SIGNAL AND SYSTEM

Signal e. Doublet Function

A signal is function representing a physical
quantity or variable, and typically itFirst derivative of impulse function is called doublet
function
contains information about the behaviour or
nature of the phenomenon. • Properties of unit doublet function
System System is a medium which process a signal. dδ ( t ) 1
δ '( t ) =
It converts signal one form to another form. , δ ' ( at ) = δ '( t )
dt a|a|
Basic continuous-time signals All even derivatives of δ ( t ) are even functions of
a. The unit step Function The unit step function u(t)
also known as the Heaviside unit function, is defined time while all odd derivatives are odd functions.
as If f(t) is any function, then
1 t>0 f ( t ) δ '( t − a ) = f ( a ) δ '( t − a ) − f '( a ) δ ( t − a )
 −df ( t )
u(t) = 1/ 2 t = 0 ∞
∫ f ( t ) δ '( t − a ) =
0 t<0 −∞ dt t = a

f. Ramp Signal

• Properties of Unit-step function

u (t) + u (-t) =1,
1 + sgn(t)
u (t) – u (–t) = signum function {
r(t) = t
0
t≥0
t<0 }
u (t) =
2 d
u(t) is a power signal r (t) = t u (t), r (at) = a r (t), r(t) = u(t)
b. Signum function dt
g. Parabolic function
1 t>0
  t2
Sgn (t) = 0 t =0  t≥0
p (t) =  2
 −1 t < 0
 0 t<0


d 3p(t) d 2 r(t) du(t)

= = = δ(t)
c. Rectangular Function/Gate function dt 3 dt 2 dt
t h. Sinc function
x(t) = A rect  
T sin πt
sinc ( t ) =
 −T T πt
x(t) = A <t< 
2 2 Area [sinc(t)] = 1
0 elsewhere  Energy [sinc(t)]
=1
d. The unit impulse function/Dirac delta function i. Sampling function
∞ t = 0 sin t
δ(t) =  Sa ( t ) =
0 otherwise t
Sa ( πt ) = sinc ( t )

• Properties of impulse function

∞ Basic Discrete-time signals
Area under impulse function is 1 ∫ −∞
δ(t) dt = 1 a. The unit step sequence
Impulse function is even function for all time The unit step sequence u(n) is defined as
δ(t) = δ(− t) 1 n≥0
u(n) =  ,
1  0 n <0
Scaling properties : δ(at) = δ(t)
a
f (t) δ (t) = f (0) δ(t) f (t) δ (t − a) = f (a) δ(t − a)
 u(n) + u(−n) = 1 + δ(n), n = 0 
∞ ∞  n ≠ 0 
∫−∞
f (t) δ(t)dt = f (0) ∫−∞
f (t) δ(t − a)dt = f (a)  1

Electronic Engineering Capsule 193 YCT

 The Unit impulse sequence E
If x (t) → Energy signal (E), Then- x(at) →
1 n=0 a
δ(n) =  ;
0 n≠0 x (t – t0) → E, x(– t) → E
Ax (t) → |A|2 E
Signal Identification:
Properties of Discrete time impulse Energy signal
i. Discrete time impulse is an even function of time For finite duration of energy signal, magnitude of signal
δ[n] = δ[− n] should be finite
ii. Scaling property
δ [mn] = δ [n]
The scaling property does not exist in discrete time
impulse function because it depends on the
amplitude.
iii. Product property: For infinite extension energy signal, peak magnitude of
(i) x [n] δ [n] = x (0) δ [n] signal should be either left side or right side or both
(ii) x [n] δ [n – n0] = x [n0] δ [n –n0] sided decreasing in nature. (From a finite value)
∞
iv. ∑ x [n] δ [n − n
n =−∞
0 ] = x[n 0 ]
n2

∑ x[n] δ [n − n
n = n1
0 ] = x[n 0 ] n1 ≤ n 0 ≤ n 2
Periodic signal are not energy signal because their
Classification of signal energy contains is infinite.
a. Energy and power signal
Energy Power
For continuous time  1 T
T →∞ 2T ∫− T | x(t) | dt;for periodicsignal
2
lim
∞ 
E = ∫ | x(t) | dt 2 Power = 
 lim 1
T/2

