CHAPTER 6

Problem 6.1 :
Using the relationship r(t) =

N
n=1
r
n
f
n
(t) and s(t; ) =

N
n=1
s
n
()f
n
(t) we have :
1
N
0
_
[r(t) s(t; )]
2
dt =
1
N
0
_
_

N
n=1
(r
n
s
n
())f
n
(t)
_
2
dt
=
1
N
0
_
N
n=1

N
m=1
(r
n
s
n
())(r
m
s
m
())f
n
(t)f
m
(t)dt
=
1
N
0

N
n=1

N
m=1
(r
n
s
n
())(r
m
s
m
())
mn
=
1
N
0

N
n=1
(r
n
s
n
())(r
n
s
n
())
=
1
2
2

N
n=1
[r
n
s
n
()]
2
where we have exploited the orthonormality of the basis functions f
n
(t) :
_
T
0
f
n
(t)f
m
(t)dt =
mn
and
2
=
N
0
2
.
Problem 6.2 :
A block diagram of a binary PSK receiver that employs match ltering is given in the following
gure :

Carrier phase
recovery
Symbol
synchronization
Matched lter
h(t) = g(T t)
X
Sampler and
Detector
Output data
Received signal
cos(2f
c
t +

)

As we note, the received signal is, rst, multiplied with cos(2f

c
t +

) and then fed the matched
lter. This allows us to have the lter matched to the baseband pulse g(t) and not to the
passband signal.
122
If we want to have the lter matched to the passband signal, then the carrier phase estimate is
fed into the matched lter, which should have an impulse response:
h(t) = s(T t) = g(T t)cos(2f
c
(T t) +

)
= g(T t)[cos(2f
c
T)cos(2f
c
t +

) +sin(2f
c
T)sin(2f
c
t +

)
= g(T t)cos(2f
c
t +

) = g(T t)cos(2f
c
t

)
where we have assumed that f
c
T is an integer so that : cos(2f
c
T) = 1, sin(2f
c
T) = 0. As we
note, in this case the impulse response of the lter should change according to the carrier phase
estimate, something that is dicult to implement in practise. Hence, the initial realization
(shown in the gure) is preferable.
Problem 6.3 :
(a) The closed loop transfer function is :
H(s) =
G(s)/s
1 +G(s)/s
=
G(s)
s +G(s)
=
1
s
2
+

2s + 1
The poles of the system are the roots of the denominator, that is

1,2
=

2 4
2
=
1

2
j
1

2
Since the real part of the roots is negative, the poles lie in the left half plane and therefore, the
system is stable.
(b) Writing the denominator in the form :
D = s
2
+ 2
n
s +
2
n
we identify the natural frequency of the loop as
n
= 1 and the damping factor as =
1

2
Problem 6.4 :
(a) The closed loop transfer function is :
H(s) =
G(s)/s
1 +G(s)/s
=
G(s)
s +G(s)
=
K

1
s
2
+s +K
=
K

1
s
2
+
1

1
s +
K

1
The gain of the system at f = 0 is :
|H(0)| = |H(s)|
s=0
= 1
123
(b) The poles of the system are the roots of the denominator, that is

1,2
=
1

1 4K
1
2
1
In order for the system to be stable the real part of the poles must be negative. Since K
is greater than zero, the latter implies that
1
is positive. If in addition we require that the
damping factor =
1
2

1
K
is less than 1, then the gain K should satisfy the condition :
K >
1
4
1
Problem 6.5 :
The transfer function of the RC circuit is :
G(s) =
R
2
+
1
Cs
R
1
+R
2
+
1
Cs
=
1 +R
2
Cs
1 + (R
1
+R
2
)Cs
=
1 +
2
s
1 +
1
s
From the last equality we identify the time constants as :

2
= R
2
C,
1
= (R
1
+R
2
)C
Problem 6.6 :
Assuming that the input resistance of the operational amplier is high so that no current ows
through it, then the voltage-current equations of the circuit are :
V
2
= AV
1
V
1
V
2
=
_
R
1
+
1
Cs
_
i
V
1
V
0
= iR
where, V
1
, V
2
is the input and output voltage of the amplier respectively, and V
0
is the signal
at the input of the lter. Eliminating i and V
1
, we obtain :
V
2
V
1
=
R
1
+
1
Cs
R
1 +
1
A

