AGNEL INSTITUTE OF TECHNOLOGY AND DESIGN

AGNEL TECHNICAL EDUCATIONAL COMPLEX

ASSAGAO, BARDEZ-GOA. 403 507
(Approved by AICTE, Affiliated to Goa University)
DEPARTMENT OF ECE
ECE 4.2 SIGNALS AND SYSTEMS
Module 2- COURSE MATERIALS

CT-Fourier Series

Fourier series is used for analyzing the periodic signal which gives the spectral
component of the signal, the resultant spectrum is discrete.
Fourier transform is used for analyzing the aperiodic signal which gives the
spectral component of the signal, the resultant spectrum is continuous.
In fourier series the signal is represented by a set of harmonically related complex
exponential,

k= jkωo t
x(t) =  ak e ωo= fundamental frequency
k= -
T = fundamental period

This representation is called fourier series representation,

If k=0, gives the constant(dc) term of the signal,
k=±1 which has fundamental frequency ω o and period T called first (fundamental)
harmonic components,
k=±2 which frequency 2ωo and period T/2 called second harmonic components,
k=±N which frequency Nωo and period T/N called Nth harmonic components.

Procedure for determining the Fourier coefficients- Analysis and synthesis equation:

In fourier series the signal is represented by a set of harmonically related complex

exponential,

k= jkωo t -jnωo t

Let x(t) =  ak e multiplying both side by e
k= -

-jnωo t k= jkωo t -jnωo t

x(t) e =  a e e
k=- k
Integrating both sides from over one period of interval 0 to T
T
-jnωo t
T
k= -jnωo t jkωo t
0
x(t) e dt =
0
 k=-
 ake

e dt----------(1)

changing the order of integration and summation

Page | 1
 k= T j(k-n)ωo t
  ake dt
k=- 0
T
j(k-n)ωo t
 ake
0
dt = T, k=n

0, k≠n.
the equation (1) reduces to
T
-jnω t
0
 x(t) e o dt= a n T
T T
1 -jnωo t 1 -jkωo t
an 
T0 x(t) e dt by replacing n by k we get a k   x(t) e
T0
dt -----(2)

k= jkωo t
x(t) =  a e ------(3)
k=- k
Equation 2 is called analysis equation 3 is called synthesis equation

Note: The coefficient ak is called spectral or fourier series coefficients, which gives the
amplitude of the signal.
Eg-1

Find the fourier series coefficients for the given signal x(t)= sinω ot
k= jkωo t jkωo t
x(t) =  ak e expand the x(t ) and equate the coefficients of e
k= -
jω -jω
e ot -e ot
sin0 t =
2j
jω -jω
e ot - e ot jω -jω
 1+a1e o t + a -1e o t
2j
when equating the coefficients
ao=0, a1=1/2j , a-1= -1/2j
magnitude spectrum
| a1| = 0.5, | a-1| = 0.5

Page | 2
Eg.2
2) x(t)= 1+sinωot + 2cosωot + cos(2ωot + π/4)

k= jkωo t jkωo t

x(t) =  ak e expand the x(t ) and equate the coefficients of e
k= -

jωo t -jωo t jωo t -jωo t

e -e e +e
sin0 t = cos 0 t =
2j 2
j2ωo t j4 -2jωo t -j4
  e e +e e
cos  20 t +  =
 4 2
jωo t -jωo t  jωo t -jωo t  j2ωo t j4 -2jωo t -j4
e -e e +e  e e +e e
x(t)= 1   2 =
2j  2  2
 
jω -jω jω -jω
a +a e o t + a e o t +a e o t + a e o t
0 1 -1 2 -2

coefficients are
 
j4 -j 4
ao=1, a1=1+1/2j , a-1= 1-1/2j , a2=1/2( e ), a -2 = 1/2( e ),

magnitude spectrum

| a0| = 1, | a-1| = 1.12, | a1| = 1.12, | a-2| = 0.5 | a2| = 0.5

phase spectrum
   
a 0 = 0, a -1 = , a1 = - , a -2 = - a 2 =
7 7 4 4

Page | 3
Eg 3
Find the coefficient of the complex exponential fourier series for a halfwave sine
wave.