T →∞ T − T∫/ 2
2
−∞ x(t) dt; for non periodic signal

For discrete time 1 N

∑
2
∞ Pavg x(n) = lim x(n)
N→∞ 2N + 1
Ex(n) = ∑ x(n)
2
n =− N
n=−∞
Energy = Finite, Energy = ∞,
Power = 0 Power = Finite
•
Signal Area Energy
x ( t ) = sin c ( t ) 1 1
x ( t ) = πt 2 1 1/ 2
t Aτ A2τ
x ( t ) = Arect  
τ b. Even & odd Signal
t Aτ 2 2 • Even signal : f (– t) = f (t) • Odd signal: f (– t) = –f (t)
x ( t ) = Atri   A τ • Any signal can be divided into two part in which one
τ 3
part will be even signal (fe (t)) and other part will be
x (t) = e u (t)
− at 1 1
odd signal (f0 (t)).
a 2a i.e., f (t) = fe (t) + f0 (t),
x ( t ) = sgn ( t ) 0 1/a f (t) − f (− t) f (t) + f (− t)
f0 (t) = fe (t) =
• For an energy signal the average power is zero 2 2
• For a power signal the energy is ∞. Even × Even = Even , Odd × Odd = Even
Odd × Even = Odd
• The frequency & phase of the signal will not effect a a
the power calculations. For even signal = ∫ f (t)dt = 2 ∫ f (t)dt
−a 0
Energy/Power calculation is independent of time a
shifting, time reversal and amplitude reversal. For Odd signal = ∫ f (t)dt = 0
−a
Power signal is also known as ‘‘means square value
of signal.’’ Derivative of even signal = Odd signal
Derivative of Odd signal = even signal
Power = (R.M.S.) 2 Integral of Odd signal = even signal

Electronic Engineering Capsule 194 YCT

• Conjugate symmetric (CS) and conjugate • Anti causal Signal A continuous time signal x(t) is
anti-symmetric (CAS) signal said to be anti causal-
If x(t) = 0 t > 0, x (n) = 0, n≥0
CSSignal CASSignal
• Non causal Signal A signal which exists both
x(t) = x * (− t) x(t) = x * (− t) positive and negative time is said to be non-causal
signal.
Any signal can be divided into two part in which one part
will be C.S. and other part will be conjugate A.C. • Area of signal
∞
i.e. x (t) = xCS (t) + xCAS (t)
where,
Area of signal x (t) = ∫ x(t)dt
−∞
 x(t) + x * (− t)  Area of signal over range (t1, t2)
 x CA(t ) = 2  t2
  = ∫ x (t) dt per period signal
x x(t) − x * (− t) 
 CAS( t) =  t1
 2  • Average of signal
1
T∫
x(t)dt; for periodic signal
T/2
1
T →∞ T ∫
lim x (t); for non-periodic signal
Periodic and Non-periodic signal −T / 2
• Periodic and Non-periodic signal a periodic Area
signal Average =
A signal is said to be periodic if it repeat itself after Time - Period
sometime. Average value also called ‘D.C. Value’
Average calculation only depends upon Amplitude
i.e. x(t) = x(t ± nTo ) where n = 1, 2, 3....... scaling and amplitude shifting
To = fundamental Time Period (FTP) Signal Average
≠0
f (t) → Ao
≠∞
• Fundamental Time Period (FTP): It is smallest kf (t) → kA o
positive volume of time for which signal is periodic. For finite duration signal having finite amplitude,
2π average value is zero.
To = for sinωt, cosωt,
W Linear Time Invariant Systems •
Notes: Fundamental time period calculation depends upon LTI System–When any system follows two
only scaling. basic properties i.e. linearity and time-invariance are
known as LTI system.
FTP
x(t)  → TO
T
i.e. x(at) 
→ O
a≠0 a

• Steps for finding the fundamental period of the

signal y (t) = x (t) * h (t) y [n] = x [n] * h [n]
x ( t ) = x1 ( t ) + x 2 ( t ) + x 3 ( t ) ........ Y (s) = X (s) .H (t) Y [z] = X [z] . H [z]
• Identify the individual periods (T1, T2, T3…..) Y(S) Y(Z)
H(s) = H(Z) =
T T T X(S) X(Z)
• Calculate the ratio of 1 , 1 , 1 .........
T2 T3 T4 Transfer function Transfer function
• If the ratio is rational number Classification of systems
Period (T)= LCM of [T1,T2,T3……….] i. Linear systems and Non- Linear systems A
• For discrete time signal – system that obeys the superposition and
N 2π homogeneity principle is said to be linear system,
=
K ω0 else non linear system.
c. Causal, anti causal and Non-causal signal • Superposition (Additivity)
• Causal Signal- A signal is said to be causal if it is If x1(t) → y1(t), x2(t) → y2(t)
defined for only positive value. Then x1(t) +x2(t) → y1(t) + y2(t)
• Homogeneity (scaling)
x(t) = 0, t < 0 x[n] = 0, n<0
ax1(t) + bx2(t) → ay1(t) + by2(t)
Electronic Engineering Capsule 195 YCT
• Some Examples- system, then the system is invertible, otherwise it is
System System System System called non-invertible system.
t
Nature Nature Ex. y(t) = ∫ x(t) dt (Invertible)
y(t) = x (ℓnt) Linear y(t) = e x (ℓnt) t Linear −∞

