Unit I: Discrete Time Signals and Systems

Dr. Raghu Indrakanti, Ph.D.

Assistant Professor
E.C.E Department

Department of Electronics and Communication Engineering

CVR College of Engineering
OUTLINE

Convolution of Linear Time-invariant Systems

Discrete Time Signals
Discrete Time Systems
Linear Time Invariant Systems
Linear Convolution
Stability and Causality
Linear Constant Coefficient Difference Equations
Frequency domain representation of Discrete Time Signals and
Systems
Introduction to DSP

Digital Signal Processing

The rapid evolution of digital computing technology which started in the 1960s, marked
the transition from analog to digital signal processing. DSP is concerned with the repre-
sentation of analog signals by sequences of numbers, the processing of these sequences
by numerical computation techniques, and the conversion of such sequences into analog
signals. DSP has evolved through parallel advances in signal processing theory and the
technology that allows its practical application.

Figure: Simplified block diagram of idealized system for (a) continuous-time processing of discrete-time signals, and (b) its
practical counterpart for digital processing of analog signals.

1/45
Introduction to DSP: Mathematical representation of signals
Consider the continuous-time signal defined by
 −2t
e cos(3πt), t≥0
s(t) = (1)
0, t<0

If sample s(t) with a sampling periodT = 0.1s, the discrete time signal
e −0.2n cos(0.3πn), n ≥ 0
s[n] = s(t)|t=nT = (2)
0, n<0

2/45
Introduction to DSP: Systems

A system is defined as a process where a signal called input is transformed into

another signal called output. Systems are classified based on the category of input
and output signals.

Figure: Block-diagram representation of a continuous-time system (a) and a discrete-time system (b).

3/45
Introduction to DSP: Systems

Continuous-time systems
A continuous-time system is a system which transforms a continuous-time input signal
x (t) into a continuous-time output signal y (t).
Z t
y (t) = x (τ )dτ (3)
−∞
Symbolically, the input-output relation of a continuous-time system is represented by
H
x (t) 7→ y (t) or y (t) = H {x (t)} (4)

Discrete-time systems
A system that transforms a discrete-time input signal x [n] into a discrete-time output
signal y [n], is called a discrete-time system.

H
x [n] 7→ y [n] or y [n] = H {x [n]} (5)
The discrete-time equivalent of the continuous-time integrator system is the
accumulator system
n
X
y [n] = x [k] (6)
k=−∞
4/45
Introduction to DSP: ADC

Analog-to-Digital Converter (ADC)

The conversion of an analog (continuous-time, continuous-amplitude) signal into a digital
(discrete-time, discrete-amplitude) signal is called as Analog-to-Digital Converter (ADC).

Figure: Block diagram representation of an ideal D/A converter (a) and a practical D/A converter (b) with the
corresponding input and output signals. In most practical applications, the staircase output of the D/A converter is subsequently
smoothed using an analog reconstruction filter.

5/45
Introduction to DSP: DAC

Digital-to-Analog Converter (DAC)

The conversion of a discrete-time signal into continuous time form is done with an interface
system called digital-to-analog (D/A) converter (DAC).

Figure: Block diagram representation of the analog-to-digital conversion process. (b) Examples of the signals x (t), x [n],
and xd [n] involved in the process. The amplitude of x [n] is known with infinite precision, whereas the amplitude of xd [n] is
known with finite precision ∆ (quantization step or resolution).

6/45
Introduction to DSP: Classes of Systems

Figure: The three classes of system: analog systems, digital systems, and interface
systems from analog-to-digital and digital-to-analog.

7/45
Introduction to DSP: Applications

Figure: Simplified block diagram of a digital cellular phone.

8/45
Discrete Time Systems

Causal System
A system is called causal if the present value of the output does not depend on future
values of the input, that is, y [n0 ] is determined by the values of x [n] for n ≤ n0 , only.

Stable System
A system is said to be stable, in the Bounded-Input Bounded-Output sense, if every
bounded input signal results in a bounded output signal, that is

|x [n]| ≤ Mx < ∞ ⇒ |y [n]| ≤ My < ∞ (7)

A signal x [n] is bounded if there exists a positive finite constant Mx such that |x [n]| ≤ Mx
for all n.

