You are on page 1of 51

Digital Signal Processing

Dr. Sanjay Sharma


Professor, ECED
Thapar Institute of Engineering and Technology
Patiala
1 ∞
𝒙(𝒕) = න 𝑿( Ω)𝒆𝒋Ω𝒕 𝒅Ω Inverse Fourier Transform
2𝝅 −∞

1 ∞
𝒙 𝒏𝑻 = න 𝑿( Ω)𝒆𝒋Ω𝒏𝑻 𝒅Ω
2𝝅 −∞
Or, simply

1 ∞
𝒙𝒏 = න 𝑿( Ω)𝒆𝒋Ω𝒏𝑻 𝒅Ω (1)
2𝝅 −∞
1 ∞
𝒙[𝒏𝑻] = න 𝑿( Ω)𝒆𝒋Ω𝒏𝑻 𝒅Ω (1)
2𝝅 −∞

𝟏 ∞𝟏 𝟐𝝅 𝒋𝝎𝒏
𝒙𝒏 = න ෍ 𝑿(Ω + 𝒌 )𝒆 𝒅𝝎൭ (𝟐)
𝟐𝝅 −∞ 𝑻 𝑻
𝑲=−∞
We know that Inverse Fourier Transform is given by
𝟏 ∞
𝒙𝒏 = න 𝑿( 𝝎)𝒆𝒋𝝎𝒏 𝒅𝝎 (𝟑)
𝟐𝝅 −∞
By comparing (2) and (3), we get

𝟏 𝟐𝝅
𝑿 𝝎 = ෍ 𝑿(Ω + 𝒌 ൱ (𝟒)
𝑻 𝑻
𝒌=−∞

Digital Analog Repeats itself


Spectrum Spectrum
𝝅 𝒇𝒔
Ω = 𝒄𝒐𝒓𝒓𝒆𝒔𝒑𝒐𝒏𝒅𝒔 𝒕𝒐 𝝎 = 𝝅 𝒐𝒓 (𝒇 = ቇ
𝑻 𝟐
If the sampling frequency 𝒇𝒔 is not sufficiently
high, the spectrum centered on 𝒇𝒔 will fold over
or alias into the base band frequencies (figure 8).
Equation (4) tells us that aliasing can only be
avoided if the analog signal is band-limited such
that
𝝅 𝝅 𝒇𝒔
𝑿 Ω = 𝟎, Ω > → 𝟐𝝅𝒇 ≥ → 𝒇 ≥
𝑻 𝑻 𝟐
This results in familiar sampling theorem. The
minimum sampling frequency for which equation
(4) holds is called Nyquist frequency.
function y = step_fun(n)
% We assume a scalar input
% Our default output value is 0
y = 0;
% We change our output to 1 if the argument is greater
% than or equal to 0
if n >= 0
y = 1;
end

Let's call the function from another script. For example:


% We iterate from -5 to 5 using only integers
for n = -5 : 5
y = step_fun(n);
stem(n, y)
hold on
end
% We adjust our axis values just to visualize better
axis([-5 5 -1 2])
and we get
Digital Frequency

You might also like