T0
x(t) = A sin  0 t 0  t 
2
T0
0  t  T0
2
T
1 -jkωo t
a k   x(t) e dt
T0
T0
2
1 -jkωo t
ak =
T0  A sinω
0
0 te dt

T0
 T20 T0

A 2
jωo t -jωo t -jkωo t A  jωo (1-k)t 2
-jωo (1+k)t 
=
2jT0  (e -e ) e dt =
2jT0  0
e dt -  e dt 
0
 0


To To
jω (1-k)t -jωo (1+k)t  2 jω (1-k)t -jω (1+k)t  2
A  e o e  A  e o e o 
= - =- +
2jT0  jωo (1-k) -jωo (1+k)  2T0 ω o  (1-k) (1+k) 
 0  0
 jωo (1-k) To2 -jωo (1+k) To2 
A e -1 e -1
=- +
2T0 ω o  (1-k) (1+k) 
 
-j(1+k)π  (1+k)e j(1-k)π -1 (1-k)e-j(1+k)π -1
A e A  
j(1-k)π
-1 e -1 +
=-  +  =-  
4π  (1-k) (1+k)  4π  1-k2 
 

for any value of k , k ej(1-k)π and k e-j(1+k)π are equal.

A  e j(1-k)π + e-j(1+k)π - 2 
4π  
=-
1- k2 
A
ak  for k=even
4 (1-k 2 )
= 0 k = odd

Page | 4
 If k = ± 1 ak is in-determinant, we have to find separately,
T0
2
1 -jkωo t
ak =
T0  A sinω
0
0 te dt, if k = 1

T0 T0
2 2
A jωo t -jωo t -jω A -j2ωo t
=
2jT0  (e
0
-e ) e o t dt =
2jT0  (1- e
0
)dt

To

A  e-j2ωo t  2
A  To 1 1 
t -   2jωo 
= = + -0-
2jT0  -2jωo  2jT0  2 2jωo 
 0
A  To  A A
=- = , a -1 = -
2jT0  2  4j 4j

Eg -4 Find the exponential FS coefficients

x(t) = Et/2π 0 ≤ t ≤ 2π
1
T
-jkωo t k= jkωo t
a k   x(t) e dt, x(t)=  a e
T0 k= - k
2
E -jkωo t
ak  2
4 te
0
dt,

use the formulae  uv dx = uv1  u ' v 2 + u '' v3

-jkωo t -jkωo t
-jkωo t e e
u= t , u ' = 1 , v =e , v1  , v2  
-jkωo k 2ωo2
2

E  te-jkωo t e-jkωo t  E  2 e-j2 k e-j2 k 1

ak  2   2 2  = 2    
4  -jkωo k ωo  4  -jk k2 k2 
 0 
-j2 k 
E  2 e jE
ak  2   =
4  -jk  2 k


Page | 5
Eg-5 The periodic wave sketched in fig defined over one period. This signal is periodic
over fundamental period To and fundamental freq ωo= 2π/ To determine the fourier series
coefficients for To= 4 T1.

x(t) = 1 |t| < T1

= 0 T1 ≤ |t| ≤ T0/2
T
1 -jkωo t
a k   x(t) e dt,
T0
T1
1 2T1
a0 
To  dt =
-T1
To
T1

1
T1
-jkωo t  e-jkωo t 
dt = -  
To -T1
ak  e
 jkωo To 
  -T1

2  e
jkωo T1 -jkωo T1 
e  = 2sinkωo T1 = sinkωo T1
ak 
kωo To  2j  kωo To k
 
sin( k / 2)
for To  4T1 , a k  for k  0, a k  0 for k even,
k
1 1 1
ao  , a1  a -1  , a 3  a -3   and so on
2  3

Page | 6
Properties of fourier series
The following are the important properties of FS
1-a) Linearity-CTFS

Let x(t) and y(t) are two periodic signals with period T. Let the fourier series
coefficient of x(t) =ak and y(t) =bk
x(t) FS ak y(t) FS bk
A x(t)+By(t) = Aak +Bbk
T T
1 -jkωo t 1 -jkωo t
a k   x(t) e dt, b k   y(t) e dt
T0 T0
T
1 -jkωo t
FS  Ax(t) + By(t)     Ax(t) + By(t)  e dt
T0
T T
1 -jkωo t B -jkωo t

