Professional Documents
Culture Documents
Signals Notes
Signals Notes
Signals Notes
𝐶 𝑒 𝑟𝑡
𝐶 𝑒 𝑟𝑡
𝐶
0<<1
< -1
DT Complex Exponential Signals (contd.)
• Periodic Complex Exponential Signals
• If in Eqn. (2), C is real and 𝛼 = 𝑗𝜔𝑜 is purely imaginary, then
𝑥[n] = C𝑒 𝑗𝜔𝑜 𝑛 (3)
2𝜋
is a periodic complex exponential signal with period 𝑁 = 𝑚 .
𝜔𝑜
• Sinusoidal Signals
Special case of periodic complex exponential signal of Eqn. (3)
𝑥[n] = A cos 𝜔𝑜 𝑛 + ∅ Real part of 𝑥[𝑛], 𝑅𝑒 𝑥[𝑛]
𝑥[n] = A 𝑠𝑖𝑛 𝜔𝑜 𝑛 + ∅ Imaginary part of 𝑥 𝑛 , 𝐼𝑚 𝑥[𝑛]
DT Complex Exponential Signals (contd.)
• General Complex Exponential Signals
From Eqn. (1)
𝑥[𝑛] = 𝐶𝛼 𝑛 , where C and are complex numbers
Expressing C in polar form and a in rectangular form
C = 𝐶 𝑒 𝑗𝜃 and 𝛼 = 𝛼 𝑒 𝑗𝜔𝑜 , then
𝑥[𝑛] = 𝐶 𝛼 𝑛 𝑒 𝑗(𝜔𝑜 𝑛+𝜃)
𝑥[𝑛] = 𝐶 𝛼 𝑛 cos(𝜔𝑜 𝑛 + 𝜃) + 𝑗 𝐶 𝛼 𝑛 sin(𝜔𝑜 𝑛 + 𝜃)
• For 𝛼 > 1 real and imaginary parts of 𝑥[𝑛] are growing sinusoidal signals.
• For 𝛼 < 1 real and imaginary parts of 𝑥[𝑛] are decaying sinusoidal signals.
• For 𝛼 = 1 real and imaginary parts of of 𝑥[𝑛] are sinusoidal signals.
Growing DT Sinusoid
Decaying DT Sinusoid
Unit Step Signal
1, 𝑡>0
• A CT unit step signal is defined as 𝑢 𝑡 = ቊ
0 𝑡<0
𝑢 𝑡 is discontinuous at t = 0.
1, 𝑛≥0
• A DT unit step signal is defined as 𝑢[𝑛] = ቊ
0 𝑛<0
Unit Impulse Signal
1, 𝑛=0
• A DT unit impulse signal is defined as 𝛿[𝑛] = ቊ
0 𝑛≠0
• DT unit impulse is the first difference
of the DT unit step signal. i.e.,
𝛿 𝑛 = 𝑢 𝑛 − 𝑢[𝑛 − 1]
• DT unit step is the running sum of the
DT unit impulse. i.e.,
𝑢 𝑛 = σ𝑛𝑚=−∞ 𝛿[𝑚]
• As evident from the following figure, running sum is 0 for n < 0 and 1 for 𝑛 ≥ 0.
Unit Impulse Signal (contd.)
• Parallel Interconnection
Interconnection of Systems (contd.)
• Series-Parallel Interconnection
• Feedback Interconnection
Basic System Properties
• Systems with and without memory
• Invertibility and Inverse Systems
• Causality
• Stability
• Time invariance
• Linearity
• On basis of the above properties, systems are also classified.
Systems without memory (memoryless)
• A system is said to be memoryless if its output at any instant of time is
dependent only on the input at that same time.
• Examples: The systems specified by the following relationships
𝑦 𝑡 = 𝐴𝑥(𝑡); an amplifier
𝑦 𝑡 = 𝐴𝑐 1 + 𝑘𝑥 𝑡 cos(2𝜋𝑓𝑐 𝑡); AM signal
𝑦 𝑛 = 2𝑥 𝑛 − 𝑥 2 [𝑛]
Systems with memory
• A system is said to be with memory if its current output is dependent
on the past or future input and output.
• Examples: The systems specified by the following relationships
𝑡
𝑦 𝑡 = −∞ 𝑥𝜏 𝑑𝜏
𝑦 𝑛 = 𝑥 𝑛 − 1 ; time delay
𝑦 𝑛 = 𝑥[𝑛 + 1]; time advance
• In many physical systems, memory is directly associated with the
storage of energy.
Invertibility
• A system is invertible if distinct input leads to distinct output.
• If a system is invertible, then an inverse system exists that when
cascaded with the original system, yields an output equal to the
input.
• Examples:
𝑡
• 𝑦 𝑡 = 𝑥 2𝑡 ; time compression and 𝑤 𝑡 = 𝑦( ); time expansion
2
• Encoding and decoding
• Modulation and demodulation
• 𝑦 𝑛 = σ𝑛−∞ 𝑥 𝑘 = 𝑦 𝑛 − 1 + 𝑥[𝑛] and 𝑥 𝑛 = 𝑦 𝑛 − 𝑦[𝑛 − 1]
Invertibility (contd.)
(b). 𝑦 𝑛 = 𝑥 𝑛 + 1 − 𝑥[𝑛 − 1]
𝑑𝑥(𝑡)
(c). 𝑦 𝑡 =
𝑑𝑡
𝑥 𝑛 = σ∞
𝑘=−∞ 𝑥 𝑘 𝛿[𝑛 − 𝑘]
An arbitrary sequence 𝑥[𝑛]can be represented as a linear combination of shifted
unit impulses 𝛿 𝑛 − 𝑘 , where the weights in this linear combination are 𝑥[𝑘].
The above equation is called the sifting property of the DT unit impulse.
Superposition Sum for DT LTI System
Graphic View
Derivation of Superposition Sum
Impulse Response
h[k] 2 0 2
h[-k] 2 0 2 0 y[0]= 4
h[-k-1] 2 0 2 -1 y[-1]= 2
h[-k-2] 2 0 2 -2 y[-2]= 0
h[-k+1] 2 0 2 1 y[1]= 2
h[-k+2] 2 0 2 2 y[2]= 2
h[-k+3] 2 0 2 3 y[3]= 0
h[-k+4] 2 0 2 4 y[4]= -2
h[-k+5] 2 0 2 5 y[5]= 0
Example 2: Tabular Method: 𝑦 𝑛 = σ𝑘 𝑥 𝑘 ℎ[𝑛 − 𝑘]
k -4 -3 -2 -1 0 1 2 3 4 5 n y[n]
x[k] -1 1 2
h[k] 3 1 -2 3 -2
h[-k] -2 3 -2 1 3 0 y[0]= 9
h[-k-1] -2 3 -2 1 3 -1 y[-1]= 2
h[-k-2] -2 3 -2 1 3 -2 y[-2]= -3
h[-k-3] -2 3 -2 1 3 -3 y[-3]= 0
h[-k+1] -2 3 -2 1 3 1 y[1]= -3
h[-k+2] -2 3 -2 1 3 2 y[2]= 1
h[-k+3] -2 3 -2 1 3 3 y[3]= 4
h[-k+4] -2 3 -2 1 3 4 y[4]= -4
h[-k+5] -2 3 -2 1 3 5 y[5]= 0
Example 2: Tabular Method: 𝑦 𝑛 = σ𝑘 ℎ 𝑘 𝑥[𝑛 − 𝑘]
k -4 -3 -2 -1 0 1 2 3 4 5 n y[n]
h[k] 3 1 -2 3 -2
x[k] -1 1 2
x[-k] 2 1 -1 0 y[0]= 9
x[-k-1] 2 1 -1 -1 y[-1]= 2
x[-k-2] 2 1 -1 -2 y[-2]= -3
x[-k-3] 2 1 -1 -3 y[-3]= 0
x[-k+1] 2 1 -1 1 y[1]= -3
h[-k+2] 2 1 -1 2 y[2]= 1
h[-k+3] 2 1 -1 3 y[3]= 4
h[-k+4] 2 1 -1 4 y[4]= -4
h[-k+5] 2 1 -1 5 y[5]= 0
Example 3
• Given ℎ 𝑛 = 0.5𝑛 𝑢[n] and 𝑥 𝑛 = 0.8𝑛 𝑢[n]. Determine 𝑦 𝑛 = 𝑥[𝑛] ∗ ℎ[𝑛].
