Exercise 5.5: Fast Fourier Transform

From LNTwww

FFT algorithm for  $N=8$

The graph shows the signal flow diagram of the Fast Fourier Transform  $\rm (FFT)$  for  $N = 8$.

The associated spectral coefficients  $D(0), \hspace{0.03cm}\text{...} \hspace{0.1cm} , D(7)$  are determined from the time coefficients  $d(0), \hspace{0.03cm}\text{...} \hspace{0.1cm}, d(7)$.  The following applies to these with  $0 ≤ μ ≤ 7$:

$$D(\mu) = \frac{1}{N}\cdot \sum_{\nu = 0 }^{N-1} d(\nu) \cdot {w}^{\hspace{0.03cm}\nu \hspace{0.05cm} \cdot \hspace{0.05cm}\mu}\hspace{0.05cm},$$

where the complex rotation factor  $w = \text{e}^{-\text{j}\hspace{0.05cm} \cdot \hspace{0.05cm}2\pi /N}$  is to be used, i.e.  $w = \text{e}^{-\text{j}\hspace{0.05cm} \cdot \hspace{0.05cm}\pi /4}$  für  $N = 8$.

  • The alternating $±1$ sequence  $\langle\hspace{0.05cm} d(ν)\hspace{0.05cm}\rangle$  is applied to the input.
  • After the bit reversal operation, this results in the sequence  $\langle \hspace{0.05cm}b(\kappa)\hspace{0.05cm}\rangle$.


It holds that  $b(κ) = d(ν)$, if  $ν$  is represented as a dual number and the resulting three bits are written as  $κ$  in reverse order.  For example

  •   $ν = 1$  $($binary  $001)$  is followed by  $κ = 4$  $($binary  $100)$,
  •   $d(2)$  remains at the same position  $2$  $($binary  $010)$.


The actual FFT algorithm happens for the example  $N = 8$  in  $\log_2 N = 3$  stages, denoted  $L = 1$,  $L =2$  and  $L = 3$.  Further:

  • In each stage, four basic operations - so-called  butterflies  - are to be performed.
  • The values at the output of the first stage are designated in this task as  $X(0),\hspace{0.03cm}\text{...} \hspace{0.1cm} , X(7)$,
    those of the second as  $Y(0), \hspace{0.03cm}\text{...} \hspace{0.1cm} , Y(7)$.
  • After the third and last stage, all values must be divided by  $N$. 
    The final result  $D(0), \hspace{0.03cm}\text{...} \hspace{0.1cm} , D(7)$  is available here.




Hint:



Questions

1

Calculate the DFT coefficient  $D(\mu=3)$.

$D(\mu=3) \ = \ $

2

Calculate the DFT coefficient  $D(\mu=4)$.

$D(\mu=4) \ = \ $

3

Determine the initial values  $X(0)$, ... , $X(7)$  of the first stage.  Which of the following statements are true?

All  $X$ values with even indices are equal to  $2$.
All  $X$ values with odd indices are equal to  $0$.

4

Determine the initial values  $Y(0)$, ... , $Y(7)$  of the second stage.  Enter the values  $Y(0)$  and  $Y(4)$  as a check.

$Y(0) \ = \ $

$Y(4) \ = \ $

5

Calculate all  $N$  spectral values  $D(\mu)$, in particular

$D(\mu =3) \ = \ $

$D(\mu = 4) \ = \ $

6

What would be the spectral coefficients for  $d(ν = 4) = 1$  and  $d(ν \neq 4) = 0$ ?
Enter the values  $D(\mu =3)$  and  $D(\mu =4)$  as a check.