T0 Ax(t) e dt+  y(t) e
T0
dt = Aa k  Bbk

1-b Linearity-DTFS

Let x(n) and y(n) are two periodic signals with period N. Let the fourier series
coefficient of x(n) =ak and y(n) =bk
x(n) FS ak y(n) FS bk

A x(n)+By(n) = Aak +Bbk

 2   2 
-jk n -jk n
N 1
 N  N 1
N
 y[n]e  
1 1
a k   x[n]e , bk 
N n 0 N n 0

 2 
N 1
-jk n
N
   x[n]  y[n] e  
1
FS  Ax[n] + By[n]
N n 0
 2   2 
N 1
-jk n 1 N 1
-jk n
N N
  Ax[ n]e   +
 By[n]e   = Aa  Bb
1
k k
N n 0 N n0

2-a)Time shifting-CTFS
-jkωo t 0
ak is the fourier series coefficient of x(t) , e a k is the fourier coefficient of x(t-to).
-jkωo t 0
x(t) F.S ak , x(t-to) FS e ak
Proof
T
1 -jkωo t
FS  x(t-t o )    x(t-t o ) e dt, let  =t-t o ,d  =dt , t=  t o
T0
T
1 -jkωo (  t o ) -jkωo t 0
  x( ) e d = e a k
T0

Page | 7
2-b)Time shifting-DTFS
 2 
-jk n 0
N
ak is the fourier series coefficient of x(n) , a k e   is the fourier coefficient of
x(n-no).
 2 
-jk n 0
N
x(n) F.S ak , x(n-no) F.S ake  
Proof
 2 
-jk
N 1 n
N
FS  x(n-n o )    x[n  no ]e  
1
N n 0
let p=n-n o  n=p+n o ,n=0 p=-n o , n=N-1 p=N-1-n o
 2   2   2   2 
N 1 no
-jk
N ( p  n o ) -jk N n o 1 N 1 no
-jk
N n -jk n o
1
x[ p ]e   e   x[ p ]e   a e  N 
 
N p  no

p  no N
k

3.a Time reversal-CTFS

Let x(t) is the given signal with f.s coefficient ak, the time reversal of x(t) is x(-t)
which has the F.S coefficient a-k.
x(t) F.S ak , x(-t) FS a-k
Proof
T
1 -jkωo t
a k   x(t) e dt,
T0
T
1 -jkωo t
FS  x(-t)    x(-t) e dt, let  =-t ,-d =dt ,
T0
-T T
1 jkωo t 1 -j(-k)ωot
   x( ) e d =  x( ) e d = a -k
T0 T0
3.b Time reversal-DTFS
Let x(n) is the given signal with f.s coefficient ak, the time reversal of x(n) is x(-n)
which has the F.S coefficient a-k.

x(n) F.S ak , x(-n) FS a-k

Proof
 2 
-jk  N n
1 N 1
FS  x(-n)    x[ n]e   , let p=-n  n=-p ,n=0 p=0 , n=N-1 p=-(N-1)
N n 0
 2   2   2 
 ( N 1)
-jk
N ( p )  ( N 1) jk
N  p  ( N 1) -j(-k)
N p
x[ p ]e     x[ p ]e   
 x[ p]e   a
1

N
p 0 p 0 p 0
-k

Page | 8
 Note: If x(t) is even, ak=a*-k, ak is even. If x(t) is odd, ak=-a*-k, ak is also odd.
4-aTime scaling-CTFS
In the time scaling property, x(ut) which gives the same fourier coefficient of x(t) but the
period is changed by an amount To/u

x(t) F.S ak with period To , x(ut) F.S ak with period To/u

Proof:
1
T
-jkωo t k= jkωo t
a k   x(t) e dt , x(t)=  a e
T0 k=- k
2
2 t jk
jk ut To
k= jkωou t k= To k=
x(ut)=  a e =  a e   a e u
k=- k k=- k k=- k