9/45
Discrete Time Systems

Linear System
A system is called linear if and only if for every real or complex constant a1 , a2 and every
input signal x1 [n] and x1 [n]

H{a1 x1 [n] + a2 x2 [n]} = H{a1 x1 [n]} + H{a2 x2 [n]} (8)

for all values of n.

Example: Test whether the following square-law system is linear or nonlinear:
y [n] = x 2 [n]
Sol:
1 Apply the input x1 [n] and the output y1 [n] = x12 [n].
2 Apply the input x2 [n] and the output y2 [n] = x22 [n].
3 Apply the input x [n] = a1 x1 [n] + a2 x2 [n]. The result is
y [n] = a12 x12 [n] + a22 x22 [n] + 2a1 a2 x1 [n]x2 [n].
4 Form the signal y 3[n] = a1 y1 [n] + a2 y2 [n] = a1 x12 [n] + a2 x22 [n] and check whether it
is equal to y [n]. If the answer is yes, then the system is linear; otherwise it is
nonlinear.

Since y3 [n] 6= y [n], the system is nonlinear.

10/45
Discrete Time Systems

Time Invariant System

A system is called time-invariant or fixed if and only if

y [n] = H{x [n]} ⇒ y [n − n0 ] = H{x [n − n0 ]} (9)

for every input x [n] and every time shift n0 . That is, a time shift in the input results in a
corresponding time shift in the output.

Example: Test whether the following system is time-invariant or time-varying:

y [n] = x [n]cos[ω0 n].
Sol:
1 Apply an input x1 [n] = x [n] and the output y1 [n] = x [n]cos[ω0 n].
2 Apply the shifted input x2 [n] = x [n − n0 ] and the output y2 [n] = x [n − n0 ]cos[ω0 n].
3 Check whether the shifted sequence y1 [n − n0 ] is equal to y2 [n]. If the answer is
yes the system is time-invariant; otherwise it is time-varying.

Since y1 [n − n0 ] = x [n − n0 ]cosω0 ([n − n0 ]) 6= y2 [n] the system is time-varying.

11/45
Discrete Time Systems: Test for linearity and time invariance

Example: A down-sampler is a system, y [n] = H{x [n]} = x [nM]

Linearity
let y1 [n] = x1 [nM] and y2 [n] = x2 [nM]. Consider the down-sampler output due to the
input x [n] = a1 x1 [n] + a2 x2 [n] given by

y [n] = H{x [n]} = x [nM] = a1 x1 [nM] + a2 x2 [nM] (10)

y [n] = a1 y1 [n] + a2 y2 [n] (11)

Hence the down-sampler is linear.

Time Invariant
Consider the output y2 [n] due to the input x [n − n0 ]

y2 [n] = H{x [n − n0 ]} = x [nM − n0 ]

6= x [n − n0 ] = x [(n − n0 )M] = x [nM − n0 M]

12/45
Convolution of Linear Time-invariant Systems

Impulse Response
The response of a linear time-invariant (LTI) system to any input can be determined
from its response h[n] to the unit sample sequence δ[n], using a formula known as
convolution summation.
The sequence h[n], which is known as impulse response.

Figure: The impulse response of a linear time-invariant system.

13/45
Convolution of Linear Time-invariant Systems

Signal decomposition into Impulses

Let us define the sequence

x [n], n=k
xk [n] = (12)
0, n 6= k

The sequence xk [n] can be obtained by multiplying the sequence x [n] by the sequence

1, n=k
δ[n − k] = (13)
0, n 6= k

Hence, the sequence x [n] can be expressed as

∞
X
x [n] = x [k]δ[n − k] −∞<n <∞ (14)
k=−∞

14/45
Convolution of Linear Time-invariant Systems

Signal decomposition into Impulses: Example

Let us define the unit step sequence

u[n] = δ[n] + δ[n − 1] + δ[n − 2] + δ[n − 3] + .... (15)

