lecture_12

# lecture_12 - MIT OpenCourseWare http/ocw.mit.edu 2.161...

This preview shows pages 1–4. Sign up to view the full content.

MIT OpenCourseWare http://ocw.mit.edu 2.161 Signal Processing: Continuous and Discrete Fall 2008 For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms .

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
1 Massachusetts Institute of Technology Department of Mechanical Engineering 2.161 Signal Processing - Continuous and Discrete Fall Term 2008 Lecture 12 1 Reading: Class Handout: The Fast Fourier Transform Proakis and Manolakis (4th Ed.): Secs. 8.1 8.3 Oppemheim Schafer & Buck (2nd Ed.): Secs. 9.0 9.3 The Fast Fourier Transform (contd.) In Lecture 11 we saw that we could write the DFT of a length N sequence as N 1 N 1 j 2 πmn F m = f n e N = f n W N mn , m =0 ,...,N 1 n =0 n =0 where W N =e j2 π/N . We noted that the number of complex multiplication operations to compute the DFT is N 2 , but if we divided the original sequence into two length N/ 2 sequences (based on even and odd samples) and computed the DFT of each shorter sequence, they could be combined F m = A m + W N m B m for m ... ( 2 1) , and F m = A m N/ 2 W N m N/ 2 B m N/ 2 for m = 2 ( N 1) where { A m } is the DFT of the even-numbered samples, and { B m } is the DFT of the odd- numbered samples. 0 1 2 3 4 5 6 7 A A A A B B B B 0 0 1 1 2 2 3 3 4 - p o i n t D F T f r o m e v e n s a m p l e s 4 - p o i n t D F T f r o m o d d s a m p l e s { f , f , f , f } 0 2 4 6 { f , f , f , f } 1 3 5 7 F 0 = A 0 + W 8 0 B 0 F 1 = A 1 + W 8 1 B 1 F 2 = A 2 + W 8 2 B 2 F 3 = A 3 + W 8 3 B 3 F 4 = A 4 + W 8 4 B 4 = A 0 - W 8 0 B 0 F 5 = A 5 + W 8 5 B 5 = A 1 8 1 B 1 F 6 = A 6 + W 8 6 B 6 = A 2 8 2 B 2 F 7 = A 7 + W 8 7 B 7 = A 3 8 3 B 3 1 copyright ± c D.Rowell 2008 0–1
F F F F 3 F 4 F F 6 F 0 1 2 5 7 The total number of required complex multiplications is ( N/ 2) 2 for each shorter DFT, and 2 to combine the two, giving a total of N ( N +1) / 2, which is less than N 2 . If N is divisible by 4, the process may be repeated, and each length 2 DFT may be formed by decimating the two 2 sequences into even and odd components, forming the length 4 DFTs, and combining these back into a length 2 DFT, as is shown for N =8 below: 4 2 - p o i n t D F T s 2 4 - p o i n t D F T s 8 - p o i n t D F T 0 1 2 3 4 5 6 7 2 - p o i n t D F T f r o m s a m p l e s { f , f } 0 4 0 0 2 2 4 4 6 6 2 - p o i n t D F T f r o m s a m p l e s { f , f } 2 6 2 - p o i n t D F T f r o m s a m p l e s { f , f } 1 5 2 - p o i n t D F T f r o m s a m p l e s { f , f } 3 7 Notice that all weights in the ﬁgure are expressed by convention as exponents of W 8 . In general, if the length of the data sequence is an integer power of 2, that is N =2 q for integer q , the DFT sequence { F m } may be formed by adding additional columns to the left and halving the length of the DFT at each step, until the length is two. For example if N =256=2 8 a total of seven column operations would be required. The ﬁnal step is to evaluate the N/2 length-2 DFTs. Each one may be written F 0 = f 0 + W 2 0 f 1 = f 0 + f 1 F 1 = f 0 + W 2 1 f 1 = f 0 f 1 , which is simply the sum and diﬀerence of the two sample points. No complex multiplications are necessary. The 2-point DFT is shown in signal-ﬂow graph form below, and is known as the FFT butterﬂy .

This preview has intentionally blurred sections. Sign up to view the full version.

View Full Document
This is the end of the preview. Sign up to access the rest of the document.

## This note was uploaded on 02/27/2012 for the course MECHANICAL 2.161 taught by Professor Derekrowell during the Fall '08 term at MIT.

### Page1 / 11

lecture_12 - MIT OpenCourseWare http/ocw.mit.edu 2.161...

This preview shows document pages 1 - 4. Sign up to view the full document.

View Full Document
Ask a homework question - tutors are online