Академический Документы
Профессиональный Документы
Культура Документы
N
x( n) W nk
X( k)
n 0
n1
N
X( k) W
1 nk
x( n)
N
k 0
Matrix Relations
• The DFT samples defined by
N 1
X [k ] kn
x[n]WN , 0 k N 1
n 0
W
k 0
nk
N
can be expanded as NXN DFT matrix
1 1 1 1
1 W 1 W 2
WN ( N 1)
N 1 N N
W
k 0
nk
N 1 WN2
WN4 WN2( N 1)
1 WN( N 1) WN2( N 1) WN( N 1) 2
n 1
N
x( n) W nk
X( k)
DFT: n 0
X(2) = 1 -1 1 -1
X(3) = 1 j -1 -j
Matrix Relations
• Likewise, the IDFT is
N 1
kn
x[n] X [k ]WN , 0 n N 1
k 0
xn
N 1
nk
n0 WN X(k)
Matrix Relations
N 1
N
W nk can also be expanded as NXN DFT matrix
k 0 1 1 1 1
1 W 1 W 2
WN( N 1)
N 1 N N
N
W nk
k 0
1 WN2
WN4 WN2( N 1)
W
N 1
nk
N
WN
*
N 1
N 1
nk
Observe:
N
W nk
k 0
1 W
N n0 N
The inversion can be had by Hermitian conjugating j by –j and dividing by N.
x(0) x(1) x(2) x(3)
X(0) = 1 1 1 1
X(1) = 1 -j -1 j
X(2) = 1 -1 1 -1
X(3) = 1 j -1 -j
X( 0) 1 1 1 1 x( 0)
X( 1) 1 j 1 j x( 1)
X( 2) 1 1 1 1 x( 2)
1 j 1 j x( 3)
X( 3)
x( 0) 1 1 1 1 X( 0)
x( 1) 1 1 j 1 j X( 1)
x( 2) 4 1 1 1 1 X( 2)
x( 3) 1 j 1 j X( 3)
FOR 8 point DFT: N=8
• The system Matrix is:
W80x0 W80x1 W80x2 W80x3 W80x4 W80x5 W80x6 W80x7
DFT Effective
IDFT
determinant
Calculation advantage in radix-2
algorithms for various values of N
No. of DIRECT COMPUTATION FFT computaton Improvement
computation complex Complex complex Complex
points multiplication additions Total multiplication additions Total A/B
N N^2 N^2-N A (N/2)ln(N) Nln(N) B
8 64 56 120 8.32 16.64 24.95 4.81
16 256 240 496 22.18 44.36 66.54 7.45
32 1024 992 2016 55.45 110.90 166.36 12.12
64 4096 4032 8128 133.08 266.17 399.25 20.36
128 16384 16256 32640 310.53 621.06 931.59 35.04
256 65536 65280 130816 709.78 1419.57 2129.35 61.43
512 262144 261632 523776 1597.01 3194.02 4791.03 109.32
1024 1048576 1047552 2096128 3548.91 7097.83 10646.74 196.88
2048 4194304 4192256 8386560 7807.61 15615.22 23422.83 358.05
4096 16777216 16773120 33550336 17034.79 34069.57 51104.36 656.51
8192 67108864 67100672 134209536 36908.70 73817.40 110726.10 1212.09
16384 268435456 268419072 536854528 79495.66 158991.33 238486.99 2251.09
The Process of Decimation
• First step of process of decimation is splitting a
sequence in smaller sequences.
• A sequence of 15 can be splitted in five
sequences of threes or three sequences of fives.
• A sequence of 16 numbers can be splitted in 2
sequences of 8. Further,
each sequence of 8 can be be splitted in two
sequences of 4;
Subsequently each sequence of 4 can be splitted in
two sequences of two;
There can be various combinations and varied
complexities.
4-point DFT
X( 0) 1 1 1 x( 0)
1
X( 1) 1 j 1 j x( 1)
X( 2) 1 1 1 1 x( 2)
X( 3) 1 j 1 j x( 3)
Can be, by interchanging col. 2 and 3, seen as
X( 0) 1 1 1 1 x( 0)
X( 1) 1 1 j j x( 2)
X( 2) 1 1 1 1 x( 1)
X( 3) 1 1 j j x( 3)
4-point DFT
X( 0) 1 1 1 1 x( 0)
X( 1) 1 1 j j x( 2)
X( 2) 1 1 1 1 x( 1)
1 1 j j x( 3)
X( 3)
The system matrix is rewritten as
X( 0) 1 0 1 1
0 1 0 0 x( 0)
X( 1) 0 1 0 j 1 1 0 0 x( 2)
X( 2) 1 0 1 0 0 0 1 1 x( 1)
X( 3) 0
1 0 j 0 0 1 1 x( 3)
Matrix manipulation to get the desired
input sequence from the actual
x( 0) 1 0 0 0
x( 0)
x( 2) 0 0 1 0 x( 1)
x( 1) 0 1 0 0 x( 2)
x( 3) 0
0 0 1 x( 3)
Process of decimation: example
X[n] X[3] X[4]
X[2]
X[6]
X[0]
-2 -1 1 5 7 n
2 3 4 6 X[7]
X[1]
X[5]
Separating the above sequence for +ve ‘n’ in even and odd sequence numbers .
