IT2101

Introduction to DSP
Digital Signal Processing represents signals by a sequence of numbers like sampling or analog-to-digital conversions. They
perform processing on these numbers with a digital processor.
• A signal is a function of independent variables, with time being perhaps the most prevalent single variable. The
signal itself carries some kind of information available for observation.
• By the term processing, we mean operating in some fashion on a signal to extract useful information.
• Also, the word digital means that the processing is done with a digital computer or particular purpose digital
hardware.

DSPs reconstruct the analog signal from processed numbers through reconstruction or digital-to-analog conversion.

Classification of Signals and Systems

The signals and systems are classified into:
• Continuous-Time and Discrete-Time Signals
o The continuous-time signal is the signals or quantities that can be defined and represented at specific
time instants of the sequence.
▪ It is denoted by x(t), where the time interval may be bounded (finite) or infinite.
▪ It can be a value of a signal that exists at every instance of time. e.g., voltage, velocity.
o A discrete-time signal is an indexed sequence of real, imaginary, or complex numbers, and it is a
function of an integer-valued n that is denoted by x(n).
▪ These are signals that have values at regular or discrete intervals of time.
▪ The amplitude can be any value.
• Analog and Digital Signals
o If a continuous-time signal s(t) can take on any values in a continuous-time interval, then s(t) is called
an analog signal.
o If a discrete-time signal can take on only a finite number of distinct values, {s(n)}, then the signal is
called a digital signal.
• Deterministic and Random Signals
o Deterministic signals are those signals whose values are entirely specified for any given time.
o Random signals are those signals that take random values at any given time.
• Periodic and Nonperiodic Signals
o A periodic signal completes a pattern within a measurable time frame, called a period, and repeats
that pattern over identical subsequent periods.
▪ A signal 𝑠(𝑡) is a periodic signal if 𝑠(𝑡) = 𝑠(𝑡 + 𝑛𝑇0 ), where 𝑇0 is called the period and the
integer 𝑛 > 0.
o A nonperiodic signal does not repeat its pattern over a period.
▪ If 𝑠(𝑡) ≠ 𝑠(𝑡 + 𝑇0 ) for all 𝑡 and any 𝑇0 , then 𝑠(𝑡) is a nonperiodic signal.
• Power and Energy Signals
o A complex signal 𝑠(𝑡) is a power signal if the average normalized power P is finite.
1 𝑇/2
𝑃 = lim ∫−𝑇/2 𝑠(𝑡)𝑠 ∗ (𝑡)𝑑𝑡 0 < 𝑃 < ∞ and 𝑠 ∗ (𝑡) is the complex conjugate of 𝑠(𝑡).
𝑇→∞𝑇

o A complicated signal 𝑠(𝑡) is an energy signal if the normalized energy E is finite.

∞ ∞
∗ (𝑡)𝑑𝑡
𝐸 = ∫ 𝑠(𝑡)𝑠 = ∫ |𝑠(𝑡)|2 𝑑𝑡 0<𝐸<∞
−∞ −∞
o In communication systems, the received waveform is usually categorized into the desired part that
contains the information signal and the undesired part, which is called noise.

01 Handout 1 *Property of STI

 [email protected] Page 1 of 7
 IT2101

Shifting and Scaling Function Transformation

• Amplitude Scaling - All corresponding amplitudes of the original signal must be multiplied by the scaling factor.

• Time Shifting – It means there is a movement done in the function concerning the x-axis.

• Time Scaling - The function of the signal g(t) is multiplied/scaled by the given time (signal stretching).

• Time Reversal - The negative variable inside the parenthesis denotes the time-reversal.