R
1
+
1
Cs
AR
If we let A (ideal amplier), then :
V
2
V
1
=
1 +R
1
Cs
RCs
=
1 +
2
s

1
s
124
Hence, the constants
1
,
2
of the active lter are given by :

1
= RC,
2
= R
1
C
Problem 6.7 :
In the non decision-directed timing recovery method we maximize the function :

L
() =

m
y
2
m
()
with respect to . Thus, we obtain the condition :
d
L
()
d
= 2

m
y
m
()
dy
m
()
d
= 0
Suppose now that we approximate the derivative of the log-likelihood
L
() by the nite dier-
ence :
d
L
()
d


L
( +)
L
( )
2
Then, if we substitute the expression of
L
() in the previous approximation, we obtain :
d
L
()
d
=

m
y
2
m
( +)

m
y
2
m
( )
2
=
1
2

m
_
__
r(t)g(t mT )dt
_
2

__
r(t)g(t mT +)dt
_
2
_
where g(t) is the impulse response of the matched lter in the receiver. However, this is the
expression of the early-late gate synchronizer, where the lowpass lter has been substituted by
the summation operator. Thus, the early-late gate synchronizer is a close approximation to the
timing recovery system.
Problem 6.8 :
An on-o keying signal is represented as :
s
1
(t) = Acos(2f
c
t +
c
), 0 t T (binary 1)
s
2
(t) = 0, 0 t T (binary 0)
Let r(t) be the received signal, that is r(t) = s(t;
c
) + n(t) where s(t;
c
) is either s
1
(t) or
s
2
(t) and n(t) is white Gaussian noise with variance
N
0
2
. The likelihood function, that is to be
maximized with respect to
c
over the inteval [0, T], is proportional to :
(
c
) = exp
_

2
N
0
_
T
0
[r(t) s(t;
c
)]
2
dt
_
125
Maximization of (
c
) is equivalent to the maximization of the log-likelihood function :

L
(
c
) =
2
N
0
_
T
0
[r(t) s(t;
c
)]
2
dt
=
2
N
0
_
T
0
r
2
(t)dt +
4
N
0
_
T
0
r(t)s(t;
c
)dt
2
N
0
_
T
0
s
2
(t;
c
)dt
Since the rst term does not involve the parameter of interest
c
and the last term is simply a
constant equal to the signal energy of the signal over [0, T] which is independent of the carrier
phase, we can carry the maximization over the function :
V (
c
) =
_
T
0
r(t)s(t;
c
)dt
Note that s(t;
c
) can take two dierent values, s
1
(t) and s
2
(t), depending on the transmission
of a binary 1 or 0. Thus, a more appropriate function to maximize is the average log-likelihood

V (
c
) =
1
2
_
T
0
r(t)s
1
(t)dt +
1
2
_
T
0
r(t)s
2
(t)dt
Since s
2
(t) = 0, the function

V (
c
) takes the form :

V (
c
) =
1
2
_
T
0
r(t)Acos(2f
c
t +
c
)dt
Setting the derivative of

V (
c
) with respect to
c
equal to zero, we obtain :

V (
c
)

c
= 0 =
1
2
_
T
0
r(t)Asin(2f
c
t +
c
)dt
= cos
c
1
2
_
T
0
r(t)Asin(2f
c
t)dt + sin
c
1
2
_
T
0
r(t)Acos(2f
c
t)dt
Thus, the maximum likelihood estimate of the carrier phase is :

c,ML
= arctan
_ _
T
0
r(t) sin(2f
c
t)dt
_
T
0
r(t) cos(2f
c
t)dt
_
Problem 6.9 :
(a) The wavelength is :
=
3 10
8
10
9
m =
3
10
m
Hence, the Doppler frequency shift is :
f
D
=
u

=
100 Km/hr
3
10
m
=
100 10
3
10
3 3600
Hz = 92.5926 Hz
126
The plus sign holds when the vehicle travels towards the transmitter whereas the minus sign
holds when the vehicle moves away from the transmitter.
(b) The maximum dierence in the Doppler frequency shift, when the vehicle travels at speed
100 km/hr and f = 1 GHz, is :
f
D
max
= 2f
D
= 185.1852 Hz
This should be the bandwith of the Doppler frequency tracking loop.
(c) The maximum Doppler frequency shift is obtained when f = 1 GHz + 1 MHz and the
vehicle moves towards the transmitter. In this case :

min
=
3 10
8
10
9
+ 10
6
m = 0.2997 m
and therefore :
f
D
max
=
100 10
3
0.2997 3600
= 92.6853 Hz
Thus, the Doppler frequency spread is B
d
= 2f
D
max
= 185.3706 Hz.
Problem 6.10 :
The maximum likelihood phase estimate given by (6-2-38) is :