𝑦 𝑛 = σ∞ 𝑘=−∞ 𝑥 𝑘 ℎ[𝑛 − 𝑘]
= σ∞ 𝑘
𝑘=−∞ 0.8 𝑢[𝑘] × 0.5
𝑛−𝑘 𝑢[n − k]
= σ∞ 𝑘
𝑘=−∞ 0.8 0.5
𝑛−𝑘 𝑢[𝑘]𝑢[𝑛 − k]
𝑢 𝑘 = 1, 𝑘 ≥ 0 and 𝑢 𝑛 − 𝑘 = 1, 𝑛 − 𝑘 ≥ 0 or 𝑘 ≤ 𝑛.
∴ 0 ≤ 𝑘 ≤ 𝑛 and 𝑛 ≥ 0.
For 𝑛 < 0, 𝑦 𝑛 = 0.
For 𝑛 ≥ 0,
𝑦[𝑛] = σ𝑛𝑘=0 0.8𝑘 0.5𝑛−𝑘
𝑛 σ𝑛 0.8 𝑘
= 0.5 𝑘=0
0.5
CT LTI SYSTEMS
• DT signal
𝑥 𝑛 = σ∞ 𝑘=−∞ 𝑥 𝑘 𝛿[𝑛 − 𝑘] using sifting property of unit impulse
• CT signal
∞
𝑥 𝑡 = −∞ 𝑥 𝜏 𝛿 𝑡 − 𝜏 𝑑𝜏
• CT LTI system
If 𝛿(𝑡) → ℎ(𝑡) impulse response, then
∞ ∞
−∞ 𝑥 𝜏 𝛿 𝑡 − 𝜏 𝑑𝜏 → −∞ 𝑥 𝜏 ℎ 𝑡 − 𝜏 𝑑𝜏
• For input 𝑥 𝑡 and IR ℎ(𝑡), output is
∞
y(t)=−∞ 𝑥 𝜏 ℎ 𝑡 − 𝜏 𝑑𝜏
Convolution Properties
• Commutative Property
• For CT system
𝑥 𝑡 ∗ ℎ 𝑡 = ℎ(𝑡) ∗ 𝑥(𝑡)
Convolution Properties (contd.)
• Associative Property
• For CT system
𝑥 𝑡 ∗ ℎ1 (𝑡) ∗ ℎ2 (𝑡) = 𝑥 𝑡 ∗ ℎ1 (𝑡) ∗ ℎ2 (𝑡)
• For CT system
𝑥 𝑡 ∗ ℎ1 𝑡 + ℎ2 𝑡 = 𝑥 𝑡 ∗ ℎ1 𝑡 + 𝑥 𝑡 ∗ ℎ2 𝑡
Convolution Properties (contd.)
• Combined Commutative and Distributive Properties
• DT System
𝑥1 [𝑛] + 𝑥2 [𝑛] ∗ ℎ[𝑛] = 𝑥1 [𝑛] ∗ ℎ[𝑛] + 𝑥2 [𝑛] ∗ ℎ[𝑛]
• CT System
𝑥1 𝑡 + 𝑥2 𝑡 ∗ ℎ 𝑡 = 𝑥1 𝑡 ∗ ℎ 𝑡 + 𝑥2 𝑡 ∗ ℎ 𝑡
The response of an LTI system to the sum of input signals is equal to the
sum of the responses of the system to these signals individually.
Convolution Properties (contd.)
• Time Shifting Property
• DT System
If 𝑥 𝑛 ∗ ℎ 𝑛 = 𝑦[𝑛]
then 𝑥 𝑛 ∗ ℎ 𝑛 − 𝑛𝑜 = 𝑥 𝑛 − 𝑛𝑜 ∗ ℎ 𝑛 = 𝑦[𝑛 − 𝑛𝑜 ]
and 𝑥 𝑛 − 𝑛1 ∗ ℎ 𝑛 − 𝑛2 = 𝑦[𝑛 − 𝑛1 − 𝑛2 ]
• CT System
If 𝑥(𝑡) ∗ ℎ(𝑡) = 𝑦(𝑡)
then 𝑥 𝑡 ∗ ℎ 𝑡 − 𝑡𝑜 = 𝑥 𝑡 − 𝑡𝑜 ∗ ℎ 𝑡 = 𝑦 𝑡 − 𝑡𝑜
and ℎ 𝑡 − 𝑡1 ∗ ℎ 𝑡 − 𝑡2 = ℎ 𝑡 − 𝑡1 − 𝑡2
Convolution Properties (contd.)
• Convolution with an Impulse
• DT Signal
𝑥 𝑛 ∗ 𝛿 𝑛 = 𝑥[𝑛] [From sifting property]
• CT Signal
𝑥(𝑡) ∗ 𝛿(𝑡) = 𝑥(𝑡) [From sifting property]
LTI System Characterization using IR
• Since an LTI system is completely specified by its impulse response (IR),
therefore, system properties can also be determined from its IR.
• System Properties
• Memory
• Invertibility
• Causality
• Stability
Memory
• A system is memoryless if output at any instant of time depends on the value of input at that
time.
• Consider a DT LTI system with input 𝑥[𝑛] and IR ℎ[𝑛], then the output
𝑦 𝑛 = σ∞
𝑘=−∞ 𝑥 𝑘 ℎ[𝑛 − 𝑘]
𝑦 𝑛 = ⋯ + 𝑥 −1 ℎ 𝑛 + 1 + 𝑥 0 ℎ 𝑛 + 𝑥 1 ℎ 𝑛 − 1 + ⋯
𝑦 −1 = ⋯ + 𝑥 −1 ℎ 0 + 𝑥 0 ℎ −1 + 𝑥 1 ℎ −2 + ⋯
𝑦 0 = ⋯ + 𝑥 −1 ℎ 1 + 𝑥 0 ℎ 0 + 𝑥 1 ℎ −1 + ⋯
𝑦 1 = ⋯ + 𝑥 −1 ℎ 2 + 𝑥 0 ℎ 1 + 𝑥 1 ℎ 0 + ⋯
• By definition, for memoryless system
𝑦 −1 = 𝑥 −1 ℎ 0
𝑦0 =𝑥 0ℎ 0
𝑦1 =𝑥 1ℎ 0
Memory (contd.)
for memoryless system
ℎ 𝑛 = 0 for 𝑛 ≠ 0
or ℎ 𝑛 = 𝐾𝛿[𝑛], where 𝐾 = ℎ[0] is a constant
and 𝑦 𝑛 = 𝐾𝑥[𝑛]
• For memoryless CT system
ℎ(𝑡) = 0 for 𝑡 ≠ 0
or ℎ(𝑡) = 𝐾𝛿(𝑡), where 𝐾 = ℎ(0) is a constant
and 𝑦(𝑡) = 𝐾𝑥(𝑡),
Memory (contd.)
• Example
For each of the following IRs, determine whether the corresponding
system is memoryless.
(a). ℎ 𝑡 = 𝑢 𝑡 + 1 − 𝑢(𝑡 − 1)
(b). ℎ 𝑡 = 𝑒 𝑎𝑡 𝑢(𝑡)
(c). ℎ 𝑛 = 𝛿 𝑛 − 𝛿[𝑛 − 1]
Invertibility
• DT System
(b). ℎ 𝑛 = 𝛿 𝑛 − 𝛿[𝑛 − 1]
(c). ℎ 𝑡 = 𝛿(𝑡 − 𝑡𝑜 )
Stability
• System is BIBO stable, if a bounded input produces a bounded output.
Consider a DT LTI system with IR ℎ 𝑛 and a bounded input 𝑥[𝑛].