The fourier coefficient is same but period is changed by the amount To/u.
4-bTime scaling-DTFS
In the time scaling property of x(n/m) which gives the same fourier coefficient but period
changed by the amount mN
Proof:
 2   2 
N 1
-jk  n k  N 1
jk n
N N
a k   x[n]e   , x(n)= ake  
1
N n 0

k 0

 2 n
k  N 1
jk 
n
ake 
N  m   1  a (periodic with period mN)
x(
m
)= 
k 0
  k
m

n
where x ( m ) (n)  x   if n is a multiple of m
m
0 if n is not a multiple of m

5-a) Frequency Shifting-CTFS

jkωm
ak is the fourier series coefficient of x(t) , FS coefficient of e x(t) is ak-m
Proof:
1
T
-jkω t k= jkω t
a k   x(t) e dt , x(t)=  a e
T0 k=- k
T T
jωmt 1 jωmt -jkω t 1 -j(k-m)ω t
FS[e x(t)]=
T0 x(t) e e dt   x(t) e
T0
dt  ak  m

5-b) Frequency Shifting-DTFS

jkωm
ak is the fourier series coefficient of x(n) , FS coefficient of e x(n) is ak-m

Page | 9
Proof:
 2   2 
N 1
-jk n k  N 1
jk n
N N
a k   x[n]e   , x(n)=
 ake  
1
N n 0 k 0

 2   2   2   2 
j mn j
N 1  mn -jk n 1 N 1
-j(k-m) N n
N N  x[n]e  N
FS[e   x(n)]= 1
 e  
 x[n]e   a
k m
N n 0 N n0

6-a Convolution Property-CTFS

If x(t) and y(t) is a periodic signal with period ‘T’ with FS coefficients a k and bk. The
FS coefficient of convolution of x(t) and y(t) i.e. x(t)*y(t) is T ak bk
x(t) F.S ak , y(t) FS bk
x(t)*y(t) F.S Takbk

Proof.
k= jkωt y(t)= k= b e jkωt
x(t)=  a e 
k=- k k=- k
T
we know that x(t)*y(t)=  x( ) y (t   ) d      (1)
 0

k= jkω and y(t- )= k= b e jkω(t- ) substitute in equation 1

x( )=  a e  k
k=- k k=-
T k = jkω k= b e jkω(t- ) d  T k= a e jkω k= b e-jkω e jkωt d
1   a e  k  0 k=-
 k  k
 0
k=- k k=-  k=-
change the order of integration and summation
k= p= jpωt T d  k  Ta b e jkωt
 
k=- p=-
 a k bpe  0 k 
k k
k 
x(t)*y(t)   Ta k b k e
jkωt is in the form of x(t)= k= a e jkωt

k  k=- k
so FS(x(t)*y(t))=Ta k b k
6-b Convolution Property-DTFS
If x(n) and y(n) is a periodic signal with period ‘N’ with FS coefficients a k and bk. The
FS coefficient of convolution of x(n) and y(n) i.e. x(n)*y(n) is N ak bk
x(n) F.S ak , y(n) FS bk
x(n)*y(n) F.S Nakbk

Proof.
 2   2 
N 1
-jk n k  N 1
jk n
 N  , x(n)= N
 ake  
1
We know that, a k   x[n]e
N n0 k 0

Page | 10
 N 1
Convolution of [x(n) * y(n)]   x[k ] y[n  k ]        (1)
k 0

 2   2 
k  N 1
jk n k  N 1
jk ( n  k )
 N  and y(n-k)= N
Let x(k)= 
k 0
ake  bk e 
k 0
 substitute in equation 1

 2   2   2 
N 1 N 1
jk n k  N 1 jk N n -jk N n
1  ake  N  be   e  
k 0 k 0

k 0
k

change the order of summation

 2 
N 1
jk n k  N 1
  a k bk e  N 
k 0
1
k 0

 2   2 
k  N 1
jk n k  N 1
jk n
 N  , This is equation in the form of x[n]= N
x(n)*y(n)  
k 0
Na k b k e  ake 
k 0


So x(n)*y(n) F.S Nakbk

7-a Conjugate and conjugate symmetry property CTFS
If ak is the fourier series coefficient of x(t). For the conjugation of x*(t) the coefficient is
a*-k
x(t) F.S ak , x*(t) FS a*-k
Proof
k= jkωo t
x(t)=  a e
k=- k
k= -jkωo t k= * -j(-k)ωo t
x * (t)=  a *e =  a e (replace k by -k) =a *
k=- k k=- -k -k
Note: If x(t) is real and even, ak=a -k, ak is real and even. If x(t) is real and odd, ak=-a*-k,
*

ak is purely imaginary and odd.

