Signals and Systems

Lakshmi Narasimhan
IIT Madras

Disclaimer: There could be errors and typos in these slides. Use with caution.
Systems

2
 Systems (Flashback)

●
A box that processes an input (single) signal and produces an output (single) signal

Input x(t) Output y(t)

System

●
Example

●
Input: a voltage signal

●
Output: signal without DC component

●
Analytically: f{x(t)} = y(t)

3
 Standard Systems

●
Integrator: y(t) = f{x(t)} =

●
y[n] = Σk = -∞ to n x[k]

●
Differentiator: y(t) = f{x(t)} = dx(t) / dt

●
y[n] = x[n] – x[n – 1]

●
Time shifter: y(t) = f{x(t)} = x(t – T)

●
Amplifier: y(t) = f{x(t)} = Ax(t)

4
Properties of Systems

5
 Causality

●
Causal:

●
Output at time t depends only on the input at time t or earlier

●
Present output depends only on past and present inputs

●
Non-causal:

●
Output at time t depends on future inputs as well

●
All practical systems are causal

●
Some non-causal systems can be implemented with delay

6
 Causality - Examples

●
Determine causality of the following systems

●
y(t) = 0.3x(t)

●
y(t) = (x(t) + x(t – 1))/2

●
y(t) = (x(t) + x(t + 1))/2

●
y[n] = x[-n]

●
y(t) = x(t + T)

●
y(t + T) = x(t)

7
 Memory of Systems

●
Memory less:

●
Output at time t depends only on the input at time t

●
Example: amplifier, y(t) = (t +1)2x(t)

●
System with memory:

●
Output at time t depends on the past (or future) input

●
Example: integrator/differentiator (CT and DT)

●
y(t) = x(t – 1), y(t) = x(at) for a ≠ 1, y[n] = y[n – 1] + x[n]

8
 Invertibility

●
Invertible system

●
Given an output signal, the corresponding input can be identified

●
For f{x(t)} = y(t), f -1{ } exists such that f -1{y(t)} = x(t)

●
The inverse operation is another (invertible) system

●
Invertible systems are lossless transformations

x(t) y(t) x(t)

System System -1

9
 Invertibility - Examples

●
Check which of the following are invertible and non-invertible systems

●
y(t) = x(t) + c

●
y(t) = A x(at + b)

●
y(t) = t2 x(t) + c

●
y(t) = cos(x(t) + c)

●
y(t) = dx(t)/dt

●
y(t) = x(t) x(t + T)

10
 Stability

●
Stable systems

●
BIBO – bounded input produces bounded output

●
For |x(t)| < ∞, a stable system f{} gives an output such that |f{x(t)}|< ∞

●
For |x[n]| < ∞, a stable system f{} gives an output such that |f{x[n]}|< ∞

●
Critical to ensure practical systems are stable

●
An invertible system is not necessarily stable and vice versa

●
Examples: y(t) = (t2 + 1) x(t), y(t) = x2(t)

11
 Stability - Examples

●
Check which of the following are stable and unstable systems

●
y(t) = x(t) + c

●
y(t) = A x(at + b)

●
y(t) = t2 x(t) + c

●
y(t) = tan(x(t) + c)

●
y(t) = dx(t)/dt

●
y(t) = x(t) x(t + T)

12
 Time Invariance

●
Time invariance

●
The system description (input-output mapping) does not change with time

●
A delayed or advanced input should produce an output which is also delayed or
advanced, respectively

●
For time invariant f{ }, if f{x(t)} = y(t), then f{x(t + T)} = y(t + T)

●
Equivalent setups:

y(t + T)
x(t) y(t) y(t + T) x(t) x(t + T)
System Delay Delay System

13
 Time Invariance

●
Time varying systems

●
Systems have different input-output mapping at different times

●
For time varying f{ }, f{t, x(t)} = y(t) and f{t + T, x(t + T)} ≠ f{t, x(t)}

●
Examples:

