ECE 414 Tutorial 1

Review:
Random process
Fourier transform ou e t a s o
2012/1/19 1
I f ti Information
Tutorials: Friday 1:30-2:20 p.m., E5-4106
Office hour: Friday 2:30-3:30 p.m., EIT 4126
Contact: [email protected]
Tutorial materials available on-line before tutorial
2012/1/19 2
R d P (1) Random Process (1)
Random variable: outcome of the experiment is a Random variable: outcome of the experiment is a
value;
Random process: outcome of the experiment is a Random process: outcome of the experiment is a
time function
1. A family of deterministic functions
2. Given , is a random variable
Joint distribution:
0
t t =
) (
0
t x
) ) ( , ) ( ( ) , ; , (
2 2 1 1 2 1 2 1
x t X x t X P t t x x F s s =
Density function:
Statistical description:
) ) ( , ) ( ( ) , ; , (
2 2 1 1 2 1 2 1
x t X x t X P t t x x F s s
2 1
2 1 2 1
2
2 1 2 1
) , ; , (
) , ; , (
x x
t t x x F
t t x x f
c c
c
=
2012/1/19 3
i n t x t x t x
t n known x x x f
n
, ) , , , (
2 1 ) ( , ), ( ), (
2 1
=

R d P (2) Random Process(2)

Mean: at each time instant equals the mean of Mean: at each time instant, equals the mean of
the random variable
ti f f ti ti d t i i )] ( [ ) ( X E
(Auto-)correlation function: describing the
time of function tic determinis a )] ( [ ) ( = t X E t m
X
( ) g
interdependence between
)] ( ) ( [ ) ( t X t X E t t R =
) ( & ) (
2 1
t X t X
)] ( ) ( [ ) , (
2 1 2 1
t X t X E t t R
X
=
2 1 2 1 ) ( ) ( 2 1
) , (
2 1
x d dx x x f x x
t X t X
}}
=
2012/1/19 4
R d P (3) Random Process(3)
Stationary processes: Stationary processes:
Strictly stationary processes
) ( ) ( f f
what does it mean?
) , , , ( ) , , , (
2 1 ) ( ), ( ), ( 2 1 ) ( ), ( ), (
2 1 2 1
n t X t X t X n t X t X t X
x x x f x x x f
n n

A + A + A +
=
Wide-sense stationary processes (WSS)
statistics order second and first the ) (
`

=
X
constnat t m
shifts origin time the as change not do ) ( ) , (
) (

)
`

= t t
X X
X
R t t R
2012/1/19 5
R d P (4) Random Process(4)
Properties of (auto )correlation function ) (t R Properties of (auto-)correlation function ) (t
X
R
0 about function even ) ( ) ( = = t t t
X X
R R
Power spectral density (PSD):
0 at value maximum ) 0 ( | ) ( | = s t t
X X
R R
) ( f S Power spectral density (PSD):
) ( f S
X
Relationship between and
(Winerer-Khintchine Theorem):
) (t
X
R
) ( f S
X
( )

=
}
}

d e R f S
f
f j
X X
t t
t t
2
2
) ( ) (
2012/1/19 6

=
}

df e f S R
f j
X X
t t
t
2
) ( ) (
R d P (5) Random Process(5)
Properties of PSD
) ( d f S
Properties of PSD

) ( d f S
X
}

= = df f S R P
X X X
) ( ) 0 (
d ) ( ) 0 (
}

d R S

component dc ) ( ) 0 (
}

= t t d R S
X X
function even ) ( ) ( f S f S
X X
=
f f S ll f 0 ) ( >

Response of LTI (Linear Time-Invariant)

systems to random input
f f S
X
all for 0 ) ( >
systems to random input

WSS is ) ( WSS, is ) ( if t Y t X
) ( | ) ( | ) (
2
f S f H f S
2012/1/19 7

) ( | ) ( | ) (
2
f S f H f S
X Y
=
S stem System
input
Impulse response
output
) (t X
) (t Y
) (t h
input
Impulse response
output
) (t
X
R ) ( f H
f
) (t
Y
R
f
) ( f S
X
f
) ( f S
Y
f
2012/1/19 8
F i T f ti (1) Fourier Transformation(1)

=
}

dt e t x f X t x
ft j t 2
) ( ) ( ) (

=
}
}

df e f X t x f X
ft j t 2
) ( ) ( ) (
Example:
1
T T
t

s s

1
Given ( ) , calculate ( ) ?
2 2
0
else
t
x t X f
s s

2012/1/19 9
F i T f ti (2) Fourier Transformation(2)
}

d f
ft j t 2
) ( ) (
}
2
2
T
ft j
d
t
}

= dt e t x f X
ft j t 2
) ( ) (
}

=
2
2
2
T
T
ft j
dt e
t
2
2
2
1
T
ft j
e
f

=
t
2
2
T
f j

t
( )
2 / 2 2 / 2
2
1
fT j fT j
e e
f j
t t
t
=

f j t
) sin( 2
2
fT j e e
f j
e e
fT j fT j
fT j fT j
t
t
t t
t t
=

T
fT
fT
t
t ) sin(
=
x t ) sin(
2012/1/19 10
x
x
x T fT
t
t ) sin(
) sinc( ) sinc( = =
Representations of Bandpass Signals
It i d i bl t d ll b d i l t i l t l i l It is desirable to reduce all bandpass signals to equivalent lowpass signals.
The resulting signals are therefore independent of the carrier frequency and
channel frequency bands.
Bandpass
Lowpass
-f
c
f
c
0 0
Such a representation is useful for
Mathematical convenience: Avoid the use of trigonometric identities in
theoretical derivations This representation can be also viewed as a complex theoretical derivations. This representation can be also viewed as a complex
number or vector and allows efficient system simulation.
Practical implementation: Most practical receivers first convert the
received signal down to complex baseband and then perform all the required g p p q
signal processing at the baseband.
Suppose that a real valued signal has a frequency content concentrated in a Suppose that a real-valued signal has a frequency content concentrated in a
narrowband of frequencies in the neighbourhood of a carrier frequency f
c
( ) ( ) f S t s
FT