Let 𝑥[𝑛] < 𝐵 for all 𝑛
𝑦 𝑛 = σ∞ 𝑘=−∞ 𝑥 𝑘 ℎ 𝑛 − 𝑘
= σ∞𝑘=−∞ ℎ 𝑘 𝑥 𝑛 − 𝑘
Magnitude of 𝑦 𝑛
𝑦 𝑛 = σ∞ 𝑘=−∞ ℎ 𝑘 𝑥 𝑛 − 𝑘
≤ σ∞−∞ ℎ 𝑘 𝑥 𝑛 − 𝑘 𝑎+𝑏 ≤ 𝑎 + 𝑏
𝑎𝑏 ≤ 𝑎 𝑏
≤ σ∞−∞ ℎ 𝑘 𝑥 𝑛 − 𝑘
Since 𝑥[𝑛 − 𝑘] < 𝐵 for all 𝑛 and k
∴ 𝑦[𝑛] ≤ 𝐵 σ∞ −∞ ℎ 𝑘 ≤ ∞, if σ∞ ℎ 𝑘 ≤ ∞.
−∞
Stability (contd.)
• For DT LTI system to be BIBO stable
σ∞𝑘=−∞ ℎ[𝑘] < ∞
i.e., IR is absolutely summable.
• For CT LTI system
∞
−∞ ℎ(𝑡) dt <
i.e., IR is absolutely integrable.
System Properties: Summary
Unit 2 Fourier Analysis
Outline
• Complex Exponentials as Eigenfunctions of LTI Systems
• Fourier Series representation of CT periodic signals
• Determining the Fourier coefficients
• Convergence of Fourier Series and Gibbs’ Phenomenon
• Fourier Transform representation of CT signals
• Properties of Fourier Transform
• Hilbert Transform and Representation of Bandpass Signals
Eigenfunction
• An input 𝑥 to a system is said to be an eigenfunction of the system
with the eigenvalue λ if the corresponding output 𝑦 is of the form
𝑦= 𝜆𝑥
where λ is a (complex) constant.
• Different systems have different eigenfunction.
• Of particular interest are the eigenfunctions of LTI systems.
Eigenfunction of LTI System
eigenvalue eigenfunction
s is complex
s = +j
eigenvalue eigenfunction
eigenvalue eigenfunction
Eigenfunction of LTI System (contd.)
• Complex Exponentials are the only Eigenfunctions of any LTI Systems
CT
Eigenfunction of LTI System (contd.)
DT:
Signal Representation as Sum of Eigenfunction
• What kinds of signals can we represent as “sums” of complex
exponentials?
CT:
Magnitude 1
DT:
CT & DT Fourier Series and Transforms
Periodic Signals
2.2 Fourier Series
Signal Representation as Sum of Eigenfunction
• What kinds of signals can we represent as “sums” of complex
exponentials?
CT:
Magnitude 1
DT:
CT & DT Fourier Series and Transforms
Periodic Signals
Fourier Series Representation of CT Periodic Signals
Euler's relation
0 – no dc component
0
0
4
Special Case: Real Signals
𝑥(𝑡) = σ∞
𝑘=−∞ 𝑎𝑘 𝑒
𝑗𝑘𝜔0 𝑡
(1) Complex exponential Fourier series
If 𝑥(𝑡) is real, then 𝑥 ∗ 𝑡 = 𝑥(𝑡)
Taking the complex conjugate of (1)
𝑥 ∗ (𝑡) = σ∞ ∗ −𝑗𝑘𝜔0 𝑡
𝑘=−∞ 𝑎𝑘 𝑒
𝑥 𝑡 = 𝑎0 + σ∞
𝑘=1 𝑎𝑘 𝑒
𝑗𝑘𝜔0 𝑡 + 𝑎 ∗ 𝑒 −𝑗𝑘𝜔0 𝑡
𝑘
Real Signals (contd.)
𝑥 𝑡 = 𝑎0 + 2 σ∞
𝑘=1 𝑅𝑒 𝑎𝑘 𝑒
𝑗𝑘𝜔0 𝑡 (4)
Writing 𝑎𝑘 in polar form
𝑎𝑘 = 𝐴𝑘 𝑒 𝑗𝜃𝑘 , where 𝐴𝑘 = 𝑎𝑘
𝑥 𝑡 = 𝑎0 + 2 σ∞
𝑘=1 𝑅𝑒 𝐴 𝑘 𝑒 𝑗(𝑘𝜔0 𝑡+𝜃𝑘 )
𝑥 𝑡 = 𝑎0 + 2 σ∞
𝑘=1 𝐴𝑘 𝑐𝑜𝑠 𝑘𝜔0 𝑡 + 𝜃𝑘 (5)
8
Examples 3.3 and 3.4 [1]
Example 3.5 [1]: Periodic Square Wave
2𝑇1
= Duty Cycle
𝑇
10
Periodic Square Wave (contd.)
• Plot for Duty cycle = 0.5, 0.75 and 1.0
1 2𝑇1
𝑎𝑘 = sin(𝑘𝜋 )
𝑘𝜋 𝑇
𝑎𝑘 = real, so in trigonometric FS, only cosine terms are present
Even symmetry → Only cosine terms
Odd symmetry → Only sine terms
Convergence of Fourier Series
𝑥(𝑡) = σ∞
𝑘=−∞ 𝑎𝑘 𝑒
𝑗𝑘𝜔0 𝑡 (1)
• FS of a periodic signal 𝑥(𝑡) converges, if 𝑥(𝑡) satisfies the following Dirichlet Conditions:
1. Over any period, 𝑥(𝑡) must be absolutely integrable; i.e.,
∞ < 𝑡𝑑 )𝑡(𝑥 𝑇
• A periodic signal that violates the above condition is
1
𝑥 𝑡 = , 0<t≤1
𝑡
Convergence of Fourier Series……
• Dirichlet conditions……..
2. In any finite interval of time, 𝑥(𝑡) has finite number of maxima and minima.
• A signal that violates this condition is
2𝜋
𝑥 𝑡 = 𝑠𝑖𝑛 , 0<𝑡≤1
𝑡
The signal has infinite number of maxima and minima in a period.
Convergence of Fourier Series……
• Dirichlet conditions……..
3. In any finite interval of time, 𝑥(𝑡) has finite number of discontinuities.
Furthermore, each of these discontinuities is finite.
