Scribd
Upload
73 views3 pages

Baseband Signal Representations: Analytic-Equivalent Signals

1) The input signal x(t) is represented as a sum of shifted and scaled pulse shapes modulated by symbols xk. 2) The output signal y(t) is equal to the input signal plus additive white Gaussian noise z(t). 3) The signal is transformed to the complex baseband representation xBB(t) which has half the bandwidth of the original signal x(t). This is done by shifting the positive frequency components of the analytic signal xA(t) to baseband.

Uploaded by

maheshwaran
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
73 views3 pages

Baseband Signal Representations: Analytic-Equivalent Signals

1) The input signal x(t) is represented as a sum of shifted and scaled pulse shapes modulated by symbols xk. 2) The output signal y(t) is equal to the input signal plus additive white Gaussian noise z(t). 3) The signal is transformed to the complex baseband representation xBB(t) which has half the bandwidth of the original signal x(t). This is done by shifting the positive frequency components of the analytic signal xA(t) to baseband.

Uploaded by

maheshwaran
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Baseband Signal Representations

At the lecture we used the following signal at the channel input:


x(t) =
N1

k=0
x
k
p(t kT) ,
where N
Ttot/T is the number of transmitted symbols and
p(t)
sin
_
t
T
_
t
T
F
P(f) =
_
1
2W
|f| W
0 o.w.
,
W
1
2T
.
The corresponding output signal is, therefore,
y(t) = x(t) + z(t)
=
N1

k=0
x
k
p(t kT) + z(t) ,
where z(t) is additive white Gaussian noise (AWGN) of power-spectrum density (PSD) height
N
0/2.
Note that for the representation of the signal X(f), we use a (double-sided) band of width
2W. Nonetheless, since x(t) is real, its Fourier transform X(f) is complex-conjugate, viz. satises
X(f) = X

(f) . (1)
This implies, in turn, that the spectrum over (all) positive frequencies contains all the infor-
mation needed for the reconstruction (and hence also for the representation) of X(f) and x(t).
Thus, it suces to specify the values over only the non-negative frequencies, which occupy a
band of width W. This observation is discussed formally in the next section.
Analytic-Equivalent Signals
Let us now dene the lter to be used in the sequel.
Denition (Phase splitter). The phase splitter, denoted by (f), is dened as
(f)
_

2 f > 0

2
2
f = 0
0 f < 0
.
1
Remark. The phase splitter lter is closely related to the Hilbert lter H
H
(f) j sign(f),
where the later, being conjugate-symmetric, is real-valued in time domain.
Denition (Analytic-equivalent signal). Let x(t) be a real-valued process. Then, its analytic-
equivalent x
A
(t) is dened, via its Fourier transform, as
X
A
(f) (f)X(f)
_

2X(f) f > 0

2
2
X(f) f = 0
0 f < 0
. (2)
Remark. The

2 factor was added so that X


A
(f) and X(f) have the same power.
Using the fact that X(f) is conjugate-symmetric (1) and the denition of the analytic-
equivalent (2) we arrive at the following relation

2X(f) =
_

2X(f) f > 0

2X

(f) f < 0

2
2
X(f) +

2
2
X

(f)
. .
=X(f=0)
f = 0
=
_

2X(f) +
=0
..

2X

(f) f > 0

2X

(f) +
=0
..

2X(f) f < 0

2
2
X(f) +

2
2
X

(f) f = 0
=
_

_
X
A
(f) f > 0
X

A
(f) f < 0
X
A
(f) + X

A
(f) f = 0
= X
A
(f) + X

A
(f) .
(3)
Remark. From (3) it is evident why the DC component of the analytic-equivalent signal, X
A
(0),
was dened with an additional
1
/2 factor.
Thus, using the Fourier property
a

(t)
F
A

(f) ,
we obtain

2x(t) = x
A
(t) + x

A
(t) = 2 Re {x
A
(t)}
x(t) =

2 Re {x
A
(t)} . (4)
Complex Baseband Representation
We now aim at representing x(t) as a complex baseband signal of half of the bandwidth of
X(f). For this, we treat X
A
(f) as a passband signal (of single-side bandwidth W) with carrier
frequency f
C
=
W
/2, and shift it to baseband.
2
Denition (Complex baseband signal). Dene the complex baseband signal as
X
BB
(f)

2X
A
(f + f
C
)
=
_

2X(f + f
c
) f > f
C

2
2
X(f + f
c
) f = f
C
0 o.w.
.
(5)
In time domain, (5) reduces to
x
BB
(t) = x
A
(t)e
j2f
C
(6)
= [(t) x(t)] e
j2f
C
x
I
BB
(t) + jx
Q
BB
(t) ,
where x
I
BB
(t) and x
Q
BB
(t) are real-valued process and denotes convolution.
Using (4) and (6), we attain
x(t) =

2 Re {x
A
(t)} (7)
=

2 Re
_
x
BB
(t)e
+j2f
C
_
=

2 Re
__
x
I
BB
(t) + jx
Q
BB
(t)
__
cos(2f
C
t) + j sin(2f
C
t)
__
=

2x
I
BB
(t) cos(2f
C
t)
. .
phase=0

2x
Q
BB
(t) sin(2f
C
t)
. .
phase=

/2
.
We see from (7) why x
I
BB
(t) and x
Q
BB
(t) are referred to as the in-phase and quadrature
components, respectively.
Thus, x
BB
(t) is a complex representation of (single-side) bandwidth W
BB
=
W
/2, i.e., half of
the bandwidth of x(t). Hence, instead of the N real channel uses, we exploit only N
complex
=
1
2
N =
Ttot/2T complex channel uses.
Remark. Note that the analytic-equivalent representation of the noise z(t) is complex Gaussian
with PSD height of N
0
.
Finally, we note that the derivation in this section applies also to passband signals, with
an appropriate choice of of f
C
. In fact, as was mentioned at the beginning of this section, we
treated the real baseband signal as a passband signal of bandwidth W and carrier frequency
f
C
=
W
/2.
3

You might also like