Introduction

Signal Analysis
BEX III/II
(Chapter 4. Sampling)
Department of Computer and Electronics & Comm. Engg.  A continuous-time signal can be processed by processing its sam-
Kathford Int’l College of Engg. & Mgmt. ples through a discrete-time system.
Balkumari, Lalitpur-08, Nepal  For this purpose, the sampling rate should be sufficiently high so
that the original signal can be reconstructed from these samples
without any error.(or with an error within a given tolerance.)
by
Bikal Adhikari  Sampling theory is the bridge between the continuous-time and
(B.E., M.Sc. Engg.) discrete-time worlds.

July 26, 2019

Typeset using LATEX

by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling) by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling)

The Sampling Theorem Types of Sampling

Theorem (Sampling Theorem)

A real signal x(t) whose spectrum is bandlimited to B Hz
{X(jw) = 0 for |w| > 2πB} can be reconstructed exactly (without Following are the types of sampling:
any error) from its samples taken uniformly at a rate fs > 2B sam-
⇒ Ideal Sampling or Impulse Train Sampling
ples per second. In other words, the minimum sampling fre-
quency is fs = 2B or {ws = 2πB rad/s}. ⇒ Practical Sampling
 Natural Sampling
Notes:  Flat-top Sampling
 The minimum sampling rate fs = 2B required to recover x(t) from
its samples f (t) is called the Nyquist rate for x(t)
 The corresponding sampling interval Ts = 2B
1
is called the Nyquist
interval for x(t)

by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling) by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling)
Ideal Sampling or Impulse Train Sampling

Since, the periodic impulse train δT (t) is used as sampling function,

this type of sampling is also known as impulse train sampling.

by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling) by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling)

 A real signal x(t) whose spectrum is bandlimited to B Hz

{X(jw) = 0 for |w| > 2πB} can be reconstructed exactly (without
any error) from its samples taken uniformly at a rate fs > 2B
samples per second. In other words, the minimum sampling fre-
quency is fs = 2B or {ws = 2πB rad/s}.
 Consider a signal x(t) whose spectrum is bandlimited to B Hz. It
can be reconstructed exactly without any error from the samples
taken uniformly at a rate fs > 2B Hz
 For convenience, spectra are shown as functions of w as well as
f (Hz)
 Since x(t) is bandlimited, it has no frequency components beyond
B Hz or 2πB rad/s

∴ X(jw) = 0, |w| > 2πB or

X(jf ) = 0, |f | > B
or as shown in figure a and b.
by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling) by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling)
  Because the impulse train δT (t) is a periodic signal of period Ts ,
 Sampling x(t) at a rate of fs Hz(fs samples per second) may be it can be expressed as a trigonometric fourier series as:
accomplished by multiplying x(t) by an impulse-train δT (t) of
1
figure c, consisting of unit impulses repeating periodically every δT (t) = {1 + 2 cos ws t + 2 cos 2ws t + 2 cos 3ws t + ......} (2)
Ts
Ts seconds, where
1 where,
Ts = 2π
fs = 2πfs
ws =
 The result is the sampled signal f(t)as shown in figure e. The Ts
resulting of sampled signal may be written as:  Therefore, from equations 1 and 2, we have
f (t) = x(t).δT (t)
f (t) = x(t).δT (t) (1)
1
∴ f (t) = ∑ x(nTs ).δ (t − nTs ) ∴ f (t) = {x(t) + 2x(t) cos ws t + 2x(t) cos 2ws t + ......} (3)
Ts
n
 Taking the fourier transform of equation 3, we get
 That is, the sampled signal consists of impulses spaced every Ts
seconds(the sampling interval). The nth impulse, located at t = 1
F(jw) = [X(jw) + {X(j(w + ws )) + X(j(w − ws ))} +
nTs , has strength x(nTs ), the value of x(t) at t = nTs . Ts
{X(j(w + 2ws )) + X(j(w − 2ws ))} + ... ... ...] (4)
by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling) by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling)

Natural Sampling
1 ∞
∴ F(jw) = ∑ X(j(w − kws )) (5)
Ts k=−∞
 From equation 4 and 5, it is clear that the spectrum F(jw) consist
of X(jw) repeating periodically with period ws = 2π 1
Ts or fs = Ts Hz
as shown in figure f.
 To reconstruct x(t) from f (t), we must be able to recover X(jw)
from F(jw). This recovery is possible if there is no overlap be-
tween the successive cycles of F(jw). Figure f shows that this
requires.
fs > 2B (6)
 Also, the sampling interval T = 1
FS . Therefore,

1
Ts < (7)
2B
 To recover the original signal x(t) from its samples f (t), the signal
f (t) must pass through an ideal lowpass filter of Bandwidth B Hz.
by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling) by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling)
Natural Sampling Contd... Natural Sampling Contd...
Mathematically,
s(t) = c(t) x(t) (8)
c(t) may be mathematically expressed in the form of complex Fourier The fourier transform of the sampled signal is given below:
series as:
+∞
c(t) = fs TA ∑ sinc(nfs T) ej2πnfs t (9)
n=−∞
where fs Ts = 1 defines sampling rate fs Therefore, substituting equation
9 in 8, we get,
+∞
s(t) = fs TA ∑ sinc(nfs T) ej2πnfs t .x(t) (10)
n=−∞
Taking Fourier Transform on both sides of above equation and using
the frequency shifting property of fourier transform, we get,
+∞
S(f ) = fs TA ∑ sinc(mfs T) X(f − mfs ) (11)
m=−∞
where, S(f ) = F [s(t)] and X(f ) = F [x(t)]
by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling) by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling)

Flat-top Sampling Flat-top Sampling contd...

The flat-top sampled signal is expressed as:

s(t) = xδ (t) ∗ h(t) (12)

Therefore, the width of s(t) is decided by h(t) and the amplitudes of

s(t) depends upon xδ (t) The ideally sampled signal xδ (t) is expressed
mathematically as under
+∞
xδ (t) = ∑ x(nTs ) δ (t − nTs ) (13)
n=−∞

By using the definition of convolution,

Z +∞
xδ (t) ∗ h(t) = xδ (τ) h(t − τ) dτ (14)
−∞

by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling) by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling)
Flat-top Sampling contd... Flat-top Sampling contd...
Substituting equation 13 into the equation 14, we get,
Z +∞ +∞ Taking Fourier Transform on both sides of equation 12, we get,
s(t) = ∑ x(nTs ) δ (τ − nTs ) h(t − τ) dτ
−∞ n=−∞ S(f ) = Xδ (f ) × H(f ) (16)
Interchanging the order of summation and integration and re-arranging,
As we know,
we get, ∞

+∞ Z +∞ Xδ (f ) = fs ∑ X(f − nfs ) (17)

n=−∞
s(t) = ∑ x(nTs ) δ (τ − nTs ) h(t − τ) dτ (15)
n=−∞ −∞ Therefore, equation becomes,
Now using the sifting property of delta function, i.e. ∞
Z ∞ S(f ) = fs X(f − nfs ).H(f ) (18)
∑
f (t).δ (t − td ) dt = f (td ) n=−∞
−∞

Applying this property to the equation 15, we have, This is the required expression for the spectrum of a flat top sampled
n=∞
signal. Also, H(f ) is the spectrum of h(t). The original signal can be
s(t) = ∑ x(nTs ) h(t − nTs ) recovered by using appropriate filter circuit.
n=−∞

by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling) by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling)

Practical Issues of Sampling Aliasing

Following are practical issues of sampling:

 If we sample at too low of a rate (below the Nyquist rate), then
 The sampled waveform consists of the finite amplitude and dura-
problems will arise that will make perfect reconstruction impos-
tion pulses rather than ideal pulses. This leads to aperture effect.
sible - this problem is known as aliasing.
 Reconstruction filters are not ideal. It leads to distortion.
 Aliasing occurs when there is an overlap in the shifted, perioidic
 The input output waveforms are rather time limited than bandlim- copies of our original signal’s FT, i.e. spectrum.
ited. It leads to aliasing.

by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling) by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling)
Aliasing Contd... Numericals

Question No. 1. Find the Nyquist rate and the Nyquist interval for the
signal.
1
x(t) = cos(4000πt) cos(1000πt)
2π
Solution: Given signal is:
1
x(t) = cos(4000πt) cos(1000πt)
2π
1
or, x(t) = [cos(4000πt + 1000πt) + cos(4000πt − 1000πt)]
4π
∵ 2 cos(A) cos(B) = cos(A + B) + cos(A − B)
1
∴ x(t) = [cos(5000πt) + cos(3000πt)]
Figure: Fig a: Spectrum of a Bandlimited signal, Fig b: Spectrum of a 4π
signal sampled at fs < 2B, Fig c: Spectrum of a signal sampled at
fs > 2B
by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling) by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling)

Numericals contd... Numericals contd...

Above equation is in the form:
∴ Nyquist Rate, fs = 2 × fm = 2 × 2500 = 5000 Hz = 5 kHz
1
x(t) = [cos(w1 t) + cos(w2 t)] Nyquist interval is given as:
4π
1 1 1
∴ w1 = 5000π Ts = = = = 0.2 × 10−3 seconds = 0.2 ms
2fm 2 × 2500 5000
or, 2πf1 = 5000π Question No. 2. Determine the Nyquist rate for a continuous-time
signal x(t) = 6 cos(50πt) + 20 sin(300πt) − 10 cos(100πt)
∴ f1 = 2500 Hz
Solution:
∴ w2 = 3000π Comparing the given signal with
or, 2πf2 = 3000π x(t) = A1 cos(w1 t) + A2 cos(w2 t) + A3 cos(w3 t)

∴ f2 = 1500 Hz we get,
f1 = 25 Hz, f2 = 150 Hz and f3 = 50 Hz
Since, the maximum frequency present in the signal is 2500Hz
Thus the maximum frequence present in the given signal is
∴ Maximum frequency, fm = f1 = 2500 Hz fm = f2 = 150 Hz
∴ Nyquist Rate = 2 × 150 = 300 Hz
by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling) by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling)
 Numericals contd...