01 Handout 1 *Property of STI

 [email protected] Page 2 of 7
 IT2101

• Multiple Transformations - Amplitude scaling, time scaling, and time-shifting can be applied simultaneously.
𝑡 − 𝑡0
𝑔(𝑡) → 𝐴𝑔 ( )
𝑎
amplitude 𝑡
scaling 𝐴 𝑡→ 𝑡 𝑡→𝑡−𝑡0 𝑡 − 𝑡0 𝑡 − 𝑡0
𝑎
𝑔(𝑡) → 𝐴𝑔(𝑡) → 𝐴𝑔 ( ) → 𝐴𝑔 ( ) ≠ 𝐴𝑔 ( )
𝑎 𝑎 𝑎

Even and Odd Continuous Time Functions

Even Functions Odd Functions
𝒈(𝒕) = 𝒈(−𝒕) 𝒈(𝒕) = −𝒈(−𝒕)

These figures are sine

These are both even
waves or odd functions.
functions. Even functions
These signals are
are symmetric concerning
symmetric to the origin.
the y-axis. Cosine signals
are classified as even
functions.
The odd part of the
Even part of the function
function
𝒈(𝒕) + 𝒈(−𝒕)
𝒈𝒆 (𝒕) = 𝒈(𝒕) − 𝒈(−𝒕)
𝟐 𝒈𝒆 (𝒕) =
𝟐

A function whose even part is zero is odd, and a function whose odd part is zero is even.

01 Handout 1 *Property of STI

 [email protected] Page 3 of 7
 IT2101

Combination of Even and Odd Functions

• Product of Two Even Functions - shows the multiplication of two even functions, a cosine wave, and a triangular
wave, resulting in an even function.

• Product of Even and Odd Functions - shows the multiplication of an even function and an odd function which
results in an odd function.

• Product of Two Odd Signals – shows the multiplication of two odd functions, which results in an even function.

Signal Sampling
Any type of manipulation done in the signal is called signal processing. A typical reason for signal processing is to eliminate
or reduce an undesirable signal. These are the common types of processing in signal sampling: filtering, noise cancellation,
encryption, and echo suppression.

The analog filter processes analog input to obtain the band-limited signal. Afterward, the signal is sent to the analog-to-
digital conversion (ADC) unit.
• The ADC unit samples the analog signal, quantizes the sampled signal, and encodes the quantized signal level to
the digital signal.

01 Handout 1 *Property of STI

 [email protected] Page 4 of 7
 IT2101

• This is an analog (continuous-time) signal (solid line)

defined at every point over the time axis (horizontal
line) and amplitude axis (vertical line). Hence, the
analog signal contains an infinite number of points.
• Sampling takes samples at the fixed time interval where
T represents the sampling interval or sampling period in
seconds.
• Each sample maintains its voltage level during the
sampling interval T to give the ADC enough time to
convert it. This process is called sample and hold.
• For a given sampling interval T, which is defined as the
period between two neighboring sample points, the
sampling rate is therefore given by
1
𝑓𝑠 = 𝑠𝑎𝑚𝑝𝑙𝑒𝑠 𝑝𝑒𝑟 𝑠𝑒𝑐𝑜𝑛𝑑 (𝐻𝑧)
𝑇
The sampling theorem guarantees that an analog signal can be, in theory, perfectly recovered as long as the sampling
rate is at least twice the highest-frequency component of the analog signal to be sampled.
𝑓𝑠 ≥ 2𝑓𝑚𝑎𝑥 where 𝑓𝑚𝑎𝑥 is the maximum frequency component of the analog signal to be sampled.

Anti-aliasing Filtering
To satisfy the sampling theorem condition, we apply an The method can also be extended to other filter types,
anti-aliasing filter to limit the input analog signal. All the such as the Chebyshev filter. The Butterworth magnitude
frequency components are less than the folding frequency response with an order of n is given by
frequency (half of the sampling rate or Nyquist 1
|𝐻(𝑓)| =
frequency). √1 + (𝑓/𝑓𝑐 )2𝑛

It was deriving the percentage of the aliasing noise level 2𝑛

using the symmetry of the Butterworth magnitude √1 + (𝑓𝑎 )
𝑓𝑐
function and its first replica. 𝐴𝑙𝑖𝑎𝑠𝑖𝑛𝑔 𝑙𝑒𝑣𝑒𝑙 % = , for 0 ≤ 𝑓 ≤ 𝑓𝑐
𝑋𝑎 |𝐻(𝑓)|𝑓=𝑓𝑠−𝑓𝑎 2𝑛
𝐴𝑙𝑖𝑎𝑠𝑖𝑛𝑔 𝑙𝑒𝑣𝑒𝑙 % = = √1 + (𝑓𝑠 − 𝑓𝑎 )
𝑋(𝑓)|𝑓=𝑓𝑎 |𝐻(𝑓)|𝑓=𝑓𝑎 𝑓𝑐