y(t) = x (et) Linear y(t) = xk (t) Non- d

y(t) = x(t) (Inversesystem)
Linear dt
y(t) = x (sint) Linear y(t) = x (t) + k Non-
Linear Convolution
y(t) = x (| t |) Linear y(t) = sin {x(t)} Non- • Convolution in continuous time
Linear ∞
r (t) = f (t) * h (t) = ∫ f (τ) h (t − τ)dτ
2 x(t)
y(t) = x (t ) Linear y(t) = e Non-
−∞
Linear
y(t) = sint x (t) Linear y(t) = | x (t)| Non- Properties
Linear f (t) * δ (t) = f(t)
y(t) = ℓnt {x (t)} Linear y(t) = [x(t)]¼ Non-
Linear f (t – t0) * δ (t) = f (t– t0)
y(t) = t2 x (t) Linear y(t) = sgn{x(t)} Non- f (t) * δ (t – t0) = f (t– t0)
Linear f (t – t0) * δ (t – t1) = f (t– t0 – t1)
ii. Time-Invariant and time variant systems- A If x1(t) * x2 (t) = y(t) then
system is called time-invariant if a time shift in the 1
input signal causes the same time- shift in the output x1(at)*x2 (at) = y ( at ) ,a ≠ 0
signal. |a|
If x(t) → y(t), then x (t – t0) → y (t – t0) • Convolution in discrete time-
System System System System ∞

y(t)= tx(t)
Nature
T.V. y(t) = Re{x(t)}
Nature
T.V.
r [n] = f [n] * h[n] = ∑ f (k) h (n − k)
K =−∞
y(t) = x(–t) T.V. y(t) = Im{x(t)} T.I.V.
y[n]= x[2n] T.V. T.I.V. • Properties
d
y(t) = x (t) f [n] *δ[n] = f [n]
dt
y(t) = x[sin(t)] T.V. y[n] = x[n]–x[n–1] T.I.V. f [n] * δ[n] = δ [n]
y(t) = 2t + x(t) T.V. y(t) = T.V. f [n-n0] * δ[n-n1] = f [n-n0-n1]
t
δ [n-n0] * δ[n-n1] = δ [n-n0-n1]
∫−∞ x ( 3τ ) dτ
y(t) = e–tx(t+2) T.V. t T.I.V. • Commutative property : f(t) * h(t) = h(t) * f(t)
y(t) = ∫−∞ x ( τ )dτ • Distributive property : Parallel combination
y(t) = Even{x(t)} T.V.
y[t] = e −2k
.x ( t ) T.I.V. f(t) * [h1(t) + h2(t)] = f(t) * h1(t) + f(t) * h2(t)
y[n] = Odd{x(t)} T.V. |n| T.V. • Associates property: Series or cascaded
y[n] = ∑ x(k) combination f(t)*[h1(t) * h2 (t)] = [f(t) * h1 (t)] * h2
(t)
k =−∞
iii. Causal and non-causal system- A system is said to The Laplace Transform
be causal, if the present output depends upon the The Bilateral (Two-sided) Laplace Transform
present input and past values of input but not on the ∞
future values. A system is called non-causal if is not
causal.
X(s) = ∫
−∞
x(t)e −st dt

Ex. of causal system → y(t) = x ( t ) Inverse Laplace Transform

1 c+ j∞
• All memory-less systems are causal, but not vice-
versa
X(t) =
2πj c− j∞ ∫
X(s)est ds
iv. Static and dynamic systems- A system is said to be Unilateral Laplace Transform
static or memory less if the output at any instant ∞
depends only on the input at that instant otherwise,
the system is a dynamic system with memory.
X(s) = ∫
0
x(t)e −st dt

Ex. of static system→ y(t) = 2x(t) • Region of convergence (ROC)

v. Stable and unstable systems- A system is bounded The range variation of σ for which the Laplace
input bounded output i.e. BIBO stable if for any transform converges is called region of convergence.
bounded input X the corresponding output Y is also ∞
i.e. ∫ f ( t ) e −σt < ∞
bounded otherwise it is unstable system. −∞

Ex. stable system→ y(t) = ex(t) • Properties of ROC of Laplace Transform-

vi. Invertible and Non-Invertible system -If it is ROC contains strip lines parallel to jω axis in s-
possible to define an inverse system of a given plane.
Electronic Engineering Capsule 196 YCT
 ROC does not contain any poles, it is bounded by e − at u ( t ) 1 Re ( s ) > − Re ( a )
the poles. s+a
If f(t) is the finite duration and the absolutely
integrable then the ROC is entire s-plane. −e − at u ( − t ) 1 Re ( s ) < − Re ( a )
If f(t) s a right sided and if the line R e {s} = σ0 is in s+a
the ROC then all values of s for which R e {s} > σ0 te − at u ( t ) 1 Re ( s ) > − Re ( a )
(s + a )
2
will also be in ROC.
Note :
− te u ( − t )
− at
1 Re ( s ) < − Re ( a )
For stability ROC must include imaginary axis.
(s + a )
2
Solutions of Laplace transform is unique only when
the ROC is given.
[ cos ω0 t ] u ( t ) s Re(s)>0
• Laplace Transforms Pairs s + ω02
2
x(t) X(s) ROC
δ(t) 1 All s [sin ω0 t ] u ( t ) ω0 Re(s)>0
u(t) 1 Re(s)>0 s + ω02
2

s e − at cos ω0 t  u ( t ) s+a Re ( s ) > − Re ( a )