Question No. 3. The given signal is m(t) = 10 cos(2000πt) cos(8000πt). Thank You!
What is the minimum sampling rate?
Answer: 10 kHz
Any Queries?

by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling) by Bikal Adhikari (B.E., M.Sc. Engg.) Signal Analysis BEX III/II (Chapter 4. Sampling)

Tribhuvan University
Institute of Engineering
Chapter Outline
Kathford Int’l College of Engineering and Management CONTINUOUS -TIME SYSTEM
Department of Computer and Electronics & Communication Engineering -9 HOURS, 15 MARKS

Bachelor’s Degree in Electronics & Communication Engineering System definition, properties of system, Linear time
[BEX III/II] invariant (LTI) system, convolution integral,
Properties of LTI system, frequency response of LTI
Signal Analysis [EX 651] system, bode plot, conditions for distortion less
Chapter Five Session 1 transmission, ideal low pass filter, impulse response
Continuous-time System and step response of ideal low pass filter, impulse
by: response and frequency response of first order
Bikal Adhikari system and second order system.
(B.E., M.Sc. Engg.)

Date:- 16th July, 2019

2
 System Definition System Definition Contd…
 A system is defined as an entity that manipulates one or x(t) y(t)
Input Signal Output Signal
more signals to accomplish a function thereby yielding or System or
or Response
or Excitation
a new signal. x[n] y[n]

 In other words, a system is defined as any physical Fig: Block diagram of a system
device that performs an operation on a signal.  Generally, a system is viewed as an operation or set of
 A system is defined as a process in which input signals operations performed on the input signal to produce the
are transformed by the combination of components output signal. We say that the input signal is
(devices) or cause the device to respond in the some transformed by the system into the output signal.
way resulting in other signal as output.  The general relationship between the input signal and
 For example, a filter is used to reduce noise and output signal is expressed as:
interference. y(t) = T{x(t)}
where,
x(t) = input signal, y(t) = output signal,
T = Transformation operator
3 4

System Definition Contd… Continuous-time system

 And Continuous-time
y[n] = T{x[n]} x(t) y(t)
System
where,
x[n] = discrete-time input signal, Fig: Block diagram of a continuous-time system

y[n] = discrete-time output signal,  In continuous-time system, the associated signals are
T = Transformation operator continuous-time signals. i.e. the input and output signals
are both continuous-time signals.
 For example: Audio amplifier, Video amplifier, Power
Classification of System supplies, etc.
 There are two types of system. They are:
 Continuous-time system
 Discrete-time system

5 6
 Discrete-time system Properties of System
x[n]
Discrete-time
y[n]  There are number of basic system properties of both
System continuous-time and discrete-time systems. Note that
Fig: Block diagram of a discrete-time system the systems are classified on the basis of system
 In discrete-time system, the associated signals are properties that they satisfy.
discrete-time signals. i.e. the input and output signals  Following are the properties:
are both discrete-time signals.  Systems with and without memory
 For example: Registers, Counters, Microprocessors, etc  Invertibility and Inverse Systems
 Causality(Causal and Non-Causal systems)
 Time-Invariance(Time-Variant and Time-Invariant
Systems)
 Stability(Stable and Unstable Systems)
 Linearity(Linear and Non-linear Systems)

7 8

Systems with and Without Memory Systems with and Without Memory Contd…
 A system is said to be memoryless if its output signal  Similarly, the identity system is also a memoryless.
value depends only upon the present value of the input That is
signal. y(t) = x(t) for continuous-time identity system
 For example: y[n] = x[n] for discrete-time identity system
 A memoryless system is also known as static system
y[n]  (2 x[n]  x 2 [n]) 2  A system has memory if its output signal value depends
in which the value of y[n] at any particular time n0 upon the past(also may be future) values of the input
depends only on the value of x[n] at that time. signal.
 Similarly, a resistor is a memoryless system where  For Example:
y(t) = R.x(t)  A capacitor is an example of system with memory

Where, 1
x(t) = current y (t )   x( )d
C 
y(t) = voltage
9 10
 Systems with and Without Memory Contd… Invertibility and Inverse System
 An accumulator or a summer is an example of a  A system is said to be invertible if distinct input leads to
discrete-time system with memory. distinct outputs.
n
 In other words, a system is said to be invertible system
y[n]   x[k ]
k  
if the input of the system is or can be recovered from
the output signal.
 A system having memory is also known as dynamic y[n] Inverse
system. x[n] System w[n] = x[n]
System
Fig: a

y(t) y(t) 11 w(t)

x(t) y(t)=2x(t)
x(t) y(t)=2x(t) w(t)=
w(t)= y(t) w(t) ==x(t)
x(t)
2 2 y(t)
Fig: Fig:
b b
11 12

Invertibility
n and Inverse System Causality(Causal and Non-Causal Systems)
y[n] n  x[k ]
y[n]
x[n] w[n]= y[n]- y[n-1] w[n] = x[n]
 A causal system is one whose output at any time
 x[k ]
k  y[n]
x[n] y[n]  w[n]= y[n]- y[n-1] w[n] = x[n] depends only on the present and past inputs but not on
k  
Fig: c
the future inputs.
 All the real-time systems are causal systems since the
 If a system is invertible
Fig: c then there exists an inverse
systems cannot know the future values of the input
system that when cascaded with the original system will
signals.
yield an output signal {w[n] or w(t)} equal to the input
 Note that only the causal systems are physically
signal{x[n] or x(t)}
realizable.
 The system which produces zero output sequence
 For Example:
{y[n] = 0 or y(t) = 0} for any input signal and 1


y(t) = x2(t) {or y[n] = x2[n]}, in which we cannot i. y (t )   x( )d ii. y[n]  x[n]  x[n  1]
C 
determine the sign of the input from the knowledge of
ii. y[n]  2 a x[n] iv. v(t )  R i (t )
the output are non-invertible systems. In all of the systems described by the above input-output relationship, the output
at any time depends only on the present and past inputs.
13 14
 Causality(Causal and Non-Causal Systems) Time-invariance
 A system is non-causal if its output at any time depends (Time-Variant and Time-Invariant Systems)
on the future value of the input.  A system is said to be time-invariant or shift-invariant if
 For Example: its input-output characteristics do not change with time.
i. y[n] = x[n]+2x[n-1]+4x[n-1]+5x[n+1] That is if
ii. y(t) = x(t+1) y(t) = T{x(t)} for continuous-time system
and y[n] = T{x[n]} for discrete-time system
Then
y(t-t0) = T{x(t-t0)} for continuous-time system
and y[n-k] = T{x[n-k]} for discrete-time system
 That is, for a time-invariant system, a time shift in the
input signal results a corresponding time shift in the
output signal.
 If a system does not satisfy the above equations then
the system is called time-variant system.
15 16

Example Numericals on Time-Invariance Example Numericals on Time-Invariance

Determine whether the systems describe by the input

output relationships given below are time-variant or a. y[n] = n x[n]
time-invariant Solution:
a. y[n] = n x[n] First we delay input by k units in time, we get,
b. y(t) = sin (x(t)) y[n, k] = n x[n-k]
c. y[n] = x[-n] Again if we delay the output by k units in time, we get,
d. y[n] = x[n] cos w0n y[n-k] = (n-k) x[n-k]
e. x[n] + y[n] Since, y[n, k] ≠ y[n-k], the given system is a time-variant
system.
z-1

17 18
 Contd… Contd…

b. y(t) = sin (x(t)) c. y[n]=x[-n]

Solution: Solution:
First we delay input by t0 units in time, we get, First we delay input by k units in time, we get,
y(t, t0) = sin(x(t- t0)) y[n, k] = x[-n-k]
Again if we delay the output by t0 units in time, we get, Again if we delay the output by k units in time, we get,
y(t- t0) = sin(x(t- t0)) y[n-k] = x[-(n-k)]= x[-n+k]
Since, y(t, t0) = y(t- t0) , the given system is time-invariant Since, y[n, k] ≠ y[n-k], the given system is a time-variant
system. system.

19 20

Contd… Contd…
e. y[n] = x[n]
x[n] + y[n]

d. y[n] = x[n] cos w0n z-1

Solution:
Solution: From the diagram, we can find that,
First we delay input by k units in time, we get, y[n] = x[n] + x[n-1]
y[n, k] = x[n-k] cos w0n First we delay input by k units in time, we get,
Again if we delay the output by k units in time, we get, y[n, k] = x[n-k] + x[n-k-1]
y[n-k] = x[n-k] cos w0(n-k) Again if we delay the output by k units in time, we get,
Since, y[n, k] ≠ y[n-k], the given system is a time-variant y[n-k] = x[n-k] + x[n-k-1]
system. Since, y[n, k] = y[n-k], the given system is a time-invariant
system.

21 22
 Stability Stability contd…
Bounded and Unbounded Signal Definition:
If a signal f(t) is bounded, then its magnitude is always a  A system is said to be Bounded Input-Bounded
finite value and we can write, Output(BIBO) stable if and only if every bounded input
…. … … … (i) produces a bounded output. The output of a stable
f (t )  M
Where, M = positive real finite number system does not diverge or does not grow unreasonably
For example: A cos w0 t  A , since, cos w0 t  1 large.
That is, a cosine wave is a bounded signal since its  If an input signal and output signal are bounded and
magnitude is always less than or equal to its peak translated mathematically to mean that there exists
amplitude. some finite numbers Mx and My such that
A signal which does not satisfy equation (i) is called as x(t )  M x   , y(t )  M y   for CT signal
unbounded signal.  Similarly,
x[n]  M x   , y[n]  M y   for DT signal
 If for some bounded input signal, if the output is
unbounded, the system is said to be unstable.
23 24

