmin d2
Pb (e) erfc min
2 k 4 N0
One has Amin wmin
12
GRAY LABELLING
“adjacent” si ( minimum distance dmin to si) has its binary vector with
Hamming distance 1 compared wih vi ..
13
EXAMPLE
011
001
111
101 000
010
100
110
14
Signal Spectrum
15
PROPERTIES OF SIGNAL
SPECTRUM
16
ANTIPODAL SIGNAL
Consider the signal space:
1
B = b1 (t ) = PT (t )
T
17
ANTIPODAL SIGNAL
18
ANTIPODAL SIGNAL
The transmitted signals can be formulated as follows
+
s (t ) = a[n]p (t − nT )
n =0
19
ANTIPODAL SIGNAL
For all the 1D signal space:
+
s (t ) = a[n]p (t − nT )
n =0
20
ANTIPODAL SIGNAL
a[n] has the stationary property of random variables, all symbols are
independent with the probability:
1
P ( a[n] = i ) =
m
1 m
Mean a = i
m i =1
Variance 1 m
= ( i − a )
2 2
a
m i =1
21
WAVEFORM
We now compute the spectral density function for the signal in the form
+
s(t ) = a[n]p(t − nT )
n =−
s(t) is a random process, the spectral density function is computed as
+
Gs ( f ) =
−
Rs ( ) e− j 2 f d
This value provides the information of the power distribution in the frequency
domain: +
P( s) = Rs (0) = G ( f ) df
−
s
22
THEORY OF SPECTRAL DENSITY
+
Consider a random process s(t ) = a[n] p(t − nT )
n =−
Where
➢ a[n] is a stationary sequence of random variables with
Ma(i)=E[a[n]a[n+i]]
➢ p(t) is real signal with its Fourier transform version P(f)
Where S a ( f ) = M a (i )e − j 2 fiT
i
23
PROOF (FOR REFERENCE)
+
Gs ( f ) =
−
Rs ( )e− j 2 f d
T
1
Rs ( ) = M SS (t + , t ) dt
T0
24
PROOF (FOR REFERENCE)
+ +
M SS (t + , t ) = E[ S (t + ) S (t )] = E a[m]p(t + − mT ) a[n] p(t − nT ) =
m=− n =−
+ +
= E a[m]a[n] p(t + − mT ) p(t − nT ) =
m =− n =−
+ +
= M
m =− n =−
a (m − n) p(t + − mT ) p(t − nT ) =
+ +
M SS (t + , t ) = M
i =− n =−
a (i) p(t + − nT − iT ) p(t − nT )
25
PROOF (FOR REFERENCE)
+ +
M SS (t + , t ) = M
i =− n =−
a (i) p(t + − nT − iT ) p(t − nT )
1 + +
T T
1
Rs ( ) = M SS (t + , t ) dt = M a (i ) p (t + − nT − iT ) p (t − nT ) dt =
T0 T 0 i =− n =−
1 + + T
= M a (i)
T i =− n =−
p(t + − nT − iT ) p(t − nT ) dt =
0
For t’=(t-nT)
− ( n −1)T
1 + +
= M a (i)
T i =− n =−
p(t '+ − iT ) p(t ') dt '
− nT
+
1 +
Rs ( ) = M a (i) p(t '+ − iT ) p(t ') dt '
T i =− −
26
PROOF (FOR REFERENCE)
+
1 +
Rs ( ) = M a (i) p(t '+ − iT ) p(t ') dt '
T i =− −
+ +
1 + +
− j 2 f
Gs ( f ) = R ( )e
s
− j 2 f
d = M a (i) p(t '+ − iT ) p(t ') dt ' e d =
− − T i =− −
+ +
1 +
= M a (i) p(t '+ − iT ) p(t ')e− j 2 f dt ' d =
T i =− − −
+ +
1 +
Gs ( f ) = M a (i) p(t ") p(t ')e − j 2 ft " e j 2 ft ' e − j 2 fiT dt ' dt " =
T i =− − −
+ +
1 +
= M a (i)e− j 2 fiT p(t ") e− j 2 ft "dt " p(t ') e j 2 ft ' dt ' =
T i =− − −
1
= S a ( f ) P ( f ) P *( f )
T
2
P( f )
Gv ( f ) = Sa ( f )
T
28
STATISTICALLY INDEPENDENT