–u(–t) Re(s)<0 (s + a )
2
1 +ω
2
0
s
tu(t) Re(s)>0 e − at sin ω0 t  u ( t ) ω0 Re ( s ) > − Re ( a )
1
( s + a ) + ω20
2
s2
k
t u(t) k! Re(s)>0
k +1
s
• Properties of the Laplace Transform
Property Signal Transform ROC
Linearity a1 x 1 ( t ) + a 2 x 2 ( t ) a1X1 ( s ) + a 2 X 2 ( s ) R ' ⊃ R1 ∩ R 2
Time shifting x(t – t0) e − st0 X ( s ) R’ = R
Shifting in s e x (t)
s0 t
X (s − s0 ) R ' = R + Re ( s0 )
Time scaling x(at) 1 s R’ = aR
X 
|a| a 
Time reversal x(–t) X(–s) R’ = –R
Differentiation in t dx ( t ) sX(s)–x(0) R'= R
dt
tx ( t ) dX ( s )
−
ds
Differentiation in s R’=R
tn x ( t ) n d X (s)
n

( −1)
ds n
Integration R ' = R ∩ {Re ( s ) > 0}
t
∫ x ( τ ) dτ
1
0
X (s )
s
Convolution x1 ( t ) * x 2 ( t ) X1 ( s ) X 2 ( s ) R ' ⊃ R1 ∩ R 2
• Initial value theorem x(0) = limsX ( s ) • Final value theorem x(∞ ) = limsX ( s )
s →∞ s →0

Note : ∞
Initial value theorem is applicable only for proper Unilateral Z-Transform X(z) =
n =0
x(n)z − n ∑
Laplace i.e. denominator power of function is more
than numerator power. 1
∫
n −1
Final value theorem is not valid when poles lies in Inverse Z-Transform X(n) = 2πj c X(z)z dz
right side of s-plane.
∞ Stability
Z-Transform- X(z) = x(n)z
n =−∞
−n
∑ An LTI system is stable if and only if the ROC of its
system function H(z) includes the unit circles, |Z| =1.

Electronic Engineering Capsule 197 YCT

 A causal LTI system with rational system function −a n u [ − n − 1] 1 |z|<|a|
H(z) is stable if and only if all of the poles of H(z)
1 − az −1
lie inside the unit circle i.e., the must all have
magnitude smaller than 1. na n u [ n ] az −1 |z|>|a|
Some Common Z-Transform pairs
(1 − az ) −1 2

x(n) X(z) ROC

δ[n ] 1 All z − na n u [ − n − 1] az −1 |z|<|a|

u[n] 1 |z|>1 (1 − az ) −1 2

1 − z −1 [cos ω0 n ] u [ n] z 2 − ( cos ω0 ) z |z|>1

–u[–n-1] 1 |z|<1 z 2 − ( 2 cos ω0 ) z + 1
1 − z −1
δ[n − m] z–m All z except [sin ω0 n ] u [ n] ( sin ω0 ) z |z|>1
0 if (m>0) or z 2
− ( 2 cos ω0 ) z + 1
∞ if (m<0)
a n u [n] 1 |z|>|a|
1 − az −1
Some properties of the Z- Transform
Property Sequence Transform ROC
Linearity a1 x 1 [ n ] + a 2 x 2 [ n ] a1X1 ( z ) + a 2 X 2 ( z ) R ' = R1 ∩ R 2
Time shifting x(n – n0) z − n0 X ( z ) R'= R
Multiplication by Z n
0 z x [n]
n
0  Z R ' =| Z0 | R
X 
 Z0 
Multiplication by e jω0 n e jω0 n x [ n ] (
X e − jω0 Z ) R’ = R
Time reversal x[-n] 1 1
X  R'=
Z R
Multiplication by n nx [n] dX ( z ) R' = R
−z
dz
Accumulation n 1 R ' = R ∩ {| z |> 1}
∑ x [k] X (z )
k =−∞ 1 − z −1
Convolution x1 [ n ] * x 2 [ n ] X1 ( z ) X 2 ( z ) R ' = R1 ∩ R 2
Initial value theorem f ( 0 ) = lim F ( z )
z →∞