●
y(t) = e-t x(t)

●
y(t) = ∫−∞
t
x(v) dv

14
 Time Invariance - Examples

●
Check which of the following are time invariant and time varying systems

●
y(t) = x(t) u(t)

●
y(t) = A x(at + b)

●
y(t) = t2 x(t) + c

●
y(t) = tan(x(t) + c)

●
y(t) = dx(t)/dt

●
y(t) = x(t) x(t + T)

15
 Time Invariance - Examples

●
Check which of the following are time invariant and time varying systems

●
y[n] = x[n] cos (wn)

●
y[n] = A x[n2 + N]

●
y[n] = n x[n]

●
y[n] = A x[n] + c

16
 Linearity

●
Linear systems

●
A system that satisfied superposition

●
Sum of scaled inputs produce sum of scaled corresponding outputs

●
f{ax1(t) + bx2(t)} = a f{x1(t)} + b f{x2(t)}

17
 Linearity - Examples

●
Check which of the following (and their DT equivalents) are linear systems

●
y(t) = x(t) u(t)

●
y(t) = A x(at + b)

●
y(t) = dx(t)/dt

18
 Linearity - Examples

●
Check which of the following are linear systems

●
y(t) = t2 x(t) + c

●
y(t) = tan(x(t) + c)

●
y(t) = x(t) x(t + T)

●
y(t) = Re{x(t)}

19
 Properties of RLC Systems

Causal Yes Yes Yes

Memory No Yes Yes

Invertible Yes No No

Time invariant Yes Yes Yes

Linear Yes Yes Yes

20
LTI Systems

21
 Representing a Signal with Dirac Deltas

●
Recall that ∫ δ(t – v) x(t) dt = x(v)

●
Equivalent to representing x(t) as an infinite sum of scaled Dirac deltas

●
What if a system is linear and we know its output to be δ(t – v) for all v?

22
 Linearity is Good

●
If hv(t) is the output of a linear system for input δ(t – v), then for input x(t), we get

●
Linearity simplifies analysis; but, can this be made even simpler?

23
 Linearity & Time Invariance is Better

●
If hv(t) is the output of a linear & time invariant system for input δ(t – v), then
hv(t) = h0(t – v) and

●
Thus, only the output of the system for δ(t) needs to be known

●
Impulse response of an LTI system f{δ(t)} = h(t)

●
The output of any another signal x(t) can be computed as ∫ x(τ) h(t – τ) dτ = y(t)

24
 The Convolution Integral

●
If h(t) is the impulse response of a system, then the output of only LTI systems for
input x(t) can be given by y(t) = x(t) * h(t)

●
* represents the convolution operation ∫ x(τ) h(t – τ) dτ

●
Given x(t) and h(t), how to convolve them?

●
Step 1: Time reversal: h(– τ)

●
Step 2: Time shift: for each value of t, h(t – τ)

●
Step 3: Multiply and add: x(τ) h(t – τ) and ∫ x(τ) h(t – τ) dτ

25
 The Convolution Integral - Example

●
Let x(t) = rect(t) and h(t) = tri(t)

At t = 0, y(0) = 0.75 At t = 1, y(1) = 0.125

26
 The Convolution Integral - Demo

●
Convolution video for rect(t) * tri(t)

27
 The Convolution Integral - Demo

●
Convolution video for rect(t) * x(t)

28
 Convolution - Examples

●
For an LTI system with impulse response rec(t), find the output for input rect(t)

29
 Convolution - Examples

●
For an LTI system with impulse response e-t u(t), find the output for input u(t)

30
 Convolution - Examples

●
For an LTI system with impulse response rec(t), find the output for input tri(t)

31
 Convolution - Examples

●
For an LTI system with impulse response rec(t), find the output for input tri(t)