7-b Conjugation property DTFS
If ak is the fourier series coefficient of x(n). For the conjugation of x *(t) the coefficient is
a*-k
x(n) F.S ak , x*(n) FS a*-k
Proof
 2 
jk n
k=N-1  N 
x(n)=  a e
k=0 k
 2   2 
-jk  n jk n
k=N-1 *  N  k=N-1 *  N 
x (n)=  a e
*
=  a e  a*
k=0 k k=0 -k -k
Note: If x(n) is real and even, ak=a -k, ak is real and even. If x(n) is real and odd, ak=-a*-k, ak
*

is purely imaginary and odd.

Page | 11
8-a Multiplication Property-CTFS
If x(t) and y(t) is a periodic signal with period ‘T’ with FS coefficients a k and bk, the
l= 
FS coefficient of multiplication of x(t) and y(t) i.e. x(t)y(t) is ab
l=-
l k l

l= 
x(t) F.S ak , y(t) FS bk, x(t)y(t) F.S ab
l=-
l k l

Proof:
T T
1 -jkωt dt, 1 -jkωt dt        (1)
FS(x(t)) = a k 
T0 x(t) e Let FS(x(t)y(t))=
T0 x(t)y(t)e

jlωt and y(t)= p= b e jpωt substitute in equation 1

l=
Let x(t)=  a l e 
l=- p=- p
1 l= jlωt p= b e jpωt e-jkωt dt
T
1    ale 
T 0 l=- p=- p
change the order of integration and summation
1 l= p= j(p-(k-l)ωt dt  1 T l  a b { T e j(p-(k-l)ωt dt  T when p=k-l
 k l t 0
T
   a l b p t 0 e
T l=- p=- T l  k
l 
x(t)y(t)  a b
l 
l k-l

8-b Multiplication Property-DTFS

If x(n) and y(n) is a periodic signal with period ‘N’ with FS coefficients a k and bk. The
l= 
FS coefficient of multiplication of x(n) and y(n) i.e. x(n)y(n) is ab
l=-
l k l

l= 
x(n) F.S ak, y(n) b k, x(n)y(n) F.S ab
l=-
l k l

Proof:
 2   2 
N 1
-jk n N 1
-jk n
1  N  1  N         (1)
FS(x(n)) = a k   x[n]e
N n0
and FS(x(n)y(n))=  x[n] y[n]e
N n0
 2   2 
l  N 1
jl n p  N 1
jp n
 N  and y(n)= N
Let x(n)= ae
l 0
l  bp e 
p 0
 substitute in equation 1

 2   2   2 
N 1 l  N 1
jl n p  N 1 jp n -jk n
 N  N N
 bp e   e   change the order of summation
1
1 
N n0
ae l 0
l
p 0

 2 
N 1 l  N 1 p  N 1
j(p-(k-l)) n l 
1  N  { N when p=k-l} so x(n)y(n) 
 
N n0
 
l 0 p 0
a l bp e a b
l 
l k-l

Page | 12
 9-a)Parsevals theorem-CTFS
The parsevals relation states that the average power in a signal is equal to the sum of
individual components of the normalized fourier series approximation.
k= 

a
2
Pavg = k
k=-
Proof:
The average normalised power of a periodic waveform of x(t) is
1 T
Pavg =  x(t) dt
2

T 0

2
x(t) = x(t)x * (t), x * (t) is the complex conjucate of x(t)
k= jkωo t k= -jkωo t
x(t)=  a e , x * (t)=  a *e
k=- k k=- k
1 T 1 T k= -jkωo t
Pavg =  x(t)x * (t)dt   x(t)  a *e dt
T 0 T 0
k=- k
Interchange the order of integration and summation
k= 1 T -jkωo t k= 1 T -jkωo t k= k= 2
Pavg =  a *  x(t)e dt   a *  x(t)e dt   a *a   a
k=- k T 0 k=- k T 0 k=- k k k=- k
9-b)Parsevals theorem-DTFS
The parsevals relation states that the average power in a signal is equal to the sum of
individual components of the normalized fourier series approximation.
k= 

a
2
Pavg = k
k=-
Proof:
The average normalised power of a periodic waveform of x(n) is
1 N 1

