Introduction to Digital Signal Processing

(OE601A)
 What is a Signal?

• Mathematically, a signal is a function of one or more

independent variables.
f (x1, x2, ...)
• At present, we are interested in 1D signal which is a
function of integer numbers (+ve/-ve)
• Signals can be Analog or Digital
 Are “Analog” and “Continuous” synonymous?
• No, not always.
• If time is discretized, but not the amplitude - the signal is still
analog.
Then what is digital signal?
Continuous
Time Quantized Coded
Analog
Discrete A/D
Time Converter
Digital

In case of digital signal, both time and amplitude are discretized.

 Why do you need Processing?

• We need to process the original signal to convert it to more

desirable form.

• Eg. Noise Filtering, Increasing/decreasing brightness, contrast

or any other properties of a picture, sending multiple signals at
a time (MUX and DEMUX)

• Processing means mainly, Addition, Multiplication and

Recalling through delay.
 Types of Signal Processing

Mixed Recent Times

Signal
Processing

Before ‘70s Analog Digital After ‘70s

 Advantages of DSP

• Less erroneous
• Can be Integrated easily
• Accuracy
• Multi-functional
• Flexible
• No Loading Problem
• Easy storage
• Very Low Frequency Processing possible
• Offline Processing
 Disadvantages of DSP

• Complexity
• Sampling and quantizing results in distortion
• Limited frequency range
• Limited speed
• Power dissipation
 Basic Elements of DSP System

Analog
Signal

Sample
A/D D/A Low-Pass
Low Pass
Pre-filter and Hold
Converter
DSP D/A Converter
Circuit Converter Filter

Analog
Signal
Digital Signal
MAKAUT Questions
 Significance of Sampling Theorem
• Analog signals can be digitized through sampling and quantization.
• Analog-to-digital (A/D) conversion is the foundation of modern digital communication
systems.
• In the A/D converter, the sampling rate must be large enough to permit the analog signal
to be reconstructed from the samples with sufficient accuracy.
• The sampling theorem is the basis for determining the proper (lossless) sampling rate
for a given signal.
• The sampling process necessarily discards much of the original signal.
• If the sampling rate is sufficiently high, however, the original signal is completely
recoverable, either exactly or within some error tolerance, from its samples.
• The necessary quantitative framework for ideal sampling is provided by the sampling
theorem.
 Sampling Theorem
In uniform sampling, sample values are equally spaced from one another by a fixed sampling interval T.

The reciprocal of the sampling interval is called the sampling frequency (or sampling rate) Fs = 1/T, which
has units of hertz.
Sampling Examples
Sampling Examples
Block Representation of Sampling
 Aliasing
• The sampling theorem assumes that the signal x(t) is bandlimited.
• However, all practical signals are timelimited, meaning that they are of finite
duration or width.
• Now, a signal cannot be timelimited and bandlimited simultaneously. If a signal
is timelimited, it cannot be bandlimited, and vice versa.
• Clearly, all practical signals, being timelimited, possess infinite bandwidth,
as shown in Fig. a.
• The sampled signal’s spectrum thus consists of overlapping cycles of X(ω)
repeating every Fs Hz (the sampling frequency), as illustrated in Fig. 3.13b.
• Sampling at a higher rate reduces but does not eliminate the overlap between
repeating spectral cycles.
• Because of the overlapping tails, Xδ˜(ω) no longer has complete information
about X(ω), and it is no longer possible, even theoretically, to recover x(t) exactly
from the sampled signal.
• If the sampled signal is passed through an ideal lowpass reconstruction filter with
cutoff frequency Fs/2 Hz, the output Xˆ(ω) does not equal X(ω), as shown in Fig. c.
 Aliasing

1. The tail of X(ω) beyond |f| > Fs/2 Hz, shown shaded light gray in Fig. before, is lost. This results in the
loss of high-frequency (|f| > Fs/2) signal content.

2. The lost tail reappears with lower frequency, shown shaded dark gray in Fig. before. For the spectra of
real signals, the tail appears to fold or invert about Fs/2. Thus, a component of frequency Fs/2 + f1 shows
up as a component of lower frequency Fs/2 − f1 in the reconstructed signal. This causes distortion of low-
frequency (|f| < Fs/2) signal content.

• The frequency Fs/2 is called the folding frequency, and the tail inversion is known as spectral folding or
aliasing. When aliasing occurs, we not only lose components above the folding frequency, but these very
components also reappear (alias) at lower frequencies.
Anti-aliasing Filter
Anti-aliasing Filter
 What happens if sampling is done below Nyquist Rate?

• Now let us consider three different signals:

f1=3 Hz f2 = 7 Hz f3 = 13 Hz
• Let us take a sampling interval of T=0.1 sec (fs=10Hz)
• Thus, after sampling we get,

• So, after sampling the signals become indistinguishable as high frequency signal appears as low
frequency. This causes aliazing effect.
• In order to get rid off this effect, we need to make ωo ≤π
• So, the minimum frequency required for proper sampling is fs = 26 Hz.
• This will make the highest frequency signal x3 to be equal to cos πn
MAKAUT Questions
MAKAUT Questions
Properties of Nyquist Rate
1000
Hz
Numerical