∞ n
X X
u[n] = δ[n − k] = δ[k] (16)
k=0 k=−∞

Convolution Summation Equation

Hence, the response of the LTI system y [n] can be expressed as
∞
X
y [n] = x [k]h[n − k] −∞<n <∞ (17)
k=−∞

15/45
Convolution of Linear Time-invariant Systems

Convolution Summation Steps:

1 Change the index of the sequences h[n], x [n] from n to k and plot them as a function
of k.
2 Flip or fold the sequence h[k] about k = 0 to obtain the sequence h[−k].
3 Shift the flipped sequence h[−k] by n samples to the right, if n > 0, or to the left,
if n < 0.
4 Multiply the sequences x [k] and h[n−k] to obtain the sequence zn [k] = x [k]h[n−k].
5 Sum all non-zero samples of zn [k] to determine the output sample at the given value
of the shift n.
6 Repeat steps 3 to 5 for all desired values of n.

16/45
Convolution of Linear Time-invariant Systems

The computation of convolution in tabular form

The input sequence is given x [n] = {1, 2, 3, 4, 5, }
The LTI system is given by h[n] = {−1, 2, 1}

17/45
Convolution of Linear Time-invariant Systems
The input sequence is given x [n] = {1, 1, 1, 1, 1, 1} and h[n] = {1, 0.5, 0.25, 0.125}

The y [n] = {1, 1.5, 1.75, 1.875, 1.875, 1.875, 0.875, 0.375, 0.125}
18/45
Convolution of LTI Systems: Example

19/45
Convolution of LTI Systems: Example

Sol:

20/45
Convolution of LTI Systems: Example

21/45
Convolution of Linear Time-invariant Systems

Causality
linear time-invariant system with impulse response h[n] is causal if

h[n] = 0 forn < 0 (18)

22/45
Convolution of Linear Time-invariant Systems

Stability
A linear time-invariant system with impulse response h[n] is stable, in the bounded-input
bounded-output sense, if and only if the impulse response is absolutely summable, that
is, if
∞
X
|h[n]| < ∞ (19)
n=−∞

Proof:

23/45
Convolution of LTI Systems: Stability

Example: Consider the system with impulse response h[n] = ban u[n]. To test whether
the system is stable.
Sol: check if the following sum is finite
∞ ∞
X X
Sh = |h[k]| = |ban u[n]| (20)
k=−∞ k=−∞
∞
X
Sh = |b||an | (21)
k=0
∞
X
Sh = |b| |an | (22)
k=0

1
Sh = |b| (23)
1 − |a|

|b|
1 If |a| < 1, the sum converges to 1−|a|

2 Therefore, the system is stable only when |a| < 1.

3 The impulse response decays asymptotically to zero.

24/45
Important Topics in Unit I
1 Test the given system is:
Energy Signal
Power signal
2 Test the given system is:
Linearity
Time Invariant
Stability
Causality
memory
3 Derivation of Stability condition of an LTI system
4 Properties of LTI System
5 Convolution of LTI System
Use Tabular or matric method for 2 marks
Graphical method or mathematical method for 5 or 10 marks
Using Frequency domain theorem: Multiplication
6 Solution of linear constant-coefficient difference equations
Using time domain
Frequency domain
Impulse response of the LTI system
25/45
Convolution of LTI Systems

Properties of Convolution
The response of the LTI system y [n] can be expressed as
∞ ∞
X X
y [n] = x [k]h[n − k] = h[k]x [n − k] (24)
k=−∞ k=−∞

y [n] = x [n] ∗ h[n] = h[n] ∗ x [n] (25)

1 Commutative: x [n] ∗ h[n] = h[n] ∗ x [n]

2 Associative: (x [n] ∗ h1 [n]) ∗ h2 [n] = x [n] ∗ (h1 [n] ∗ h2 [n])
3 Distributive: x [n] ∗ (h1 [n] + h2 [n]) = x [n] ∗ h1 [n] + x [n] ∗ h2 [n]

26/45
Convolution of LTI Systems

27/45
Z-Transform

The z-transform of a sequence x[n] is a function X(z) defined by

∞
X
X (z) = x [n]z −n (26)
n=−∞

where the independent variable z can represent any complex number.