• A signal that violates this condition is
4
Example: Unit Impulse
(a) (t) 1
0 t 0
𝛿 𝑡 ⇔ 1
1/2 ()
(b) t 0
0
1 ∞ 1
𝐹 −1 𝛿 𝜔 = 𝑗𝜔𝑡
න 𝛿 𝜔 𝑒 𝑑𝜔 =
2𝜋 −∞ 2𝜋
1 ⇔ 2𝜋 𝛿 𝜔
𝛿 𝑡 ∓ 𝑡𝑜 ⇔ 𝑒 ∓𝑗𝜔𝑡𝑜
5
Example: Right-Sided Exponential
𝑥(𝑡)
0 𝜔𝑜
7
Fourier Transform of Cosine Signal
8
Fourier Transform of General Periodic Signal
𝑒 𝑗𝜔𝑜 𝑡 ↔ 2𝜋 𝛿 𝜔 −𝜔𝑜
𝑒 𝑗𝑘𝜔𝑜 𝑡 ↔ 2𝜋 𝛿 𝜔 −𝑘𝜔𝑜
∞ ∞
Note: (period in t) T
(period in ) 2/T
10
Example: Periodic Square Wave
g(t)
𝜔𝑇1
↔ 𝐺 𝑗𝜔 = 2𝑇1 𝑠𝑖𝑛𝑐
𝜋
2𝑇1 𝑘2𝑇1
𝑋 𝑗𝜔 = σ∞ ∞
𝑘=−∞ 2𝜋𝑎𝑘 𝛿 𝜔 −𝑘𝜔𝑜 = σ𝑘=−∞ 2𝜋 𝑠𝑖𝑛𝑐 𝛿 𝜔 −𝑘𝜔𝑜
𝑇 𝑇
1. Linearity
If 𝑥 𝑡 ↔ 𝑋 𝑗𝜔 𝑎𝑛𝑑 𝑦 𝑡 ↔ 𝑌 𝑗𝜔 , then
a 𝑥 𝑡 + 𝑏 𝑦 𝑡 ↔ 𝑎 𝑋 𝑗𝜔 + 𝑏 𝑌 𝑗𝜔
If 𝑥 𝑡 ↔ 𝑋 𝑗𝜔 , then
FT magnitude unchanged
14
Time Shifting: Example
x(t) 1
𝑡 𝜔𝑇
𝑥 𝑡 = 𝑟𝑒𝑐𝑡 ⟷ 𝑇𝑠𝑖𝑛𝑐
𝑇 2𝜋
-T/2 0 T/2
y(t) 1
𝑡 1 𝜔𝑇 −𝑗𝜔𝑇/2
𝑦 𝑡 = 𝑥 𝑡 − 𝑇/2 = 𝑟𝑒𝑐𝑡 − ⟷ 𝑇𝑠𝑖𝑛𝑐 𝑒
0 T 𝑇 2 2𝜋
If 𝑥 𝑡 ↔ 𝑋 𝑗𝜔 , then
𝑒 −𝑗𝜔𝑜 𝑡 𝑥 𝑡 ↔ 𝑋 𝑗(𝜔 − 𝜔𝑜 )
X(j)
Also known as Modulation Property
Example:
-
2𝜋 2𝜋
𝑇 𝑇
𝑡 𝜔𝑇
If 𝑥 𝑡 = 𝑟𝑒𝑡 ↔ 𝑋 𝑗𝜔 = 𝑇𝑠𝑖𝑛𝑐 , then X(j(- 𝜔𝑜 ))
𝑇 2𝜋
−𝑗𝜔𝑜 𝑡 𝑡 (𝜔−𝜔𝑜 )𝑇
𝑒 𝑟𝑒𝑡 ↔ 𝑋 𝑗(𝜔 − 𝜔𝑜 ) = 𝑇𝑠𝑖𝑛𝑐
𝑇 2𝜋
0 𝜔𝑜
𝜔𝑜
[1] Problem 4.21-4.22 16
Properties of Fourier Transform……….
If 𝑥 𝑡 is real and even, then
4. Conjugate Symmetry
𝑋 −𝑗𝜔 = 𝑋 𝑗𝜔
If 𝑥 𝑡 ↔ 𝑋 𝑗𝜔 , then From (1)
∗
Even
𝑥 ∗ 𝑡 ↔ 𝑋 ∗ −𝑗𝜔 𝑋 −𝑗𝜔 = 𝑋 𝑗𝜔
∴ 𝑋 −𝑗𝜔 = 𝑋 𝑗𝜔 = 𝑋 ∗ 𝑗𝜔
If 𝑥 𝑡 is real, then
𝑋 𝑗𝜔 is real
𝑋 ∗ −𝑗𝜔 = 𝑋 𝑗𝜔 , or If 𝑥 𝑡 is real and even, then 𝑋 𝑗𝜔 is
Or
𝑋 −𝑗𝜔 = 𝑋 ∗ 𝑗𝜔 (1) Conjugate Symmetry also real and even.
Even If 𝑥 𝑡 is real and odd, then 𝑋 𝑗𝜔 is
purely imaginary and odd.
Odd 𝑥 𝑡 = 𝑥𝑒 𝑡 + 𝑥𝑜 (𝑡), then
F[𝑥 𝑡 ] = 𝐹[𝑥𝑒 𝑡 ] + F[𝑥𝑜 𝑡 ]
Even
𝐹[𝑥𝑒 𝑡 ] = Re 𝑋 𝑗𝜔
Odd 𝐹[𝑥𝑜 𝑡 ] = jIm 𝑋 𝑗𝜔 18
Example 4.10 [1]
Evaluate the FT of 𝑥 𝑡 = 𝑒 −𝑎 𝑡 , 𝑎 > 0 using the symmetry property.
Fig. 𝑥 𝑡 = 𝑒 −𝑎 𝑡
c)
20
Properties of Fourier Transform……….
6. Time Differentiation Frequency Differentiation
If 𝑥 𝑡 ↔ 𝑋 𝑗𝜔 , then If 𝑥 𝑡 ↔ 𝑋 𝑗𝜔 , then
𝑑𝑥(𝑡) 𝑑𝑋 𝑗𝜔
↔ 𝑗𝜔 𝑋 𝑗𝜔 , and 𝑡𝑥 𝑡 ↔𝑗 , and
𝑑𝑡 𝑑𝜔
𝑑 𝑛 𝑥(𝑡) 𝑑 𝑛 𝑋 𝑗𝜔
↔ 𝑗𝜔 𝑛 𝑋 𝑗𝜔 𝑡 𝑛 𝑥(𝑡) ↔ 𝑗 𝑛
𝑑𝑡 𝑛 𝑑𝜔𝑛
21
Example
x(t) 1
𝑡 𝜔𝑇
𝑥 𝑡 = 𝑟𝑒𝑐𝑡 ⟷ 𝑋 𝑗𝜔 = 𝑇𝑠𝑖𝑛𝑐
t 𝑇 2𝜋
-T/2 0 T/2
𝑑𝑥(𝑡)
𝑑𝑥(𝑡) 𝜔𝑇
𝑑𝑡 FT = jω 𝑋 𝑗𝜔 = 2 𝑗 𝑠𝑖𝑛( )
𝑑𝑡 2
T/2
-T/2 0 t Directly
𝑗𝜔𝑇 𝑗𝜔𝑇
𝑑𝑥(𝑡) 𝑇 𝑇 − 𝜔𝑇
= δ 𝑡+ − δ 𝑡− ⟷ 𝑒 2 −𝑒 2 = 2 𝑗 𝑠𝑖𝑛( )
𝑑𝑡 2 2 2
Properties of Fourier Transform……….
7. Time Integration
If 𝑥 𝑡 ↔ 𝑋 𝑗𝜔 , then
𝑡 1
−∞ 𝑥 𝜏 𝑑𝜏 ↔
𝑗𝜔
𝑋 𝑗𝜔 + π𝑋 0 𝛿(𝜔)and
DC value due to integration
23
Properties of Fourier Transform……….
8. Duality
If 𝑥 𝑡 ↔ 𝑋 𝑗𝜔 , then
𝑋 𝑡 ↔ 2𝜋 𝑥 −𝑗𝜔
X(j)
x(t) 1
Example 1:
𝜔
If 𝑥 𝑡 = 𝑟𝑒𝑡 𝑡 ↔ 𝑋 𝑗𝜔 = 𝑠𝑖𝑛𝑐 , then t
2𝜋 -1/2 0 1/2 -2 2
𝑡
𝑋 𝑡 = 𝑠𝑖𝑛𝑐 ↔ 2𝜋𝑥 −jω = 2𝜋 𝑟𝑒𝑡 𝜔 Even function
2𝜋
x(t)
X(j) 1
Scaling by a = 2𝜋,
𝜔
𝑠𝑖𝑛𝑐 𝑡 ↔ 𝑟𝑒𝑡
2𝜋 - 0
-1 1 t
Useful for computation of FT of typical signals 24
Example 4.13 1 :
2
Determine the FT of signal 𝑥 𝑡 =
1+𝑡 2
2𝑎
𝑒 −𝑎 𝑡 ↔ 2 , a>0
𝑎+𝜔
If a = 1
2
𝑒− 𝑡 ↔
1+𝜔2
1 𝜋
= 1 2 𝑑𝜔 = 1
2𝜋 −𝜋
Easy to compute in frequency domain for some signals Example 4.14 [1]
26
Properties of Fourier Transform……….
10. Convolution in Time Domain
If 𝑥 𝑡 ↔ 𝑋 𝑗𝜔 and 𝑦 𝑡 ↔ 𝑌 𝑗𝜔 then,
𝑥 𝑡 ∗ 𝑦(𝑡) ↔ 𝑋 𝑗𝜔 𝑌 𝑗𝜔
Applications:
1. Determining the response of an LTI system
𝑦 t = 𝑥 𝑡 ∗ ℎ 𝑡 ↔ 𝑌 𝑗𝜔 = 𝑋 𝑗𝜔 𝐻 𝑗𝜔
𝑌 𝑗𝜔 ↔ 𝑦(𝑡)
2. Design of Inverse system
𝑦(𝑡)
ℎ(𝑡) ℎ1 (𝑡)
ℎ 𝑡 ∗ ℎ1 𝑡 = 𝛿 𝑡 ↔ 𝐻 𝑗𝜔 𝐻1 𝑗𝜔 = 1
1
𝐻1 𝑗𝜔 =
𝐻 𝑗𝜔 27
Convolution in Time Domain…………..