X[2] X[6]
X[0]
-2 -1 1 5 7 n -2 -1 1 5 7 n
2 3 4 6 2 3 4 6 X[7]
X[1]
X[5]
Process of decimation: example
X[n] X[3]
X[n] X[4]
X[2] X[6]
X[0]
- 1 5 n - - 1 5 7 n
7
2 3 4 6 2 3 4 6 X[7]
X[1]
X[5]
Shift the sequence to left by one and
Compress the even sequence by two.
compress by two
X[4] X[n] X[3]
X[n]
X[2] X[6]
X[0]
4 6 n
- n
2 1 3 2 X[7]
X[1]
The compression is also called decimation X[5]
X( 0) 1 0 1 0 1 1 x( 0)
0 0
X( 1) 0 1 0 j 1 1 0 0 x( 2)
First stage of
X( 2) 1 0 1 0 0 0 1 1 x( 1) realization
X( 3) 0
1 0 j 0 0 1 1 x( 3)
x( 0) x( 2) 1 1 x( 0)
0 0
Second x( 0) x( 2) 1 1 0 0 x( 2)
x( 1) x( 3) 0 0 1 1 x( 1)
stage
x( 1) 0
x( 3) 0 1 1 x( 3)
X( 0) 1 0 1 x( 0)
0 x( 2)
X( 1) 0 1 0 j x( 0) x( 2)
X( 2) 1 0 1 0 x( 1) x( 3)
0 1
X( 3) 0 j x( 1) x( 3)
Decimation of 4 point DFT into 2xradix-2
x0 x0+x2 X[0]
Wo
x2 -1 xo -x2 X[1]
w1
x1+x3 X[2]
x1 W 2
x1-x3 W 3
X[3]
x3 -1
• The values of
W40= 1; W42 = -1; W41= -j; and W 43 = j
Decimation of 4 point DFT into 2xradix-2
x0 x0+x2 X[0]
N/4 point
DFT Wo
x2 xo -x2 X[1]
even
w1
x1+x3 X[2]
x1 W 2
N/4 point
DFT x1-x3 W 3
X[3]
x3
odd
• The values of
W40= 1; W42 = -1; W41= -j; and W 43 = j
N=8-point radix-4 DIT-FFT:
X0[0]
X[0]
X(0)
1
X0[1]
X(2) N/2 point X[1]
DFT X0[2] a
X(4) X[2]
[EVEN]
X0[3] -j
X(6) X[3]
-b
X1 [0]
-1 X[4]
X(1)
X1 [1]
-a X[5]
X(3) N/2 point
DFT X1[2]
X(5) [ODD] j X[6]
X1[3] b X[7]
X(7)
x[0] X[0]
x[4] X[1]
-1
x[2] X[2]
-1
x[6] w2
X[3]
-1 -1
x[1] X[4]
-1
X[5]
x[5] w1 -1
-1
x[3] w2 X[6]
-1 -1
x[7] w2 w3 X[7]
-1 -1 -1
System Matrix for N/2 point DFT
Block
0
X ( 0)
X ( 1) 1 0 1 0 0 0 0 0 x( 0) x( 4)
o
0 1 0 j 0 0 0 0
x( 0) x( 4)
Xo( 2)
1 0 1 0 0 0 0 0 x( 2) x( 6)
X0( 3) 0 1 0 j 0 0 0 0 x( 2) x( 6)
x( 5)
1
X ( 0)
0 0 0 0 1 0 1 0
x( 1)
X ( 1) 0 0 0 0 0 1 0 j x( 1) x( 5)
1 0 0 0 0 1 0 1 0 x( 3) x( 7)
X ( 2)
1 0 0 0 0 0 1 0 j x( 3) x( 7)
X1( 3)
Note that partitions 2 and 3 are null matrices while partitions 1 and 4 are identical.
Also note that subpartitions 1st and 3rd are I matrices and
2nd and 4th sub-partitions have sign difference.
Cross diagonals in either case are zero.
System Matrix for N/4 block
x( 0) x( 4) 1 1 0 0 0 0 0 0 x( 1)
x( 0) x( 4)
1 1 0 0 0 0 0 0
x( 4)
x( 2) x( 6) 0 0 1 1 0 0 0 0 x( 2)
x( 2) x( 6) 0 0 1 1 0 0 0 0 x( 6)
x( 5)
x( 1) 0 0 0 0 1 1 0 0
x( 1)
x( 1) x( 5) 0 0 0 0 1 1 0 0 x( 5)
x( 3) x( 7) 0 0 0 0 0 0 1 1 x( 3)
x( 3) x( 7) 0 0 0 0 0 0 1 1 x( 7)
Again see the simile between 1and 4 partitions. Note that 2 and 3 s are null matrices.
Further note the sub-partition matrices. They also follow the same rule.
All the elements are real.
Signal flow graph for decimation of 8 point DFT
N/4 point DFT N/2 point DFT N-point DFT
x[0] X[0]
x[4] X[1]
-1
x[2] X[2]
-1
x[6] w2
X[3]
-1 -1
x[1] X[4]
-1
X[5]
x[5] w1 -1
-1
x[3] w2 X[6]
-1 -1
x[7] w2 w3 X[7]
-1 -1 -1