Sampling Theorem

SIGNAL PROCESSING BASICS
A/D AND D/A CONVERSION

SAMPLING
QUANTISATION
CODING

SAMPLER

QUANTIZER

CODER
x
a
(t)
x(n) x
q
(n)
010110- - -
- -

Analog
signal
Discrete
time signal
Quantised
signal
Digital
signal
SAMPLING

Linear
Filter
Sampling
y(nT)
0 0.5 1 1.5 2 2.5 3 3.5 4
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0 0.5 1 1.5 2 2.5 3 3.5 4
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
x(t)
SAMPLING THEOREM
Let x(t) be a band-limited signal with
X(j) = 0 for || >
m
. Then x(t) is uniquely
determined by its samples x(nT),
n = 0, 1, 2, . . .,
if
s
> 2
m
where,
s
= 2/T
NYQUIST RATE
The frequency 2
m
, which, under
the sampling theorem, must be
exceeded by the sampling
frequency, is commonly referred
to as the Nyquist rate.
FILTERS
To eliminate some frequency
components.

IIR Filters

FIR Filters
MULTIRATE DSP
CONCEPT

DECIMATION

INTERPOLATION

SAMPLING RATE CONVERSION

DECIMATION
Decimation operation is implemented by
keeping every M-th sample of x[n] and
removing in-between samples to generate
y[n].

Input-output relation
y[n] = x[nM]

DECIMATION
Figure below shows the decimation by a
factor of 3 of a sinusoidal sequence
0 10 20 30 40 50
-1
-0.5
0
0.5
1
Input Sequence
Time index n
A
m
p
l
i
t
u
d
e
0 10 20 30 40 50
-1
-0.5
0
0.5
1
Output sequence down-sampled by 3
A
m
p
l
i
t
u
d
e
Time index n
INTERPOLATION
Interpolation operation is implemented
by inserting equidistant zero-valued
samples between two consecutive
samples of x[n]
Input-output relation

=
=
otherwise , 0
, 2 , , 0 ], / [
] [
L
L L n L n x
n x
u
INTERPOLATION
Figure below shows the interpolation by
a factor of 3 of a sinusoidal sequence
0 10 20 30 40 50
-1
-0.5
0
0.5
1
Output sequence up-sampled by 3
Time index n
A
m
p
l
i
t
u
d
e
Low pass and band pass
14
I. Phasors (complex envelope) representation for
sinusoidal signal
narrow band signal
II. Complex Representation of Linear Modulated Signals & Bandpass
System
Band Pass Systems, Phasors and Complex Representation of Systems
KEY LEARNING OBJECTIVES
Phasors and Complex Representation are useful for analyzing
baseband component of a signal
eliminates high frequency carrier components
15
x(t) is a narrowband signal (aka bandpass signal) if
X(f) 0 in some small neighborhood of f
0
,

a high frequency
X(f) 0 for | f f
0
| W where W < f
0

f
0
is usually referred to as center frequency, but need not be
center frequency or in signal bandwidth at all
X(f)
2W
-f
0
-W -f
0 -
f
0
+W f
0
-W f
0
f
0
+W
I. Phasors for monochromatic & narrow band signals
h(t) is a Bandpass System,, that passes signals with frequency components in
the neighborhood of some frequency, f
0

H(f) = 1 for | f f
0
| W otherwise H(f) 0
bandpass system h(t) passes a bandpass signal x(t)
X(f) X(f) H(f)
16
output determined by multiplying X & frequency response of
system computed at input frequency, f
0
input & output frequencies are same output phasor gives output
signal
Consider LTI system driven by input x(t)
H(f) X(f)
Y(f)
determine the phasor for sinusoida1 signal and narrowband signal
capture phase and magnitude of base band signal
ignore effects of the carrier
17
z(t) = Aexp(j(2f
0
t + ))
= Acos(2f
0
t + ) + jAsin(2f
0
t + )
= x(t) + jx
q
(t)
(i) define a signal z(t) as a vector rotating with angular frequency 2f
0

1. determination of phasor, X for sinusoidal input signal x(t)
x(t) = Acos(2f
0
t + )
x
q
(t) = Asin(2f
0
t + )
quadrature component shifted 90
o
from x(t)
(ii) obtain phasor X from z(t) by eliminating 2f
0
rotation
- rotate z(t) at an angular frequency = 2f
0
in opposite direction
- equivalent to multiplying z(t) by exp(2f
0
t)
X = z(t) exp(-j2f
0
t ) = Aexp(j(2f
0
t + ))exp(-j2f
0
t )
= Aexp(j)
2f
0
Aexp(j)

