You are on page 1of 43

EE330 Digital Signal Processing

Discrete Time Signals and Systems

Arbab Latif
Spring 2023

Resources:
Discrete Time Signal Processing, A. V. Oppenheim and R. W. Schaffer, 3rd Edition, 2010 (Chapter 2.1-2.7)
2 Today’s Class
• Signal
• Discrete Time Signals
• Discrete Time Systems
– Types / Properties
• Linear Time Invariant Systems
• Properties of LTI Systems
Discrete Time Signals

Section 2.1
4 Signal Types

Signals

Continuous-time Discrete-time

Continuous-value Continuous-value Discrete-value

Analog Discrete Digital


5 Signal Types
• Analog signals: continuous in time and amplitude
– Example: voltage, current, temperature,…
• Digital signals: discrete both in time and amplitude
– Example: attendance of this class, digitizes analog signals,…
• Discrete-time signals: discrete in time, continuous in
amplitude
– Example: hourly change of temperature
• Theory of digital signals would be too complicated
– Requires inclusion of nonlinearities into theory
• Theory is based on discrete-time continuous-amplitude
signals
– Most convenient to develop theory
– Good enough approximation to practice with some care
• In practice we mostly process digital signals on processors
– Need to take into account finite precision effects
6 Signal Types

Continuous time –
Continuous amplitude

Continuous time –
Discrete amplitude

Discrete time –
Continuous amplitude

Discrete time –
Discrete amplitude
7 Discrete-Time Signals: Sequences
• Discrete-time signals are represented by sequence of numbers
– The nth number in the sequence is represented with x[n]
• Often times sequences are obtained by sampling of continuous-
time signals
– In this case x[n] is value of the analog signal at xc(nT)
– Where T is the sampling period
10

-10
0 20 40 60 80 100 t (ms)
10

-10
0 10 20 30 40 50 n (samples)
8 Example of a discrete-time Signal:
9 A sampled signal:
10 Unit Sample Sequence:

1 𝑛=0
• 𝑥 𝑛 =𝛿 𝑛 =ቊ
0 𝑛≠0

Any sequence can be expressed as a sum of scaled


and delayed impulses. ?
𝑥 𝑛 = σ∞
𝑘=−∞ 𝑥 𝑘 𝛿[𝑛 − 𝑘]
11
A sequence as a sum of scaled and
delayed impulses
Any sequence can be expressed as a sum of scaled
and delayed impulses.
𝑥 𝑛 = σ∞𝑘=−∞ 𝑥 𝑘 𝛿[𝑛 − 𝑘]

So, can you express 𝑝 𝑛 ?


12 Unit-Step Sequence:
1 𝑛≥0
• 𝑢 𝑛 =ቊ
0 𝑛 < 0.
• Can you express 𝑢 𝑛 in terms of 𝛿[𝑛] ?
• 𝑢 𝑛 =𝛿 𝑛 +𝛿 𝑛−1 +𝛿 𝑛−2 + …
• 𝑢 𝑛 = σ∞ 𝑘=0 𝛿[𝑛 − 𝑘]

• Also, 𝑢 𝑛 = σ𝑛𝑘=−∞ 𝛿[𝑘]


• Can you express 𝛿 𝑛 in terms of 𝑢[𝑛] ?
• 𝛿 𝑛 =𝑢 𝑛 −𝑢 𝑛−1
13 Real Exponential Signal:
• 𝒙 𝒏 = 𝑨𝜶𝒏
• For 𝐴 ∈ ℝ+ and 0 < 𝛼 < 1.
– Monotonically decaying
• For 𝐴 ∈ ℝ− and 0 < 𝛼 < 1
– Negative sequence decaying
• For 𝐴 ∈ ℝ and 𝛼 = 1
– Constant Signal
• For 𝐴 ∈ ℝ+ and 𝛼 > 1
– Positive sequence and magnitude increasing
• 𝛼 < −1
– Alternating between positive and negative values
– Increasing in magnitude
• −1 < 𝛼 < 0
– Alternating between positive and negative values
– Decreasing in magnitude
14 Complex Exponential Signal
• 𝑥 𝑛 = 𝐴𝛼 𝑛 = 𝐴 𝑒 𝑗𝜙 𝛼 𝑛 𝑒 𝑗𝜔0 𝑛
• 𝑥 𝑛 = 𝐴 𝛼 𝑛 𝑒 𝑗 𝜔0 𝑛+𝜙
• 𝑥 𝑛 = 𝐴 𝛼 𝑛 cos 𝜔0 𝑛 + 𝜙 + 𝑗 𝐴 𝛼 𝑛 sin 𝜔0 𝑛 + 𝜙
• Special Cases:
– 𝛼 > 1 : sequence oscillates with exponentially growing
envelope
– 𝛼 < 1 : sequence oscillates with exponentially decaying
envelope
– 𝛼 = 1 : sinusoid with amplitude |𝐴|,frequency 𝜔0 and phase
𝜙.
17 Demo