2
Pavg = x[n]
N n 0
2
x(n) = x(n)x * (n), x * (n) is the complex conjucate of x(n)
 2   2 
jk n jk n
k=N-1 N k=N-1 *  N 
x(n)=  a e   ,x (n)=  a k e
*

k=0 k k =0
 2 
jk n
1 N 1
1 N 1 k=N-1 *  N 
Pavg =  x[n]x [n] 
*
 x[n] k=0 ake change the order of summation
N n 0 N n0
 2 
jk  N n
1 N 1 k=N-1 * 1 N 1
Pavg =  x[n]x [n]   a k  x[ n]e 
* 
N n 0 k=0 N n0
k=N-1 k=N-1 2
Pavg =  a *a   a
k=0 k k k=0 k

Page | 13
DT-Fourier Series

Fourier series is used for analyzing the periodic signal which gives the spectral
component of the signal, the resultant spectrum is discrete.
Fourier transform is used for analyzing the aperiodic signal which gives the
spectral component of the signal, the resultant spectrum is continuous.
In fourier series the signal is represented by a set of harmonically related complex
exponential,
Let x[n] is a discrete time periodic signal and it can be expressed as linear
combination of complex exponential as follows
 2 
jk n
k=N-1  N  2
x(n)=  a e = fundamental frequency
k=0 k N
N = fundamental period

This representation is called fourier series representation, For N successive values

of ‘n’ corresponding to one period N linear equations are obtained. ‘N’ unknown ‘ak’
coefficients, so finding of ak is complex and time consuming process. The following
procedure gives a closed form expression to find the coefficients ‘‘a k’

Procedure for determining the Fourier coefficients- Analysis and synthesis equation:

In fourier series the signal x[n] is represented by a set of harmonically related complex
exponential,

 2 
jk n
k=N-1 N
Let x(n)=  a e   -------------(1)
k=0 k
-jr  2πN  n
 
multiplying both side by e and summing over N terms,

N 1 -jr  2πN  n -jr  2πN  n jk  2πN  n

N 1 N 1

 x[n]e
n 0
 
=  a k e   e  
n 0 k 0
By changing the order of summation,
N 1 -jr  2πN  n N 1 N 1 j(k-r) 2πN  n
n 0
x[ n ]e  
  ak  e
k 0 k 0
 
,

N 1 j(k-r) 2πN  n
and the equation gives e
k 0
 
 N if k=r

= 0 if k  r

Page | 14
 N 1 -jr  2πN  n
 x[n]e
n 0
 
 ar N

1 N 1 -jr  2πN  n
ar 
N
 x[n]e
n 0
 
, after replacing 'r' by 'k'

1 N 1 -jk  2πN  n
ak 
N
 x[n]e
n 0
 
           (2)

Equation 2 is called analysis equation 1 is called synthesis equation of Fourier

series.

LTI System Response to the Complex exponential signal- Eigen Function

A function is said to be an eigen function, if any operation is performed on that
function it has to give the same function with some constant value. That function is said to
be “Eigen function”, that constant value is called “Eigen value” and the operation is called
“Eigen operator”.

If a system has linearity (additive and Homogeneity) and time invariance property then the
system is said to be LTI System. In LTI system, signals are represented as linear
combinations of basic signals which gives convenient representation for the response of the
system.

The importance of complex exponential in the study of LTI system is that the response
of a LTI system to a complex exponential input is same complex exponential with only
change in amplitude.
For continuous time est H(s) est
n
For Discrete time z H(z) zn

The show the complex exponentials are eigen function of LTI system, let us consider
CT LTI system with impulse response h(t). For an input x(t), output response of a
Continuous LTI system is ,
 
y(t)=  x( )h(t  )d =  h( )x(t  )d
 
st
For the input x(t)=e ,
  
y(t)=  h( )e s(t  )
d   h( )e e d  e st s st
 h( )e
s
d  est H(s)
  
So the complex exponential is an eigen function of a LTI system and H(s) is an eigen

Page | 15
 value.
In a same manner , the complex exponential sequences are eigen functions of discrete
time LTI systems.
Consider a discrete time LTI system with impulse response h(n) and its input sequence
x[n]=zn
 
y[n]= x(k )h(n  k )  h(k ) x(n  k )
 
n
For the input x[n]=z ,
  
y[n]=h(k ) z nk
 h(k )z z  z
n k n
h(k)z k
 z n H ( z)
  
If the input x[n] is the complex exponential, the output is the same complex exponential
multiplied by a constant H(z).
So the complex exponential ‘z n’ is an eigen function of a Discrete LTI system and H(z)
is an eigen value.