ML
= tan
1
Im
_

K1
n=0
I

n
y
n
_
Re
_

K1
n=0
I

n
y
n
_
where y
n
=
_
(n+1)T
nT
r(t)g

(t nT)dt. The Re(y

n
), Im(y
n
) are statistically independent compo-
nents of y
n
. Since r(t) = e
j

n
I
n
g(t nT) + z(t) it follows that y
n
= I
n
e
j
+z
n
, where the
pulse energy is normalized to unity. Then :
K1

n=0
I

n
y
n
=
K1

n=0
_
|I
n
|
2
e
j
+I

n
z
n
_
Hence :
E
_
Im
_
K1

n=0
_
|I
n
|
2
e
j
+I

n
z
n
_
__
= K

I
n

2
sin
and
E
_
Re
_
K1

n=0
_
|I
n
|
2
e
j
+I

n
z
n
_
__
= K

I
n

2
cos
127
Consequently : E
_

ML
_
= tan
1 sin
cos
= , and hence,

ML
is an unbiased estimate of the
true phase .
Problem 6.11 :
The procedure that is used in Sec. 5-2-7 to derive the pdf p(
r
) for the phase of a PSK signal
may be used to determine the pdf p(

ML
). Specically, we have :

ML
= tan
1
Im
_

K1
n=0
I

n
y
n
_
Re
_

K1
n=0
I

n
y
n
_
where y
n
=
_
(n+1)T
nT
r(t)g

(t nT)dt and r(t) = e

j

n
I
n
g(t nT) +z(t). Substitution of r(t)
into y
n
yields : y
n
= I
n
e
j
+z
n
. Hence :
K1

n=0
I

n
y
n
= e
j
K1

n=0
|I
n
|
2
+
K1

n=0
I

n
z
n
U +jV = Ce
j
+z = C cos +x + j(y C sin )
where C =

K1
n=0
|I
n
|
2
and z =

K1
n=0
I

n
z
n
= x +jy. The random variables (x,y) are zero-mean,
Gaussian random variables with variances
2
.Hence :
p(U, V ) =
1
2
2
e
[(UC cos )
2
(V C sin )
2
]
By dening R =

U
2
+V
2
and

ML
= tan
1 V
U
and making the change in variables, we obtain
p(R,

ML
) and nally, p(

ML
) =
_

0
p(r,

ML
)dr. Upon performing the integration over R, as in
Sec. 5-2-7, we obtain :
p(

ML
) =
1
2
e
2 sin
2

ML
_

0
re
(r

4 cos

ML)
2
/2
dr
where = C
2
/2
2
. The graph of p(

ML
) is identical to that given on page 271, Fig. 5-2-9. We
observe that E(

ML
) = , so that the estimate is unbiased.
Problem 6.12 :
We begin with the log-likelihood function given in (6-2-35), namely :

L
() = Re
__
1
N
0
_
T
0
r(t)s

l
(t)dt
_
e
j
_
128
where s
l
(t) is given as : s
l
(t) =

n
I
n
g(t nT) + j

n
J
n
u(t nT T/2). Again we dene
y
n
=
_
(n+1)T
nT
r(t)s

l
(t nT)dt. Also, let : x
n
=
_
(n+3/2)T
(n+1/2)T
r(t)s

l
(t nT T/2)dt. Then :

L
() = Re
_
e
j
N
0
_

K1
n=0
I

n
y
n
j

K1
n=0
J

n
x
n
__
= Re [Acos +jAsin ]
where A =

K1
n=0
I

n
y
n
j

K1
n=0
J

n
x
n
. Thus :
L
() = Re(A) cos Im(A) sin and :
d
L
()
d
= Re(A) sin Im(A) cos = 0

ML
= tan
1
Im
_

K1
n=0
I

n
y
n
j

K1
n=0
J

n
x
n
_
Re
_

K1
n=0
I

n
y
n
j

K1
n=0
J

n
x
n
_
Problem 6.13 :
Assume that the signal u
m
(t) is the input to the Costas loop. Then u
m
(t) is multiplied by
cos(2f
c
t +