28
Properties of Fourier Transform……….
11. Multiplication in Time Domain
If 𝑥 𝑡 ↔ 𝑋 𝑗𝜔 and 𝑦 𝑡 ↔ 𝑌 𝑗𝜔 then,
1
𝑥 𝑡 𝑦(𝑡) ↔ 𝑋 𝑗𝜔 ∗ 𝑌 𝑗𝜔
2𝜋
29
LTI System Analysis using Fourier Transform
An LTI system can be chracterized in terms of
• Linear-constant coefficient differential equation
𝑑𝑦(𝑡) 1 1
+ 𝑦 𝑡 = 𝑥(𝑡)
𝑑𝑡 𝑅𝐶 𝑅𝐶 𝑦(𝑡)
𝑥(𝑡)
• Impulse Response
𝑦 t =𝑥 𝑡 ∗ℎ 𝑡
• Transfer function or Frequency response of the system
𝑌 𝑗𝜔 = 𝑋 𝑗𝜔 𝐻 𝑗𝜔
or 𝐻 𝑗𝜔 = 𝑋𝑌 𝑗𝜔
𝑗𝜔
, System Transfer function or Frequency response of the system
1
Example 4.16 [1]
Determining System IR
6
Frequency Selective Filter Design
• Filter: Selects the desired frequency and rejects all other frequencies
• Ideal Low Pass Filter (LPF)
• It passes frequencies from 0 to 𝜔𝑐 and rejects other frequencies
• Frequency response 𝐻 𝑗𝜔 is
𝜔
𝐻 𝑗𝜔 = 𝑟𝑒𝑐𝑡
2𝜔𝑐 h(t)
𝜔 𝜔
∴ℎ 𝑡 = 𝑐 𝑠𝑖𝑛𝑐 𝑐 𝑡
𝜋 𝜋
• Perfect frequency selectivity, but IR
• ℎ 𝑡 ≠ 0 for t < 0 non-causal, not applicable 𝜋
-𝜔 𝜋 t
for real-time applications. Also, non-realizable. 𝑐 𝜔𝑐
8
Hilbert Transform
Hilbert Transformer is a system whose frequency response is characterized as follows :
• Magnitude response is unity for all frequencies
• Phase response is -90⁰ for +𝜔 and +90⁰ for −𝜔
Magnitude Response
∴ H jω = −j. sgn(ω)
h t =?
1
Hilbert Transform……….
t
1
න x(t)dt ⟺ X jω + πδ ω X(0)
jω
−∞
u t = න δ(t) dt
−∞
since, δ t ⟺ 1
t
1
x t = u t = න δ(t)dt ⟺ + πδ ω = X(jω)
jω
−∞
2
sgn t = 2odd u t ⟺ [xodd t ⟺ Im X jω ]
jω
j 1
sgn(t) ⟺
2π πω
Hilbert Transform……….
Using duality
1 j
⟺ 2π. sgn −ω = −j. sgn ω = H(jω)
πt 2π
y t = x t ∗ h(t)
+∞
y(t) = න x τ h t − τ dτ
−∞
+∞
1 x τ
y(t) = න dτ
π (t − τ)
−∞
Hilbert Transform……….
Examples:
1 1
where, gො t = ∗g t =g t ∗
πt πt
• Given signal g t is the real part of pre-envelope g + t
• H.T of g t , gො t is the imaginary part of pre-envelope g + t
One important feature of pre-envelope g + t is the behavior of its FT.
Let , g + (t) ⟺ G+ (jω)
Taking FT of eq (1)
G+ jω = G jω + j −j sgn ω G(jω)
G+ jω = 2G jω , ω>0
=G 0 , ω=0 (2)
=0, 5
ω < 0 ⇒ pre envelope of a signal has no frequency content for negative frequencies
Pre-Envelope………….
g + t = 2 න G jω ejωt dω (3)
0
Band Pass Signal
ωc = Carrier frequency
2ω1 = Bandwidth
Let the pre-envelope of narrowband signal g t with its FT G(jω) centerd around
±ωc is expressed as
g + t = g t ejωc t (4)
where g (t) is complex envelope of g t .
From Eqn. (2), we have
G+ jω = 2G jω , ω>0
=G 0 , ω=0
=0, ω<0
Band Pass Signal………
−1 gQ t
And 𝜑 𝑡 = tan ( ൗgI t ), phase of BP signal g(t)
s is complex
s = +j
eigenvalue eigenfunction
Since X (s) is infinite at a pole, therefore 𝐹 𝑒 −𝜎𝑡 𝑥(𝑡) clearly does not converge at
a pole, and thus the ROC cannot contain values of s that are poles.
ROC of Laplace Transform……
• Property 3:
If 𝑥(𝑡) is of finite duration and is absolutely integrable, then the ROC is the entire
s-plane.
If 𝑥(𝑡) is right sided, and if the line ℜ 𝑠 = 𝜎0 is in the ROC, then all values of s for
which ℜ 𝑠 > 𝜎0 will also be in the ROC.
If 𝑒 −𝜎0 𝑡 𝑥(𝑡) is integrable, then for 𝜎1 > 𝜎0 , 𝑒 −𝜎1 𝑡 𝑥(𝑡) is also integrable.
ROC of Laplace Transform……
• Property 5:
If 𝑥(𝑡) is left sided, and if the line ℜ 𝑠 = 𝜎0 is in the ROC, then all values of s for
which ℜ 𝑠 < 𝜎0 will also be in the ROC.
• Property 6:
If 𝑥(𝑡) is two sided, and if the line ℜ 𝑠 = 𝜎0 is in the ROC, then the ROC will consist
of a strip in the s-plane that includes the ℜ 𝑠 = 𝜎0 .
Example 9.7
ROC for 𝑥𝑅 (t) ROC for 𝑥𝐿 (t) ROC for 𝑥 𝑡 = 𝑥𝑅 (t) + 𝑥𝑅 (t)
ROC of Laplace Transform……
• Property 7:
If the Laplace transform X(s) of 𝑥(𝑡) is rational, then its ROC is bounded by poles or
extends to infinity. In addition, no poles of X(s) are contained in the ROC.
• Property 8:
If the Laplace transform X(s) of 𝑥(𝑡) is rational, then if 𝑥(𝑡) is right sided, the ROC is
the region in the s-plane to the right of the rightmost pole. If 𝑥(𝑡) is left sided, the
ROC is the region in the s-plane to the left of the leftmost pole.
Example 9.8 [1]
1
Let 𝑋 𝑠 =
𝑠+1 (𝑠+2)
Pole-zero pattern
1
Example:
• Determine the Transfer function of the following systems:
(i). An ideal delay of T second
𝑦 𝑡 = 𝑥(𝑡 − 𝑇)
Taking LT
𝑌 𝑠 = 𝑋 𝑠 𝑒 −𝑠𝑇 [Time shifting property of LT]
𝑌(𝑠)
or H s = = 𝑒 −𝑠𝑇
𝑋(𝑠)
The ROC associated with the system function for a causal system is a
right-half plane.
However, an ROC to the right of the rightmost pole does not guarantee that the
system is causal.
However,
For a system with a rational system function, causality of the system is
equivalent to the ROC being the right-half plane to the right of the
rightmost pole.
Example:
1
𝐻 𝑠 = , 𝑅𝑒 𝑠 > 1
(𝑠−1)(𝑠+2)
Causal
1
𝐻 𝑠 = , 𝑅𝑒 𝑠 < −𝑎
(𝑠+𝑎)
Non-causal
Examples 9.17, 9.18 and 9.19 [1]
System Characterization…..