Dirichlet conditions for the existence of FS-Convergence conditions

 Fourier maintained that any periodic signal can be represented by FS coefficients,
but this is not true.
 An alternative set of conditions developed by dirichlet should satisfied by all the
periodic signals guarantees that it has fourier series representation.
The following conditions, under which periodic function can be expanded in a
convergent fourier series is called Dirichlet Conditions
Condition 1: Signal should have a finite number of maxima and minima over the range of
time period.
For example consider a signal x1(t) is periodic with period ‘T’ which has one maxima
value ‘1’ and one minima value ‘0’ over the period ‘T’. As per Dirichlet, FS coefficient is
exist for this signal. On other hand the signal x 2(t)=sin(2π/t) 0≤t≤1 is periodic signal but it
has infinite number of maxima and minima violates the Dirichlet condition, so FS
coefficient does not exist for this signal.

Condition 2: Signal should have a finite number of discontinuities over the range of time
period.
For example consider a square signal x1(t) shown in figure is periodic with period ‘T’
which has finite number of discontinuities over the period ‘T’. As per Dirichlet, FS
coefficient is exist for this signal. On other hand the signal x2(t) is also periodic signal but it

Page | 16
violates dirichlet condition because it has infinite number of discontinuities over the time
period. so FS coefficient does not exist for this signal.

Condition 3: Signal should be absolutely integrable over the range of time period.
i.e.  x(t)dt 
T

For example consider a square signal x1(t) shown in figure is periodic with period ‘T’ The
value of integration over one period is 2. i.e. absolutely integrable. As per Dirichlet, FS
coefficient is existing for this signal. On other hand the signal x2(t)=1/t 0≤t≤1 is also
periodic signal but it gives infinite value of integration over the time period. so FS
coefficient does not exist for this signal.
Gibbs Phenomenon

Consider a periodic signal x(t), the fourier coefficients of this signal is infinite. To process
the coefficients the infinite fourier series coefficients is truncated into finite coefficients let
xN(t) is the truncated signal.

k= jkωo t k=N jkωo t

x(t)=  a e and x N (t )   a e
k=- k k=-N k

Page | 17
Since the square wave satisfies the dirichlet conditions, the increase in N ∞ of xN(t) at
the discontinuities should be the average value of discontinuity,

lt xN (t )  x(t )
N 
Thus as N increases, the ripples are compressed towards the discontinuity, but for
any finite value the peak value of the ripple is constant. This effect is known as Gibbs
Phenomenon.
 The implications is that the truncated fourier series approximation x N(t) of a
discontinuous signal x(t) in general exhibit high frequency ripples and overshoot x(t)
near the discontinuities. Illustration of Gibbs phenomenon is shown in Fig for different
values of N.

From the figure as N increases ripples get compressed at discontinuities. For any finite N
peak amplitude of the ripples is constant, this behavior is known as Gibbs phenomenon.

Eg-5 Find the fourier series coefficients for the given signal x(n)= cosωon
N 1 jk  2π  n
we know that, x[ n]   a k e  N 
k 0

jk  2πN  n
To find a k expand the x  n  and equate the coefficients of e  

 2   2   2   2 
 j 2   j  j  j 2 
x[n]  ...  a2 e  N 
 a1e  N 
 a0  a1e  N 
 a2 e  N 
 ......
 2   2   2   2 
j   j  j   j 
 N   N   N   N 
e + e e e
Given signal x[n]=coso n   
2 2 2
 2   2 
j   j 
 N   N   2   2 
e e  j  j 
  a1e  N 
 a0  a1e  N 
2 2
1 1
From the above equation, a0  0, a1  , a1 
2 2

Page | 18
 No imaginary parts in ak so it has only magnitude spectrum
magnitude spectrum
| a1| = 0.5, | a-1| = 0.5