Final value theorem f ( ∞ ) = lim ( z − 1) F ( z )

z →1
(If final value exists)
Fourier Series ∞

• It is an approximation process by which any non- ∫ x (t).x

−∞
1 2 (t) = 0
sinusoidal waveform is converted into sinusoidal
waveform. i.e. area of [x1 (t). x2 (t)] graph = 0
• A Fourier analysis also called as a harmonic Fourier series representation of continuous time
analysis. periodic signals
• Dirichlet's conditions : conditions for Fourier series 1. Trigonometric Fourier series
∞

∑ (a
to exist.
f(t) = a0 + n cos ω0 t + b n sin ω0 t)
1. ∫
T
f (t) dt < ∞ n =1
Where, ω0 = Fundamental frequency
2. There must be finite number of maxima and minima a0 , an , bn = Trigonometric Fourier series coefficient
in a given time period for f (t). nω0 = nth harmonic of ω0
3. There must be finite number of discontinuities in a 1 T 2 T
given time period for f (t). a0 =
T 0 ∫
f (t)dt , an =
T 0 ∫
f (t)cos nω0 t dt
• Condition for orthogonality 2 T
Signal x1(t) and x2(t) are said to be orthogonal if bn =
T 0 ∫
f (t)sin nω0 t dt

Electronic Engineering Capsule 198 YCT

• Here a0 is the value of constant component of the an b
signal f(t). an = (Cn + C–n), −j n
Cn =
2 2
• The Fourier coefficient an and bn are maximum an bn
amplitude of nth harmonic component. bn = j (Cn – C–n), C–n = + j
2 2
2. Polar form of Trigonometric Fourier series
∞
Effect of symmetry of Fourier coefficients
f(t) = C0 + Σ Cn cos(nω0 t − φn ) Odd
n =1 Symmetry
b 
Where, C0 = a0 , Cn = a n2 + b n2 , φn = tan −1  n 
 an 

x(t) = –x(–t)
For signals with odd symmetry, the
Fourier coefficients a0 and an are zero.
Even
symmetry-

3. Exponential Fourier series-

∞ T x(t) = x (–t)
1
f(t) = Σ Cn e jnω0 t Where, C0 = a0 =
n =−∞ T ∫
0
f (t)dt For signals with even symmetry, the
Fourier coefficient bn is zero.
1 T Half-wave
Cn =
T ∫
0
f (t)e− jnω0 t dt symmetry
 T 
x(t) = − x  t ± o 
 2
Note- For HWS signal, average value will
Exponential Fourier series is compact form of Fourier equal to zero, but converse of this
series. statement is not true

Positive and negative frequency indicate different x (t ± T/2) = –x(t)

phase of rotation, they maintain same magnitude but For signals with half-wave symmetry,
different phase. the Fourier series will consist of odd
Relation between exponential and Trigonometric harmonic terms of sine and cosine
Fourier series– signals..
Function Cn Fourier Coefficient Trigonometric Fourier series
Real (Neither even nor Generally complex a0 ≠ 0 DC term, sine terms and cosine terms
odd) Cn = C*− n an = b n ≠ 0 are present

Even Real (Even in nature) a0 ≠ 0 DC term and cosine terms are present
an ≠ 0
bn = 0
Odd Imaginary a0 = 0 Only sine terms are present
(odd in nature) an = 0
bn ≠ 0
Half wave
symmetry
Cn = 0 ; For n = even an = 0
}
;a =0
bn = 0 0
Odd sine and odd cosine terms are
present
;n = even
Even and half wave
symmetry
Cn = Real & even
Cn = 0 ; n = even
an = 0
}
;a =0
bn = 0 0
Only odd cosine terms are present

n = even
Odd and half wave
symmetry
Cn = imaginary &
odd
an = 0
bn = 0 }
; n = even
Only odd sine terms are present

Cn = 0 ; n = even
Electronic Engineering Capsule 199 YCT
 Properties of Exponential form of Fourier series Properties of Discrete Time Fourier Series
coefficients
Property Continuous Fourier series
Property Continuous time Fourier series time periodic coefficient
periodic signal coefficients signal
Linearity Ax (t) + By (t) ACn + Bdn
Linearity Ax[n] + By[n] ACk + Bdk
Time shifting x (t ± t0) Cn e ± jmω0 t 0
Time shifting x[n ± n0]  2π 
± jk  n 0
Frequency ± jmω0 t Ck e N
e x(t) Cn∓ m
shifting
Frequency  2π 
Conjugation x*(t) C*–n ± jM   n Ck∓m
shifting e  N  x[n]
Time reversal x(–t) C–n
Time scaling x(αt); α > 0 [x(t) Cn (No change Conjugation x*[n] C*–k
is period with in Fourier Time reversal x[–n] C–k
period T/α] coefficient)
Time scaling x(m)[n] 1
Multiplication x(t) y(t) +∞ Ck (Period mN)
∑c md n − m
 n
=  x   ; if n
m
m =−∞  m
Differentiatio d jnω0cn is a multiple of
n x(t) m
dt
Integration t 1 0 ; otherwise
∫ −∞
x (t) dt
jnω0
cn Multiplication x[n] y[n] N −1