1 The infinite sum in above equation may or may not be finite for all sequences or all

values of z.
2 For any given sequence, the set of values of z for which the series converges is

known as the region of convergence (ROC) of the z-transform.

3 The values of z for which X (z) = 0 are called zeros of X (z), and the values of z

for which X (z) is infinite are known as poles. By definition, the ROC cannot
include any poles.

28/45
Z-Transform

Unit sample sequence

linear time-invariant system with impulse response h[n] is causal if
∞
X X
X (z) = δ[n]z −n = 1z −n = z 0 = 1 ROC : All z (27)
n=−∞ n=0

Causal exponential sequence

The z-transform of the causal exponential sequence x [n] = an u[n] is given by
∞ ∞ ∞
X X X 1
X (z) = an u[n]z −n = an z −n = (az −1 )n = ROC : |z| > |a|
1 − az −1
n=−∞ n=0 n=0
(28)
−1
1 The infinite geometric series converges if |az | < 1 or vertz| > |a|.
1 z
2 Since X (z) = 1−az −1
= (z−a)
, there is a zero at z = 0 and a pole at p = a.

29/45
Z-Transform

30/45
Z-Transform

31/45
Z-Transform: Some common z -transform pairs

32/45
Z-Transform

Region of Convergence (ROC)

1 The ROC cannot include any poles.
2 The ROC is a connected (that is, a single contiguous) region.
3 For finite duration sequences the ROC is the entire z-plane, with the possible
exception of z = 0 or z = ∞.
4 For infinite duration sequences the ROC can have one of the following shapes:

33/45
Z-Transform

System function of LTI systems

The response of the LTI system y [n] can be expressed as
∞ ∞
X X
y [n] = x [k]h[n − k] = h[k]x [n − k] (29)
k=−∞ k=−∞

Input-output relationship is given by

Y (z) = X (z)H(z) = H(z)X (z) (30)

where H(z) is known as the system function or transfer function of the system.

34/45
Z-Transform: Convolution

System function of LTI systems: Example

Determine the response of a system with impulse response h[n] = an u[n],|a| < 1 to the
input x [n] = u[n] using the convolution theorem.
Sol:

35/45
LTI systems characterized by linear constant-coefficient difference equations
(LCCDE)

LTI systems characterized by LCCDE

A class of LTI systems whose input and output sequences satisfy a linear constant-
coefficient difference equation of the form
N M
X X
ak y [n − k] = bk x [n − k] a0 = 1 (31)
k=−N k=−M

If we assume that the system is causal, we have

N
X M
X
y [n] + ak y [n − k] = bk x [n − k] (32)
k=1 k=0

1 In signal processing applications, we assume that the system is initially at rest,

that is, x [n] = y [n] = 0 for n < 0.
2 Therefore, we can set the initial conditions at n = 0 to zero, that is,
x [−1] = ... = x [−M] = 0 and y [−1] = ... = y [−N] = 0, and then recursively
compute the output values y [0], y [1], ..., y [L].
36/45
LTI systems characterized by LCCDE

N
X M
X
y [n] + ak y [n − k] = bk x [n − k] (33)
k=1 k=0

The solution of the LTI system is given by

y [n] = yh [n] + yp [n] (34)

Where yh [n] is the homogeneous solution and yp [n] is the particular solution.
1 The homogeneous solution is the response of the system when there is no input
i.e. x [n] = 0. It is also called as zero input response.
2 The particular solution is the solution of the difference equation for specific input
signal x [n] for n ≥ 0

37/45
LTI systems characterized by LCCDE: Homogeneous Solution

The homogeneous solution is the response of the system when there is no input i.e.
x [n] = 0. It is also called as zero input response.
N
X M
X
ak y [n − k] = bk x [n − k] (35)
k=0 k=0
N
X
ak y [n − k] = 0 (36)
k=0
Let y [n] = λn
N
X
ak λ(n−k) = 0 (37)
k=0
Assume a0 = 1, expanding the above expression