) and sin(2f
c
t +

), where cos(2f
c
t +

) is the output of the VCO. Hence :
u
mc
(t)
= A
m
g
T
(t) cos(2f
c
t) cos(2f
c
t +

) A
m
g
T
(t) sin(2f
c
t) cos(2f
c
t +

)
=
A
m
g
T
(t)
2
_
cos(22f
c
t +

) + cos(

)
_

A
m
g
T
(t)
2
_
sin(22f
c
t +

) sin(

)
_
u
ms
(t)
= A
m
g
T
(t) cos(2f
c
t) sin(2f
c
t +

) A
m
g
T
(t) sin(2f
c
t) sin(2f
c
t +

)
=
A
m
g
T
(t)
2
_
sin(22f
c
t +

) + sin(

)
_

A
m
g
T
(t)
2
_
cos(

) cos(22f
c
t +

)
_
The lowpass lters of the Costas loop will reject the double frequency components, so that :
y
mc
(t) =
A
m
g
T
(t)
2
cos(

) +
A
m
g
T
(t)
2
sin(

)
y
ms
(t) =
A
m
g
T
(t)
2
sin(

)
A
m
g
T
(t)
2
cos(

)
Note that when the carrier phase has been extracted correctly,

= 0 and therefore :
y
mc
(t) =
A
m
g
T
(t)
2
, y
ms
(t) =
A
m
g
T
(t)
2
If the second signal, y
ms
(t) is passed through a Hilbert transformer, then :
y
ms
(t) =
A
m

g
T
(t)
2
=
A
m
g
T
(t)
2
129
and by adding this signal to y
mc
(t) we obtain the original unmodulated signal.
Problem 6.14 :
(a) The signal r(t) can be written as :
r(t) =
_
2P
s
cos(2f
c
t +) +
_
2P
c
sin(2f
c
t +)
=
_
2(P
c
+P
s
) sin
_
2f
c
t + +a
n
tan
1
_

P
s
P
c
__
=
_
2P
T
sin
_
2f
c
t + +a
n
cos
1
_

P
c
P
T
__
where a
n
= 1 are the information symbols and P
T
is the total transmitted power. As it is
observed the signal has the form of a PM signal where :

n
= a
n
cos
1
_

P
c
P
T
_
Any method used to extract the carrier phase from the received signal can be employed at the
receiver. The following gure shows the structure of a receiver that employs a decision-feedback
PLL. The operation of the PLL is described in the next part.

Threshold
t = T
b
_
T
b
0
()dt
cos(2f
c
t +

)
DFPLL
v(t)
(b) At the receiver (DFPLL) the signal is demodulated by crosscorrelating the received signal :
r(t) =
_
2P
T
sin
_
2f
c
t + +a
n
cos
1
_

P
c
P
T
__
+n(t)
with cos(2f
c
t +

) and sin(2f
c
t +

). The sampled values at the ouput of the correlators are :
r
1
=
1
2
_
_
2P
T
n
s
(t)
_
sin(

+
n
) +
1
2
n
c
(t) cos(

+
n
)
r
2
=
1
2
_
_
2P
T
n
s
(t)
_
cos(

+
n
) +
1
2
n
c
(t) sin(

n
)
130
where n
c
(t), n
s
(t) are the in-phase and quadrature components of the noise n(t). If the detector
has made the correct decision on the transmitted point, then by multiplying r
1
by cos(
n
) and
r
2
by sin(
n
) and subtracting the results, we obtain (after ignoring the noise) :
r
1
cos(
n
) =
1
2
_
2P
T
_
sin(

) cos
2
(
n
) + cos(

) sin(
n
) cos(
n
)
_
r
2
sin(
n
) =
1
2
_
2P
T
_
cos(

) cos(
n
) sin(
n
) sin(

) sin
2
(
n
)
_
e(t) = r
1
cos(
n
) r
2
sin(
n
) =
1
2
_
2P
T
sin(

)
The error e(t) is passed to the loop lter of the DFPLL that drives the VCO. As it is seen only
the phase
n
is used to estimate the carrier phase.
(c) Having a correct carrier phase estimate, the output of the lowpass lter sampled at t = T
b
is :
r =
1
2
_
2P
T
sin cos
1
_