Linearity(Linear and Non-Linear Systems) Example Numericals on Linearity

 A system is said to linear if it satisfies the principle of Determine whether the systems described by the following
superposition. input output equations are linear or non-linear.
 A linear system may also be defined as a system whose i. y (t )  t x(t ) ii. y[n]  x[n2 ] iii. y(t )  x 2 (t )
response to the sum of weighted inputs is equal to the iv. y[n]  e x[ n ] iv. y[n]  A x[n]  B
sum of weighted responses. i. y (t )  t x(t )
 A system is linear if and only if: Solution:
T[a1x1 (t )  a2 x2 (t )]  a1T[ x1 (t )]  a2T[ x2 (t )] Let the output of the system to the two input signals x1(t)
for continuous-time system. and x2(t) are y1(t) and y2(t).
and i.e. y1(t) = t x1(t) and y2(t) = t x2(t)
T{a1x1[n]  a2 x2[n]}  a1T{x1[n]}  a2T{x2[n]} Thus, the linear combination of two outputs is given as:
for discrete-time system. ylo(t) = a1 y1(t) + a2 y2(t) = a1 t x1(t) + a2 t x2(t)
where a1 and a2 are constants. Let x3(t) be the linear combination of two inputs x1(t) and
If a system is non-linear, the principle of superposition x2(t). That is:
does not hold true and above equations are not x3(t) = a1 x1(t) + a2 x2(t)
satisfied. 25 26
 Contd… Contd…
If x3(t) is the input to the system then the corresponding ii. y[n]  x[n ]
2

output y3(t) will be : Solution:

y3(t) = T[x3(t)] = t x3(t) = t [a1 x1(t) + a2 x2(t)] Let the output of the system to the two input signals x1[n]
= t a1 x1(t) + t a2 x2(t) and x2[n] are y1[n] and y2[n].
= a1 tx1(t) + a2 t x2(t) i.e. y1[n] = x1[n2] and y2[n] = x2[n2]
Since, the sum of linear combination of the outputs i.e. Thus, the linear combination of two outputs is given as:
ylo(t) is equal to the output to the linear combination of ylo[n] = a1 y1[n] + a2 y2[n] = a1 x1[n2] + a2 x2[n2]
inputs i.e y3(t), the given system is linear. Let x3[n] be the linear combination of two inputs x1[n] and
i.e. Since, ylo (t )  y3 (t ) , the given system is linear. x2[n]. That is:
x3[n] = a1 x1[n] + a2 x2[n]
If x3[n] is the input to the system then the corresponding
output y3[n] will be :
y3[n] = T{x3[n]} = x3[n2] = a1 x1[n2] + a2 x2[n2]
 ylo [n]  y3[n] . Thus, the given system is linear.
27 28

Contd… Contd…
iii. y(t )  x (t )
2
iv. y[n]  A x[n]  B
Solution: Solution:
Let the output of the system to the two input signals x1(t) Let the output of the system to the two input signals x1[n]
and x2(t) are y1(t) and y2(t). and x2[n] are y1[n] and y2[n].
i.e. y1(t) = x12(t) and y2(t) = x22(t) i.e. y1[n] = A x1[n] + B and y2[n] = A x2[n] + B
Thus, the linear combination of two outputs is given as: Thus, the linear combination of two outputs is given as:
ylo(t) = a1 y1(t) + a2 y2(t) = a1 x12(t) + a2 x22(t) ylo[n] = a1 y1[n] + a2 y2[n]
Let x3(t) be the linear combination of two inputs x1(t) and = a1 {A x1[n] + B } + a2 {A x2[n] + B }
x2(t). That is: x3(t) = a1 x1(t) + a2 x2(t) = a1 A x1[n] + a2 A x2[n] + (a1 + a2)B
If x3(t) is the input to the system then the corresponding Let x3[n] be the linear combination of two inputs x1[n] and
output y3(t) will be : x2[n]. That is:
y3(t) = T[x3(t)] = x32(t) = [a1 x1(t) + a2 x2(t)]2 x3[n] = a1 x1[n] + a2 x2[n]
= a12 x12(t) + 2 a1 a2 x1(t) x2(t) + a22 x22(t) If x3[n] is the input to the system then the corresponding
Since, ylo (t )  y3 (t ) , the given system is non-linear. output y3[n] will be :
29 30
 Contd… Contd…
If x3[n] is the input to the system then the corresponding Let x3[n] be the linear combination of two inputs x1[n] and
output y3[n] will be : x2[n]. That is:
y3[n] = A x3[n] + B x3[n] = a1 x1[n] + a2 x2[n]
= A {a1 x1[n] + a2 x2[n]} + B If x3[n] is the input to the system then the corresponding
= a1 A x1[n] + a2 A x2[n] + B output y3[n] will be :
Since, ylo [n]  y3[n] the given system is non-linear. y3[n]  e a1 x1 [ n ] a2 x2 [ n ]
v. y[n]  e
x[ n ]
 y3[n]  e a1 x1 [ n ] .e a2 x2 [ n ]
Solution:
Let the output of the system to the two input signals x1[n] Since, ylo [n]  y3[n] , the given system is non-linear.
and x2[n] are y1[n] and y2[n].
i.e. y1[n]  e and y2 [n]  e
1x [ n] 2 x [ n]

Thus, the linear combination of two outputs is given as:

ylo[n] = a1 e  a2 e
x [ n]
1 2 x [ n]

31 32

Linear-Time Invariant(LTI) Systems Linear-Time Invariant(LTI) Systems Contd…

 Thus a continuous-time LTI system can be represented
 A time-invariant system which holds the linearity h(t)
as shown in figure:
property in its input-output relationship is called LTI CT LTI
system. x(t) y(t)
System
where,
 A LTI system can be characterized by its impulse x(t) = continuous-time input signal or excitation
response. h(t) = impulse response of the continuous-time LTI system
 The impulse response of a LTI system is its response of y(t) = output/response of the continuous-time LTI system
the system when it is excited by delta function(i.e.  Similarly, a discrete-time LTI system can be represented
h[n]
impulse function). as:
T x[n]
DT LTI
y[n]
System
 (t ) LTI y(t )  T [ (t )]  h(t ) where,
or or
 [n] System y[n]  T { [n]}  h[n] x[n] = discrete-time input signal or excitation
h[n] = impulse response of the discrete-time LTI system
where h(t) is the continuous-time impulse response y(t) = output/response of the discrete-time LTI system
and h[n] is the discrete-time impulse response.
33 34
 Linear-Time Invariant(LTI) Systems
Notes:
 One of the important characteristics of the unit
impulse(impulse function), both in continuous-time and
discrete-time , is that very general signals can be
represented as a linear combination of delayed
impulses.
 This fact together with the properties of superposition
Thank You!!!
and time-invariance, will allow us to develop a
complete characterization of any LTI system in terms
of its response to a unit impulse. Such a representation
Any Queries??
referred to as convolution integral in the continuous-
time case, and the convolution sum in discrete-time,
provides considerable analytical convenience in dealing
with LTI systems.
35 36
Tribhuvan University
Institute of Engineering
Linear-Time Invariant(LTI) Systems
Kathford Int’l College of Engineering and Management  A time-invariant system which holds the linearity
Department of Computer and Electronics & Communication Engineering
property in its input-output relationship is called LTI
system.
Bachelor’s Degree in Electronics & Communication Engineering
[BEX III/II]  A LTI system can be characterized by its impulse
response.
Signal Analysis [EX 651]  The impulse response of a LTI system is the response of
Chapter Five Session 2 the system when it is excited by delta function(i.e.
impulse function).
Continuous-time System T
by:  (t ) LTI y(t )  T [ (t )]  h(t )
or or
Bikal Adhikari  [n] System y[n]  T { [n]}  h[n]
(B.E., M.Sc. Engg.)
where h(t) is the continuous-time impulse response
and h[n] is the discrete-time impulse response.
Date:- 24th July, 2019
2
Linear-Time Invariant(LTI) Systems Contd… Linear-Time Invariant(LTI) Systems
 Thus a continuous-time LTI system can be represented Notes:
as shown in figure: h(t)  One of the important characteristics of the unit
CT LTI impulse(impulse function), both in continuous-time and
x(t) y(t)
System
where, discrete-time , is that very general signals can be
x(t) = continuous-time input signal or excitation represented as a linear combination of delayed
h(t) = impulse response of the continuous-time LTI system
impulses.
y(t) = output/response of the continuous-time LTI system
 This fact together with the properties of superposition
 Similarly, a discrete-time LTI system can be represented
h[n] and time-invariance, will allow us to develop a
as:
x[n]
DT LTI complete characterization of any LTI system in terms
System y[n]
where, of its response to a unit impulse. Such a representation
x[n] = discrete-time input signal or excitation referred to as convolution integral in the continuous-
h[n] = impulse response of the discrete-time LTI system time case, and the convolution sum in discrete-time,
y(t) = output/response of the discrete-time LTI system provides considerable analytical convenience in dealing
with LTI systems.
3 4

Representation of Continuous-time Signals in Contd…

terms of impulses  The approximation gets better and better as   0
 A continuous-time signal can be approximated by a linear 
combination of staircase as shown in figure a. x(t )  lim  x(k)   (t  k)  ………(ii)
 0
 If we define,  1 ,0  t   k  
  (t )   
 0 , otherwise
 We can say that under the limiting condition,   0  As   0 ,  becomes d and k becomes 
 (t ) becomes an impulse  (t )

 Then the summation in equation (ii) approaches an integral and
 Since,   (t ) has unit amplitude, we can express x(t) as: can be written as: 

x(t ) 


 x(k)   (t  k)  ………(i)

 x(t )   x( )  (t   ) d

k  
 x(t )  x(t ) *  (t ) ………(iii)
 x(t )  ......  x(2)   (t  2)   x()   (t  )   
 x ( 0)  (t )   x (  )  (t   )     Equation (iii) is known as sifting property of continuous-time
    impulse.
 x(2)   (t  2)   ...... 
5 6
Representation of Continuous-time Signals in
terms of impulses

Fig: Approximation of a continuous-time signal as a

linear combination of weighted, delayed, rectangular
pulses

7 8

Convolution Integral Convolution Integral Contd…

 We know that the input signal x(t) can be expressed in  Substituting equation (iii) into equation(ii), we obtain,
terms of shifted impulses as: 

 x( ) h(t   ) d

y (t )  ………(iv)
x(t )   x( )  (t   ) d

………(i) 

 Since the system is linear, the response y(t) of the  y(t )  x(t ) * h(t ) ………(v)
system to an arbitrary input x(t) can be expressed as:  Where ‘*’ represents convolution operation.
 