(Finite valued ∑C b
l =0
l k −l
and periodic
only if a0 = 0) First difference x[n] –x [n –1]   2π 

Periodic TCndn  − j  k 
convolution ∫ x(τ)y(t − τ)dτ
T
1 − e    k

N

C

Symmetry of x(t) is real Cn = C*− n |Cn|=|

real signals Running sum  
C–n|; ∠Cn = –∠ n

C–n Re {Cn} = ∑ x[k] 


1 
 2π  C
− jk    k
Re {C-n } k =−∞  1− e  N  
 
Im {Cn} = –Im Convolution N −1 Nakdk

Real and even x(t) real and

{C–n}
Cn are real and
∑
r =0
x[r]y[n − r]

even even
Parseval's Power Theorem- Parseval's power
Real and odd x(t) real and odd Cn are
theorem is given by
imaginary and
odd ⇒ For continuous time signals,
Imaginary x(t) imaginary Cn, imaginary T ∞ ∞

∑ ∑
1 1
∫
2 2
and even and even and even P= x(t) dt = C n = a 02 + (a n2 + b n2 )
T n =−∞
2 n =1
Imaginary x(t), imaginary Cn, real and odd 0
and odd and odd ⇒ For discrete signals,
Where, Cn → Fourier series coefficient of x(t) N −1 N −1

∑ ∑
1 2 2
dn → Fourier series coefficients of y(t) P= x[n] = Ck
Fourier Series Representation of Discrete Time N n =0 k =0
Periodic Signals-
Fourier series representation of x[n], with period 'N'
The Fourier Transform-
is given as- ∞ 1 ∞
N−1  2π 
jk n
X(ω) =
−∞ ∫
x(t)e − jωt dt , x(t) =
2 π −∞ ∫
X( ω)e jωt dω

∑
x[n] = Ck e  N  ,Where,
k =0
Common Fourier Transform pairs
x(t) X ( ω) x(t) X ( ω)
N−1  2π 
− jk n
∑
1
Ck = x[n]e  N  δ(t) 1 e − at u ( t ) ,a > 0 1
N n=0
jω + a
Ck is periodic with period 'N', Ck = Ck+N
Electronic Engineering Capsule 200 YCT
 δ ( t − t0 ) e − jωt0 te − at u ( t ) ,a > 0 1 Time reversal x(–t) X ( −ω)
( jω + a )
2
Duality x(t) 2πX ( −ω)
Time dx ( t ) jωX ( ω)
1 2πδ ( ω) e − a|t| , a > 0 2a differentiation
a 2 + ω2 dt
Frequency ( − jt ) x ( t ) dX ( ω)
differentiation
e jω0 t
2πδ ( ω − ω0 ) 1 e − a|ω|
dω
a + t2
2
Integration t 1
u(t) 1 sgn(t) 2 ∫−∞ x ( τ )dτ πX( 0) δ( ω) +
jω
X( ω)
πδ ( ω) +
jω jω Convolution x1 ( t ) * x 2 ( t ) X1 ( ω ) X 2 ( ω )
u ( −t ) 1 x1 ( t ) x 2 ( t )
πδ ( ω) − Multiplication 1
X1 ( ω) *X2 ( ω)
jω 2π
cos ω0 t π  δ ( ω − ω0 ) + δ ( ω + ω0 ) Discrete time Fourier Transform
(DTFT)
+∞
sin ω0 t − jπ  δ ( ω − ω0 ) − δ ( ω + ω0 ) DTFT{x [ n ]} = X [ Ω ] = ∑ x [ n ] e − jΩn
n =−∞
Where ' Ω ' is discrete time frequency in
Properties of the Fourier transform
‘rad/number’. The transform X ( Ω ) is frequency
Property Signal Fourier
transform domain description of x[n]
Note:
Linearity a1x1 ( t ) + a 2 x 2 ( t a1X1 ( ω) + a 2 X2 ( ω) x [n] 
DTFT
→ X [Ω]
( Discrete aperiodic ) (Periodic continuous)