$D(\mu = 3) \ = \ $

$D(\mu = 4) \ = \ $


Solution

(1)  According to the general DFT equation given on the specification sheet, with  $w = \text{e}^{-\text{j}\hspace{0.05cm} \cdot \hspace{0.05cm}\pi /4}$  taking into account the alternating time coefficients:

$$8 \cdot D(3) = w^0 - w^3 + w^6- w^9+ w^{12}- w^{15}+ w^{18}- w^{21} = w^0 - w^3 + w^2- w^1+ w^{4}- w^{7}+ w^{6}- w^{5}\hspace{0.05cm}.$$
  • Here it is taken into account that due to the periodicity  $w_9 = w_1$,  $w_{12} = w_4$,  $w_{15} = w_7$,  $w_{18} = w_2$  und  $w_{21} = w_5$  ist.
  • After re-sorting, the same applies:
$$8 \cdot D(3) = (w^0 + w^4) - (w^1 + w^5)+ (w^2 + w^6) - (w^3 + w^7) = (1 + w + w^2+ w^3) \cdot (w^0 + w^4)\hspace{0.05cm}.$$
  • Thus, because  $w_0 = 1$  and  $w_4 = \text{e}^{-\text{j}\pi } = \hspace{0.08cm} - \hspace{-0.08cm}1$ , we obtain  $\underline {D(\mu=3) = 0}$.


(2)  In analogy to sub-taske  (1) , we now get:

$$ 8 \cdot D(4) = w^0 - w^4 + w^8- w^{12}+ w^{16}- w^{20}+ w^{24}- w^{28}= 4 \cdot (w^0 - w^4)= 8 \hspace{0.3cm} \Rightarrow \hspace{0.3cm}\hspace{0.15 cm}\underline{D(\mu=4) = 1}\hspace{0.05cm}.$$


Example for the FFT algorithm

(3)  Proposed solution 2 is correct:

  • The term  $w^0 = 1$  does not have to be taken into account.
  • All output values with odd indices are zero by subtracting two identical input values.
  • The first statement is not true:   It holds  $X(0) = X(2) = +2$  and  $X(4) = X(6) = - 2$.


(4)  The multiplication with  $w^{2} = -{\rm j}$  can be dispensed with, since in the signal flow diagram the corresponding input values are zero.

  • One thus obtains  $Y(0) \;\underline{= 4}$  and  $Y(4) \;\underline{= - \hspace{-0.03cm}4}$.
  • All other values are zero.


(5)  Because of  $Y(5) = Y(6) =Y(7) = 0$, the multiplications with  $w$,  $w^2$  and  $w^3$  do not matter in the third stage either.  All spectral coefficients  $D(\mu)$  therefore result in zero with the exception of

$$\hspace{0.15 cm}\underline{D(\mu=4)} = {1}/{N}\cdot \left[Y(0) - Y(4) \right ] \hspace{0.15 cm}\underline{= 1} \hspace{0.05cm},$$
$$\hspace{0.15 cm}\underline{D(\mu=3)} = D(\mu\ne 4) \hspace{0.15 cm}\underline{= 0} \hspace{0.05cm}.$$

This result agrees with the results from  (1)  und  (2).


(6)  Since both the time coefficients  $d(ν)$  and all spectral coefficients  $D(\mu)$  are purely real, there is no difference between the FFT and the IFFT.

  • This means at the same time:  The input and output values can be interchanged.
  • Subtask  (5)  gave the following result:
$$d({\rm even}\hspace{0.15cm}\nu) = +1, \hspace{0.2cm}d({\rm odd}\hspace{0.15cm}\nu)= -1$$
$$\Rightarrow \hspace{0.3cm}D(\mu = 4)= 1,\hspace{0.2cm}D(\mu \ne 4)= 0.$$
  • By swapping the input and output values, we arrive at problem  (6):
$$d(\nu = 4)= 1, \hspace{0.2cm}d(\nu \ne 4)= 0 \hspace{0.3cm} \Rightarrow \hspace{0.3cm}D({\rm gerades}\hspace{0.15cm}\mu) = +1, \hspace{0.2cm}D({\rm ungerades}\hspace{0.15cm}\mu)= -1 \hspace{0.05cm}.$$
  • In particular, this results in  $D(\mu=3) \; \underline{= -1}$  und  $D(\mu=4) \; \underline{= +1}$.