  Equation (v) indicates that a continuous-time LTI
y (t )  T {x(t )}  T   x( )  (t   ) d 
  
system is completely characterized by its impulse

response h(t). This equation is known as convolution
  x( ) [T { (t   )}] d ………(ii)

integral.
Since the system is time-invariant, we have,
h(t   )  T [ (t   )] ………(iii)

9 10
 Numericals on Convolution Integral Numericals on Convolution Integral contd…
Q 1. Let x(t) be the input to an LTI system with unit a. Solution:
x( ) h( )
impulse response h(t), where, 1
x(t )  e at u(t ), a  0 1

and h(t )  u (t )  
0 0
a.) Compute the output y(t) such that
 h( ) h(t   )
y (t )  x(t ) * h(t )   x( ) h(t   ) d

1 1

t<0

b.) Compute the output y(t) such that  

0 t 0

y (t )  h(t ) * x(t )   h( ) x(t   ) d h(t   )

 1
t>0

t 
0

11 12

Numericals on Convolution Integral Contd… Numericals on Convolution Integral Contd…


b. Solution:
y (t )  x(t ) * h(t )   x( ) h(t   ) d
 y (t )  h(t ) * x(t )   h( ) x(t   ) d


e  a ,0    t 
We have, x( ) h(t   )   Now,
 0 , otherwise e  a (t  ) ,0    t
h( ) x(t   )  
From the figure, we see that,  0 , otherwise
y (t )  0 for t  0 Then,
t
For t > 0, we have, e a
t t
t  a t
e y (t )   e  a ( t  )
d  e  at
e
a
d  e  at

y (t )   e  a d  y (t ) a
a
0 0 0
 at
e   
0 0
1 e 1
y(t )  at
1  1  e at
1
a

  e  at  e 0  a a a

1
 
 y (t )  1  e at u (t ) t
 y (t ) 
1
a

1  e  at u (t ) 
0
a
13 14
 Properties of LTI System 1. Commutative Property
 We know that the convolution integral is given by:  The commutative property is a basic property of

convolution.
y (t )  x(t ) * h(t )   x( ) h(t   ) d

 According to this property, we have,
where, asterisk(*) is used to denote the convolution
y(t )  x(t ) * h(t )  h(t ) * x(t )
operation.
 Following are the properties of LTI system:  That is
 
1. Commutative Property
2. Distributive Property
y (t )   x( ) h(t   ) d   h( ) x(t   ) d
 
3. Associative Property
4. Causality for LTI Systems  It can be interpreted as:
5. Stability of LTI Systems
6. LTI Systems with memory and without memory x(t) h(t) y(t) h(t) x(t) y(t)
7. Invertibility of LTI Systems
Figure: Interpretation of Commutative Property
15 16

2. The Distributive Property

Proof:  This is another basic property of convolution.
We have,  According to this property, we have,

x(t ) *[h1 (t )  h2 (t )]  x(t ) * h1 (t )  x(t ) * h2 (t )
x(t ) * y (t )   x( ) h(t   ) d

 It can be interpreted as:
Let, t   s , then   t  s , then,
  h1(t)
x(t ) * y (t )   x(t  s) h(s) ds   h(s) x(t  s) ds  h(t ) * x(t )
 
x(t) + y(t) x(t) h1(t) + h2(t) y(t)
 x(t ) * y(t )  y(t ) * x(t )
h2(t)
Figure: Interpretation of Distributive Property

17 18
 3. The Associative Property
Proof:  This is another important and useful property of
We have, convolution.

 According to this property, we have,
x(t ) * [h1 (t )  h2 (t )]   x( ) [ h1 (t   )  h2 (t   )] d
 x(t ) *[h1 (t ) * h2 (t )]  [ x(t ) * h1 (t )]* h2 (t )
   It can be interpreted as:
  x( ) h1 (t   ) d   x( ) h2 (t   ) d
z(t) h(t)
  x(t) h1(t) h2(t) y(t) x(t) y(t)
= h1(t) * h2(t)
 x(t ) * h1 (t )  x(t ) * h2 (t )
Figure: Interpretation of Associative Property
 x(t ) *[h1 (t )  h2 (t )]  x(t ) * h1 (t )  x(t ) * h2 (t )
Proof:
From first figure, we have,
y(t) = z(t) * h2(t)
where, z(t) = x(t) * h1(t)
 y(t )  [ x(t ) * h1 (t )]* h2 (t ) ………(i)
19 20

4. Causality of LTI Systems

Again from second figure,  In LTI Systems, causality can be translated in the
y(t) = x(t) * h(t) condition of impulse response. For this, let us consider
where, h(t) = h1(t) * h2(t) a LTI system having output at t = t0 given by the
convolution formula:
 y(t )  x(t ) *[h1 (t ) * h2 (t )] ………(ii) 
y (t0 )   h( ) x(t0   ) d
From equations (i) and (ii),we get, 

 x(t ) *[h1 (t ) * h2 (t )]  [ x(t ) * h1 (t )]* h2 (t )
0
or, y(t0 )   h( ) x(t0   ) d   h( ) x(t0   ) d
 0
(I) (II)
 In the (I) term, the integration involves future
components only. And, in the (II) term, the integration
involves present and past values of inputs.
 Therefore, it is clearly, seen that for a system to be
causal, the impulse response must satisfy the condition
21 22
 Contd… 5. Stability of LTI Systems
h(t) = 0 for t < 0  A system is BIBO(Bounded Input-Bounded Output)
Which is both necessary and sufficient condition for stable if and only if every bounded input produces a
causality. bounded output.
 If input signal x(t) is bounded, then there exists a
constant Mx such that
x(t )  M X   for all t
 Similarly, if output signal y(t) is bounded, then there
exists a constant My such that
y(t )  M Y   for all t
 Now, if x(t) is bounded input to a LTI system, then
using convolution formula,

y (t )   h( ) x(t   ) d ………(i)


23 24

Contd…
 Taking absolute value on both sides of this equation,
we get, 
y (t )   h( ) x(t   ) d

 But
y (t ) 




h( ) x(t   ) d

Thank You!!!
or, y (t )  M X  h( ) d

 From this expression, it is clear that the output is
bounded if the impulse response of the system is
Any Queries??
absolutely integrable, i.e., if




h( ) d  

25 26
 Tribhuvan University
Institute of Engineering
5. LTI Systems with and without memory:
Kathford Int’l College of Engineering and Management  A continuous-time LTI system is memoryless if:
Department of Computer and Electronics & Communication Engineering
h(t )  0, t  0 ………(i)
Bachelor’s Degree in Electronics & Communication Engineering  

[BEX III/II] 

 y (t )  

x ( ) h (t   ) d 

Signal Analysis [EX 651]
Chapter Five Session 3  Such a memoryless system has the form
Continuous-time System y(t )  k x(t ) ………(ii)

by:
 For some constant k it has the impulse response
Bikal Adhikari ………(iii)
h(t )  k  (t )
(B.E., M.Sc. Engg.)  For k = 1, in equation (iii), the system is then called
identity system and we can write convolution integral
Date:- 26th July, 2019 formula as: 2

contd… 6. Invertibility of LTI System


 Consider h(t) be the impulse response of LTI system.
y (t )   x( )  (t   ) d  x(t ) *  (t )

………(iv)
Then the LTI system is invertible only when an inverse
system exists that when cascaded with the original
 Which is also the shifting property of continuous-time system produces an output equal to the input to the first
unit impulse system. In figure, hI(t) is the impulse response of the
 If the impulse response h(t )  0, for t  0 , the system inverse system.
is called dynamic LTI or LTI system with memory.  h(t) and hI(t) are related as: h(t )* hI (t )   (t )
y(t)
x(t) h(t) hI(t) w(t) = x(t)

Fig: a

x(t) h(t)* hI(t) w(t) = x(t)

Fig: b
3 4
 Numerical Example: Contd…
Determine the inverse system of the delay system given To find the inverse system,
below: y(t )  x(t  t0 ) Let t  t0  s then we get,
Solution: y ( s  t 0 )  x( s )
Given, y(t )  x(t  t0 ) ………(i)
or, x( s)  y( s) *  ( s  t0 )
Then the impulse response can be obtained when
x(t )   (t ) In general,
Therefore, x(t )  y(t ) * (t  t0 ) ………(ii)
h(t )   (t  t0 ) From equation (ii), we have,
And hI (t )   (t  t0 )
x(t  t0 )  x(t ) * (t  t0 )
That is, the convolution of a signal with a shifted impulse Which is the required expression for the inverse system’s
simply shifts the signal. impulse response.

5 6

Frequency Response of LTI System Contd…

 If x(t) is input signal and h(t) is the unit impulse  Applying convolution property of CTFT to equation(i),
response of LTI system, then the response(or output) we get,
y(t) of the LTI system is expressed in terms of
Y ( jw)  X ( jw) H ( jw)
convolution integral as:
 Y ( jw)
H ( jw)  ………(ii)
y (t )   x( ) h(t   ) d

………(i) X ( jw)

 x(t ) * h(t )  Where H(jw) is the frequency response of LTI system,

 To analyze LTI system, it is convenient to utilize the It is also known as transfer function or system
frequency domain because differential function.
equations(difference equation for discrete-time) and
convolution operation in the time-domain become
algebraic operation in frequency domain.
7 8
 Contd… Magnitude and Phase Representation of
 Since, frequency response H(jw) is given by: Frequency Response
 The frequency response H(jw) can be written in polar
Y ( jw)
H ( jw)  form as:
X ( jw)
H ( jw)  H ( jw) e jH ( jw)
 Now, if X(jw) =1 then H(jw) = Y(jw)
 Since X(jw) = 1 corresponds to an unit impulse signal Where,
  (t ) in time-domain, therefore, H(jw) is defined as H ( jw) = amplitude response
“transform of the response of the LTI system when its
input is an unit impulse signal.” H ( jw) = phase response
Note that phase response does not affect the amplitude of
the individual frequency components but only provides
information concerning relative phases of exponentials
that make up h(t).
9 10

Condition for Distortionless Transmission Condition for Distortionless Transmission

 In several applications, such as signal amplification and y(t )  k x(t  td ) ………(i)
message signal transmission over a communication
channel, we require the output wave to be a replica of Where, k = constant(gain or attenuation)
the input waveform. td = constant time delay
 In such cases, we need to minimize distortion caused by  Taking Fourier transform of the equation (i) using time
the amplifier or the communication channel. shifting property, we get,
 Therefore, we have to determine the characteristics of a Y ( jw)  k X ( jw) e  jwtd
system that allows a signal to pass without distortion. Y ( jw)
 Transmission is said to be distortionless if the input and or,  k e  jwtd
X ( jw)
output have identical wave shapes within a
 Y ( jw) 
multiplicative constant. A delayed output that retains  H ( jw)  k e  jwtd  H ( jw)  is freq. response 
the input waveform is considered distortionless.  X ( jw)
………(ii) 
 Thus, in distortionless transmission input x(t) and  We also know that:
output y(t) satisfy the condition: H ( jw)  H ( jw) e jH ( jw) ………(iii)
11 12
 Condition for Distortionless Transmission Condition for Distortionless Transmission
 Comparing equation(ii) and equation(iii), we get, H ( jw)
H ( jw)  k
k
H ( jw)  wt d or n  wt d
Where n is any integer.
 This shows that for distortionless transmission; the
magnitude response must be a constant. Similarly, the
phase response must be a linear function of w. w
 That is the phase spectrum should be directly
proportional to the frequency. A time delay introduced
H ( jw)
in the system is equivalent to the phase shift of wtd