R

I

x
q
(t)
x(t)
18
1a. determine Frequency Domain equivalent of z(t) and X
Z(f) =
[cos()(ff
0
) + jsin()(ff
0
)]
2
2A
x(t) = Acos(2f
0
t + ) = Acos()cos(2f
0
t) + Asin()sin(2f
0
t)
X(f) = cos()[(ff
0
) + (f+f
0
)]
2
A
sin()[(f+f
0
) - (f-f
0
)]
2
A
- j
(1) determine X(f) = F[x(t)], delete negative frequencies & multiply by 2
X = Aexp(j)
(ii) then shift Z(f) by f
0

(i) obtain Z(f), using either or two methods
z(t) = Aexp(j(2f
0
t + )) = Aexp(j)exp(j2f
0
t )
Z(f) = Aexp(j)(f f
0
) since F[exp(j2t)] = {(f-)}
(2) determine Z(f) = F[z(t)]
19
z(t) is known as the analytic signal or pre-envelope of x(t)
2. determine phasor for a narrowband signal, x(t)
Z(f) = 2u
-1
(f)X(f)
based on definition of z(t) in sinusoid case: z(t) = x(t) + jx
q
(t)
find Z(f) by deleting negative frequencies of X(f) & multiply result by 2
find z(t) using IFT find signal whose Fourier transform = u
-1
(f)
we know that F[u
-1
(t)] =
f j
f
t
o
2
1
) (
2
1
+
by duality = u
-1
(f)
(
+
t
j
t
t
o
2
) (
2
1
F
by convolution ) ( ) ( t x
t
j
t -
|
.
|
\
|
+
t
o z(t) =
) ( ) ( t x j t x + then z(t) = let
) (
1
) ( t x
t
t x - =
t
20
2
t
phase shift x(t) by for positive frequencies

2
t
+
phase shift x(t) by for negative frequencies
Hilbert Transform of x(t) is given by ) (
1
) ( t x
t
t x - =
t
pre-envelope for two types of signals
(ii) narrowband case
) ( t x z(t) = x(t) + j
z(t) = x(t) + jx
q
(t) (i) sinusoid case
x(t) = Acos(2f
0
t+)
x
q
(t)= Asin(2f
0
t+)
21
determine phasor, x
l
(t) of bandpass signal x(t)
x
l
(t) = low pass representation of x(t)
determined by shifting spectrum of z(t) left by f
0
X
l
(f) = Z(f + f
0
) = 2u
-1
(f + f
0
)X(f + f
0
)
x
l
(t) = z(t)exp(-j2f
0
t)
x
l
(t) is a low pass signal
X
l
(f) 0 for all | f | W
phasor for band pass signal
X(f)
f
0
f
0

f
0

f
Z(f)
2A
f
f
X
l
(f)
2A
A
22
x
l
(t) = x
c
(t) + jx
s
(t)
Generally x
l
(t) is complex signal with real (in phase) & imaginary (quadrature)
components
z(t) = x
l
(t)exp(j2f
0
t)
= [x
c
(t) + jx
s
(t)]exp(j2f
0
t)
= x
c
(t)cos(2f
0
t) - x
s
(t)sin(2f
0
t) + j[x
c
(t)sin(2f
0
t)+x
s
(t)cos(2f
0
t)]
z(t) =
) ( ) ( t x j t x +
rewrite in terms of quadrature & in-phase components
equate real & imaginary parts of z(t) and x
l
(t)
= Im{z(t)} = x
c
(t)sin(2f
0
t)+x
s
(t)cos(2f
0
t)
x(t) = Re{z(t)} = x
c
(t)cos(2f
0
t) - x
s
(t)sin(2f
0
t)
) ( t x
bandpass to lowpass transform describes relationship of x(t) & in terms
of x
c
(t) & x
s
(t)
) ( t x
23
x
l
(t)

