Signals & Systems: Part III
Satyajit Thakor
IIT Mandi
Fourier series
I Fourier series expansion: A periodic signal with period T0 = 1
f0
can be expanded in terms of complex exponential signals as
1
X
x(t) = xn ej2⇡nf0 t (synthesis eq.)
n= 1
Z ↵+T0
1 j2⇡nf0 t
where xn = x(t)e dt (analysis eq.)
T0 ↵
if Dirichlet conditions, x(t) in a period interval (i) is absolutely
integrable (ii) has finite number of extrema and (iii) has finite
discontinuities, are satisfied. (sufficient conditions for the
existence of FS)
Parseval’s relation
I Parseval’s relation: the power content of a periodic signal is the
sum of the power contents of its components in the
Fourier-series representation of that signal.
Z ↵+T0 1
X
1
|x(t)|2 dt = |xn |2
T0 ↵ n= 1
I That is, power content of a signal is the sum of power contents
of its harmonics.
I Proof uses the orthogonality of the basis used for the
Fourier-series expansion: recall
Z ↵+T0
ej2⇡(m n)f0 t
dt = T0 if n = m,
↵
=0 otherwise, n, m 2 Z.
Fourier series for real signals
I For real x(t), the positive and negative Fourier series coefficients
are conjugates (since x(t) = x⇤ (t)):
Z ↵+T0
1
x n = x(t)ej2⇡nf0 t dt
T0 ↵
Z ↵+T0 ⇤
1
= x(t)e j2⇡nf0 t dt = x⇤n
T0 ↵
I Let xn = an 2jbn . Then, for real signal, x n = an +jb
2
n
.
I If x(t) is a real signal then its Fourier series expansion can be
alternatively presented as the trigonometric FS expansion:
1
a0 X
x(t) = + [an cos(2⇡nf0 t) + bn sin(2⇡nf0 t)] .
2
n=1
Fourier series for real signals
1
X
I Proof: x(t) = xn ej2⇡nf0 t
1
1
X
= x0 + [xn ej2⇡nf0 t + x ne
j2⇡nf0 t
]
n=1
X1
a0 an jbn j2⇡nf0 t an + jbn j2⇡nf0 t
= + [ e + e ]
2 2 2
n=1
Now simplify using Euler’s identity to get
1
a0 X
x(t) = + [an cos(2⇡nf0 t) + bn sin(2⇡nf0 t)] .
2
n=1
I an , bn in terms of x(t):
Z ↵+T0
an jbn 1 j2⇡nf0 t
xn = = x(t)e dt
2 T0 ↵
Fourier series for real signals
Z ↵+T0
1
= x(t)[cos(2⇡nf0 t) j sin(2⇡nf0 t)]dt
T0 ↵
Z ↵+T0
2
=) an = x(t) cos(2⇡nf0 t)dt,
T0 ↵
Z ↵+T0
2
bn = x(t) sin(2⇡nf0 t)dt
p et of even and
T0 ↵
odd signals is odd
I If a real periodic signal is even (odd) then all bn (an ) coefficients
C integral of Todd signal is zero
are zero.
I Also, for real x(t) (using the polar form for xn ’s)
1
X
x(t) = x0 + 2 |xn | cos(2⇡nf0 t + \xn ).
n=1
Fourier series for real signals
1
X
I Proof: x(t) = x0 + [xn ej2⇡nf0 t + x ne
j2⇡nf0 t
]
n=1
X1
= x0 + [xn ej2⇡nf0 t + (xn ej2⇡nf0 t )⇤ ]
n=1
X1
= x0 + 2Re(xn ej2⇡nf0 t )
n=1
X1
= x0 + 2Re(|xn |ej\xn ej2⇡nf0 t )
n=1
X1
= x0 + 2 |xn | cos(2⇡nf0 t + \xn )
n=1
Fourier transform
I We saw that viewing a signal in terms of complex exponentials
is very useful for analyzing LTI systems.
I Fourier transform: The spectrum of a signal x(t) (not
necessarily periodic) is defined as
Z 1
X(f ) = x(t)e j2⇡f t dt (analysis eq.)
1
Z 1
where x(t) = X(f )ej2⇡f t df (synthesis eq.)
1
if Dirichlet conditions, x(t) (1) is absolutely integrable (2) has
finite number of extrema in any finite interval and (3) has finite
discontinuities in any finite interval, are satisfied.
I Note that the variable of integration in the synthesis equation is
f (in place of ! as we used in IC260 and hence the multiplying
factor of 1/2⇡ disappears).
FT of Real, even and odd signals
I Sometimes X(f ) is referred to as a voltage spectrum (as
opposed to a power/energy spectrum - defined later)
I Notation:
1
X(f ) = F[x(t)], x(t) = F [X(f )], x(t) , X(f )
Z 1
j2⇡nf t
F[x(t)] = x(t)e dt
1
Z 1 Z 1
= x(t) cos(2⇡nf t)dt j x(t) sin(2⇡nf t)dt
1 1
it
I If x(t) is real then both the integrals are real. T.it
I Hence, for real x(t), the real part of X(f ) is even and the
imaginary part is odd function of f .
Xr f net cos stuff dt Imt cos 2 nft dt f
FT of Real, even and odd signals
I This means, X(f ) is a Hermitian function:
X( f ) = X ⇤ (f )
I In other words,
Re[X( f )] = Re[X(f )], Im[X( f )] = Im[X(f )]
|X( f )| = |X(f )|, \X( f ) = \X(f )
I What if x(t) is real and even? f is real
I What if x(t) is real and odd? imaginary
f is purely
Fourier transform of periodic signals
I For periodic x(t),
1
X
x(t) = xn ej2⇡nf0 t (FS systhesis eq.)
n= 1
Z 1 1
X
F[x(t)] = xn ej2⇡nf0 t e j2⇡f t
dt
1 n= 1
1
X
= xn F(ej2⇡nf0 t )
n= 1
X1
= xn (f nf0 ) (* recall: ej2⇡f0 t , (f f0 ))
n= 1
I The spectrum consists of impulses in frequency at multiples of
the fundamental frequency of the periodic function.
I The weights of these impulses are the Fourier series coefficients.
Revisiting complex exponentials (part 3)
I Find Fourier transform of x(t) = cos(!0 t): cos not
e Ex
f S f to S f to
I !0 = 2⇡f0 is angular frequency in radiance per second
I Imaginary parts in the complex exponentials cancel out.
I Positive and negative frequencies
Ét
I Positive frequencies: terms of the form ej!t , phasor rotating
counter-clockwise
I Negative frequencies: terms of the form e j!t , phasor rotating
clockwise
Properties of Fourier transform
I Homework: prove the properties.
Fourier transform pairs
I Homework: prove the FT-IFT pairs.