Fig:Frequency response of LTI system for distortionless transmission

13 14

Notes: Ideal Low Pass Filter:

 For distortionless transmission, the frequency  A filter is a frequency selective circuit(or network) that
component in x(t) are equally attenuated or amplified passes a specified band of frequencies through it with
and the system has infinite bandwidth. little attenuation. However , the attenuation increases
 But no system has infinite bandwidth and therefore seriously/sharply for the remaining frequencies.
distortionless conditions are never met exactly.  The allowed band of frequencies is known as pass band
 The following distortion may occur during transmission and the attenuated or rejected band of frequencies is
of signal through the linear system. known as stop band.
 Amplitude Distortion: Amplitude distortion occurs when  An ideal Low Pass Filter(LPF) if a type of filter which
magnitude response is not constant within the frequency allows frequency component below w = wm(i.e. pass
band of interest and frequency components present in band = 0 to wm) rad/s to pass without any distortion and
the input signal. suppresses all components above w = wm. It has linear
 Phase Distortion: Phase distortion occurs when phase
phase of slope –td, which results in a time delay td
response is not linear with respect to w and different
frequency components in the input are subjected to
seconds for all of its input component frequencies
different time delay during the transmission below wm rad/s.
15 16
 Contd… Contd…
H ( jw)
1 x(t) y(t)
a Ideal a
LPF
-wm wm w t t

H ( jw)  wt d
Fig:Frequency response of an ideal LPF
 In frequency domain, for the LPF with frequency
response H(jw), the spectrum of output is modified by
the nature of H(jw), allowing X(jw) components with
frequency w < wm
We know,
Y ( jw)  H ( jw) X ( jw)
17 18

Unit Impulse Response of Ideal LPF Contd…

wm
 An ideal LPF allows distortionless transmission within 1
e
jw( t t d )
or, h(t )  dw
w  wm and completely attenuates for w  wm 2  wm
 The frequency response of an ideal LPF with cut-off jw( t t d ) wm
frequency wm is defined by: or, h(t ) 
1 e
e  jwtd , w  wm 2 j (t  t d )  w
H ( jw)  
m

………(i)
 0 , w  wm or, h(t ) 
1  e jwm (t td ) e  jwm ( t td ) 

 
2  j (t  t d ) j (t  t d ) 
 The impulse response h(t) of the ideal LPF is:
1
wm
or, h(t ) 
1  e jwm (t td )  e  jwm (t td ) 
h(t )   H ( jw) e
jwt
dw  
2 2  j (t  t d ) 
 wm
wm 1  e jwm ( t td )  e  jwm (t td ) 
1 or, h(t )   
e
 jwtd
or, h(t )  e jwt dw
2  2 j (t  t d ) 
 wm
19 20
 Contd… Contd…
wm  e jwm ( t td )  e  jwm ( t td )  h(t)
or, h(t ) 
  2 j wm (t  t d ) 
w sin{wm (t  t d )}
h(t )  m td
t
 wm (t  t d )
wm Fig:Impulse response of an ideal LPF
h(t )  sin c{wm (t  t d )} ………(ii)

 This is the impulse response of an ideal LPF.  From the figure, it is clear that h(t )  0 for t < 0, Hence,
the ideal LPF is not causal and practically can not be
realized.

21 22

Unit Step Response of an Ideal LPF Contd…

h(t) or H(jw)
 Again, we know that the frequency response of an ideal
Ideal LPF is:
x(t)=u(t) y(t) e  jwtd , w  wm
LPF H ( jw)   ………(ii)
 0 , w  wm
 The step response is given by:
 When the input signal to the ideal low pass filter is unit
step signal, i.e. when x(t) = u(t) , then the response or s(t )  u (t ) * h(t )
 Taking Fourier transform on either side, we get,
output of the ideal LPF is the unit step response.
 We know, that the unit step signal u(t) and its Fourier S ( jw)  U ( jw) H ( jw)
transform U(jw) are related as: 1 
S ( jw)     ( jw) e  jwtd ………(iii)
1  jw 
u (t ) 
U ( jw)    ( jw) ………(i)
jw  Taking inverse Fourier Transform on either side of
equation(iii), we get,
23 24
 Contd… Contd…
 
1 1 1 2  jw(t td )
 jw e
1
s (t )   S ( jw) e dw
jwt
or, s (t )   dw
2  2 2 2
We know,
1

 1   jwtd  jwt
or, s (t ) 
2  jw
   ( jw) e

 e dw


sgn(t ) 
2
 1
jw
1

  jwtd  jwt
or, s (t ) 
2  jw   ( jw) e

 e dw
 2  jwtd
 sgn(t  t d )  e
1

1

1 jw
  ( jw) e  jw e
 jw( t t d )  jw( t t d )
or, s (t )  dw  dw
2 

  jw(t t 1

2 

2  jw(t td )
 s(t )  12  12 sgn(t  t d )
or, s (t )    jw e
)
e d
dw
2 w0 4  Which is the required expression for the unit step
response of an ideal LPF
25 26

Remaining Topics of this Chapter

 Bode Plot
(Self Study from Control Systems Note)
 Frequency Response and impulse response of first
order and second order system
 (See from handwritten lecture note from drive
folder.)
Thank You!!!
Any Queries??
27 28
 Tribhuvan University
Institute of Engineering
Chapter Outline
Kathford Int’l College of Engineering and Management DISCRETE -TIME SYSTEM
Department of Computer and Electronics & Communication Engineering -9 HOURS, 15 MARKS

Bachelor’s Degree in Electronics & Communication Engineering System definition, properties of system, Linear time
[BEX III/II] invariant (LTI) system, convolution sum, properties
of LTI system, difference equation, transfer
Signal Analysis [EX 651] function, frequency response of LTI system, bode
Chapter Six Session 1 plot, conditions for distortion less transmission,
Discrete-time System impulse response and frequency response of first
by: order system and second order system.
Bikal Adhikari
(B.E., M.Sc. Engg.)

Date:- 28th July, 2019

2

System Definition Discrete-time system

 A discrete-time system is a device or algorithm that Discrete-time
operates on a discrete-time signal, called input or x[n] y[n]
System
excitation, according to some well defined rule, to
produce another discrete-time signal, called the output Fig: Block diagram of a discrete-time system

or response.  In discrete-time system, the associated signals are

 We say that the input signal x[n] is transformed by the discrete-time signals. i.e. the input and output signals
system into a signal y[n] and express the general are both discrete-time signals.
relationship between x[n] and y[n] as:  For example: Registers, Counters, Microprocessors, etc
y[n]=T{x[n]} …. … … … (i)
Where
T= transformation operator
The input-output relation of equation (i) can be written as:
x[n] 

T
y[n] …. … … … (ii)
3 4
 Properties of System Systems with and Without Memory
 There are number of basic system properties of both  A system is said to be memoryless if its output signal
continuous-time and discrete-time systems. Note that value depends only upon the present value of the input
the systems are classified on the basis of system signal.
properties that they satisfy.  For example:
 Following are the properties: y[n]  (2 x[n]  x 2 [n]) 2
 Systems with and without memory
 Invertibility and Inverse Systems in which the value of y[n] at any particular time n0
 Causality(Causal and Non-Causal systems) depends only on the value of x[n] at that time.
 Time-Invariance(Time-Variant and Time-Invariant  Similarly, the identity system is also a memoryless. For
Systems) Example:
 Stability(Stable and Unstable Systems) y[n] = x[n]
 Linearity(Linear and Non-linear Systems)

5 6

Systems with and Without Memory Contd… Invertibility and Inverse System
 A memoryless system is also known as static system.  A system is said to be invertible if distinct input leads to
 A system has memory if its output signal value depends distinct outputs.
upon the past(also may be future) values of the input  In other words, a system is said to be invertible system
signal. if the input of the system is or can be recovered from
 An accumulator or a summer is an example of a the output signal.
discrete-time system with memory. y[n] Inverse w[n] = x[n]
n x[n] System
y[n]   x[k ]
k  
System
Fig: a
 A system having memory is also known as dynamic
system.

7 8
 Invertibility
n and Inverse System Causality(Causal and Non-Causal Systems)
y[n] n  x[k ]
y[n]
x[n] w[n]= y[n]- y[n-1] w[n] = x[n]
 A causal system is one whose output at any time
 x[k ]
k  y[n]
x[n] y[n]  w[n]= y[n]- y[n-1] w[n] = x[n] depends only on the present and past inputs but not on
k  
Fig: c
the future inputs.
 All the real-time systems are causal systems since the
 If a system is invertible
Fig: c then there exists an inverse
systems cannot know the future values of the input
system that when cascaded with the original system will
signals.
yield an output signal {w[n]} equal to the input
 Note that only the causal systems are physically
signal{x[n]}
realizable.
 The system which produces zero output sequence
 For Example:
{y[n] = 0} for any input signal and {y[n] = x2[n]}, in
which we cannot determine the sign of the input from i. y[n]  2 a x[n]
the knowledge of the output are non-invertible systems.
ii. y[n]  x[n]  x[n  1]
In all of the systems described by the above input-output relationship, the output
at any time depends only on the present and past inputs.
9 10

Causality(Causal and Non-Causal Systems) Time-invariance

 A system is non-causal if its output at any time depends (Time-Variant and Time-Invariant Systems)
on the future value of the input.  A system is said to be time-invariant or shift-invariant if
 For Example: its input-output characteristics do not change with time.
y[n] = x[n]+2x[n-1]+4x[n-1]+5x[n+1] That is if
y[n] = T{x[n]}
then
y[n-k] = T{x[n-k]}
 That is, for a time-invariant system, a time shift in the
input signal results a corresponding time shift in the
output signal.
 If a system does not satisfy the above equations then
the system is called time-variant system.