https://youtu.be/LznjC4Lo7lE
18 Periodic Signals:
• A signal 𝑥 𝑛 is considered periodic if ?

• 𝑥 𝑛 = 𝑥 𝑛 + 𝑁 , for all 𝑛 ∈ ℤ, and N = integer

• For discrete-time sinusoid, this property implies that


• 𝐴 cos 𝜔0 𝑛 + 𝜙 = cos 𝜔0 𝑛 + 𝜔0 𝑁 + 𝜙
• Which requires that
• 𝜔0 𝑁 = 2𝜋𝑘 ,
• where 𝑘 is an integer.
EE330 Digital Signal Processing

Discrete Time Signals and Systems

Arbab Latif
Spring 2023

Resources:
Discrete Time Signal Processing, A. V. Oppenheim and R. W. Schaffer, 3rd Edition, 2010 (Chapter 2.1-2.7)
Discrete Time Systems

• Section 2.2
22 Discrete-time System
• A discrete-time system is a transformation or mapping
from an input sequence 𝑥[𝑛] to the output sequence with
values 𝑦 𝑛 .

• 𝑦 𝑛 =𝑇 𝑥 𝑛 .

x[n]
T{.} y[n]
23 Example 2.3: Ideal Delay System
• The ideal delay system is defined by the equation: ?
𝑦 𝑛 = 𝑥[𝑛 − 𝑛𝑑 ], −∞ < 𝑛 < ∞.
• When 𝑛𝑑 > 0, the system provides a time delay of 𝑛𝑑
samples
• When 𝑛𝑑 < 0, the system provides a time advance of 𝑛𝑑
samples.
• Impulse Response: ?
𝛿[𝑛 − 𝑛𝑑 ]
• Frequency Response: ?
𝑒 −𝑗𝜔𝑛𝑑
24 Ex. 2.4: Moving Average System:
• The moving average system is defined by the equation:
1 𝑀2
• 𝑦𝑛 = σ𝑘=−𝑀 𝑥[𝑛 − 𝑘]
𝑀1 +𝑀2 +1 1
1
• 𝑦𝑛 = ሼ𝑥 𝑛 + 𝑀1 + 𝑥 𝑛 + 𝑀1 − 1 + ⋯ 𝑥 𝑛 − 1 + 𝑥 𝑛 + ⋯
𝑀1 +𝑀2 +1
+ 𝑥 𝑛 − 𝑀2 ሽ.
• The system is non-causal when ?
• −M1 < 0
25 Memoryless Systems:
• A memoryless system requires only the current value of the
input sequence 𝑥[𝑛] to compute the current value of the
output 𝑦 𝑛 .

• Example 2.5: 𝑦 𝑛 = 𝑥 2 [𝑛].

• Is ideal delay system memoryless ?

• Ideal delay system is not memoryless for 𝑛𝑑 ≠ 0.


26 Linear Systems
• Systems that satisfy the two properties of additivity and
scaling (or homogeneity)
• Additivity: 𝑇 𝑥1 𝑛 + 𝑥2 𝑛 = 𝑇 𝑥1 𝑛 + 𝑇ሼ𝑥2 [𝑛]ሽ.
• Scaling: 𝑇 𝛼𝑥1 𝑛 = 𝛼𝑇ሼ𝑥[𝑛]ሽ

• Is accumulator system: 𝑦 𝑛 = σ𝑛𝑘=−∞ 𝑥[𝑘] linear ?