Time shifting x(t – t0) e − jωt 0

X ( ω) Inverse Discrete Time Fourier Transform
(IDTFT)
Frequency e jω0 t x ( t ) X ( ω − ω0 ) 1
IDTFT {X ( Ω )} = x [ n ] = X ( Ω ) e jΩn dΩ
shifting 2π 2∫π
Time scaling x ( at ) 1  ω Properties of DTFT
X 
|a |  a 
Property Sequence Fourier Transform
x[n] X(Ω)
x1[n] X1(Ω)
x2[n] X2(Ω)
Periodicity x[n] X(Ω+2π)=X(Ω)
Linearity a1x1[n]+a2x2[n] a1X1(Ω)+a2X2(Ω)
Time shifting x[n-n0] e − jΩn 0 X ( Ω )
Frequency shifting e jΩ0n x[n] X(Ω−Ω0)
Conjugation x*[n] X*(– Ω)
Time reversal x[–n] X(–Ω)
Time scaling  x [ n / m ] if n = km X(m Ω)
x ( m) [ n ] = 
0 if n ≠ km
Frequency nx[n] dX ( Ω )
differentiation j
dΩ
First difference x[n]–x[n–1] ( e − jΩ ) X ( Ω )
1 −
Accumulation ∞ 1
∑ x [k] πX ( 0 ) δ ( Ω ) + X ( Ω ) , | Ω |≤ π
k =−∞ 1 − e jΩ
Convolution x1[n]*x2[n] X1 ( Ω ) X 2 ( Ω )
Multiplication x1 [ n ] x 2 [ n ] 1
X1 ( Ω ) ⊗ X 2 ( Ω )
2π
Electronic Engineering Capsule 201 YCT
 Real sequence x [n ] = x e [n ] + x 0 [n ] X ( Ω ) = A ( Ω ) + jB ( Ω ) , X ( −Ω ) = X * ( Ω )
Even component x e [n ] R e {X ( Ω )} = A ( Ω )
Odd component x0[n] jIm {X ( Ω )} = jB ( Ω )
Parseval’s relations ∞
1
∑ x [ n ]x [ n ] = 2π ∫
n =−∞
1 2 2π
X1 ( Ω ) X 2 ( −Ω ) dΩ
∞
1
∑ x [n] X ( Ω ) dΩ
2 2

n =−∞
=
2π ∫2 π

Some Common Discrete Time Fourier Transform Note: x[n] ← DFT
→ X[K]
Pair Discrete periodic Discrete periodic

Sequence Fourier Transform Inverse DFT

x[n] X(Ω) 1 N −1 2 πK

[ ] [ ]
j n

δ[n ] 1 x n = ∑
N K =0
X K e N

δ[n − n0 ] e − jΩn 0
Where ‘N’ is period of the signal x[n]
x[n]=1 2πδ ( Ω ) ,| ( Ω ) |≤ π Properties of DFT
Linearity
e jΩ0 n 2πδ ( Ω − Ω0 ) ,| Ω 0 |≤ π Then x1 [ n ] ← DFT
→ X1 [ K ]
cos Ω0 n π δ( Ω−Ω0 ) +δ( Ω+Ω0 ) ,| Ω |,| Ω0 | ≤ π x [ n ] ← DFT
→ X [K] 2 2

Then αx1 [ n ] + βx 2 [ n ] ←

DFT
→αX1 [ K ] + βX 2 [ K ]
sin Ω0 n −jπδ( Ω−Ω0 ) −δ( Ω+Ω0 ) ,| Ω|,| Ω0 | ≤π
Time shifting
u[n]
πδ ( Ω ) +
1
,| Ω |≤ π If x [ n ] ←
DFT
→ X [K]
1 − e − jΩ 2π
−j
Then x [ n − n 0 ] ← X [K]
Kn 0
–u[–n–1] 1 DFT
→e N
−πδ ( Ω ) + ,| Ω |≤ π
1 − e − jΩ Shift in k-domain
n
a u[n],|a| <1 1 If x [ n ] ←
DFT
→ X [K]
1 − ae − jΩ
Then x[n]
−a u[ −n −1] ,| a |>1
n
1 2π
→ X [K − K 0 ]
j K0n
1 − ae − jΩ e N
←
DFT

( n +1) anu[ n],| a |<1 1 Conjugate symmetry

(1 − ae ) − jΩ 2
If x [ n ] ←
DFT
→ X [K]
a |n| , | a |< 1 1− a2 Then x * [ n ] ←
DFT
→ X * [−K ]
1 − 2a cos Ω + a 2 Time reversal
1 | n|≤ N1   1  If x [ n ] ←
DFT
→ X [K]
x[ n] =  sin Ω  N1 +  
0 | n|> N1   2  Then x [ −n ] ←
DFT
→ X [ −K ]
sin ( Ω / 2 )
Or x [ N − n ] ←
DFT
→ X [ N − K]
1 0 ≤| Ω |≤ W
,0 < W < π X ( Ω ) = 
sin Wn
Periodic convolution
πn 0 W <| Ω |≤ π If x1 [ n ] ←
DFT
→ X1 [ K ] (with period N0)
∞ ∞
2π And x 2 [ n ] ←
DFT
→ X 2 [ K ] (with period N0)
∑ δ [ n − kN 0 ]
k =−∞
Ω0 ∑
k =−∞
δ ( Ω − kΩ 0 ) , Ω 0 =
N0
Discrete Fourier Transform (DFT)
(with period N0)
The Discrete Fourier transform (DFT) is basically Linear Convolution
fourier transform of “sampled signal repeated If x1 [ n ] ←
DFT
→ X1 [ K ] (with period N1)
periodically”. The number of point in DFT known as
N-point DFT. And x 2 [ n ] ←
DFT
→ X 2 [ K ] (with period N2)
N −1 2 πK
DFT{x[n]} = X[K] = ∑ x[n]e
−j
N
n
Then x1 [ n ] * x 2 [ n ] ←
DFT
→ X1 [ K ] X 2 [ K ]
n =0 (with period N1+N2–1)