11 12
 Example Numericals on Time-Invariance Example Numericals on Time-Invariance

Determine whether the systems describe by the input

output relationships given below are time-variant or a. y[n] = n x[n]
time-invariant Solution:
a. y[n] = n x[n] First we delay input by k units in time, we get,
b. y[n] = x[-n] y[n, k] = n x[n-k]
c. y[n] = x[n] cos w0n Again if we delay the output by k units in time, we get,
d. x[n] y[n-k] = (n-k) x[n-k]
x[n] + y[n] Since, y[n, k] ≠ y[n-k], the given system is a time-variant
system.
z-1

13 14

Contd… Contd…

b. y[n]=x[-n] c. y[n] = x[n] cos w0n

Solution: Solution:
First we delay input by k units in time, we get, First we delay input by k units in time, we get,
y[n, k] = x[-n-k] y[n, k] = x[n-k] cos w0n
Again if we delay the output by k units in time, we get, Again if we delay the output by k units in time, we get,
y[n-k] = x[-(n-k)]= x[-n+k] y[n-k] = x[n-k] cos w0(n-k)
Since, y[n, k] ≠ y[n-k], the given system is a time-variant Since, y[n, k] ≠ y[n-k], the given system is a time-variant
system. system.

15 16
 Contd… Stability
d. y[n] =x[n]
x[n] + y[n] Bounded and Unbounded Signal
If a signal x[n] is bounded, then its magnitude is always a
z-1 finite value and we can write,
Solution: x[n]  M …. … … … (i)
From the diagram, we can find that, Where, M = positive real finite number
y[n] = x[n] + x[n-1] For example: A cos w0 n  A , since, cos w0 n  1
First we delay input by k units in time, we get, That is, a cosine wave is a bounded signal since its
y[n, k] = x[n-k] + x[n-k-1] magnitude is always less than or equal to its peak
Again if we delay the output by k units in time, we get, amplitude.
y[n-k] = x[n-k] + x[n-k-1] A signal which does not satisfy equation (i) is called as
Since, y[n, k] = y[n-k], the given system is a time-invariant unbounded signal.
system.

17 18

Stability contd… Linearity(Linear and Non-Linear Systems)

Definition:  A system is said to linear if it satisfies the principle of
 A system is said to be Bounded Input-Bounded superposition.
Output(BIBO) stable if and only if every bounded input  A linear system may also be defined as a system whose
produces a bounded output. The output of a stable response to the sum of weighted inputs is equal to the
system does not diverge or does not grow unreasonably sum of weighted responses.
large.  A discrete-time system is linear if and only if:
 If an input signal and output signal are bounded and T{a1x1[n]  a2 x2[n]}  a1T{x1[n]}  a2T{x2[n]}
translated mathematically to mean that there exists where a1 and a2 are constants.
some finite numbers Mx and My such that  If a system is non-linear, the principle of superposition
x[n]  M x   and y[n]  M y   does not hold true and above equation is not satisfied.
 If for some bounded input signal, if the output is
unbounded, the system is said to be unstable.

19 20
 Example Numericals on Linearity Contd…
Determine whether the systems described by the following i. y[n]  x[n ]
2

input output equations are linear or non-linear. Solution:

i. y[n]  x[n2 ] Let the output of the system to the two input signals x1[n]
and x2[n] are y1[n] and y2[n].
ii. y[n]  A x[n]  B i.e. y1[n] = x1[n2] and y2[n] = x2[n2]
Thus, the linear combination of two outputs is given as:
iii. y[n]  e x[ n ] ylo[n] = a1 y1[n] + a2 y2[n] = a1 x1[n2] + a2 x2[n2]
Let x3[n] be the linear combination of two inputs x1[n] and
x2[n]. That is:
x3[n] = a1 x1[n] + a2 x2[n]
If x3[n] is the input to the system then the corresponding
output y3[n] will be :
y3[n] = T{x3[n]} = x3[n2] = a1 x1[n2] + a2 x2[n2]
 ylo [n]  y3[n] . Thus, the given system is linear.
21 22

Contd… Contd…
ii. y[n]  A x[n]  B If x3[n] is the input to the system then the corresponding
Solution: output y3[n] will be :
Let the output of the system to the two input signals x1[n] y3[n] = A x3[n] + B
and x2[n] are y1[n] and y2[n]. = A {a1 x1[n] + a2 x2[n]} + B
i.e. y1[n] = A x1[n] + B and y2[n] = A x2[n] + B = a1 A x1[n] + a2 A x2[n] + B
Thus, the linear combination of two outputs is given as: Since, ylo [n]  y3[n] the given system is non-linear.
ylo[n] = a1 y1[n] + a2 y2[n] iii. y[n]  e
x[ n ]

= a1 {A x1[n] + B } + a2 {A x2[n] + B } Solution:

= a1 A x1[n] + a2 A x2[n] + (a1 + a2)B Let the output of the system to the two input signals x1[n]
Let x3[n] be the linear combination of two inputs x1[n] and and x2[n] are y1[n] and y2[n].
x2[n]. That is: i.e. y1[n]  e
x [ n]
1
and y2 [n]  e
2 x [ n]

x3[n] = a1 x1[n] + a2 x2[n] Thus, the linear combination of two outputs is given as:
If x3[n] is the input to the system then the corresponding ylo[n] = a1 e  a2 e
x [ n]
1 2 x [ n]

output y3[n] will be :

23 24
 Contd… Linear-Time Invariant(LTI) Systems
Let x3[n] be the linear combination of two inputs x1[n] and  A time-invariant system which holds the linearity
x2[n]. That is: property in its input-output relationship is called LTI
x3[n] = a1 x1[n] + a2 x2[n] system.
If x3[n] is the input to the system then the corresponding  A LTI system can be characterized by its impulse
output y3[n] will be : response.
y3[n]  e a1 x1 [ n ] a2 x2 [ n ]  The impulse response of a LTI system is its response of
 y3[n]  e a1 x1 [ n ] .e a2 x2 [ n ] the system when it is excited by delta function(i.e.
impulse function).
Since, ylo [n]  y3[n] , the given system is non-linear.

25 26

Linear-Time Invariant(LTI) Systems Contd… Linear-Time Invariant(LTI) Systems

 Thus a continuous-time LTI system can be represented Notes:
as shown in figure: h(t)  One of the important characteristics of the unit
CT LTI impulse(impulse function), is that very general signals
x(t) y(t)
System
where, can be represented as a linear combination of delayed
x(t) = continuous-time input signal or excitation impulses.
h(t) = impulse response of the continuous-time LTI system
 This fact together with the properties of superposition
y(t) = output/response of the continuous-time LTI system
and time-invariance, will allow us to develop a
 Similarly, a discrete-time LTI system can be represented
h[n] complete characterization of any LTI system in terms
as:
x[n]
DT LTI of its response to a unit impulse. Such a representation
System y[n]
where, referred to as convolution sum, provides considerable
x[n] = discrete-time input signal or excitation analytical convenience in dealing with LTI systems.
h[n] = impulse response of the discrete-time LTI system
y(t) = output/response of the discrete-time LTI system

27 28
 Tribhuvan University
Institute of Engineering
Kathford Int’l College of Engineering and Management
Department of Computer and Electronics & Communication Engineering

Bachelor’s Degree in Electronics & Communication Engineering

Thank You!!! [BEX III/II]

Signal Analysis [EX 651]

Chapter Six Session 2
Any Queries?? Discrete-time System
by:
Bikal Adhikari
(B.E., M.Sc. Engg.)

Date:- 20th August, 2019

29

Representation of Discrete-time Signals in terms Contd…

of impulses  From figure we can write,
x[n]
 A discrete-time sequence x[0]
x[1]  x[2] , n  2  x[1] , n  1
x[2]  [n  2]   x[1]  [n  1]  
x[-2]
x[2]
x[n] can be represented as x[-1]

a linear combination of -4 -3 -2 -1 0 1 2 3 4
n Fig: a  0 , n  2  0 ,n 1
weighted impulses. The x[-2] x[2]  [n  2]  x[1] , n  1
concept is illustrated in x[1]  [n  1]  
the figure alongside. -4 -3 -2 -1 0 1 2 3 4
n Fig: b  0 , n  1
x[1]  [n  1]  x[0] , n  0
x[-1] x[0]  [n]  
n Fig: c  0 ,n  0
Fig: -4 -3 -2 -1 0 1 2 3 4
x[0]
x[0]  [n]  In general, we have,
Graphical representation
x[n]  ... ...  x[3] [n  3]  x[2] [n  2]  x[1] [n  1]
n Fig: d
of decomposition of a
 x[0] [n]  x[1] [n  1]  x[2] [n  2]
-4 -3 -2 -1 0 1 2 3 4
x[1]
x[1]  [n  1]
discrete-time sequence into
 x[3] [n  3]  ... ...
n Fig: e
a sum of weighted impulses 2 -4 -3 -2 -1 0 1 2 3 4
3
 Contd… Convolution Sum
 Therefore we may write above equation in following  We know that the input signal x[n] can be expressed in
 terms of shifted impulses as:
 x[k ]  [n  k ]
form:
x[n]  

k  
………(i)
x[n]   x[k ]  [n  k ] ………(i)
k  

 Equation (i) corresponds to the representation of an  Since the system is linear, the response y[n] of the
arbitrary sequence x[n] as a linear combination of system to an arbitrary input x[n] can be expressed as:
shifted unit impulses  [n  k ] , where the weights in this   
y[n]  T {x[n]}  T   x[k ]  [n  k ]
linear combination are x[k]. k  
 For the case when, x[n]=u[n] 

u[n]    [n  k ]


1 , k  0   x[k ] T { [n  k ]} ………(ii)
 u[ k ]   k  
k 0  0 , k  0 Since the system is time-invariant, we have,
h[n  k ]  T { [n  k ]} ………(iii)
4 5

