ECT303 DIGITAL SIGNAL

PROCESSING
MODULE 2- PART III

Ms. Neethu Radha Gopan, Asst. Professor, Dept. of ECE, RSET

➢ FFT can be used for the computation of DFT of 2 real valued sequences using a
single DFT.
➢ Let 𝑥1 𝑛 & 𝑥2 𝑛 be 2 real valued sequences of length N, and let x(n) be a complex
valued sequence defined as
𝑥 𝑛 = 𝑥1 𝑛 + 𝑗𝑥2 𝑛
➢ The DFT operation is linear and hence the DFT of x(n) can be expressed as
X 𝑘 = 𝑋1 𝑘 + 𝑗𝑋2 𝑘
➢ The sequences 𝑥1 𝑛 & 𝑥2 𝑛 can be expressed in terms of x(n) as follows
𝑥 𝑛 + 𝑥 ∗ (𝑛)
𝑥1 𝑛 =
2
∗
Using complex conjugate property,
𝑥 𝑛 − 𝑥 (𝑛)
𝑥2 𝑛 =
2𝑗 DFT of 𝑥 ∗ 𝑛 is 𝑋 ∗ 𝑁 − 𝑘 . Therefore
➢ Hence the DFT’s of 𝑥1 𝑛 & 𝑥2 𝑛 are as follows
1
𝑋1 𝑘 = {𝑋(𝑘) + 𝑋 ∗ 𝑁 − 𝑘 }
1 2
𝑋1 𝑘 = {𝐷𝐹𝑇 𝑥 𝑛 + 𝐷𝐹𝑇 𝑥 ∗ 𝑛 }
2 1
𝑋 𝑘 = {𝑋 𝑘 − 𝑋 ∗ 𝑁−𝑘 }
1 2
∗ 2𝑗
𝑋2 𝑘 = {𝐷𝐹𝑇 𝑥 𝑛 − 𝐷𝐹𝑇 𝑥 𝑛 }
2𝑗
➢ Thus by performing a single DFT on the complex valued sequence x(n), we can
obtain the DFT of two real sequences with only a small amount of additional
computation that is involved in computing 𝑋1 𝑘 and 𝑋2 𝑘 from 𝑋 𝑘 .
Q. Compute the 4 point DFT of the 2 sequences 𝒙𝟏 𝒏 ={1,2,0,1} and
𝒙𝟐 𝒏 = 𝟐, 𝟐, 𝟏, 𝟏 using single DFT.

Soln:
𝑥1 𝑛 = 1, 2,0,1 & 𝑥2 𝑛 = {2,2,1,1}
Let 𝑥 𝑛 = 𝑥1 𝑛 + 𝑗𝑥2 𝑛 = 1 + 2j, 2 + 2j, j, 1 + 1j
Now find 4 point DFT, X(k) using DIT FFT or DIF FFT
j2𝜋
−
𝑊𝑁 = e N
j2𝜋
𝑊4 = e 4 .0 = 1
0 −
j2𝜋.1
1 − 4
𝑊4 = e = −j
 To find X 𝑘 , we use DIT FFT
6

x(0) = 1+2j 1+3j 4+6j

1+j 2
x(2) = j
-1
𝑊40 =1
3+3j -2
x(1) = 2+2j
𝑊40 = 1 -1

x(3) = 1+j
-1 1+j 𝑊41 = −𝑗 -1 2j
𝑊40 =1
𝑋 𝑘 = {4 + 6𝑗, 2, −2, 2𝑗}

1
𝑋1 𝑘 = [𝑋(𝑘) + 𝑋 ∗ 𝑁 − 𝑘 }
2
1
𝑋2 𝑘 = [𝑋 𝑘 − 𝑋 ∗ 𝑁 − 𝑘 }
2𝑗
Hence we find 𝑋 ∗ 𝑁 − 𝑘 . k varies from 0 to 3
k=0, 𝑋 ∗ 𝑁 − 𝑘 =𝑋 ∗ 4 = 𝑋 ∗ 0 = 4 − 6𝑗
k=1, 𝑋 ∗ 𝑁 − 𝑘 =𝑋 ∗ 3 = −2𝑗

k=2, 𝑋 ∗ 𝑁 − 𝑘 =𝑋 ∗ 2 = −2
𝑋 ∗ 𝑁 − 𝑘 = 𝑋 ∗ 4 − 𝑘 = 4 − 6𝑗, −2𝑗, −2,2
k=3, 𝑋 ∗ 𝑁 − 𝑘 =𝑋 ∗ 1 = 2
𝑋 𝑘 + 𝑋∗ 𝑁 − 𝑘 ={8, 2-2j, -4, 2+2j}