Our objective is to develop a mathematical representation for in the baseband
which includes the same exact information embedded in the original signal.
( ) t s
( ) ( ) ( ) f S f U f S 2 =
Considering the symmetrical property, first construct a signal which contains
only the positive frequencies
( ) f U : Unit step function
( ) ( ) ( ) f S f U f S 2 =
+
( ) ( ) ( )
( ) { } ( ) { } f S F f U F
t s t u t s
2
2
1 1
=
=

+
( ) f p
( ) t s
+
: Analytic signal
( ) t s : Hilbert transform of s(t)
Inverse FT
( ) { } ( ) { }
( ) ( )
( ) ( ) j
t s
t
j
t
f S F f U F

+ =

t
o
( ) ( ) t s
t
j
t s
def
=
t

( ) ( ) t s j t s + =
Example: p
s t
( )
= cos(2t f
0
t)
s
+
t
( )
= cos(2t f
0
t) + j sin(2t f
0
t) s
+
t
( )
cos(2t f
0
t) + j sin(2t f
0
t)
Where the Hilbert transform of cos(a t) is sin(a t)
2012/1/19 13
s (t) contains only the positive frequencies but still located at f s
+
(t) contains only the positive frequencies, but still located at f
c
.
Now shift the spectral content to f=0. The resulting signal s
l
(t) is called the
complex envelope of the real signal s(t) and is its equivalent lowpass signal.
( ) ( )
c l
f f S f S + =
+
Inverse FT and define x(t) and y(t) as
the real and imaginary parts of s
l
(t)
( ) ( )
( ) ( ) ( ) ( ) | |
t f j
t f j
l
c
c
e t s j t s t jy t x
e t s t s
t
t
2
2

+
+ +
=
x(t) y(t) :Quadrature components of s(t) ( ) ( ) ( ) ( ) | |
f j
c
e t s j t s t jy t x + = + x(t), y(t) :Quadrature components of s(t)
( ) ( ) ( ) ( ) ( ) t f t y t f t x t s
c c
t t 2 sin 2 cos =
( ) ( ) ( ) ( ) { }
t f j t f j
l
c c
e t jy t x e t s t s
t t 2 2
] [ Re ] Re[ + = =
( ) ( )
( ) | |
{ } ( ) ( ) | | t t f t a e t a t s
c
t t f j
c
u t
u t
+ = =
+
2 cos Re
2
( ) ( ) ( ) ( ) t s t y t x t a
l
def
= + =
2 2
( ) ( ) ( ) ( ) t x t y arctg t
def
= u ( ) ( ) ( ) ( ) t s t y t x t a
l
+ ( ) ( ) ( ) ( ) t x t y arctg t u
( ) ( ) ( ) | | { }
ft j t f j ft j
} }

t t t 2 2 2
( ) ( ) ( ) | | { } dt e e t s dt e t s f S
ft j t f j
l
ft j
c
}
=
}
=



t t t 2 2 2
Re
( ) ( ) ( ) | | f f f f f
*
1
( ) ( )
*
2
1
Re z z z + =
( ) ( ) ( ) | |
c l c l
f f S f f S f S + =
2
1

N id th f b d t t b d i l Now consider the response of a bandpass system to a bandpass signal
( ) t s ( ) t r ( ) ( ) ( ) ( ) ( ) ( ) f S f H f R t s t h t r
FT
= =
( ) t h
BPF
( ) t h
( ) t s
l
( ) t r
l
( ) ( ) ( ) ( ) ( ) ( ) f S f H f R t s t h t r
l l l
FT
l l l
= =
LPF
( ) t h
( ) t h
l
( ) t s
l
( ) t r
l
( ) ( ) ( ) ( ) ( ) ( ) f f f
l l l l l l
where
( ) ( ) ( )
c l c l
f f H f f H f H + =
*
( ) ( ) ( )
c l c l
f f f f f
( ) f S ( ) f S
w
A flat power spectrum contains all frequency
components with equal power weighting and is
designated as white, in an analogy to white light.
2
0
N
( ) f S
n
In general, the bandwidth of the receiver (i.e.
front end filter) is narrower than the bandwidth
limitations of the noise process If the noise has a limitations of the noise process. If the noise has a
flat PSD extending beyond the bandwidth of a
given system, the noise appears to the system as if
it were band-limited and white.
2
0
N
( ) ( ) f S f S
y x
n n
=
0
N
( ) ( ) ( ) ( )
LPF
c n c n n n
f f S f f S f S f S
y x
+ + = =
( ) ( ) ( ) ( ) ( )
FT
0
( ) ( ) ( ) ( ) ( ) t o
0 0
N f R f R N f S f S
y x y x
n n
FT
n n
= = = =
Some material is taken from Chapter 4 of Digital Communications by J. Proakis.