Eg.6 Find the FS coefficient of the given signal and sketch the magnitude and phase
spectrum
 2π   2π   4π 
x[n]=1+sin   n  2 cos   n  cos  n  
N N N 4
N 1 jk  2πN  n
we know that, x[ n]   a k e
k 0

jk  2πN  n

To find a k expand the x  n  and equate the coefficients of e
 2   2   2   2 
 j 2   j  j  j 2 
x[n]  ...  a2 e  N 
 a1e  N 
 a0  a1e  N 
 a2 e  N 
 ......
 2   2   2   2   2   2 
j   j  j   j  j  j 
 N   N   2   2   N   N   N   N 
2e + 2e j   j  e - e e e
2coso n  e  N 
e  N 
and sino n   
2 2j 2j 2j
 2     2   
j 2  j   j 2   j 
 4π  e  N 
e 4
+ e  N 
e 4
cos  n   
N 4 2
 2   2   2     2   
j  j  j 2  j   j 2   j 
 2   2 
 2π   2π   4π  e N e N j   j  e  N 
e 4
+ e  N 
e  4
1+sin   n  2 cos   n  cos  n    1   e  N  e  N  
 
N  
N  N 4  2 j 2 j 2
   
j   j 
 2   2   2   2 
 j  j   j   j   e  4  j 2 N  e  4   j 2 N 
=1+ 1   e  N    1   e  N   e  e
 2  2 2 2
   
j   j 
 2   2   2   2   2   2   2   2 
 j 2   j  j  j 2   j  j   j   j   e  4  j 2 N  e  4   j 2 N 
a2 e  N 
 a1e  N 
 a0  a1e  N 
 a2 e  N 
 1+ 1   e  N    1   e  N   e  e
 2  2 2 2
 2   2   2   2 
j   j  j 2   j 2 
 N   N   N   N 
compare the coefficients of e ,e ,e ,e
   
j   j 
 j  j e 4 e 4
a0  1, a1  1   , a1   1   , a2  , a2 
 2  2 2 2
magnitude spectrum

| a0| = 1, | a-1| = 1.12, | a1| = 1.12, | a-2| = 0.5 | a2| = 0.5

phase spectrum

Page | 19
    
a 0 = 0, a -1 = , a1 = - , a -2 = - a 2 =
7 7 4 4

Eg.7 Consider a Discrete time periodic signal, x[n]= 1 -N1 ≤n≤ N1. Find the FS
coefficient of the given signal
1 N 1 -jk  2π  n
The analysis Equation, a k   x[n]e  N     (1)
N n 0
given x[n]=1 -N1  n  N1 ,
1 N1 -jk  2πN  n
Equation 1  a k 
N
e
n  N1
 
     (2)

Let m=n+N1 , n=m-N1 , if n=-N1 m=0, if n=N1 m=2N1

1 2 N1 -jk  2πN (m-N1 )  2π   2π 
1 jk  N  N1 2 N1 -jk  N  m
Equation 2  a k 
N
e
m0
 
 e
N
e
m0

 -jk  2π  2 N1 1 

  1  a 1  
 2π  2 N 1
1 jk  N  N1  1  e  N  2 N1
 e   2π   ,   a m

1 a 

N  1  e-jk  N    m 0
 
k k
-j  2πN  2 N1 1 -j  2πN 
   
Take e 2 in the NR and e 2 in the DR for further simplication
 π 
 2π 
-jk  Nπ  2 N1 1  jk  Nπ  2 N1 1 -jk  Nπ  2 N1 1  sin  k    2 N1  1 
1 jk  N  N1 e  
e  
e  
 1  N 
ak  e   
N -jk   π π
-jk   -jk    2π 
 N   π 
e  N   e N  e  N   sin  k   
  N 
This a k is valid only for k  0,  N,  2N ....., if k=0,  N,  2N .....

Page | 20
 0
a k  , apply L-Hospital Rule to find a k
0
d  π   π  π 
sin  k    2 N1  1  cos  k    2 N1  1     2 N1  1
1 dk  N   1 lt  N  N  2 N1  1
ak  lt 
N k  0 d   π  N k  0   π  π  N
sin  k    cos  k     
dk   N    N  N 
 π 
sin  k    2 N1  1 
ak 
1  N  , for k  0,  N,  2N .....
N   π 
sin  k   
  N 
2 N1  1
= for k= 0,  N,  2N .....
N

Page | 21