Convolution Sum Contd… Convolution Sum Contd…

 Substituting equation (iii) into equation(ii), we obtain,  Equation (vi) can be computed in four steps. They are:
 a) Folding: Fold h[k] about k = 0 to obtain h[-k]
y[n]   x[k ] h[n  k ] ………(iv)
k  
b) Shifting: Shift h[-k] by n0 to the right(or left) if n0
is positive(negative) to obtain h[n0-k].
 y[n]  x[n] * h[n] ………(v) c) Multiplication: Multiply x[k] by h[n0-k] to obtain
 Where ‘*’ represents convolution operation. the product vn0 = x[k] h[n0-k]
 Equation (v) indicates that a discrete-time LTI system d) Summation: Sum all the values of the product
is completely characterized by its impulse response sequence vn0[k] to obtain the value of output at
h[n]. This equation is known as convolution sum. time n = n0
 Suppose we wish to find the output of the system at Note that steps b through d must be repeated for all
time n = n0 , then from equation (iv), we get, possible time shift k in the range    k  

y[n0 ]   x[k ] h[n
k  
0  k] ………(vi)

6 7
 Convolution Sum by Graphical Method
Q 1. The impulse response of a LTI system is
3 x[k]
h[n]  {1, 2, 1,  1} 2
h[k]
2
1 1
2 3 4
Determine the response of the system to the input
signal x[n]  {1, 2, 3, 1}
-3 -2 -1 0 1 k -3 -2 -1 0 1 2 3 4 5 k
-1

Solution: x[k]
Fold
3
h[k] Multiply
2 2
1
h[-k] v0 [k ]  x[k ] h[k ]
1
2 2
2 3 4
-3 -2 -1 0 1
Continued
k
-1
in next slide…
-3 -2 -1 0 1 2 3 4 5 k
-2
1

Fold
See the separate pdf file entitled -4 -3
-1
-1 0 1 2 3 k -3 -2 -1 0 1 2 3 4 5 k

“GraphicalConvolution.pdf”for
Multiply better view of the  y[0]   v [k ]  2  2  4
0
h[-k] v [k ]  x[k ] h[k ]
2 following content… 2
0 Shift k  

1 v1[k ]  x[k ] h[1  k ]

-2 4
-4 -3 -1 0 1 2 3 k -3 -2 -1 0 1 2 3 4 5 k h[1-k] 3
-1 8  9
 v [k ]  2  2  4
2
 y[0]  0
Shift k   1 1
-1 Multiply
v1[k ]  x[k ] h[1  k ] -3 -2 0 1 2 3 4 k with x[k] -3 -2 -1 0 1 v23[k ]3 x[4k ] h5[3 kk ]
4 -1 h[3-k] 3


2
h[1-k] 3
1
2  y[1] 2  v [k ]  1  4  3  8
k 
1
Shift Multiply
1 1 1 1
-1 Multiply -1 0 2 3 4 5 6 k with x[k] -2 -1 0 2 3 4 5 6 k
-3 -2 0 1 2 3 4 k with x[k] -3 -2 -1 0 1 2 3 4 5 k -1 6 v2 [k ]  x[k ] h[2  k ]
-1  -2 
 y[1]   v [k ]  1  4  3  8
k 
1 Shift
h[2-k]  y[3]   v [k ]  2  3  2  3
3
Shift k 
2 2
1 1
6 v2 [k ]  x[k ] h[2  k ] 0 h[4-k] Multiply 0 v4 [k ]  x[k ] h[4  k ]
-2 -1 1 22 3 4 5 k with x[k] -3 -2 -1 1 2 3 4 5 k
h[2-k] -1 1 -1 1
2 2 2 Multiply 2


 v [k ]  1k  2  6  1 
0 1 5 6 7 k with x[k] -3 -2 -1 0 1 3 4
0
1
Multiply 0
1
-1
34
Shift  y[2]  2
k 
-2 -1 1 2 3 4 5 k with x[k] -3 -2 -1 1 2 3 4 5 k
-1 -1
Shift -3


Shift  y[2]   v [k ]  1  2  6  1  8
k 
2  y[4]   v [k ]  3  1  2
4
k 

10 h[5-k] 11 v5[k ]  x[k ] h[5  k ]

 
 y[4]   v [k ]  3  1  2
4
k  Now shifting h[-k] to the left,
Now shifting h[-k] to the left,
h[5-k] v5[k ]  x[k ] h[5  k ]
2 h[-1-k] v1[k ]  x[k ] h[1  k ]
2
3
1
Multiply 3 1 h[-1-k] 1
-3 Multiply
1 2
-1
4 5 6 7 8 k with x[k] -3 -2 -1 0 1 2
-1
4 5 k
-5 -4 -2 -1 0 1 2 k
2 with x[k] -3 -2 -1 0 1 2 3 4 k
 -1


Shift
 y[5]   v [k ]  1
k  
5
1  y[1]  v 1 [k ]  1
Shift
-3 k  
Multiply
-5 -4h[-2-k] -2 -1 0 1 2 k with x[k]
v6 [k ]  x[k ] h[6  k ] 2 v2[k ]  x[k ] h[2  k ]
2
h[6-k]
1
-1
1 -4 Multiply
4 Multiply -6 -5 -3 -2 -1 0 1 k with x[k] -3 -2 -1 0 1 2 3 4 5 k
with x[k] k -1
2 3 5 6 7 8 9 k -3 -2 -1 0 1 2 3 4 5
-1 Shift 

  y[2]  v 2 [k ]  0
 y[6]   v6 [k ]  0
k  
k  

h[-2-k]
12 2 13
1
Convolution Sum: Analytical Method
y[n]  -4 Multiply
 v [k ]
8 Q. No. 1. Determine the output y[n] of a relaxed-linear
y[n]  n -6 -5 -1 impulse
-3 -2 with 0 k
1 response
k  
time-invariant system with x[k]
n -1
h[n] = a u[n], |a|<1, when x[n] = u[n]
Solution:
4 We know that the convolution is given by the formula:
3 
y[n]  h[n] * x[n]   h[k ] x[n  k ]
k  
1 
4 5 6 7 8   a k u[k ] u[n  k ]
-5 -4 -3 -2 -1 0 1 2 3 k  
n
-1
We know, 1 , k  0
-2 u[k ]  
0 , k  0
y[n]  {..., 0, 0,1, 4, 8, 8, 3,  2,  1, 0, 0,...} 1 , n  k  0  k  n
 u[n  k ]  
14
0 , n  k  0  k  n 15
 Contd… Contd…
We know, a k ,0  k  n Q. No. 2. Determine by the convolution and sketch the
h[k ] x[n  k ]   response of an LTI system having input signal.
 0 , otherwise
1 ,0  n  4
n 1  a n1 x[n]  
 y[n]   a k
, a  1and n  0 0 , otherwise
k 0 1 a
Which is a geometric series and can be expressed as: and impulse response
y[n] = 1+ a + a2 + a3 + a4+ … … … … + an  n ,0  n  6
y[n] h[n]  
1  0 , otherwise
1 a 1 a  a2 1 a  a  a Solution:
2 3

1 a
We know that the convolution sum is given by:
1 
y[n]   x[k ] h[n  k ]
k  
n
0 16 17

Contd… Contd…
In order to calculate the convolution of two signals, it is convenient
to consider various intervals for n. This is shown in figure

Fig: Graphical Representation of

convolution being performed

Fig: Signals to be convolved

18 19
Contd… Contd…

20 21

Contd… Contd…

22 23
 Convolution Sum: Mathematical Method Contd…
Q. No. 1. Find the output sequence y[n] when a discrete- y[3]  x[0]h[3]  x[1]h[2]  x[2]h[1]  x[3]h[0]  18
time LTI system has: y[4]  x[0]h[4]  x[1]h[3]  x[2]h[2]  x[3]h[1]
x[n]  1,1,1,1
 x[4]h[0]  14
h[n]  {6, 5, 4, 3, 2,1} y[5]  x[0]h[5]  x[1]h[4]  x[2]h[3]  x[3]h[2]
Using mathematical method.
Solution:  x[4]h[1]  x[5]h[0]  10

y[6]  x[1]h[5]  x[2]h[4]  x[3]h[2]  6
y[n]   x[k ] h[n  k ]
k   y[7]  x[2]h[5]  x[3]h[4]  3
y[0]  x[0]h[0]  6,
y[8]  x[3]h[5]  1
y[1]  x[0]h[1]  x[1]h[0]  11
y[n]  0, n  9
y[2]  x[0]h[2]  x[1]h[1]  x[2]h[0]  15
y[n]  {6,11,15,18,14,10, 6, 3,1}
24 25

Contd…
Q No. 2. The impulse response of a LTI system is Q. No. 3. Find the output sequence y[n] when a discrete-
h[n]  {1, 2, 1,  1} time LTI system has:
 1 2 3 4 5 6
h[n] = {0,0,
x[n1,] 1,1,
0, 1,, 1,, 0,, 0}, , 
Determine the response of the system to the input  3 3 3 3 3 3
signal x[n]  {1, 2, 3, 1}
h[n] = {0, 0, 1, 1, 1, 1, 1, 0, 0}
using mathematical method.
Answer: Use

y[n]  {..., 0, 0,1, 4, 8, 8, 3,  2,  1, 0, 0,...}

a) Graphical Method
b) Mathematical Method
Answer:
y[n] = {0, 1/3, 1, 2, 10/3, 5, 20/3, 6, 5, 11/3,2}

26 27
 Short-Cut Method Short-Cut Method Contd…
Q. The impulse response of a LTI system is:
h[n]  {1, 2, 1,  1} First tabulate the values as shown:

Determine the response of the system to the input signal

Input Sequence
x[n]  {1, 2, 3, 1} x[n]
using short-cut method
Solution: 1 2 3 1

Sequence
nx = length of input sequence = 4

Impulse
nh = length of impulse sequence = 4 1 1 2 3 1

h[n]
xmin = minimum index of input sequence = 0
hmin = minimum index of impulse sequence = -1 2 2 4 6 2
nmin = minimum index of output sequence
= min(xmin, hmin) = min(0, -1) = -1 1 1 2 3 1
nmax = maximum index of output sequence
= nx + nh + nmin -2 -1 -1 -2 -3 -1
= 4 + 4 + (-1) -2
Computation
= 8 - 3 =of
5 Discrete-time 28 Convolution by Graphical 29
Method and Short-Cut Method
Short-Cut Method Contd… Properties of Discrete-time LTI System
Input Sequence  We know that the convolution sum is given by:

x[n] y[n]  x[n] * h[n]   x[k ] h[n  k ]
k  
1 2 3 1 where, asterisk(*) is used to denote the convolution
Sequence
Impulse