2. Stability
System is stable if Fourie transform of its IR converges. And
𝐹𝑇 ℎ(𝑡) = 𝐿𝑇[ℎ 𝑡 ]𝑠=𝑗𝜔
An LTI system is stable if and only if the ROC of its system function H(s)
includes the 𝑗𝜔 − axis.
1
Let 𝑋 𝑠 =
𝑠+1 (𝑠+2)
Pole-zero pattern
IR of the system
1 −𝑡ൗ
ℎ 𝑡 = 𝑒 𝜏 𝑢(𝑡)
𝜏
Step Response
𝑡
1
𝑥 𝑡 = 𝑢 𝑡 = න 𝛿 𝜏 𝑑𝜏 ↔ 𝑋 𝑠 =
−∞ 𝑠
1 𝐴 𝐵
𝑌 𝑠 =𝑋 𝑠 𝐻 𝑠 = = +
𝑠(𝜏𝑠 + 1) 𝑠 (𝜏𝑠 + 1)
A = 1 and B = −𝜏
1 𝜏 1 1
𝑌 𝑠 = − = −
𝑠 𝜏𝑠+1 𝑠 (𝑠+1/𝜏)
𝑡 𝑡
𝑦 𝑡 = 𝑢 𝑡 − 𝑒 − Τ𝜏 𝑢 𝑡 = 1 − 𝑒 − Τ𝜏 𝑢 𝑡
Final value of the step response
lim 𝑦 𝑡 = 1
𝑡→∞
−2 𝜉 𝜔𝑛 ± 4 𝜉 2 𝜔𝑛 2 −4 𝜔𝑛 2
𝑠1 , 𝑠2 =
2
𝑠1 , 𝑠2 = − 𝜉 𝜔𝑛 ± 𝜔𝑛 𝜉2 − 1
1. 𝜉 = 0,
𝑠1 = 𝑗𝜔𝑛 and 𝑠2 = −𝑗𝜔𝑛 = 𝑠1 ∗ (lie on Im axis)
2. 𝜉 = 1,
𝑠1 = 𝑠2 = −𝜔𝑛 (lie on real axis)
3. 𝜉 > 1,
𝑠1 = − 𝜉 𝜔𝑛 + 𝜔𝑛 𝜉 2 − 1 and 𝑠2 = − 𝜉 𝜔𝑛 − 𝜔𝑛 𝜉2 − 1
𝑠1 and 𝑠2 are real and negative
4. 0 < 𝜉 < 1,
𝑠1 = − 𝜉 𝜔𝑛 + 𝑗𝜔𝑛 1 − 𝜉 2 and 𝑠2 = − 𝜉 𝜔𝑛 − 𝑗 𝜔𝑛 1 − 𝜉2
𝑠1 and 𝑠2 are complex conjugate
1
𝑋 𝑠 =
𝑠
𝜔𝑛 2
𝑌 𝑠 =𝑋 𝑠 𝐻 𝑠 =
𝑠 (𝑠 − 𝑠1 )(𝑠 − 𝑠2 )
1. 𝜉 = 0
𝜔𝑛 2 𝐴 𝐵 𝐶
𝑌 𝑠 = = + +
𝑠 (𝑠 − 𝑗𝜔𝑛 )(𝑠 + 𝑗𝜔𝑛 ) 𝑠 (𝑠 − 𝑗𝜔𝑛 ) (𝑠 + 𝑗𝜔𝑛 )
1 𝑗𝜔 𝑡 1 −𝑗𝜔 𝑡
𝑦 𝑡 =𝑢 𝑡 − 𝑒 𝑛 𝑢 𝑡 − 𝑒 𝑛 𝑢 𝑡
2 2
𝑦 𝑡 = 1 − cos 𝜔𝑛 𝑡 𝑢(𝑡)
2. 0 < 𝜉 < 1, under damped system
Step response
𝑒 −𝜉𝜔𝑛 𝑡 1 − 𝜉2
𝑦 𝑡 = 1− sin 𝜔𝑛 1 − 𝜉 2 𝑡 + 𝑡𝑎𝑛−1 𝑢(𝑡
1 − 𝜉2 𝜉
Step response is plotted below for different values of 𝜉.
𝜔𝑛 2 𝜔𝑛 2
𝐻 𝑗𝜔 = 𝑗𝜔 2 𝑗𝜔
= 𝜔 2 𝜔
+2 𝜉 +1 1− 𝜔 +𝑗 2 𝜉 𝜔
𝜔𝑛 𝜔𝑛 𝑛 𝑛
1
𝐻 𝑗𝜔 =
2 2 2
𝜔 𝜔
1− + 4 𝜉2
𝜔𝑛 𝜔𝑛
𝜔
➢ ≪ 1 or 𝜔 ≪ 𝜔𝑛 , low frequencies
𝜔𝑛
𝜔𝑛 2
𝐻 𝑗𝜔 ≈ , or 20log 𝐻 𝑗𝜔 = −40 log 𝜔 + 40 log 𝜔𝑛 (in dB) (2)
𝜔
decreases 40 dB/decade
➢ At 𝜔 = 𝜔𝑛 , low and high frequency asymptotes meet.
➢ Actual magnitude response is a function of 𝜉
▪ For 𝜉 > 0.707, no peak
• Aavailability of low cost power full digital processors, digital storage media and digital
transmission requires a discrete-time (DT) signal
• Sampling converts a CT signal into corresponding sequence of samples that are usually
spaced uniformly in time domain (DT signal).
• The sampling duration/period should be chosen such that the sequence of samples
uniquely defined the original analog signal.
Impulse-Train (Ideal) Sampling
• Let 𝑥(𝑡) be the CT signal and 𝑥[𝑛𝑇] = value of 𝑥(𝑡) at
time 𝑡 = 𝑛𝑇.
• 𝑥[𝑛𝑇] is the nth sample value and is obtained by
performing 𝑥(𝑡)𝛿(𝑡 − 𝑛𝑇)
(sampling property of delta function).
• Therefore, the signal to be sampled is multiplied with
a periodic impulse train with period 𝑇 as shown in
Fig.1.
• The period T is called sampling period (or interval)
2𝜋
and 𝜔𝑆 = is called sampling frequency.
𝑇
➢ How to choose sampling period T such that
sequence of samples {𝑥[𝑛𝑇]} uniquely specify 𝑥(𝑡)?
Fig. 1
Ideal Sampling……..
The sampled signal is given as
𝑥𝑝 𝑡 = 𝑥 𝑡 𝑝(𝑡) (1)
where,
𝑝(𝑡) = σ∞
𝑛=−∞ 𝛿 (𝑡 − 𝑛𝑇); impulse train (2)
Putting (2) in (1), we have
𝑥𝑝 (𝑡) = 𝑥(𝑡) σ∞ ∞
𝑛=−∞ 𝛿 (𝑡 − 𝑛𝑇) = σ𝑛=−∞ 𝑥[𝑛𝑇] 𝛿 (𝑡 − 𝑛𝑇) (3)
We know that
σ∞ σ ∞
𝑛=−∞ 𝛿(𝑡 − 𝑛𝑇) ⇔ 𝜔𝑆 𝑘=−∞ 𝛿(𝜔 − 𝑘𝜔𝑆 ) (4)
If 𝑥(𝑡) ⇔ 𝑋(𝑗𝜔), then by multiplication in time domain property, Fourier transform
(FT) of the sampled signal (3) is given as
Ideal Sampling……..
∞
1
𝑋𝑝 = [𝑋(𝑗𝜔) ⊕ 𝜔𝑆 𝛿(𝜔 − 𝑘𝜔𝑆 )]
2𝜋
𝑘=−∞
1
= σ∞
𝑘=−∞ 𝑋(𝑗(𝜔 − 𝑘 𝜔𝑆 )) (5)
𝑇
Fig. 2
Ideal Sampling……..
• Periodic signal Discrete spectra (from Fourier series)
• Discrete time signal Periodic spectra (from sampling)
• For a signal with band limited spectra as shown in Fig. 2(a)
• There is no overlap between the shifted replicas of X(j) for S ≥ 2M as
shown in Fig. 2(c).
• X(j) is reproduced at integer multiples of the sampling frequency S.