• Yes
• Is the system 𝑦 𝑛 = sin(𝑥[𝑛]) linear system ?
• No
27 Time-invariant Systems:
• A shifted input causes an output shifted by the
same amount.
• Let 𝑦 𝑛 = 𝑇ሼ𝑥[𝑛]ሽ , then for a time-invariant
system it must hold that 𝑇 𝑥 𝑛 − 𝑛0 =
𝑦 𝑛 − 𝑛0 , ∀𝑛0 ∈ ℤ.
• Is accumulator a time-invariant system ?
• Yes
28 Example: The Compressor System
• 𝑦 𝑛 = 𝑥[𝑀𝑛], −∞ < 𝑛 < ∞.
• It retains every 𝑀-th sample.
• System is not time-invariant for 𝑀 > 1

𝜹[𝒏] x[2n] 𝜹[𝒏]

𝜹[𝒏 − 𝟏]
x[2n] 0
29 Causality:
• For a causal system, the computation of the output must not
require any future value.
• The compressor system 𝑦 𝑛 = 𝑥[2𝑛] is non-causal because
𝑦 1 = 𝑥[2].
• Forward-difference system, i.e.,
𝑦 𝑛 = 𝑥 𝑛 + 1 − 𝑥[𝑛]
is non-causal as 𝑦 1 = 𝑥 2 − 𝑥 1 .
• Backward-difference system, i.e.,
𝑦 𝑛 = 𝑥 𝑛 − 𝑥[𝑛 − 1]
is causal.
• All memoryless systems are causal.
30 Stability:
• A bounded input must produce a bounded output,
• i.e., if there exists a positive real number 𝐵𝑥 < ∞ such that
𝑥 𝑛 ≤ 𝐵𝑥 for all 𝑛, then there must exist 𝐵𝑦 ∈ ℝ+ such that
𝑦 𝑛 ≤ 𝐵𝑦 < ∞.
• Is an accumulator a stable system ?
• For example, if 𝑥 𝑛 = 𝑢[𝑛], then we can choose 𝐵𝑥 = 1, i.e.,
the input is bounded.
• Now the output is
0, 𝑛 < 0
𝑦𝑛 =ቊ
𝑛 + 1, 𝑛 ≥ 0.
• Thus, the output is unbounded, hence, its an unstable system.
Linear Time-invariant Systems

Section 2.3
32 LTI Systems:
• Linear and Time-invariant Systems.

• Characterized completely by the response to the unit impulse


sequence 𝛿[𝑛].

• As 𝑦 𝑛 = 𝑇 𝑥 𝑛 , so, lets define ℎ 𝑛 = 𝑇ሼ𝛿[𝑛]ሽ

• Now, as 𝑥 𝑛 = σ+∞
𝑘=−∞ 𝑥 𝑘 𝛿[𝑛 − 𝑘].

