Lec9 Error Probability
Lec9 Error Probability
Lec9 Error Probability
Digital Telecommunications
PART II
BASEBAND DIGITAL
TRANSMISSION
Lecture #9
Optimum Digital Receivers
Error Performance
Introduction
Points to be discussed in this lecture
Probability of Bit Error (BER)
Error Performance for Line Codes
Introduction
Reading Assignment
Simon Haykin, Digital Communications,
J ohn Wiley & Sons, Inc., 1988, Chapter 6,
Sec. 6.2.
Simon Haykin, Communication Systems, 4
th
Ed., J ohn Wiley & Sons, Inc., 2001, Chapter
3, Sec. 3.7.
Probability of Bit Error (BER)
One-Dimensional Signal Set
Recall that
( ) ( ) ( )
( ) ( ) ( )
1 11 1 12 2
2 21 1 22 2
s t s t s t
s t s t s t
= +
= +
0
b
t T
A signal set is one-dimensional when
( )
2
0 t = ,
12 22
0 s s = = .
2 21
s = s
1 11
s = s
( )
1
t
Probability of Bit Error (BER)
One-Dimensional Signal Set
The MAP decision rule reduces to
( )
( )
( )
1
2 1
2 1 2 1 1
1
ln
2
o
p
E E N T
p s s
+
1
1 2 1 11 21
2
1
2 ln
o
p
E E r s s N
p
r
< +
<
The receiver computes and compares it with a
threshold T for decision-making.
1
r
Probability of Bit Error (BER)
One-Dimensional Signal Set
Probability of making an error is given by
( ) ( )
( )
( )
( )
( )
1 1
1 1 1 1 2 2
1 1 1 1 2 1 2 1
| |
| |
e
T
R R
T
p P r T s t p P r T s t p
p f r s t dr p f r s t dr
= > + <
= +
11
s
21
s
T
( )
1
1 2
|
R
f r s
( )
1
1 1
|
R
f r s
1
r
Probability of Bit Error (BER)
One-Dimensional Signal Set
Substituting for
( )
( )
1
1 1
|
R
f r s t yields
( )
( )
( )
1
2
1 1 1 1 11 1
0
0
1 1
| exp
R
T T
f r s t dr r s dr
N
N
=
Letting
1 11 1
/2 /2
o o
r s dr
z dz
N N
= =
( )
( )
11
1
2
2
11
1 1 1
1
| exp
2
2 /2
T s
N
o
R
T
o
z T s
f r s t dr dz Q
N
= =
Probability of Bit Error (BER)
One-Dimensional Signal Set
Likewise
( )
( )
( )
1
2
1 2 1 1 21 1
0
0
21
1 1
| 1 exp
1
/2
T
R
T
o
f r s t dr r s dr
N
N
T s
Q
N
=
where
[ ]
2
1
exp
2
2
x
z
Q x dz
Probability of Bit Error (BER)
One-Dimensional Signal Set
For the case where
11 21
1 2
,
2
s s
p p p T
+
= = =
11 21
Distance between signal vectors
2 Noise RMS value
2 /2
e
o
s s
p Q Q
N
= =
( )
r t
0
b
T
dt
( )
1
t
b
t T =
1
r
Decision
device
1 1
2 1
if
if
b r T
b
b r T
<
=
>
Threshold T
Probability of Bit Error (BER)
One-Dimensional Signal Set
Example: Consider the signal set
{ }
1 2
( ), ( ) s t s t of
antipodal signals below. Derive the expression
for the BER for an optimum receiver assuming
an AWGN channel and
1 2
p p = .
A
( )
1
s t
t t
A
A A
b
T
0
b
T
0
( )
2
s t
Probability of Bit Error (BER)
Two-Dimensional Signal Set
For an arthogonal signal set
( ) ( ) ( )
( ) ( ) ( )
1 11 1 12 2
2 21 1 22 2
s t s t s t
s t s t s t
= +
= +
11 1 12
21 22 2
, 0
0,
s E s
s s E
= =
= =
For a special case where
1 2
p p p and
1 2
E E = = =
Probability of Bit Error (BER)
Two-Dimensional Signal Set
In this case, the MAP decision rule becomes
1
1 2
set if b b
r r
=
>
Defining a new Gaussian random variable L with
sample value
2
l r r
1
= , the decision rule becomes
1
1 2
set if
0
b b
l r r
=
= >
Probability of Bit Error (BER)
Two-Dimensional Signal Set
( )
( )
[ ]
1 11
2 22
|
|
b
b
o
E L s t s E
E L s t s E
Var L N
= =
= =
=
( )
( )
( )
2
1
1
| exp
2
2
b
L
o
o
l E
f l s t
N
N
=
( )
( )
( )
2
2
1
| exp
2
2
b
L
o
o
l E
f l s t
N
N
+
=
Probability of Bit Error (BER)
Two-Dimensional Signal Set
Probability of making an error is given by
22
s
11
s 0 T =
( ) ( )
1
|
L
f l s t
( ) ( )
2
|
L
f l s t
l
( ) ( )
( )
( )
( )
( )
1 2
0
1 2
0
1 1
0| 0|
2 2
1 1
| |
2 2
e
L L
p P l s t P l s t
f l s t dl f l s t dl
= < + >
= +
Probability of Bit Error (BER)
Two-Dimensional Signal Set
( )
( )
1
|
L
f l s t and
( )
( )
2
|
L
f l s t yields Substituting for
Distance between signal vectors
2 Noise RMS value
b
e
o
E
p Q Q
N
= =
1
r
( )
1 b
T t
( )
r t
b
t T =
2
r
Decision
device
b
t T =
Threshold T=0
l
1 1
2 1
if
if
b r T
b
b r T
<
=
>
( )
2 b
T t
Probability of Bit Error (BER)
/
b o
E N is the standard quality measure for digital
communication system performance.