P
c
P
T
_
+n
=
1
2
_
2P
T

1
P
c
P
T
+n
=
1
2

2P
T
_
1
P
c
P
T
_
+n
where n is a zero-mean Gaussian random variable with variance :

2
n
= E
_
_
T
b
0
_
T
b
0
n(t)n() cos(2f
c
t +) cos(2f
c
+)dtd
_
=
N
0
2
_
T
b
0
cos
2
(2f
c
t +)dt
=
N
0
4
Note that T
b
has been normalized to 1 since the problem has been stated in terms of the power
of the involved signals. The probability of error is given by :
P(error) = Q
_

2P
T
N
0
_
1
P
c
P
T
_
_
The loss due to the allocation of power to the pilot signal is :
SNR
loss
= 10 log
10
_
1
P
c
P
T
_
When P
c
/P
T
= 0.1, then SNR
loss
= 10 log
10
(0.9) = 0.4576 dB. The negative sign indicates
that the SNR is decreased by 0.4576 dB.
131
Problem 6.15 :
The received signal-plus-noise vector at the output of the matched lter may be represented as
(see (5-2-63) for example) :
r
n
=
_
E
s
e
j(n)
+N
n
where
n
= 0, /2, , 3/2 for QPSK, and is the carrier phase. By raising r
n
to the fourth
power and neglecting all products of noise terms, we obtain :
r
4
n

_

E
s
_
4
e
j4(n)
+ 4
_

E
s
_
3
N
n

E
s
_
3
_

E
s
e
j4
+ 4N
n
_
If the estimate is formed by averaging the received vectors {r
4
n
} over K signal intervals, we have
the resultant vector U = K

E
s
e
j
+ 4

K
n=1
N
n
. Let
4
4. Then, the estimate of
4
is :

4
= tan
1
Im(U)
Re(U)
N
n
is a complex-valued Gaussian noise component with zero mean and variance
2
= N
0
/2.
Hence, the pdf of

4
is given by (5-2-55) where :

s
=
_
K

E
s
_
2
16 (2K
2
)
=
K
2
E
s
16KN
0
=
KE
s
16N
0
To a rst approximation, the variance of the estimate is :

s
=
16
KE
s
/N
0
Problem 6.16 :
The PDF of the carrier phase error
e
, is given by :
p(
e
) =
1

2
e
2
2

Thus the average probability of error is :

P
2
=
_

P
2
(
e
)p(
e
)d
e
=
_

Q
_

2E
b
N
0
cos
2

e
_
p(
e
)d
e
=
1
2

_

_
2E
b
N
0
cos
2
e
exp
_

1
2
_
x
2
+

2
e

__
dxd
e
132
Problem 6.17:
The log-likelihood function of the symbol timing may be expressed in terms of the equivalent
low-pass signals as

L
() =
_
1
N
0
_
T
0
r(t)s
l

(t; )dt
_
=
_
1
N
0
_
T
0
r(t)

n
I
n

(t nT )dt
_
=
_
1
N
0

n
I
n

y
n
()
_
where y
n
() =
_
T
0
r(t)g

(t nT )dt.
A necessary condition for to be the ML estimate of is
d
L
()

= 0
d
d
[

n
I
n

y
n
() +

n
I
n
y
n

()] = 0

n
I
n
d
d
y
n
() +

n
I
n
d
d
y
n

() = 0
If we express y
n
() into its real and imaginary parts : y
n
() = a
n
() + jb
n
(), the above
expression simplies to the following condition for the ML estimate of the timing

n
[I
n
]
d
d
a
n
() +

n
[I
n
]
d
d
b
n
() = 0
Problem 6.18:
We follow the exact same steps of the derivation found in Sec. 6.4. For a PAM signal I
n

= I
n
and J
n
= 0. Since the pulse g(t) is real, it follows that B() in expression (6.4-6) is zero,
therefore (6.4-7) can be rewritten as

L
(, ) = A() cos
where
A() =
1
N
0

I
n
y
n
()
Then the necessary conditions for the estimates of and to be the ML estimates (6.4-8) and
(6.4-9) give

ML
= 0
and

n
I
n
d
d
[y
n
()]
=
ML
= 0
133
Problem 6.19:
The derivation for the ML estimates of and for an oset QPSK signal follow the derivation
found in Sec. 6.4, with the added simplication that, since w(t) = g(t T/2), we have that
x
n
() = y
n
( +T/2).
134