Electronic Engineering Capsule 202 YCT

 Parseval’s relation Finite Impulse Response (FIR) filter- The impulse
N −1
1 N −1 response is of finite duration. The differences
E = ∑ | x [ n ] |2 = ∑ | x [ K ] |2 equation is given by-
n =0 N k =0
N
Correlation- The correlation of two functions or y [ n ] = ∑ bi x [ n − i ]
signals or waveforms is defined as the measure of i =0
similarity between those signals. There are two types Where, x [ n ] , y [ n ] → Input and output respectively
of correlations-
(i) Cross-correlation (ii) Auto-correlation N→ Filter order
For two signals f ( t ) and g(t), correlation is defined bi→ Value of impulse response at ith instant
Properties-
as-
∞
No feedback is required
R fg ( τ ) = ∫ f ( t ) g ( t − τ ) dt =  f ( t ) ⊗ g ( − t )  t =τ They are inherently stable.
−∞
Auto correlation function is a special case where They can be easily designed to be linear phase by
both the signals are same. making the coefficient sequence symmetric.
For energy signals, Computationally complex.
∞ Basic Structure for FIR systems
R ff ( τ ) = ∫ f ( t ) f ( t − τ ) dt = f ( t ) ⊗ f ( − t )  t =τ The system function of the FIR system whose
−∞
For power signals, realization take the form of a non-recursive
1 ∞ computational algorithm may be written as-
R ff ( τ ) = lim ∫ f ( t ) f ( t − τ ) dτ M −1
H ( z ) = ∑ h ( n )z − n
s →∞ T −∞

Properties of auto-correlation function n =0

R ff ( τ ) is even, R ff ( τ ) = R ff ( −τ ) That is if the impulse response is M samples in
R ff ( 0 ) > R ff ( τ ) duration, the H(z) is a polynomial in Z–1 of degree
M–1. Thus H(z) has (M-1) poles at Z = 0 and (M-1)
For energy signals, energy = Rff (0) zeros that can be anywhere in the finite z-plane.
For power signals, power = Rff(0)
• Infinite impulse response filters.
For energy signals, Fourier transform of R ff ( τ ) The impulse response of IIR filters continues
gives energy spectral density, i.e. indefinitely. The difference equation is given as-
R ff ( τ ) → F ( ω) F* ( ω) [ ESDf ] 1 P Q

y [ n ] =  ∑ bi x [ n − i ] − ∑ a j y [ n − j] 
For power signals, Fourier transform of R ff ( τ ) a 0  i =0 j=1 
gives power spectral density, i.e. Where,
| FT ( ω) |2 P→ Feed forward filter orders
R ff ( τ ) ←
→ lim
T →∞
[ PSDf ] Q→ feedback filter order
T
Fast Fourier Transform bi→ Feedforward filter coefficients
The Fast Fourier Transform is an algorithm that computes aj→ Feedback filter
the discrete Fourier Transform (DFT) of a sequence. It Properties
reduces the number of the computations needed to find
transform. Feedback is required
For N- Point DFT Stability depends on the filter coefficients
N They are preferred over FIR filters for their
Number of multiplications = log 2 N efficiency in implementation.
2
Number of additions = N log2N Basic structures for IIR systems.
Sampling Causal IIR systems are characterized by the constant
Any band limited signal x(t) is uniquely determined coefficient difference equation or equivalently, by
by its samples if the sampling frequency (ωs) is the real rational transfer function of equation. From
greater than twice the maximum frequency (ωm) these equations, it can be seen that the realization of
present in the signal. infinite duration impulse response (IIR) systems
There are two types of sampling. involves a recursive computational algorithm. In this
Ideal/instantaneous sampling (zero pulse width) section, the most important filter structures namely
Practical sampling (Non-zero pulse width) direct forms- I and II, cascade and parallel
(i) Natural sampling (ii) Flat top sampling realizations for IIR system are discussed.
Digital filters: M

y(z) ∑
−k
b z k
Digital filter is a system that performs mathematical
H (z) = =
k =0
operations on a sampled, discrete time signal to x (z) N

reduce or enhance parameters of that signal. There ∑a z

k =1
k
−k

are two types-

Electrical Engineering Capsule 203 YCT