0% found this document useful (0 votes)
11 views32 pages

Module 2-1

This document discusses using the discrete Fourier transform (DFT) for linear filtering of signals. It explains that the DFT provides a frequency domain representation where linear filtering is equivalent to multiplication. Filtering long signals is done by segmenting into blocks and processing each block with DFT/IDFT. Two common methods are overlap-save and overlap-add. The document also covers efficient computation of the DFT using the fast Fourier transform (FFT) algorithm and properties of the twiddle factor used in the DFT calculation.

Uploaded by

Abhishek G M
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
11 views32 pages

Module 2-1

This document discusses using the discrete Fourier transform (DFT) for linear filtering of signals. It explains that the DFT provides a frequency domain representation where linear filtering is equivalent to multiplication. Filtering long signals is done by segmenting into blocks and processing each block with DFT/IDFT. Two common methods are overlap-save and overlap-add. The document also covers efficient computation of the DFT using the fast Fourier transform (FFT) algorithm and properties of the twiddle factor used in the DFT calculation.

Uploaded by

Abhishek G M
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 32

Module 2

Part 1-Linear Filtering Methods Based on


DFT
Use of DFT as a tool for linear system analysis-linear
filtering
Output spectrum
Y(ω)=X(ω)H(ω)
y(n) –Inverse FT
ω-Continuous variable
DFT- Linear Filtering in frequency domain
Alternative to time domain convolution

2 02/19/2024
Use of DFT in linear Filtering
Product of two DFTs is equivalent to circular
convolution of corresponding time domain sequences
Frequency domain methodology equivalent to linear
convolution
Finite duration sequence x(n)-Length L
FIR filter of Length M

3 02/19/2024
4 02/19/2024
DFT of size N≥L+M-1 is required to represent
{y(n)} in frequency domain

5 02/19/2024
x(n), h(n) –duration less than N- pad with zeros to
increase length to N
Does not alter X(ω) & H(ω)
Increased the number of samples that represent these
sequences in frequency domain beyond minimum
number (L or M respectively)

6 02/19/2024
N=L+M-1 point DFT of the output sequence y(n) is
sufficient to represent y(n) in the frequency domain
Implies multiplication of N point DFTs X(k) and H(k)
followed by computation of N point IDFT yields {y(n)}
Increasing the length of sequences x(n) and h(n) to N points
(by appending zeros) and then circularly convolving the
resulting sequences gives same result as linear convolution

7 02/19/2024
Filtering of long data sequences
Practical application-x(n) is very long sequence
Segmentation to fixed size blocks prior to processing
Linear FIR filtering –block by block basis-DFT
Segmentation into blocks
Each block processed via DFT & IDFT
Output blocks are fitted together –overall output
sequence

8 02/19/2024
Two Methods
Overlap Save Method
Overlap Add Method

9 02/19/2024
Overlap Save Method
Size of input data block N=L+M-1
DFTs and IDFT of length N
Each block consists of last M-1 data points of previous
data block followed by L new data points to form data
sequence of length N=L+M-1
Compute N Point DFT
Impulse response of FIR filter is increased in length by
appending L-1 Zeros

10 02/19/2024
11 02/19/2024
12 02/19/2024
13 02/19/2024
Overlap Add Method

14 02/19/2024
15 02/19/2024
16 02/19/2024
Efficient Computation of DFT
Divide & Conquer Approach
DFT of size N
Computing smaller DFTs from which
larger DFT is computed
Fast Fourier Transform (FFT) algorithm
N –power of 2 or power of 4

17 02/19/2024
DFT & IDFT Equation

18 02/19/2024
Direct Computation

Each value of k
 N complex multiplications
 N-1 complex additions

N values of DFT
 N2 Complex Multiplications
 N2-N Complex Additions

Direct Computation- Inefficient

19 02/19/2024
Example
1024 point DFT
N=1024
Complex Multiplications ≈1*106
Complex Additions ≈1*106
Assume the computer capacity
1 complex Multiplication/µs
1 complex Addition/µs
Total Time=2s-large time

20 02/19/2024
Determining number of real multiplications, real
additions & trigonometric functions

21 02/19/2024
22 02/19/2024
23 02/19/2024
24 02/19/2024
25 02/19/2024
26 02/19/2024
27 02/19/2024
28 02/19/2024
Properties of WN

29 02/19/2024
30 02/19/2024
31 02/19/2024
Thank You

32 02/19/2024

You might also like