SYMBOLS WITH ZERO MEAN
➢ Zero mean: μa = 0
29
STATISTICALLY INDEPENDENT
SYMBOLS WITH ZERO MEAN
We have
for i = 0 ( )
M a (i ) = E a[n]2 = a2
30
STATISTICALLY INDEPENDENT
SYMBOLS WITH ZERO MEAN
For:
S a ( f ) = M a (i )e − j 2 fiT = a2
i
2
P( f )
Spectral density: Gs ( f ) = Sa ( f )
T
2
P( f )
Another expression (for simplicity):
Gs ( f ) = 2
a
T
31
EXAMPLE
Assume that:
1
B = b1 (t ) = PT (t )
T
32
EXAMPLE
+
The transmitted waveform: s (t ) = a[n]p (t − nT )
n =0
33
EXAMPLE
34
EXAMPLE
We have: 1
p(t ) = b1 (t ) = PT (t )
T
sin( x)
sinc( x) =
( x)
sin( fT ) − j f
P( f ) = T sinc( fT )e− j f = T e
( fT )
35
EXAMPLE
GG
s (v(f)
f)
fT
36
CONCLUSION
37
CONSTELLATION 1
The signal space is matched well with a “low-pass response”
ssT((tt)) r (t )
H( f )
The received signal (with no interference) has the spectral density function:
Gr ( f ) = H ( f ) Gs ( f )
2
Since Gs(f) spans the entire frequency axis, only H(f)=1 can keep the
signal in its form
38
CONSTELLATION
ssT((tt)) r (t )
H( f )
Gr ( f ) = H ( f ) Gs ( f )
2
39
CONSTELLATION
ssT((tt)) r (t )
H( f )
Gr ( f ) = H ( f ) Gs ( f )
2
40
CONSTELLATION
Let us consider
2
B = b1 (t ) = PT (t ) cos ( 2 f 0t )
T
T
M = {s1 = (+ ) , s2 = (− )} =A
2
41
CONSTELLATION 2
+
The transmitted signal is given by s (t ) = a[n]p (t − nT )
n =0
variance a2 = 0.5 ( 2 + 2 ) = 2 = A2
T
2
42
CONSTELLATION 2
2
Where p(t ) = b1 (t ) = PT (t ) cos ( 2 f 0t )
T
sin( fT ) − j f 1
has Fourier transform P( f ) = 2T e ( ( f − f ) + ( f + f ) )
=
( fT ) 2
T sin( ( f − f 0 )T ) sin( ( f + f 0 )T ) − j f
= + e
2 ( ( f − f 0 )T ) ( ( f + f 0 )T )
43
CONSTELLATION 2
It follows
T sin( ( f − f 0 )T )
2
sin( ( f + f 0 )T )
2
P ( f ) = +
2
2 ( ( f − f 0 )T ) ( ( f + f 0 )T )
1 2 sin( ( f − f 0 )T ) sin( ( f + f 0 )T )
2 2
Gs ( f ) = A T +
4
( ( f − f 0 )T ) ( ( f + f 0 )T )
44
CONSTELLATION 2
1 2 sin( ( f − f 0 )T ) sin( ( f + f 0 )T )
2 2
Gs ( f ) = A T +
4 ( ( f − f 0 )T ) ( ( f + f 0 )T )
Gs ( f )
Gv(f) Gv(f)
− f0 f0
1 2 1 1
T T T T
45
CONSTELLATION 2
46
LINEAR MODULATION
In general, given s (t ) = a[n]p (t − nT )
n
2
P( f )
with spectrum Gs ( f ) = a2
T
1
Its spectrum is Gs ' ( f ) = [Gs ( f − f 0 ) + Gs ( f + f 0 )]
4
47
CONSTELLATION 3
Let us consider
sin ( t / T ) sin ( t / T )
M = s1 (t ) = + A , s2 (t ) = − A
( t / T ) ( t / T )
1 sin ( t / T )
B = b1 (t ) =
T ( t / T )
48
CONSTELLATION 3
+
The transmitted signal is given by s (t ) = a[n]p (t − nT )
n =0
49
CONSTELLATION 3
1 sin ( t / T )
p(t ) =
Where T ( t / T )
50
CONSTELLATION 3
1 sin ( t / T ) rect L ( x)
p(t ) =
T ( t / T )
P ( f ) = T rect 1 ( f )
T
x
P ( f ) = T rect 1 ( f )
2
L L
−
T 2 2
51
CONSTELLATION 3
Gv ( f )
1 1
−
2T 2T
1/ T
It is a baseband spectrum
52