1 1 2 3 1 operation.
h[n]

 Following are the properties of LTI system:

2 2 4 6 2 1. Commutative Property
1 1 2 3 1 2. Distributive Property
-1 -1 -2 -3 -1 3. Associative Property
4. Causality for LTI Systems
y[n]  {......,0, 1, 2  2, 1  4  3,  1 2  6  1, 2  3  2,  3  1,  1, 0,... ...}
5. Stability of LTI Systems
 y[n]  {......,0, 1, 4, 8, 8, 3,  2,  1, 0,... ...} 6. LTI Systems with memory and without memory
7. Invertibility of LTI Systems
Note: Add the values diagonally
30 31
 1. Commutative Property
 The commutative property is a basic property of Proof:
convolution. We have,

 According to this property, we have, x[n] * h[n]   x[k ] h[n  k ]
k  
y[n]  x[n] * h[n]  h[n] * x[n]
Let, n  k  l , then k  n  l , then,
 That is  

  x[n] * h[n]   x[n  l ] h[l ]   h[l ] x[n  l ] h[n] * x[n]

y[n]   x[k ] h[n  k ]   h[k ] x[n  k ]
k   k  
l  

 x[n] * h[n]  h[n] * x[n]

l  

 It can be interpreted as:

x[n] h[n] y[n] h[n] x[n] y[n]

Figure: Interpretation of Commutative Property

32 33

2. The Distributive Property

 This is another basic property of convolution. Proof:
 According to this property, we have, We have, 
x[n] *h1[n]  h2[n]  x[n] * h1[n]  x[n] * h2[n] x[n] * h1[n]  h2 [n]   x[k ] h1[n  k ]  h2 [n  k ]
k  
 It can be interpreted as:  
  x[k ] h1[n  k ]   x[k ] h [n  k ]
2
h1[n] k   k  

x[n] y[n] x[n] h1[n] + h2[n] y[n]  x[n] * h1[n]  x[n] * h2 [n]
+

h2[n]  x[n] *h1[n]  h2[n]  x[n] * h1[n]  x[n]* h2[n]

Figure: Interpretation of Distributive Property

34 35
 3. The Associative Property
 This is another important and useful property of Again from second figure,
convolution. y[n] = x[n] * h[n]
 According to this property, we have, where, h[n] = h1[n] * h2[n]
x[n] *h1[n] * h2[n]  x[n]* h1[n]* h2 (t )  y[n]  x[n] *h1[n] * h2[n] ………(ii)
 It can be interpreted as:
z[n]
h[n] From equations (i) and (ii),we get,
 x[n] *h1[n] * h2[n]  x[n] * h1[n]* h2[n]
x[n] h1[n] h2[n] y[n] x[n] y[n]
= h1[n] * h2[n]

Figure: Interpretation of Associative Property

Proof:
From first figure, we have,
y[n] = z[n] * h2[n]
where, z[n] = x[n] * h1[n]
 y[n]  x[n] * h1[n]* h2[n] ………(i)
36 37

4. Causality of LTI Systems Contd…

 In LTI Systems, causality can be translated in the h[n] = 0 for n < 0
condition of impulse response. For this, let us consider  Hence, causal LTI system has its impulse response zero
a LTI system having output at n = n0 given by the for negative values of n.
convolution formula:  This is both necessary and sufficient condition for
y[n0 ]   h[k ] x[n
k  
0  k] causality.
1 
or, y[n0 ]   h[k ] x[n
k  
0  k ]   h[k ] x[n0  k ]
k 0
(I) (II)
 In the (I) term, the summation involves future
components only. And, in the (II) term, the summation
involves present and past values of inputs.
 Therefore, it is clearly, seen that for a system to be
causal, the impulse response must satisfy the condition
38 39
 5. Stability of LTI Systems Contd…
 A system is BIBO(Bounded Input-Bounded Output)  Taking absolute value on both sides of this equation,
stable if and only if every bounded input produces a we get, 

bounded output. y[n]   h[k ] x[n  k ]

k  
 If input signal x[n] is bounded, then there exists a  But

constant Mx such that
x[n]  M X   for all n
y[n]   h[k ]
k  
x[n  k ]

 Similarly, if output signal y[n] is bounded, then there
exists a constant My such that
y[n]  M X  h[k ]
k  
y[n]  M Y   for all t  From this expression, it is clear that the output is
 Now, if x[n] is bounded input to a LTI system, then bounded if the impulse response of the system is
using convolution formula, absolutely summable, i.e., if
 
y[n]   h[k ] x[n  k ]
k  
………(i)  h[k ]
k  

 This is both necessary and sufficient condition.
40 41
Tribhuvan University
Institute of Engineering
Kathford Int’l College of Engineering and Management
Department of Computer and Electronics & Communication Engineering

Bachelor’s Degree in Electronics & Communication Engineering

Thank You!!! [BEX III/II]

Signal Analysis [EX 651]

Chapter Six Session 3
Any Queries?? Discrete-time System
by:
Bikal Adhikari
(B.E., M.Sc. Engg.)

Date:- 20th August, 2019

42
 5. LTI Systems with and without memory: 6. Invertibility of LTI System
 A system is memoryless if its output at any instant  Consider h[n] be the impulse response of LTI system.
depends only on the value of the input at instant. Then the LTI system is invertible only when an inverse
 In LTI system, the convolution sum is: system exists that when cascaded with the original

system produces an output equal to the input to the first
y[n]  x[n] * h[n]   x[k ] h[n  k ] ………(i)
k  
system. In figure, hI[n] is the impulse response of the
 This system is memoryless if h[n] = 0 for all n ≠ 0. In inverse system.
this case, h[n]  k  [n] ………(ii)  h[n] and hI [n] are related as: h[n]* hI [n]   [n]
y[n]
 If k = 1, then h[n]   [n]  y[n]  x[n] *  [n]  x[n………(iii)
] x[n] h[n] hI [n] w[n] = x[n]
 That is if k = 1, then the LTI system is the Identity
system. Fig: a
 If h[n] ≠0 for n ≠ 0, then the system has memory. h[n] * hI [n]   [n]
 x[n] w[n] = x[n]
 From equation (iii), we get,
y[n]   x[k ]  [n  k ] Fig: b
2 k   3

Numerical Example: Contd…

n n 1

 x[k ]   x[k ]  x[n]

Determine the inverse system of the system with.
h[n]  u[n]
 y[n] 
k   k  
Solution: ………(i)
The convolution sum is: y[n]  y[n  1]  x[n]

y[n]  x[n] * h[n]   x[k ] h[n  k ]  x[n]  y[n]  y[n  1]
k  

Given h[n]  u[n] , therefore, Then the impulse response of inverse system is given by:

hI [n]   [n]   [n 1]
y[n]   x[k ] u[n  k ]
k  
We know,
1 , n  k  0  k  n
u[n  k ]  
0 , n  k  0  k  n
4 5
 Frequency Response of LTI System Contd…
 If x[n] is input signal and h[n] is the unit impulse  Applying convolution property of DTFT to equation(i),
response of LTI system, then the response(or output) we get,
y[n] of the LTI system is expressed in terms of
Y (e jw )  X (e jw ) H (e jw )
convolution sum as:
 Y (e jw ) ………(ii)
 x[k ] h[n  k ] H (e ) 
jw
y[n]  ………(i)
X (e jw )
k  

 x[n] * h[n]  Where H(ejw) is the frequency response of discrete-time

 To analyze LTI system, it is convenient to utilize the LTI system, It is also known as transfer function or
frequency domain because the difference equation for system function.
discrete-time and convolution operation in the time-
domain becomes algebraic operation in the frequency
domain.
6 7

Magnitude and Phase Representation of Condition for Distortionless Transmission

Frequency Response  As in the continuous-time systems, the condition for
 The frequency response H(jw) can be written in polar distortionless transmission in discrete-time systems is
form as: given by:
H (e jw )  H (e jw ) e jH ( e
jw
) y[n]  k x[n  nd ] ………(i)
Where, k = a constant
Where, nd = constant time delay
H (e jw ) = amplitude response  Taking DTFT on both sides of equation (i), we get,
Y (e jw )  k X (e jw ) e  jwnd
 n (e jw )  H (e jw )
= phase response Y (e jw )
Note that phase response does not affect the amplitude of or, jw
 k e  jwnd
X (e )
the individual frequency components but only provides
information concerning relative phases of exponentials   jwnd Y (e jw ) 
 H (e )  k e
jw
 H ( e jw
)  is freq. response 
that make up h[n] ………(ii)  X (e jw ) 
8 9
 Condition for Distortionless Transmission Impulse & Freq. response of First Order System
 We know,  The first order causal LTI system is described by the
jH ( e jw )
H (e )  H (e ) e
jw jw
………(iii) difference equation,
y[n]  a y[n  1]  x[n] ………(i)
 Comparing equation(ii) and equation(iii), we get,  Taking DTFT on both sides of above eqn, we obtain,
H (e jw )  k
X (e jw )  Y (e jw )  ae jwY (e jw )
H (e jw )   wnd or n  wnd X (e jw )  Y (e jw )(1  ae jw )
Where n is any integer.
 This shows that for distortionless transmission; the Y (e jw ) 1

magnitude response must be a constant. Similarly, the X (e jw ) (1  ae jw )
phase response must be a linear function of w.
 That is the phase spectrum should be directly 1
 H (e jw )  ………(ii)
proportional to the frequency. A time delay introduced (1  ae jw )
in the system is equivalent to the phase shift of wnd  Which is the frequency response of first order system.
10 11

Contd…
 Taking inverse Fourier Transform, we obtain the
impulse response as:
h[n]  a nu[n] ………(ii)

 Expressing equation (ii) in polar form, we get,

1 1
H (e jw )  
1  a(cos w  j sin w) (1  a cos w)  j a sin w)
1
or, H (e jw ) 
(1  a cos w) 2  (a sin w) 2
1
or, H (e jw ) 
1  a 2  2a cos w
 a sin w 

H (e jw )   tan 1  
 1  a cos w 
Fig: Plot of h[n]  a u[n] for different values of a
n
12 13
 Fig: Phase Response
Fig: Magnitude Response

14 15

Impulse & Freq. response of 2nd Order Syst.

16 17
Completion of
Course!!! Thank You!!!
Any Queries??
19