R

I

(t)
V(t)
monochromatic phasor has constant amplitude & phase
bandpass signals phase & envelope vary slowly with time vector
representation moves on a curve in the complex plane
V(t) & (t) are slowly time varying
x
l
(t) = V(t)exp( j(t) ) then
=
) ( ) (
2 2
t x t x
s c
+
|
|
.
|
\
|
) (
) (
1 tan
t x
t x
c
s
define envelope of x
l
(t) as V(t) =
) ( ) (
2 2
t x t x
s c
+
(t) =
|
|
.
|
\
|
) (
) (
1 tan
t x
t x
c
s
define phase of x
l
(t) as
Define x
l
(t) in terms of phase & envelope
24
II. Complex Representation of Linear Modulated Signals & Bandpass System
s(t) = s
I
(t)cos(2f
c
t) - s
Q
(t)sin(2f
c
t)
canonical representation of any bandpass signal, s(t) has 2 components
s
I
(t) = in-phase component of s(t)
s
Q
(t) = quadrature component of s(t)
properties of s
I
(t) & s
Q
(t)
are real valued functions
are orthogonal to each other
are uniquely defined in terms of the baseband signal m(t)
two components can be used to synthesize modulated signal s(t)
25
circuit used to synthesize s(t) from s
I
(t) & s
Q
(t)
E
s(t)
cos(2tf
c
t)
sin(2tf
c
t)
90
o

oscillator
s
I
(t)
s
Q
(t)
s
I
(t)
LPF
s(t)
2cos(2tf
c
t)
-2sin(2tf
c
t)
oscillator
90
o

s
Q
(t)
LPF
circuits used to analyze s
I
(t) & s
Q
(t) based on s(t),
26
1. Complex Envelope of a Band-Pass Signal s(t) is given as
s
(t) preserves information content of s(t), except for f

c
(t)
s
(t) = s
I
(t) + js
Q
(t)
s(t) = Re{s
(t)e
(2f
c
t)
}
= s
I
(t)cos(2f
c
t) - s
Q
(t)sin(2f
c
t)
then,
s
(t)e
(2f
c
t)
= [s
I
(t) + js
Q
(t)] [cos(2f
c
t) + jsin(2f
c
t)]
= s
I
(t)cos(2f
c
t) - s
Q
(t)sin(2f
c
t) + j[s
I
(t)sin(2f
c
t)+s
Q
(t)cos(2f
c
t)]
real
imag
27
system is narrowband if bandwidth W << f
c
, the systems center
frequency
input x(t) is modulated by carrier, f
c

output = y(t)
h(t)
x(t) y(t)
2. Consider a narrowband linear band-pass system
x
(t) 2(t)
h
(t)
use equivalent complex baseband model to simplify analysis
impulse response given by
h
(t) = h
I
(t) + jh
Q
(t)
canonical representation of systems impulse response given by:
h(t) = h
I
(t)cos(2f
c
t) - h
Q
(t)sin(2f
c
t)
28
2.1 Passband Analysis of LTI System
y(t) =
}

d t h x ) ( ) (
}

[x
I
()cos(2f
c
)-x
Q
(t)sin(2f
c
)]
[h
I
(t-)cos(2f
c
t-)-h
Q
(t-)sin(2f
c
t-)]d
y(t) =
= x
I
(t) h
I
(t-) cos(2f
c
t)cos(2f
c
t-) d
}

x
I
(t)h
Q
(t-)cos(2f
c
t)sin(2f
c
t-) d
}

-
+ x
Q
(t) h
Q
(t-) sin(2f
c
t)sin(2f
c
t-) d
}

x
Q
(t)h
I
(t-)cos(2f
c
t-)sin(2f
c
t) d -
29
Passband Analysis of LTI System (continued)
}

y(t)
= x
I
(t) h
I
(t-) [ cos() + cos(4f
c
t-) ] d
}

x
I
(t)h
Q
(t-)[ sin(4f
c
t) + sin() ] d
}

-
+ x
Q
(t) h
Q
(t-) [ cos() - cos(4f
c
t-) ] d
x
Q
(t)h
I
(t-)[ sin(4f
c
t) - sin() ] d
}

-
30
complex envelopes are related by complex convolution
2.2 Equivalent Complex Baseband Model
(t) = y
I
(t) + jy
Q
(t) is the complex envelope of y(t)
complex input & output are complex envelopes of bandpass systems
input & output
x
(t) = x
I
(t) + jx
Q
(t) is the complex envelope of x(t)
}

= [x
I
(t) + jx
Q
(t)] [h
I
(t-) + jh
Q
(t-)]d
(t) =
}

d t h x ) (
~
) (
~
2
1
=
}

d h t x ) (
~
) (
~
2
1
= h
I
(t-)x
I
(t) - h
Q
(t-)x
Q
(t) + j[x
Q
(t)h
I
(t-) + h
Q
(t-)x
I
(t)]d
}