• Consequently, original signal x(t) can be recovered exactly from its sampled
version xp(t) by using a low pass filter with gain T and cutoff frequency greater
than M and less than S-M shown in Fig. 3
Fig. 3 Exact recovery of a CT signal from its samples using
an ideal LPF.
Examples
• Toll quality speech (PSTN): frequency band 300 – 3400 Hz
fM = 3400 Hz
fS = 8000 samples/sec ; standard sampling rate
• Problem 7.1 [1]
𝑥(𝑡) is sampled at 𝜔𝑠 = 10,000 π . For what values of 𝜔 is 𝑋(𝑗𝜔) guaranteed to be zero.
Soln.
𝜔𝑠 ≥ 2 𝜔𝑀
or 𝜔𝑀 ≤ 𝜔𝑆 /2
𝜔𝑀 ≤ 10000 π/2 or 𝜔𝑀 ≤ 5000 π
• Problem 7.3 [1]
Determine the Nyquist rate corresponding to each of the following signals:
(a). 𝑥 𝑡 = 1 + cos 2000𝜋𝑡 + 𝑠𝑖𝑛(4000𝜋𝑡)
𝜔0 = 0, 𝜔1 = 2000𝜋, 𝜔2 = 4000𝜋
𝜔𝑀 = 4000𝜋 (highest signal frequency)
Examples……..
sin(4000𝜋𝑡)
(b). 𝑥 𝑡 = = 4000 𝑠𝑖𝑛𝑐(4000𝑡)
𝜋𝑡
.
.
𝜔
4000 𝑠𝑖𝑛𝑐(4000𝑡) ⇔ 𝑟𝑒𝑐𝑡
8000𝜋
𝜔𝑀 = 4000𝜋 − 4000𝜋 4000𝜋
𝜔𝑆 = 2𝜔𝑀 = 8000𝜋
sin(4000𝜋𝑡) 2
(c). 𝑥 𝑡 =
𝜋𝑡
𝜔 𝜔
𝑥 𝑡 = 4000 𝑠𝑖𝑛𝑐(4000𝑡) × 4000 𝑠𝑖𝑛𝑐(4000𝑡) ⇔ 𝑟𝑒𝑐𝑡 ∗ 𝑟𝑒𝑐𝑡
8000𝜋 8000𝜋
𝜔
=∆
8000𝜋
𝜔𝑀 = 8000𝜋
𝜔𝑆 = 2𝜔𝑀 = 16000𝜋 − 8000𝜋 8000𝜋
Sampling….
• Sampling Theorem
A band limited signal of finite energy, which has no frequency components higher than 𝜔𝑀 , may be
completely recovered from the knowledge of its samples taken at the rate of 2𝜔𝑀 samples per
second. Sampling rate of 𝜔𝑆 = 2𝜔𝑀 is called the Nyquist sampling rate.
• Aliasing
If sampling rate is less than the Nyquist rate, then high frequency components in the spectrum
of the signal appear as lower frequencies in the spectrum of its sampled version as shown in
Figure 2(d). This gives distortion.
• A chance of Aliasing is there since messages to be sampled are time limited, and hence their
spectra cannot be strictly band limited.
• To avoid aliasing:
• Prior to sampling, signal to be sampled is filtered with a LPF.
• Filtered signal is sampled at rate slightly higher than the Nyquist rate.
Signal Reconstruction from Its Samples
• For reconstruction, interpolation (i.e., the fitting of a continuous signal to a set of sample values)
is commonly used.
• Reconstruction may be either approximately or exactly.
• As discussed earlier for a band-limited signal, if the sampling rate is greater than the Nyquist rate,
then the signal can be reconstructed exactly; i.e., through the use of a lowpass filter, exact
interpolation can be carried out between the sample points as shown below.
𝑋𝑟 𝑗𝜔 = 𝑋𝑝 𝑗𝜔 𝐻 𝑗𝜔
Or
𝑥𝑟 𝑡 = 𝑥𝑝 𝑡 ∗ ℎ 𝑡
𝜔 𝜔𝑐 𝑡
𝐻 𝑗𝜔 = 𝑇 𝑟𝑒𝑐𝑡 ↔ ℎ 𝑡 = 2 𝑇𝜔𝑐 𝑠𝑖𝑛𝑐
2𝜔𝑐 𝜋
Signal Reconstruction…….
∵ 𝑥𝑝 (𝑡) = σ∞
𝑛=−∞ 𝑥[𝑛𝑇] 𝛿 (𝑡 − 𝑛𝑇)
∴ 𝑥𝑟 (𝑡) = σ∞
𝑛=−∞ 𝑥 𝑛𝑇 𝛿 (𝑡 − 𝑛𝑇) ∗ ℎ(𝑡)
𝑥𝑟 (𝑡) = σ∞
𝑛=−∞ 𝑥 𝑛𝑇 ℎ (𝑡 − 𝑛𝑇)
𝜔𝑐 (𝑡−𝑛𝑇)
𝑥𝑟 (𝑡) = 2 𝑇𝜔𝑐 σ∞
𝑛=−∞ 𝑥 𝑛𝑇 𝑠𝑖𝑛𝑐 𝜋
𝜔𝑠 𝜋
For 𝜔𝑐 = =𝑇
2
𝑡
𝑥𝑟 (𝑡) = 2 𝜋 σ∞
𝑛=−∞ 𝑥 𝑛𝑇 𝑠𝑖𝑛𝑐 −𝑛
𝑇
Discrete-Time Fourier Transform (DTFT)
• CTFT 𝑥(𝑡) ⇔ 𝑋 𝑗𝜔
where
∞
F 𝑥(𝑡) = 𝑋 𝑗𝜔 = −∞ 𝑥(𝑡)𝑒 −𝑗𝜔𝑡 𝑑𝑡 (1)
and
1 ∞
𝑥 𝑡 = 𝑋 𝑗𝜔 𝑒 𝑗𝜔𝑡 𝑑𝜔 (2)
2𝜋 −∞
• FT of ideal sampled signal
1 ∞
F σ∞
𝑛=−∞ 𝑥 𝑛𝑇 𝛿 (𝑡 − 𝑛𝑇) = σ 𝑋(𝑗(𝜔 − 𝑘 𝜔𝑆 )) (3)
𝑇 𝑘=−∞
2𝜋
Periodic with period 𝜔𝑆 =
𝑇
If T = 1, then (3) becomes
∞
F σ∞ 𝑛=−∞ 𝑥 𝑛 𝛿 (𝑡 − 𝑛) = σ𝑘=−∞ 𝑋(𝑗(𝜔 − 𝑘 2𝜋)); Periodic with period = 2𝜋
σ∞ −𝑗𝜔𝑛 = σ∞
𝑛=−∞ 𝑥 𝑛 𝑒 𝑘=−∞ 𝑋(𝑗(𝜔 − 𝑘 2𝜋)) [left side is from (1)]
DTFT………
• DTFT
𝑋 𝑒 𝑗𝜔 = σ∞
𝑛=−∞ 𝑥 𝑛 𝑒 −𝑗𝜔𝑛 (4) analysis equation
Variable is 𝜔 (0 − 2π)
IDFT
1
𝑥[𝑛] = 𝑋 𝑒 𝑗𝜔 𝑒 𝑗𝜔𝑛 𝑑𝜔 (5) synthesis equation
2𝜋 2𝜋
a>0
a<0
DTFT of Periodic Signals
• CTFT of Periodic Signal
𝛿 𝑡 ↔1
1 ↔ 2𝜋 𝛿 𝜔
𝑒 𝑗𝜔𝑜 𝑡 ↔ 2𝜋 𝛿 𝜔 −𝜔𝑜
𝑒 𝑗𝑘𝜔𝑜 𝑡 ↔ 2𝜋 𝛿 𝜔 −𝑘𝜔𝑜
𝑎𝑘 𝑒 𝑗𝑘𝜔𝑜 𝑡 ↔ 2𝜋𝑎𝑘 𝛿 𝜔 −𝑘𝜔𝑜
σ∞ 𝑎
𝑘=−∞ 𝑘 𝑒 𝑗𝑘𝜔𝑜 𝑡 ↔ σ∞
𝑘=−∞ 2𝜋𝑎𝑘 𝛿 𝜔 −𝑘𝜔𝑜
∴ 𝑋 𝑒 𝑗𝜔 = σ∞ 𝑒 𝑗𝜔𝑜 𝑛 𝑒 −𝑗𝜔𝑛 = σ∞
𝑛=−∞ 𝑙=−∞ 2𝜋 𝛿 𝜔 −𝜔𝑜 −2𝜋𝑙
DTFT of Periodic Signals……
2𝜋
• Periodic sequence 𝑥[𝑛] with period N, 𝜔𝑜 =
𝑁
FT of x[n]
∞
∴ 𝑋 𝑒 𝑗𝜔 = 2𝜋𝑎𝑘 𝛿 𝜔 − 2𝜋𝑘ൗ𝑁
𝑘=−∞
Fourier transform of a DT periodic signal can be directly constructed from its Fourier
coefficients.
