Scribd
Upload
4 views12 pages

Multirate_part2

The document discusses multirate signal processing, specifically focusing on upsampling, which involves inserting zeros between input samples to increase the sampling rate. It explains the Z-transform and Fourier-transform relationships of upsampling, emphasizing the need for low-pass filters to avoid spectral replicas. Additionally, it addresses the process of changing the sampling rate by a rational number using a combination of upsampling and downsampling, highlighting the importance of filter selection to prevent information loss and distortion.

Uploaded by

rakeshdhakharwal0507
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
4 views12 pages

Multirate_part2

The document discusses multirate signal processing, specifically focusing on upsampling, which involves inserting zeros between input samples to increase the sampling rate. It explains the Z-transform and Fourier-transform relationships of upsampling, emphasizing the need for low-pass filters to avoid spectral replicas. Additionally, it addresses the process of changing the sampling rate by a rational number using a combination of upsampling and downsampling, highlighting the importance of filter selection to prevent information loss and distortion.

Uploaded by

rakeshdhakharwal0507
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 12

Department of Electronics and Communication

Engineering

Digital signal processing

Subject Code – ECN-312

Multirate Signal Processing Part-2


Upsampling: Increasing sampling rate
• Interpolation or Upsampling operation consists of introducing
M-1 zeros between successive value of the input.
 n
 x   , for n = 0, L, 2 L
y ( n) =   L  L
 0, otherwise
 For L=3, if we apply above
Time Domain Representation operation, but observe the
x(n) signal with original
sampling rate Fs, only Lth
sample has useful
information, others are
y(n)
zero.
For L=3, if we apply above
y(m) operation with
interpolation filter, and
observe output with higher
sampling rate (LFs).
Frequency Domain Representation
of Upsampling operation
Let us observe its behavior in Z-transform domain, for
simplicity, we take M=2: y (n) = x(n / 2) (1)
y (n) = {x(0), 0, x(1), 0, x(2),...,} (2)
Taking Z-transform of x(n) and y(n)
X ( z ) = x(0) + x(1) z −1 + x(2) z −2 + ... + x(n) z − n + ... (3)
Y ( z ) = y (0) + y (1) z −1 + y (2) z −2 + ... + y (n) z − n + ... (4)
By using relation between y(n) and x(n)
Y ( z ) = x(0) + x(1) z −2 + x(2) z −4 + ... + x(n) z −2 n + ... (5)

We can observe in Z-domain:


Y ( z) = X ( z 2 ) (6)
We can replace Z-transform representation by Fourier-
transform by using (z = e j )
Y (e j ) = X (e j 2 ) (11)

or Y f ( ) = DTFT { 2  x(n)} (12)


= X f (2 )

What will be the output of an upsampler, if the frequency


characteristics of the signal is given as follows?
Note: Fourier transform
of any signal is a 2π-
X f ( )

periodic function.
If we simply see eq.(12), we may conclude the Y f ( ) to be

X f (2 ) Wrong!!

But, Y f ( ) is the Fourier transform of a discrete data


sequence, which should be 2π-periodic:
Y f ( ) = X f (2 )

Correct!!
Up-sampling by factor L
Y f ( ) = DTFT { L  x(n)}
= X f ( L )
• If plotted while considering older sampling frequency
rate is considered, the signal spectrum will shrink by L
times.
• L-1 spectral images will also be created in the interval
(0, π).
• No information is lost as all original signal points are
still available.

To avoid these replicas, a low-pass filter with


bandwidth 𝜋/𝐿 is applied after up-sampling operation.
v(n)

Let us see the example of processing a signal via up-sampler


of factor I:
After upsampling by inserting zeros

Signal
spectrum
Filter Response Upsampled interpolated
Signal

Note that:
*Both down-sampling as well as up-sampling operation require filters
with low-pass bandwidth of at least π/(conversion factor).
*For up-sampling, this filter is called interpolator filter, which is
applied after upsampling operation.
*For down-sampling, this filter is called anti-aliasing filter, which
is applied to the input signal to contain its bandwidth.

What if conversion rate is a rational number L/M?


Sampling Rate change by a rational number
(L/M)
It can be done by using combination of down-sampling and up-
sampling operations. There can be two options as depicted below:
(1) Downsampling followed by upsampling: Loss of information
and distortion due to interpolation between decimated samples.

(2) Upsampling followed by downsampling: More accurate


extraction of original information

The two filters in cascade can be further designed as a


single filter with composite performance
Sampling Rate change by a rational number
(L/M)

• Composite filter is designed to both eliminate spectral


images and to avoid aliasing.
• The cascade of two ideal low-pass filter is a low-pass
filter with a cut-off frequency.

The selection of appropriate filter is an important step. Let us


see an example:
Effect of filter order
Signal to be up-sampled

Up-sampling by zero insertion By insertion of


zeroes, replicas
will be generated
in frequency
domain, which
need to be filtered
Smoother transition

21-tap FIR filter

7-tap FIR filter

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy