Digital Communications I:

Modulation and Coding Course

Spring - 2013
Jeffrey N. Denenberg
Lecture 4b: Detection of M-ary Bandpass Signals
Lecture 8 2
Last time we talked about:

Some bandpass modulation schemes
M-PAM, M-PSK, M-FSK, M-QAM

How to perform coherent and non-
coherent detection

Lecture 8 3
Example of two dim. modulation
) (
1
t
2
s
1
s
s
E
00
11
) (
2
t
3
s
4
s
10
01 QPSK
3
s
7
s
110
) (
1
t
4
s
2
s
s
E
000
) (
2
t
6
s
8
s
1
s
5
s
001
011
010
101
111
100
8PSK
) (
1
t
) (
2
t
2
s
1
s
3
s
4
s
0000 0001 0011 0010
6
s
5
s
7
s
8
s
10
s
9
s
11
s
12
s
14
s
13
s
15
s
16
s
1 3 -1 -3
1000 1001 1011 1010
1100 1101 1111 1110
0100 0101 0111 0110
1
3
-1
-3
16QAM
Lecture 8 4
Today, we are going to talk about:
How to calculate the average probability
of symbol error for different modulation
schemes that we studied?

How to compare different modulation
schemes based on their error
performances?


Lecture 8 5
Error probability of bandpass modulation
Before evaluating the error probability, it is
important to remember that:
The type of modulation and detection ( coherent or non-
coherent) determines the structure of the decision circuits
and hence the decision variable, denoted by z.
The decision variable, z, is compared with M-1 thresholds,
corresponding to M decision regions for detection purposes.
(
(
(

N
r
r

1
r =
}
T
0
) (
1
t
}
T
0
) (t
N

) (t r
1
r
N
r
r
Decision
Circuits
Compare z
with threshold.
m
Lecture 8 6
The matched filters output (observation vector= ) is
the detector input and the decision variable is a
function of , i.e.
For MPAM, MQAM and MFSK with coherent detection
For MPSK with coherent detection
For non-coherent detection (M-FSK and DPSK),
We know that for calculating the average probability of
symbol error, we need to determine


Hence, we need to know the statistics of z, which
depends on the modulation scheme and the
detection type.
Error probability
) (r f z =
r
r
r = z
r Z = z
| | r = z
sent) | C condition satisfies Pr(z sent) | Z inside lies Pr(
i i i i
s s r
Lecture 8 7
Error probability
AWGN channel model:
The signal vector is deterministic.
The elements of the noise vector are
i.i.d Gaussian random variables with zero-mean and
variance . The noise vector's pdf is


The elements of the observed vector
are independent Gaussian random variables. Its pdf is
) ,..., , (
2 1 iN i i i
a a a = s
) ,..., , (
2 1 N
r r r = r
) ,..., , (
2 1 N
n n n = n
n s r + =
i
2 /
0
N
( )
|
|
.
|

\
|
=
0
2
2 /
0
exp
1
) (
N
N
p
N
n
n
n
t
( )
|
|
.
|

\
|

=
0
2
2 /
0
exp
1
) | (
N
N
p
i
N
i
s r
s r
r
t
Lecture 8 8
Error probability
BPSK and BFSK with coherent detection:
) (
1
t
2
s
1
s
b
E
0 1
b
E
) (
2
t
b
E 2
2 1
= s s
) (
1
t
2
s
1
s
) (
2
t
b
E
b
E
b
E 2
2 1
= s s
0
1
BPSK BFSK

2
0
|
|
.
|

\
|
=
N
E
Q P
b
B

2 /
2 /
0
2 1
|
|
.
|

\
|

=
N
Q P
B
s s

0
|
|
.
|

\
|
=
N
E
Q P
b
B
Lecture 8 9
Error probability
Non-coherent detection of BFSK
}
T
0
) cos( / 2
1
t T e
}
T
0
) (t r
11
r
12
r
}
T
0
}
T
0
21
r
22
r
Decision rule:



) cos( / 2
2
t T e
) sin( / 2
2
t T e
) sin( / 2
1
t T e
( )
2

( )
2

( )
2

( )
2

+
-
z
0 , 0 ) ( if
1 , 0 ) ( if
= <
= >
m T z
m T z
m
2
12
2
11 1
r r z + =
2
22
2
21 2
r r z + =
2 1
z z z =
Decision variable:
Difference of envelopes
Lecture 8 10
Error probability contd
Non-coherent detection of BFSK





Similarly, non-coherent detection of DBPSK
| |
}} }

(

= > =
> = > =
> + > =
0
2 2 2 1 2 1
0
2 2 2 2 2 2 1
2 2 2 1 2 2 1
1 1 2 2 2 1
) | ( ) | ( ) | ( ) , | Pr(
) , | Pr( ) | Pr(
) | Pr(
2
1
) | Pr(
2
1
2
dz z p dz z p dz z p z z z
z z z E z z
z z z z P
z
B
s s s s
s s
s s
|
|
.
|

\
|
=
0
2
exp
2
1
N
E
P
b
B
|
|
.
|

\
|
=
0
exp
2
1
N
E
P
b
B
Rayleigh pdf Rician pdf
Lecture 8 11
Coherent detection of M-PAM
Decision variable:

}
T
0
) (
1
t
ML detector
(Compare with M-1 thresholds)
) (t r
1
r
m

Error probability .
) (
1
t
2
s
1
s
0 g
E 3
00 01
4
s
3
s
11 10
g
E
g
E
g
E 3
4-PAM
1
r z =
Lecture 8 12
Error probability .
Coherent detection of M-PAM .
Error happens if the noise, , exceeds in amplitude
one-half of the distance between adjacent symbols. For symbols
on the border, error can happen only in one direction. Hence:

( )
( ) ( )
g M M e g e
g m m e
E r n P E r n P
M m E r n P
< = = > = =
< < > = =
s s s s
s s
1 1 1 1 1 1
1 1
Pr ) ( and Pr ) (
; 1 2 for | | | | Pr ) (
|
|
.
|

\
|

=
0
2
2
1
log 6 ) 1 ( 2
) (
N
E
M
M
Q
M
M
M P
b
E
g b s
E
M
E M E
3
) 1 (
) (log
2
2

= =
m
r n s =
1 1
Gaussian pdf with
zero mean and variance 2 /
0
N
( ) ( ) ( )
( )
|
|
.
|

\
|

= >

=
< + > + >

= =
}

=
0
1
1 1 1
1
2
) 1 ( 2
) (
) 1 ( 2
Pr
) 1 ( 2
Pr
M
1
Pr
M
1
| | Pr
2
) (
1
) (
1
N
E
Q
M
M
dn n p
M
M
E n
M
M
E n E n E n
M
M
P
M
M P
g
E
n g
g g g
M
m
m e E
g
s
Lecture 8 13
Error probability
Coherent detection
of M-QAM
}
T
0
) (
1
t
ML detector

1
r
}
T
0
) (
2
t
ML detector

) (t r
2
r
m
Parallel-to-serial
converter
s) threshold 1 with (Compare M
s) threshold 1 with (Compare M
) (
1
t
) (
2
t
2
s
1
s
3
s
4
s
0000 0001 0011 0010
6
s
5
s
7
s
8
s
10
s
9
s
11
s
12
s
14
s
13
s
15
s
16
s
1000 1001 1011 1010
1100 1101 1111 1110
0100 0101 0111 0110
16-QAM
Lecture 8 14
Error probability
Coherent detection of M-QAM
M-QAM can be viewed as the combination of two
modulations on I and Q branches, respectively.
No error occurs if no error is detected on either the I or the Q
branch.
Considering the symmetry of the signal space and the
orthogonality of the I and Q branches:
PAM M
branches) Q and I on detected error no Pr( 1 ) ( 1 ) ( = =M P M P
C E
() ( )
2
2
1 I) on error Pr(no
Q) on error I)Pr(no on error Pr(no branches) Q and I on detected error no Pr(
M P
E
= =
=
|
|
.
|

\
|

|
.
|

\
|
=
0
2
1
log 3 1
1 4 ) (
N
E
M
M
Q
M
M P
b
E
Average probability of
symbol error for
PAM M
Lecture 8 15
Error probability
Coherent detection
of MPSK

Compute

Choose
smallest
2
1
arctan
r
r
|

| | |
i
}
T
0
) (
1
t
}
T
0
) (
2
t
) (t r
1
r
2
r
m

3
s
7
s
110
) (
1
t
4
s
2
s
s
E
000
) (
2
t
6
s
8
s
1
s
5
s
001
011
010
101
111
100
8-PSK
Decision variable
r Z = =|

z
Lecture 8 16
Error probability
Coherent detection of MPSK
The detector compares the phase of observation vector to M-1
thresholds.
Due to the circular symmetry of the signal space, we have:


where


It can be shown that

| |
t
t
|
d p P P
M
M P M P
M
M
c
M
m
m c C E
) ( 1 ) ( 1 ) (
1
1 ) ( 1 ) (
/
/
1
1
}

=
= = = = s s
|
|
.
|

\
|
|
.
|

\
|
~
M N
E
Q M P
s
E
t
sin
2
2 ) (
0
( )
|
|
.
|

\
|
|
.
|

\
|
~
M N
E M
Q M P
b
E
t
sin
log 2
2 ) (
0
2
or
2
| | ; sin exp ) cos(
2
) (
2
0 0

t
| | |
t
|
|
s
|
|
.
|

\
|
~
N
E
N
E
p
s s
Lecture 8 17
Error probability
Coherent detection of M-FSK
(
(
(

M
r
r

1
r =
}
T
0
) (
1
t
}
T
0
) (t
M

) (t r
1
r
M
r
r
ML detector:
Choose
the largest element
in the observed vector

m

Lecture 8 18
Error probability
Coherent detection of M-FSK
The dimension of the signal space is M. An upper
bound for the average symbol error probability can be
obtained by using the union bound. Hence:


or, equivalently

( )
|
|
.
|

\
|
s
0
1 ) (
N
E
Q M M P
s
E
( )
( )
|
|
.
|

\
|
s
0
2
log
1 ) (
N
E M
Q M M P
b
E
Lecture 8 19
Bit error probability versus symbol error
probability
Number of bits per symbol
For orthogonal M-ary signaling (M-FSK)




For M-PSK, M-PAM and M-QAM

2
1
lim
1
2 /
1 2
2
1
=

E
B
k
k
k
E
B
P
P
M
M
P
P
M k
2
log =

1 for << ~
E
E
B
P
k
P
P
Lecture 8 20
Probability of symbol error for binary
modulation
E
P
dB /
0
N E
b
Note!
The same average symbol
energy for different sizes of
signal space
Lecture 8 21
Probability of symbol error for M-PSK
E
P
dB /
0
N E
b
Note!
The same average symbol
energy for different sizes of
signal space
Lecture 8 22
Probability of symbol error for M-FSK
E
P
dB /
0
N E
b
Note!
The same average symbol
energy for different sizes of
signal space
Lecture 8 23
Probability of symbol error for M-PAM
E
P
dB /
0
N E
b
Note!
The same average symbol
energy for different sizes of
signal space
Lecture 8 24
Probability of symbol error for M-
QAM
E
P
dB /
0
N E
b
Note!
The same average symbol
energy for different sizes of
signal space
Lecture 8 25
Example of samples of matched filter output
for some bandpass modulation schemes