• So, 𝑦 𝑛 = 𝑇ሼσ+∞
𝑘=−∞ 𝑥 𝑘 𝛿[𝑛 − 𝑘]ሽ

• (By Linearity of 𝑇) 𝑦 𝑛 = σ+∞


𝑘=−∞ 𝑥 𝑘 𝑇ሼ𝛿 𝑛 − 𝑘 ሽ

• (By Time-invariance of 𝑇) ℎ 𝑛 − 𝑘 = 𝑇ሼ𝛿 𝑛 − 𝑘 ሽ

• Thus, 𝑦 𝑛 = σ+∞
𝑘=−∞ 𝑥 𝑘 ℎ[𝑛 − 𝑘]
33 Convolution Sum:
𝑘=∞

𝑦 𝑛 = ෍ 𝑥 𝑘 ℎ[𝑛 − 𝑘]
𝑘=−∞
• How to compute:
– Take two signals 𝑥[𝑛] and ℎ 𝑛 .
– Express 𝑥[𝑛] over the variable of summation, i.e., 𝑥[𝑘] .
– Express ℎ[𝑛] over the variable of summation, i.e., ℎ 𝑘 .
– Flip ℎ[𝑘] about the 𝑦 − 𝑎𝑥𝑖𝑠 and get ℎ[−𝑘]
– Add 𝑛 to the axis to get ℎ[𝑛 − 𝑘].
– For each 𝑛, multiply 𝑥[𝑘] and ℎ[𝑛 − 𝑘] and sum over 𝑘.
34
Example 2.13: Computation of the
Convolution Sum
• ℎ 𝑛 =𝑢 𝑛 −𝑢 𝑛−𝑁
• 𝑥 𝑛 = 𝛼 𝑛 𝑢[𝑛]
35
Example 2.13 : Computation of the
Convolution Sum (contd.)
• For 𝑛 < 0, 𝑦 𝑛 = 0. No overlap.
36
Example 2.13 : Computation of the
Convolution Sum (contd.)
• For 0 ≤ 𝑛 and 𝑛 − 𝑁 − 1 ≤ 0:
• 𝑦 𝑛 = σ𝑛𝑘=0 𝛼 𝑘 , 𝑓𝑜𝑟 0 ≤ 𝑛 ≤ 𝑁 − 1
𝛼 𝑁1 −𝛼 𝑁2+1
• Using the relation σ𝑁2
𝑘=𝑁1 𝛼 𝑘 =
1−𝛼
1−𝛼 𝑛+1
• 𝑦𝑛 = , 𝑓𝑜𝑟 0 ≤ 𝑛 ≤ 𝑁 − 1
1−𝛼
37
Example 2.13 : Computation of the
Convolution Sum (contd.)
• For 0 < 𝑛 − 𝑁 − 1 or 𝑛 > N +1
• 𝑦 𝑛 = σ𝑛𝑘=𝑛−𝑁+1 𝛼 𝑘
𝛼 𝑛−𝑁+1 −𝛼 𝑛+1 𝑛−𝑁+1 1−𝛼
𝑁
• 𝑦𝑛 = = 𝛼 ( )
1−𝛼 1−𝛼
38 Example 2.13: Solution
Properties of Linear Time-
Invariant Systems

Section 2.4
40 Properties of Convolution:
• Convolution is commutative:

– 𝑥 𝑛 ∗ ℎ 𝑛 = ℎ 𝑛 ∗ 𝑥[𝑛]
– Proof: 𝐿𝐻𝑆 = 𝑥 𝑛 ∗ ℎ 𝑛
• = σ∞𝑘=−∞ 𝑥 𝑘 ℎ[𝑛 − 𝑘]
• = σ∞𝑚=−∞ 𝑥 𝑛 − 𝑚 ℎ[𝑚]
• = σ∞𝑚=−∞ ℎ[𝑚]𝑥 𝑛 − 𝑚
• = ℎ 𝑛 ∗ 𝑥 𝑛 = 𝑅𝐻𝑆.
41 Properties of Convolution:
• Distribution over Addition [Parallel Systems]:
– 𝑥 𝑛 ∗ ℎ1 𝑛 + ℎ2 𝑛 = 𝑥 𝑛 ∗ ℎ1 𝑛 + 𝑥 𝑛 ∗ ℎ2 [𝑛]
42 Properties of Convolution:
• Associative Property /Cascade/Series Connected
Systems:
43
An LTI System is stable iff the impulse response is
absolutely summable

• LTI System Stable ↔ Absolutely Summable impulse


response.
• Proof: If
• then

• If 𝑥 𝑛 is bounded, so that

• Then,
44 Causality
• Causality ↔ ℎ 𝑛 = 0 for 𝑛 < 0.
45 Impulse Response:
45

1. Ideal delay system: ℎ 𝑛 = 𝛿[𝑛 − 𝑛𝑑 ]

1
2. Moving Average: ℎ 𝑛 = σ𝑀2
𝛿[𝑛 − 𝑘]
𝑀1 +𝑀2 +1 𝑘=−𝑀1

3. Accumulator: ℎ 𝑛 = σ𝑛𝑘=−∞ 𝛿 𝑘 = 𝑢[𝑛] .

4. Forward Difference: ℎ 𝑛 = 𝛿 𝑛 + 1 − 𝛿 𝑛 .

5. Backward Difference: ℎ 𝑛 = 𝛿 𝑛 − 𝛿 𝑛 − 1 .
46 Inverse System:

• For example, Accumulator and Backward Difference systems


are inverse of each other.

You might also like