where
DTFT of Sine Function
12
DTFT of DT Unit Sample Train
𝑋 𝑒 𝑗𝜔 = 2𝜋𝑎𝑘 𝛿 𝜔 − 2𝜋𝑘ൗ𝑁
𝑘=−∞
Fourier transform of a DT periodic signal can be directly
constructed from its Fourier coefficients. Where,
Discrete Fourier Transform (DFT)
• DTFT 𝑋 𝑒 𝑗𝜔 = σ∞
𝑛=−∞ 𝑥 𝑛 𝑒
−𝑗𝜔𝑛
0 N-1 n
For 𝑎 > 1, ROC does not contain unit circle, hence DTFT of 𝑎𝑛 𝑢 𝑛 does not exist.
Example 2
Inverse z-transform
• Partial Fraction Method
8𝑧−19
(i). 𝑋 𝑧 = , 𝑧 >3
(𝑧−2)(𝑧−3)
8−19𝑧 −1 𝐴 𝐵
𝑋 𝑧 = = +
(1−2𝑧 −1 )(1−3𝑧 −1 ) (1−2𝑧 −1 ) (1−3𝑧 −1 )
𝐴 = 𝑋 𝑧 (1 − 2𝑧 −1 )ȁ𝑧=2 = 3, 𝐵 = 𝑋 𝑧 (1 − 3𝑧 −1 )ȁ𝑧=3 = 5
3 5
𝑋 𝑧 = +
(1−2𝑧 −1 ) (1−3𝑧 −1 )
1
𝑎𝑛 𝑢 𝑛 ↔ , 𝑧 > 𝑎
1−𝑎𝑧 −1
3 5
, 𝑧 > 2 ↔ 3 2𝑛 𝑢 𝑛 and , 𝑧 > 3 ↔ 5 3𝑛 𝑢 𝑛
(1−2𝑧 −1 ) (1−3𝑧 −1 )
Inverse z-transform……..
𝑥 𝑛 = 3 2𝑛 𝑢 𝑛 + 5 3𝑛 𝑢 𝑛
Examples 10.9, 10.10, 10.11 [1].
3 5
𝑋 𝑧 = + , 2< 𝑧 <3
(1−2𝑧 −1 ) (1−3𝑧 −1 )
3
, 𝑧 > 2 ↔ 3 2𝑛 𝑢 𝑛
(1−2𝑧 −1 )
5
and , 𝑧 < 3 ↔ −5 3𝑛 𝑢 −𝑛 − 1
(1−3𝑧 −1 )
𝑥 𝑛 = 3 2𝑛 𝑢 𝑛 − 5 3𝑛 𝑢 −𝑛 − 1
Inverse z-transform…….
(b). When X(z) is not rational
By power series expansion of X(z)
𝑋 𝑧 = σ∞ 𝑛=−∞ 𝑥 𝑛 𝑧 −𝑛
X(z) = ⋯ + 𝑥 −2 𝑧 2 + 𝑥 −1 𝑧 + 𝑥 0 𝑧 0 + 𝑥 1 𝑧 −1 + 𝑥 2 𝑧 −2 + ⋯
Example:
X z = 4 𝑧 2 + 2 + 3 𝑧 −1 , 0 < 𝑧 < ∞
𝑥 −2 = 4
𝑥 0 =2
𝑥 1 =3
𝑥 𝑛 = 4 δ n + 2 + 2 δ n + 3 δ[n − 1]
Inverse z-transform…….
By long division method
Properties of z-transform
• Linearity
If 𝑥1 [𝑛] ↔ 𝑋1 (𝑧), ROC = 𝑅1
𝑥2 [𝑛] ↔ 𝑋2 (𝑧), ROC = 𝑅2
then
𝑎 𝑥1 𝑛 + 𝑏 𝑥2 𝑛 ↔ 𝑎 𝑋1 𝑧 + b 𝑋2 (𝑧), ROC = 𝑅1 ∩ 𝑅2
• Time Shifting
If 𝑥[𝑛] ↔ 𝑋 𝑧 , ROC = 𝑅, then
𝑥 𝑛 − 𝑛𝑜 ↔ 𝑧 −𝑛𝑜 𝑋 𝑧
𝑥 𝑛 + 𝑛𝑜 ↔ 𝑧 𝑛𝑜 𝑋 𝑧 , ROC = R except possible addition or
deletion of origin or ∞
Properties of z- transform…….
• Scaling in z-domain
𝑥[𝑛] ↔ 𝑋 𝑧 , ROC = 𝑅
𝑧𝑜𝑛 𝑥[𝑛] ↔ 𝑋 𝑧Τ𝑧𝑜 , ROC = 𝑧𝑜 𝑅
• Time Reversal
𝑥[𝑛] ↔ 𝑋 𝑧 , ROC = 𝑅
1 1
𝑥[−𝑛] ↔ 𝑋 , ROC =
𝑧 𝑅
• Convolution in time domain
If 𝑥1 [𝑛] ↔ 𝑋1 (𝑧), ROC = 𝑅1
𝑥2 [𝑛] ↔ 𝑋2 (𝑧), ROC = 𝑅2 , then
𝑥1 𝑛 ∗ 𝑥2 [𝑛] ↔ 𝑋1 𝑧 𝑋2 𝑧 , ROC = 𝑅1 ∩ 𝑅2
Exp 10.15 , 10.16 [1]
Properties of z- transform……..
• Differentiation in the z- domain
if 𝑥[𝑛] ↔ 𝑋 𝑧 , ROC = 𝑅
𝑑𝑋(𝑧)
then 𝑛 𝑥[𝑛] ↔ −𝑧 , ROC = 𝑅
𝑑𝑧
Exp 10.17 , 10.18 [1]
For other properties, refer to [1]
Analysis and characterization of LTI systems using z-transform
𝑦 𝑛 = 𝑥[𝑛] ∗ ℎ[𝑛]
Using convolution property
𝑌 𝑧 =𝑋 𝑧 𝐻 𝑧
𝑌(𝑧)
or 𝐻 𝑧 = ; System function or transfer function
𝑋(𝑧)
For 𝑧 = 𝑒 𝑗𝑤 , evaluated on unit circle
𝐻 𝑧 will be the frequency respons.
* Many properties of the system can be described using H(z)
Causality
IR ℎ𝑛
𝐻 𝑧 = σ∞ −∞ ℎ 𝑛 𝑧 −𝑛
1 1 1
𝐻 𝑧 = 1 + , < 𝑧 < 2 stable
1− 𝑧 −1 1−2𝑧 −1 2
2
right sided left sided
• z-transform
• To obtain system function H(z)
• To obtain frequency response 𝐻 𝑗𝜔 = 𝐻 𝑧 ȁ𝑧= 𝑒 𝑗𝜔
• To obtain time domain response
Block Diagram Representation of LTI System
• Consider an LTI system with input 𝑥(𝑡), output 𝑦(𝑡)and impulse response ℎ(𝑡)
• Feedback Interconnection
𝑦(𝑧) 𝐻1 (𝑧)
𝐻 𝑧 = =
𝑥(𝑧) 1+𝐻1 (𝑧)𝐻2 (𝑧)