[ ]
Q x is a monotonically decreasing function of
. Hence x
e
p decreases with the increase in /
b o
E N .
For fixed channel noise psd,
o
/
b
E N is increased by
increasing
b
E corresponding to an increase in the
Euclidean distance between signals
( )
1
s t and
( )
2
s t .
Different modulation methods can be compared
based on the required /
o b
E N for the fixed
e
p .
Error Performance for Line Codes
Error performance for the line codes is determined
by probability of bit error for an AWGN channel.
The signal is corrupted by additive white Gaussian
noise of zero-mean and power spectral density
(psd) /2.
o
N
The two bits are assumed to be equally likely, i.e.,
1 2
p p = .
For comparison purpose, the amplitude of each
pulse is the same for all line codes.
Error Performance for Line Codes
(i) Unipolar NZR Line Codes
Signal set
( ) ( )
{ }
1 2
, s t s t
( )
1
, 0
0, elsewhere
b
A t T
s t
( )
2
0 s t =
Orthonormal basis function
( )
1
t for
representation of the signal set is given by
( )
1
1
, 0
0, elsewhere
b
b
t T
t T
Error Performance for Line Codes
Signal vectors and
1
s
2
s
1 1
E
=
s ,
[ ]
2
2 1
0 where
b
E A T = = s
Signal constellation diagram for the unipolar NZR
line code is one-dimensional.
2
0 = s
1 1
E = s
/2 T E =
1
Z
( )
1
t
2
Z
Error Performance for Line Codes
For an AWGN channel, the probability of bit error
e
p
is expressed as
avb 1
2
e
o o
E E
p Q Q
N N
= =
,
1
avb
0
2
E
E
+
= is the average
energy per bit.
The expression for
e
p applies for the unipolar RZ
line code.
b
t T =
( )
1
t
1
r
1
Threshold T= E /2
Decision
device
1
1
bit 1 if
bit 0 if
r T
b
r T
>
=
<
( )
r t
0
b
T
dt
Error Performance for Line Codes
(ii) Polar NZR Line Codes
Signal set
( ) ( )
{ }
1 2
, s t s t
( )
1
, 0
0, elsewhere
b
A t T
s t
( )
2
, 0
0, elsewhere
b
A t T
s t
Orthonormal basis function
( )
1
t for
representation of the signal set is given by
( )
1
1
, 0
0, elsewhere
b
b
t T
s t T
Error Performance for Line Codes
Signal vectors and
1
s
2
s
1 1
E
=
s ,
2
2 2 1 2
where
b
E E E A T
= = =
s
Signal constellation diagram for the polar NZR line
code is one-dimensional.
2 2
E = s
1 1
E = s
0 T =
1
Z
( )
1
t
2
Z
Error Performance for Line Codes
For an AWGN channel, the probability of bit error
e
p
is expressed as
avb 1
2 2
e
o o
E E
p Q Q
N N
= =
The expression for
e
p applies for the polar RZ and
Manchester line codes.
b
t T =
( )
1
t
1
r
Threshold T=0
Decision
device
1
1
bit 1 if
bit 0 if
r T
b
r T
>
=
<
( )
r t
0
b
T
dt
Error Performance for Line Codes
(iii) Bipolar NZR Line Codes
Signal set
( ) ( ) ( )
{ }
1 2 3
, , s t s t s t
( )
1
, 0
0, elsewhere
b
A t T
s t
,
( )
2
0 s t = ,
( )
3
, 0
0, elsewhere
b
A t T
s t
Orthonormal basis function
( )
1
t for
representation of the signal set is given by
( )
1
1
, 0
0, elsewhere
b
b
t T
s t T
Error Performance for Line Codes
Signal vectors , and
1
s
2
s
3
s
1
E
=
s ,
[ ]
2
2 3
0 , , where
b
E E A T
= = =
s s
Signal constellation diagram for the bipolar NZR
line code is one-dimensional.
2
Z
2
0 = s
3
Z
1
Z
1
E = s
3
E = s
( )
1
t
Error Performance for Line Codes
For an AWGN channel, the probability of bit error
e
p
is expressed as
avb 1
3 3
2 2 2
e
o o
E E
p Q Q
N N
= =
( )
1
t
( )
r t
0
b
T
dt
b
t T =
1
r
Decision
device
1
1
bit 1 if
bit 0 if
r T
b
r T
>
=
<
Threshold T= /2 E