DSP Presentation - 5
DSP Presentation - 5
The DFT as a set of N samples {X(k)} of the Fourier transform X(ω) for a finite-duration
sequence {x(n)} of length L < N.
The sampling of X(ω) occurs at the N equally spaced frequencies ωk = 2πk/N, k = 0, 1,
2, ......., N-1.
The N samples {X(k)} uniquely represent the sequence {x(n)} in the frequency domain.
The DFT and inverse DFT (IDFT) for an N-point sequence {x(n)} are given as
N 1
DFT : X (k ) x(n) WNkn k 0, 1, , N 1 (1)
n0
N 1
1 (2)
IDFT: x (n)
N
X (k ) W
k 0
N
kn
n 0, 1, , N 1
where WN is defined as
WN e j 2 / N (3)
The notation used below to denote the N-point DFT pair x(n) and X(k) is
DFT
x(n)
N X (k )
These periodicities in x(n) and X(k) follow immediately from formulas Eq. (1)
and Eq. (2) for the DFT and IDFT, respectively.
In some applications it is advantageous to do this.
2. Linearity:
If
DFT DFT
x1 (n)
N X 1 (k ) and x2 (n)
N X 2 (k )
then for any real-valued or complex-valued constants a1 and a2,
DFT
a1 x1 (n) a2 x2 (n)
N a1 X 1 (k ) a2 X 2 (k ) (6)
This property follows immediately from the definition of the DFT given by Eq. (1).
X ( N k ) X ( k ) X ( k ) (13)
Consequently,
|X(N-k)| = |X(k)| and X(N - k) = - X(k).
Furthermore, xl(n) = 0 and therefore x(n) can be determined from Eq.(11), which is
another form for the IDFT.
x( n) x( N n) 0 n N 1
1 N 1
2 kn
x ( n)
N
X (k ) cos
k 0 N
0 n N 1 (15)
x( n) x( N n) 0 n N 1
1 N 1
2 kn
x (n) j
N
X (k ) sin
k 0 N
0 n N 1 (17)
(20)
Now suppose that we shift the periodic sequence xp(n) by k units to the right.
Thus, we obtain another periodic sequence
xp ( n) x p ( n k )
l
x( n k lN ) (23)
x ( N n ) x ( n) 1 n N 1 (26)
The time reversal of an N-point sequence is attained by reversing its samples about the
point zero on the circle.
Thus the sequence x((-n))N is simply given as
x((n)) N x( N n) 0 n N 1 (27)
where
x pe ( n) 12 x p ( n) xp ( N n)
(31)
x po ( n) 12 x p (n ) x p ( N n)
Let us consider that two finite-duration sequences of length N, x1(n) and x2(n).
Their respective N-point DFTs are
N 1
X 1 (k ) x1 (n)e j 2 k / N k 0, 1, , N 1 (32)
n 0
N 1
X 2 ( k ) x2 ( n)e j 2 k / N k 0, 1, , N 1 (33)
n 0
Then, multiply the two DFTs together and the result is a DFT, say of a sequence x3(n)
of length N.
Let us determine the relationship between x3(n) and the sequences x1(n) and x2(n).
We have
X 3 ( k ) X 1 ( k ) X 2 (k ) k 0, 1, , N 1 (34)
Using Eq.(38) and Eq.(39), we obtain the desired expression for x3(m) in the form
N 1
x3 (m) x1 (n) x2 ((m n)) N , m 0, 1, , N 1 (39)
n 0
(40)
where x1(n) x2(n) denotes the circular convolution of the sequence x1(n) and x2(n).
If x (n)
DFT
N
X (k )
Then
x(( n)) N x ( N n)
DFT
N
X (( k )) N X ( N k ) (41)
Hence, reversing the N-point sequence in time is equivalent to reversing the DFT
values.
Time reversal of a sequence x(n) is illustrated in Fig.3.
N 1
m0
x(m) e j 2 km / N
N 1
m 0
x ( m ) e j 2 m ( N k ) / N
X (N k)
We note that
If x(n)
DFT
N
X (k )
j 2 kl / N
Then x((n l )) N
DFT
N
X ( k ) e (42)
Proof: From the definition of the DFT we have
N 1 l 1 N 1
DFT{x ((n l )) N } x((n l ))
n0
N e j 2 kn / N
x((n l ))
n0
N e j 2 kn / N
x( n l ) e j kn / N
nl
x((n l ))
n0
N e j 2 kn / N
x ( N l n) e
n0
j 2 kn / N
m N l
x (m) e j k ( m l ) / N
Furthermore,
N 1 N 1l
x(n l ) e
nl
j kn / N
m0
x ( m ) e j 2 k ( m l ) / N
Therefore,
N 1
DFT{x ((n l ))} x ( m) e
m0
j 2 k ( m l ) / N
X ( k ) e j 2 k l / N
December 15, 2022 18
3. Circular frequency shift:
If
x ( n)
DFT
N
X (k )
Then
x ( n) e j 2 l n / N
DFT
N
X ((k l )) N (43)
Hence, the multiplication of the sequence x(n) with the complex exponential
sequence e j 2 kn / N is equivalent to the circular shift of the DFT by I units in frequency.
This is the dual to the circular time-shifting property and its proof is similar to the
latter.
If
x ( n)
DFT
N
X (k )
Then
x (n)
DFT
X
(( k )) X
(N k) (44)
N N
Proof:
The IDFT of X*(k) is
1 N 1 1 N 1
N
X
( k ) e j kn / N X (k ) e j 2 k ( N n ) / N
k 0 N k 0
Therefore,
N 1
1
N
x
k 0
(( n )) N x
( N n ) DFTN
N
X
(k ) (45)
Proof: We can write rxy (l ) as the circular convolution of x(n) with y*(–n), that is,
rxy (l ) x(l ) y * ( l )
Then, with the aid of the properties in Eq.(40) and Eq.(45), the N-point DFT of rxy (l ) is
R ( k ) X (k )Y * ( k )
xy
In the special case where y(n) = x(n), we have the corresponding expression for the
circular autocorrelation of x(n),
rxx (l )
DFT
(k ) X (k ) 2
R (47)
N Xx
DFT 1 (48)
Then x1 (n) x2 ( n)
N
X 1 (k ) X 2 (k )
N
This property is the dual of Eq.(40).
Its proof follows simply by interchanging the roles of time and frequency in the
expression for the circular convolution of two sequences.
7. Parseval’s theorem:
x ( n)
DFT
N
X (k ) and y ( n)
DFT
N
Y (k )
Then,
N 1 N 1
1
n0
x ( n ) y *
( n )
N
X (k )Y
k 0
*
(k ) (49)
x(n) y (n) r
n0
*
xy (0)
and
N 1 N 1
1 1
rxy (l )
N
R xy (k ) e
k 0
j 2 k l / N
N
k 0
X ( k ) Y ( k ) e j 2 k l / N
which expresses the energy in the finite-duration sequence x(n) in terms of the
frequency com ponents {X(k)}.
43 2 2 2 43 2
2 2 X (7) j
X (3) j 2 2
2 2