32
 Properties of Convolution

●
Time length of convolved signal

●
If x(t) is T1 long and h(t) is T2 long, then

●
x(t)*h(t) is a maximum of T1 + T2 long

●
Convolution is commutative and associative x(t)
h2(t) h1(t)
y(t)

●
x(t)*h(t) = h(t)*x(t) and [x(t)*h1(t)]*h2(t) = x(t)*[h1(t)*h2(t)]

●
Order of LTI subsystems do not matter x(t)
h1(t) h2(t)
y(t)

33
 Properties of Convolution

●
Convolution is distributive: x(t)* [h1(t) + h2(t)] = x(t)*h1(t) + x(t)*h2(t)

●
Follows from linearity

●
Convolution with impulse (Dirac delta): δ(t)*x(t) = ∫ δ(t – v) x(v) dv = x(t)

●
δ(t – T)*x(t) = x(t – T)

●
Shifting property

●
If x(t)*h(t) = y(t), then x(t)*h(t – T) = x(t)*h(t)*δ(t – T) = y(t – T)

34
 Properties of LTI Systems

●
ONLY for LTI Systems, output = input * impulse response

●
Memory

●
An LTI system is memoryless if and only if its impulse response is Aδ(t)

●
Causality

●
An LTI system is causal if and only if its impulse response is zero for t < 0

35
 Properties of LTI Systems

●
Invertibility

●
An LTI system is invertible if there exists h-1(t) such that h(t)*h-1(t) = δ(t)

●
Stability

●
An LTI system is stable if its impulse response is absolutely integrable

●
∫ |h(t)| dt < ∞

●
If h(t) is periodic, then the system is unstable

36
 LTI Systems - Examples

Impulse Memory Causal Invertible Stable

Response h(t)
e2tu(1 – t)
e-|t|
cos(wt)
7δ(t)
(1/2)tu(t)
Σn=-∞ δ(t - nT)
sin(t)/t

37
 LTI Systems - Examples

Impulse Memoryless Causal Invertible Stable

Response h(t)
e2tu(1 – t) No No Yes Yes
e-|t| No No Yes Yes
cos(wt) No No No No
7δ(t) Yes Yes Yes Yes
(1/2)tu(t) No Yes Yes Yes
Σn=-∞ δ(t - nT) No No No No
sin(t)/t No No No No

38
 Unit Step Response of LTI Systems

●
Output of an LTI system for unit step input signal y(t)
δ(t)
∫ h(t)

●
Impulse response is derivative of unit step response
y(t)
h(t)
Causality: unit step response is zero for t < 0
u(t)
●

●
Stability: unit step response is absolutely integrable δ(t)
h(t) ∫ y(t)

●
Sufficient but not necessary

39
LTI Systems - Example

40
LTI Systems - Example

41
Grand Summary of Systems

42
 Summary

●
Continuous-time single-input-single-output systems

●
Memory: Dependence of present output on past input

●
Causality: Present output depends only on past and present inputs

●
All practical systems are causal

●
Invertibility: One-to-one map between input and output signals

●
Stability: Bounded input produces bounded output (BIBO)

●
Examples: Integrator, differentiator, amplifier, delay (time shifter)

43
 Summary

●
Linear and time invariant (LTI) systems

●
Linearity: Satisfies superposition property

●
Time invariance: A delay input produces a delayed output

●
All passive electric circuits are LTI

●
Impulse response: Output of a system to Dirac delta input

●
For LTI: Causal if impulse response is zero for t < 0

●
For LTI: Stable if impulse response is absolutely integrable

44
 Summary

●
Convolution: Weighted sum of product of time reversed + shifted signal

●
Input-out relationship of an LTI system only: y(t) = x(t) * h(t)

●
Impulse response of the cascade of an LTI system and its inverse = Dirac delta

●
Commutative, associative, and distributive

●
Unit step response is the integral of impulse response

●
Defined when the response is bounded

●
Complex exponentials are eigenfunctions of convolution (& LTI systems)

45