Example: Given the DSP system where a sampling rate of 8000 Hz is used and anti-aliasing filter if a second-order
Butterworth lowpass filter with a cutoff frequency of 3.4 kHz. Determine the following:
• Percentage of aliasing level at the cutoff frequency.
• Percentage of aliasing level at the frequency of 1000 Hz.
Solution: 𝑓𝑠 = 8000, 𝑓𝑐 = 3400, and 𝑛 = 2.
If 𝑓𝑎 = 𝑓𝑐 = 3400 𝐻𝑧: If 𝑓𝑎 = 1000𝐻𝑧:
2𝑛 2𝑛
√1 + (𝑓𝑎 ) √1 + (𝑓𝑎 )
𝑓𝑐 𝑓𝑐
𝐴𝑙𝑖𝑎𝑠𝑖𝑛𝑔 𝑙𝑒𝑣𝑒𝑙 % = 𝐴𝑙𝑖𝑎𝑠𝑖𝑛𝑔 𝑙𝑒𝑣𝑒𝑙 % =
2𝑛 2𝑛
√1 + (𝑓𝑠 − 𝑓𝑎 ) √1 + (𝑓𝑠 − 𝑓𝑎 )
𝑓𝑐 𝑓𝑐

2(2) 2(2)
√1 + (3400) √1 + (1000)
3400 3400
= =
2(2) 2(2)
√1 + (8000 − 3400) √1 + (8000 − 1000)
3400 3400
1.4142 1.03007
= =
2.0858 4.3551
𝑨𝒍𝒊𝒂𝒔𝒊𝒏𝒈 𝒍𝒆𝒗𝒆𝒍 % = 𝟔𝟕. 𝟖% 𝑨𝒍𝒊𝒂𝒔𝒊𝒏𝒈 𝒍𝒆𝒗𝒆𝒍 % = 𝟐𝟑. 𝟎𝟓%

01 Handout 1 *Property of STI

 [email protected] Page 5 of 7
 IT2101

Example: Given the DSP system where a sampling rate of 40,000 Hz is used, the anti-aliasing filter is the Butterworth
lowpass filter with a cutoff frequency 8 kHz, and the percentage of aliasing level at the cutoff frequency is required to
be less than 1%, determine the order of the anti-aliasing lowpass filter.
Solution: Using fs=40, 000, fc=8000, and fa=8000 Hz.
1.4142
2(1) 𝑛 = 2: 𝐴𝐿% == = 8.82%
√1 + (8000) √1 + (4)4
8000
𝑛 = 1: 𝐴𝐿% = 1.4142
𝑛 = 3: 𝐴𝐿% == = 2.21%
40000 − 8000 2(1) √1 + (4)6
√1 + ( ) 1.4142
8000 𝑛 = 4: 𝐴𝐿% == = 0.55% < 1%
1.4142 √1 + (4)8
𝐴𝐿% = = 34.30% ∴ 𝒏 = 𝟒 has satisfied the 1% aliasing noise level.
√1 + (4)2

Anti-image Filter and Equalizer

Consider that a unit impulse function 𝛿(𝑡) is passed through the hold transfer function, 𝐻ℎ (𝑠).
The impulse response ℎ(𝑡) can be expressed as
ℎ(𝑡) = 𝑢(𝑡) − 𝑢(𝑡 − 𝑇)
where T is the sampling period, and 𝑢(𝑡) is the unit step function, that is,
1𝑡 ≥ 0
𝑢(𝑡) = {
0𝑡 < 0
The transfer function 𝐻ℎ (𝑠) can be obtained by Laplace transform of the impulse response ℎ(𝑡), that is,
1 1
𝐻ℎ (𝑠) = 𝐿{ℎ(𝑡)} = 𝐿{𝑢(𝑡) − 𝑢(𝑡 − 𝑇)} = − 𝑒 −𝑠𝑇 = 1 − 𝑒 −𝑠𝑇
𝑠 𝑠
The DAC unit converts the processed digital signal 𝑦(𝑛) to a sampled signal 𝑦𝑠 (𝑡), and then the hold circuit produces
the sample-and-hold voltage 𝑦𝐻 (𝑡). The transfer function of the hold circuit is derived as:
1 − 𝑒 −𝑠𝑇
𝐻ℎ (𝑠) =
𝑠

Signal notations at the practical

reconstruction stage:
(A) Processed digital signal,
(B) Recovered ideal sampled signal
(C) Recovered sample-and-hold voltage
(D) Recovered analog signal.