𝑋 𝑘 − 𝑋∗ 𝑁 − 𝑘 ={12j, 2+2j, 0, -2+2j}

1
𝑋1 𝑘 = 𝑋 𝑘 + 𝑋 ∗ 𝑁 − 𝑘 = 4, 1 − j, −2, 1 + j
2

1
𝑋2 𝑘 = 𝑋 𝑘 − 𝑋∗ 𝑁 − 𝑘 = {6 , 1 − j, 0, 1 + j}
2𝑗
2. Efficient Computation of DFT of 2N-Point Real Sequence

➢ Assume g(n) to be a real valued sequence of 2N points. The 2N point DFT of g(n)
can be obtained from the computation of N-point DFT involving complex valued
data.
➢ The 2N –point real sequence is now divided into two N-point real sequences.
𝑥1 𝑛 = 𝑔 2𝑛 & 𝑥2 𝑛 = 𝑔 2𝑛 + 1 , 0≤𝑛 ≤𝑁−1
➢ Let x(n) be the N point complex valued sequence
𝑥 𝑛 = 𝑥1 𝑛 + 𝑗𝑥2 𝑛
➢ The sequences 𝑥1 𝑛 & 𝑥2 𝑛 can be expressed in terms of x(n) as follows
𝑥 𝑛 + 𝑥 ∗ (𝑛) 𝑥 𝑛 − 𝑥 ∗ (𝑛)
𝑥1 𝑛 = 𝑥2 𝑛 =
2 2𝑗
➢ Hence the DFT’s of 𝑥1 𝑛 & 𝑥2 𝑛 are as follows Now the 2N point DFT of g(n) is
1 expressed in terms of the two N point
𝑋1 𝑘 = {𝐷𝐹𝑇 𝑥 𝑛 + 𝐷𝐹𝑇 𝑥 ∗ 𝑛 }
2 DFT’s, 𝑋1 𝑘 & 𝑋2 𝑘 using DIT FFT
1
𝑋2 𝑘 = {𝐷𝐹𝑇 𝑥 𝑛 − 𝐷𝐹𝑇 𝑥 ∗ 𝑛 }
2𝑗 𝐺 𝑘 = 𝑋1 𝑘 + 𝑊2𝑁 𝑘 𝑋2 𝑘
Using complex conjugate property, gives the first N DFT samples
DFT of 𝑥 ∗ 𝑛 is 𝑋 ∗ 𝑁 − 𝑘 . Therefore
1 𝐺 𝑘 + 𝑁 = 𝑋1 𝑘 − 𝑊2𝑁 𝑘 𝑋2 𝑘
𝑋1 𝑘 = [𝑋(𝑘) + 𝑋 ∗ 𝑁 − 𝑘 } gives the next N samples
2
1
𝑋2 𝑘 = [𝑋 𝑘 − 𝑋 ∗ 𝑁 − 𝑘 }
2𝑗
Q. Find the 8 point DFT of the sequence {1,2,2,2,0,1,1,1} using 4 point DFT

Soln:
𝑔 𝑛 = 1,2,2,2,0,1,1,1
𝑥1 𝑛 = 𝑔 2𝑛 = 1, 2,0,1 & 𝑥2 𝑛 = 𝑔 2𝑛 + 1 = {2,2,1,1}
𝑥 𝑛 = 𝑥1 𝑛 + 𝑗𝑥2 𝑛 = 1 + 2𝑗, 2 + 2𝑗, 𝑗, 1 + 𝑗

Find the N=4 point DFT’s of 𝑥(𝑛) using DIT FFT or DIF FFT
j2𝜋
− N
𝑊𝑁 = e
j2𝜋
0 − 4 .0
𝑊4 = e =1
j2𝜋.1
1 − 4
𝑊4 = e = −j
 To find X 𝑘 , we use DIF FFT
12