31
Equivalent Notation for complex baseband model ( - = convolution)
(t) = (x
(t)- h
(t)) = (h
(t) - x
(t))
factor added to maintain equivalence between real & complex models
f
c
is omitted from complex baseband model simplifies analysis
without loss of information
x(t) = Re{x
(t)exp(2f
c
t)}
y(t) = Re{(t)exp(2f
c
t)}
Passband signals are readily determined from (t) and x
(t)
Impulse response of band-pass system given by
h(t) = Re{h
(t)exp(2f
c
t)}
= Re{ (h
I
(t) + jh
Q
(t)) (cos(2f
c
t) + jsin(2f
c
t) ) }
= h
I
(t)cos(2f
c
t) - h
Q
(t) sin(2f
c
t)
32
Appendix: More on Complex Envelope - viewed as an extension of phasor for a real
harmonic signal x(t)
x(t) = o
x
cos(2tf
0
t + |
x
) t e R
assume o
x
> 0 and phase is 0 s |
x
< 2t, then:
(i) exp( j(2tf
0
t+|
x
)) = cos(2tf
0
t +|
x
) + jsin(2tf
0
t +|
x
)
= Re [o
x
exp(j(2tf
0
t + |
x)
)] t e R
= Re [o
x
exp(j|
x
) exp(j2tf
0
t )] t e R
(ii) x(t) = Re[o
x
( cos(2tf
0
t +|
x
) + jsin(2tf
0
t +|
x
) )] t e R
phasor representing phase & magnitude of x(t) = complex envelope:
o
x
exp(j|
x
) = o
x
cos(|
x
) + jo
x
sin(|
x
)
o
x
= magnitude
|
x
= argument (phase of x(t))
33
ii. suppress negative frequencies & multiply by 2
iii. shift left by f
0
to obtain frequency signal
= o
x
exp(j|
x
)o(f
0
)
f e R
iv. take Inverse Fourier Transform
i. Take Fourier Transform of x(t)
X(f) = F[o
x
cos(2tf
0
t+ |
x
)] = o
x
exp(j|
x
)o(f-f
0
) + o
x
exp(-j|
x
)o(f+f
0
)
2
1
2
1
derive complex envelope for any real continuous signal, x(t)
assume x(t) = Re [x
e
(t) exp(j2tf
0
t )] t e R
where x
e
(t)= o
x
exp(j|
x
),
x
e
(f)
= o
x
exp(j|
x
)o(f-f
0
) f e R x
p
(f)
= x
e
(t) = o
x
exp(j|
x
) F
-1
[x
e
(f) ]
34
x(t) = o cos(2tf
1
t + |
x
) t e R e.g. Pure Harmonic signal given by
if f
1
= f
0
complex envelope = phasor
if |f
1
-f
0
| << f
0
x
e
varies slowly compared to exp(2jtf
0
t)
where o
x
> 0
0 s |
x
< 2t
i. FT yields X(f) = o exp(j|
x
)o(f-f
1
) + o exp(-j|
x
)o(f+f
1
)
ii.
iii.
x
e
(t) = o exp(j|)exp(2jt(f
1
-f
0
))t t e R
iv
= o exp(j|)o(f-f
1
) x
p
(f)
= o exp(j|)o(f-f
1
+f
0
) x
e
(f)
35
If x(t) = real, continuous function, & F(x) has no delta function at f = 0
pre-envelope (aka analytical) of x is complex valued signal x
p
with
complex-envelope of x with respect to frequency f
0
is signal x
e

x
e
(f) = x
p
(f+f
0
) = 2X(f+f
0
) 1(f+f
0
) f e R
F[x
p
] = = 2X(f)1(f) f e R x
p
(f)
x
e
(t) = F
-1
[ x
e
(f) ]
36
Complex Envelope for let x(t) = real, band-pass, band-limited signal
f
c
= center frequency & W = bandwidth
where W < f
c
, are positive real numbers (W << f
c
x(t) is narrowband)
f e R X(f) = 0 for | f | < f
c
-W and | f | > f
c
+W
0
W
f
c
0 -f
c

W X(f)
x
p
= analytical
f
c
0
) ( f x
p
x
e
= complex envelope with respect to f
0

contains only low frequencies
f
0
e R+ x
e
is not uniquely defined
0
) ( f x
e

Sampling Theorem

Uploaded by

Copyright:

Available Formats

Sampling Theorem

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sampling Theorem

Uploaded by

Copyright:

Available Formats

SIGNAL PROCESSING BASICS

A/D AND D/A CONVERSION

phase shift x(t) by for positive frequencies

(t) preserves information content of s(t), except for f

You might also like