λn + a1 λn−1 + a2 λn−2 + a3 λn−3 + a4 λn−4 + ..... + an λ(n−N) = 0 (38)

λ(n−N) (λN + a1 λN−1 + a2 λN−2 + a3 λN−3 + a4 λN−4 + ..... + aN ) = 0 (39)

The characteristics polynomial of the system is given by
λN + a1 λN−1 + a2 λN−2 + a3 λN−3 + a4 λN−4 + ..... + aN = 0. It has N roots, which are
denoted as λ1 , λ2 , λ1 , ..., λN
38/45
LTI systems characterized by LCCDE: Particular Solution

The particular solution is the solution of the difference equation for specific input signal
x [n] for n ≥ 0.

If x [n] is constant, then yp is also a constant

Let x [n] = u[n]; then yp = Ku[n]

If x [n] is exponential, then yp is also a exponential

Let x [n] = an u[n]; then yp = Kan u[n]

If x [n] is sinusoidal, then yp is also a sinusoidal

Let x [n] = Acos[ω0 n]; then yp = K1 cos[ω0 n] + K2 sin[ω0 n]
Let x [n] = Asin[ω0 n]; then yp = K1 cos[ω0 n] + K2 sin[ω0 n]

39/45
LTI systems characterized by LCCDE

Frequency Domain
apply the z-transform on both sides. Using the properties of linearity and time shifting,
we obtain
N M
X X
(1 + ak z −k )Y (z) = ( bk z −k )X (z) (40)
k=1 k=1

The system function, is given by

PM
b z −k
k=1 k
H(z) = P N
(41)
1+ −k
ak z
k=1

where H(z) is a rational function, that is, the ratio of two polynomials in z −1 .
The form bk x [n − k] correspond to bk z −k and terms of the form ak y [n − k] correspond
to ak z −k , it is straightforward to obtain the difference equation from the system function
and vice versa.

40/45
Z-Transform

System function of LTI systems: Example

Find the difference equation corresponding to the system function
Y (z) 6 − 10z −1 + 2z −2
H(z) = = (42)
X (z) 1 − 3z −1 + 2z −2
Sol:
Cross-multiplying the numerator and denominator terms, we obtain

(1 − 3z −1 + 2z −2 )Y (z) = (6 − 10z −1 + 2z −2 )X (z) (43)

Thus, the difference equation is

y [n] − 3y [n − 1] + 2y [n − 2] = 6x [n] − 10x [n − 1] + 2x [n − 2] (44)

If we assume that the system is causal, we have

y [n] = 3y [n − 1] − 2y [n − 2] + 6x [n] − 10x [n − 1] + 2x [n − 2] (45)

With some practice, the conversion from the difference equation to system function and
vice versa can be done by simple inspection.

41/45
Important Topics in Unit I
1 Test the given system is:
Energy Signal
Power signal
2 Test the given system is:
Linearity
Time Invariant
Stability
Causality
memory
3 Derivation of Stability condition of an LTI system
4 Properties of LTI System
5 Convolution of LTI System
Use Tabular or matric method for 2 marks
Graphical method or mathematical method for 5 or 10 marks
Using Frequency domain theorem: Multiplication
6 Solution of linear constant-coefficient difference equations
Using time domain
Frequency domain
Impulse response of the LTI system
42/45
Frequency domain representation of Discrete Time Signals

Fourier series for discrete-time periodic signals

Fourier transforms for discrete-time aperiodic signals

43/45
Fourier representation of signals

Summary of Fourier representation of signals.

44/45
Important Topics in Unit I
1 Test the given system is:
Energy Signal
Power signal
2 Test the given system is:
Linearity
Time Invariant
Stability
Causality
memory
3 Derivation of Stability condition of an LTI system
4 Properties of LTI System
5 Convolution of LTI System
Use Tabular or matric method for 2 marks
Graphical method or mathematical method for 5 or 10 marks
Using Frequency domain theorem: Multiplication
6 Solution of linear constant-coefficient difference equations
Using time domain
Frequency domain
Impulse response of the LTI system
45/45