1+2j 1+3j 1+3j 4+6j

3+3j 3+3j
2+2j -2
-1 𝑊40 =1
1+j 1+j
j 2
-1 𝑊40 = 1
1+j 1-j
1+j 2j
-1 𝑊41 = −𝑗 -1 𝑊40 =1

𝑋 𝑘 = {4 + 6𝑗, 2, −2, 2𝑗}

k=2, 𝑋 ∗ 𝑁 − 𝑘 =𝑋 ∗ 2 = −2
𝑋 ∗ 𝑁 − 𝑘 = 𝑋 ∗ 4 − 𝑘 = 4 − 6𝑗, −2𝑗, −2,2
k=3, 𝑋 ∗ 𝑁 − 𝑘 =𝑋 ∗ 1 = 2
 1 1
𝑋1 𝑘 = 𝑋 𝑘 + 𝑋 ∗ 𝑁 − 𝑘 = 2 8, 2 − 2j, −4,2 + 2j = 4, 1 − j, −2, 1 + j
2

1 1
𝑋2 𝑘 = 𝑋 𝑘 − 𝑋∗ 𝑁 − 𝑘 = {12j , 2 + 2j, 0 , −2 + 2j} = {6 , 1 − j, 0, 1 + j}
2𝑗 2𝑗
The 8 point DFT is given by
𝐺 𝑘 = 𝑋1 𝑘 + 𝑊2𝑁 𝑘 𝑋2 𝑘 &
𝐺 𝑘 + 𝑁 = 𝑋1 𝑘 − 𝑊2𝑁 𝑘 𝑋2 𝑘 , where k varies from 0 to 3 and 2N=8.
➢ Substituting 2N and N we get
𝐺 𝑘 = 𝑋1 𝑘 + 𝑊8 𝑘 𝑋2 𝑘 & 𝐺 𝑘 + 4 = 𝑋1 𝑘 − 𝑊8 𝑘 𝑋2 𝑘
➢ When k=0, we get
𝐺 0 = 𝑋1 0 + 𝑊8 0 𝑋2 0 &
𝐺 4 = 𝑋1 0 − 𝑊8 0 𝑋2 0
➢ When k=1, we get
𝐺 1 = 𝑋1 1 + 𝑊81 𝑋2 1 &
𝐺 5 = 𝑋1 1 − 𝑊81 𝑋2 1
➢ Similarly when k=2, we get 𝐺 2 & 𝐺 6 ,
➢ When k=3, we get 𝐺 3 & 𝐺 7
𝐺 0 = 𝑋1 0 + 𝑊8 0 𝑋2 0 = 4 + 1. 6 = 10
𝐺 1 = 𝑋1 1 + 𝑊81 𝑋2 1 = 1 − 𝑗 + 0.707 − 0.707𝑗 1 − 𝑗 = 1 − 2.414𝑗
𝐺 2 = 𝑋1 2 + 𝑊8 2 𝑋2 2 = −2 + −j 0 = −2
𝐺 3 = 𝑋1 3 + 𝑊8 3 𝑋2 3 = 1 + j + −0.707 − 0.707j 1 + j = 1 − 0.414j
𝐺 4 = 𝑋1 0 − 𝑊8 0 𝑋2 0 = 4 − 1. 6 = −2
𝐺 5 = 𝑋1 1 − 𝑊81 𝑋2 1 = 1 − 𝑗 − 0.707 − 0.707𝑗 1 − 𝑗 = 1 + 0.414𝑗
𝐺 6 = 𝑋1 2 − 𝑊8 2 𝑋2 2 = −2 − −j 0 = −2
𝐺 7 = 𝑋1 3 − 𝑊8 3 𝑋2 3 = 1 + j − −0.707 − 0.707j 1 + j = 1 + 2.414j
G(k) ={ 10, 1-2.414j, -2, 1-0.414j, -2, 1+0.414j, -2, 1+2.414j }
 REFERENCES
17

1. Proakis J. G. and Manolakis D. G., Digital Signal Processing, 3/e,

THANK YOU!