The magnitude and phase responses are given by In terms of Hz, we have
sin(𝜔𝑇/2) sin(𝑥) sin(𝜋𝑓𝑇)
|𝐻ℎ (𝜔)| = 𝑇 | | = 𝑇| | |𝐻ℎ (𝑓)| = 𝑇 | | ∠𝐻ℎ (𝑓) = −𝜋𝑓𝑇
𝜔𝑇/2 𝑥 𝜋𝑓𝑇
∠𝐻ℎ (𝜔) = −𝜔𝑇/2 where 𝑥 = 𝜔𝑇/2.

Distortion
sin(𝜋𝑓𝑇)
The magnitude frequency response of the sampled and hold signal becomes |𝑌𝐻 (𝑓)| = 𝑇 | | |𝑌𝑆 (𝑓)|
𝜋𝑓𝑇
• The magnitude frequency response acts like lowpass filtering and shapes the sampled signal spectrum of 𝑌𝑠 (𝑓).
This shaping effect distorts the sampled signal spectrum 𝑌𝑠 (𝑓) in the desired frequency band.
• On the other hand, the spectral images are reduced due to the lowpass effect of 𝑠𝑖𝑛(𝑥)/𝑥. This sample-and-
hold effect can help us design the anti-image filter.

01 Handout 1 *Property of STI

 [email protected] Page 6 of 7
 IT2101

• Since the magnitude frequency response of the sampled signal using an ideal sampler is 𝑇|𝑌𝑠 (𝑓)|, therefore,
the spectral distortion at the recovery stage can be derived as:
𝑇|𝑌𝑠 (𝑓)| − |𝑌𝐻 (𝑓)| |𝑌𝐻 (𝑓)| sin(𝜋𝑓𝑇)
𝐷𝑖𝑠𝑡𝑜𝑟𝑡𝑖𝑜𝑛% = =1− 𝐷𝑖𝑠𝑡𝑜𝑟𝑡𝑖𝑜𝑛% = 1 − | |
𝑇|𝑌𝑠 (𝑓)| 𝑇|𝑌𝑠 (𝑓)| 𝜋𝑓𝑇
sin(𝜋𝑓𝑇)
The percentage of distortion in the desired frequency band is given by 𝐷𝑖𝑠𝑡𝑜𝑟𝑡𝑖𝑜𝑛% = (1 − | |) × 100%
𝜋𝑓𝑇

Example: Given a DSP system with a sampling rate of 8000 Hz and a hold circuit used after DAC.
• Determine the percentage of distortion at the frequency of 3400 Hz.
• Determine the percentage of distortion at the frequency of 1000 Hz.
Solution:
Since 𝑓𝑇 = 3400 × 1/8000 = 0.425 Since 𝑓𝑇 = 1000 × 1/8000 = 0.125
sin(𝜋𝑓𝑇) sin(𝜋𝑓𝑇)
𝐷% = (1 − | |) × 100% 𝐷% = (1 − | |) × 100%
𝜋𝑓𝑇 𝜋𝑓𝑇
sin(0.425𝜋) sin(0.125𝜋)
𝐷% = (1 − | |) × 100% 𝐷% = (1 − | |) × 100%
0.425𝜋 0.125𝜋
𝐷% = 27.17% 𝐷% = 2.55%

References:
Abood, S. (2020). Digital signal processing: A primer with MATLAB®. CRC Press.
Deergha Rao, K. & Swamy, M. (2018). Digital signal processing: Theory and practice. Springer.
Gazi, O. (2018). Understanding digital signal processing. Springer
Kim, K. (2021). Conceptual digital signal processing with MATLAB. Springer.
Tan, T. & Jiang, J. (2019). Digital signal processing: Fundamentals and applications (3rd ed.). Elsevier.

01 Handout 1 *Property of STI

 [email protected] Page 7 of 7