1.

Which of the following is true regarding the number of computations required to

compute an N-point DFT?
a) N2 complex multiplications and N(N-1) complex additions
b) N2 complex additions and N(N-1) complex multiplications
c) N2 complex multiplications and N(N+1) complex additions
d) N2 complex additions and N(N+1) complex multiplications

Answer: a
Explanation: The formula for calculating N point DFT is given as
X(k)=∑N−1n=0x(n)e−j2πkn/N
From the formula given at every step of computing we are performing N complex
multiplications and N-1 complex additions. So, in a total to perform N-point DFT we
perform N2 complex multiplications and N(N-1) complex additions.
2. Which of the following is true regarding the number of computations required to
compute DFT at any one value of ‘k’?
a) 4N-2 real multiplications and 4N real additions
b) 4N real multiplications and 4N-4 real additions
c) 4N-2 real multiplications and 4N+2 real additions
d) 4N real multiplications and 4N-2 real additions

Answer: d
Explanation: The formula for calculating N point DFT is given as
X(k)=∑N−1n=0x(n)e−j2πkn/N
From the formula given at every step of computing we are performing N complex
multiplications and N-1 complex additions. So, it requires 4N real multiplications and 4N-
2 real additions for any value of ‘k’ to compute DFT of the sequence.
3. WNk+N/2=?
a) WNk
b) -WNk
c) WN-k
d) None of the mentioned

Answer: b
Explanation: According to the symmetry property, we get W Nk+N/2=-WNk.
4. What is the real part of the N point DFT XR(k) of a complex valued sequence x(n)?
a) ∑N−1n=0[xR(n)cos2πknN–xI(n)sin2πknN]
b) ∑N−1n=0[xR(n)sin2πknN+xI(n)cos2πknN]
c) ∑N−1n=0[xR(n)cos2πknN+xI(n)sin2πknN]
d) None of the mentioned

Answer: c
Explanation: For a complex valued sequence x(n) of N points, the DFT may be
expressed as
XR(k)=∑N−1n=0[xR(n)cos2πknN+xI(n)sin2πknN]
5. The computation of XR(k) for a complex valued x(n) of N points requires
_____________
a) 2N2 evaluations of trigonometric functions
b) 4N2 real multiplications
c) 4N(N-1) real additions
d) All of the mentioned

Answer: d
Explanation: The expression for X R(k) is given as
XR(k)=∑N−1n=0[xR(n)cos2πknN+xI(n)sin2πknN]
So, from the equation we can tell that the computation of X R(k) requires 2N2 evaluations
of trigonometric functions, 4N2 real multiplications and 4N(N-1) real additions.
6. Divide-and-conquer approach is based on the decomposition of an N-point DFT into
successively smaller DFTs. This basic approach leads to FFT algorithms.
a) True
b) False

Answer: a
Explanation: The development of computationally efficient algorithms for the DFT is
made possible if we adopt a divide-and-conquer approach. This approach is based on
the decomposition of an N-point DFT into successively smaller DFTs. This basic
approach leads to a family of computationally efficient algorithms known collectively as
FFT algorithms.
7. If the arrangement is of the form in which the first row consists of the first M elements
of x(n), the second row consists of the next M elements of x(n), and so on, then which of
the following mapping represents the above arrangement?
a) n=l+mL
b) n=Ml+m
c) n=ML+l
d) none of the mentioned

Answer: b
Explanation: If we consider the mapping n=Ml+m, then it leads to an arrangement in
which the first row consists of the first M elements of x(n), the second row consists of the
next M elements of x(n), and so on.
8. If N=LM, then what is the value of WNmqL?
a) WMmq
b) WLmq
c) WNmq
d) None of the mentioned

Answer: a
Explanation: We know that if N=LM, then WNmqL = WN/Lmq = W M
mq
.

9. How many complex multiplications are performed in computing the N-point DFT of a
sequence using divide-and-conquer method if N=LM?
a) N(L+M+2)
b) N(L+M-2)
c) N(L+M-1)
d) N(L+M+1)

Answer: d
Explanation: The expression for N point DFT is given as
X(p,q)=∑L−1l=0{WlqN[∑M−1m=0x(l,m)WmqM]}WlpL
The first step involves L DFTs, each of M points. Hence this step requires LM 2 complex
multiplications, second require LM and finally third requires ML 2. So, Total complex
multiplications = N(L+M+1).
10. How many complex additions are performed in computing the N-point DFT of a
sequence using divide-and-conquer method if N=LM?
a) N(L+M+2)
b) N(L+M-2)
c) N(L+M-1)
d) N(L+M+1)

Answer: b
Explanation: The expression for N point DFT is given as
X(p,q)=∑L−1l=0{WlqN[∑M−1m=0x(l,m)WmqM]}WlpL
The first step involves L DFTs, each of M points. Hence this step requires LM(M-1)
complex additions, second step do not require any additions and finally third step
requires ML(L-1) complex additions. So, Total number of complex additions=N(L+M-2).
11. Which is the correct order of the following steps to be done in one of the algorithm of
divide and conquer method?
i) Store the signal column wise
ii) Compute the M-point DFT of each row
iii) Multiply the resulting array by the phase factors W Nlq.
iv) Compute the L-point DFT of each column.
v) Read the result array row wise.
a) i-ii-iv-iii-v
b) i-iii-ii-iv-v
c) i-ii-iii-iv-v
d) i-iv-iii-ii-v

Answer: c
Explanation: According to one of the algorithm describing the divide and conquer
method, if we store the signal in column wise, then compute the M-point DFT of each
row and multiply the resulting array by the phase factors W Nlq and then compute the L-
point DFT of each column and read the result row wise.
12. If we store the signal row wise then the result must be read column wise.
a) True
b) False

Answer: a
Explanation: According to the second algorithm of divide and conquer approach, if the
input signal is stored in row wise, then the result must be read column wise.
13. If we store the signal row wise and compute the L point DFT at each column, the
resulting array must be multiplied by which of the following factors?
a) WNlq
b) WNpq
c) WNlq
d) WNpm

Answer: d
Explanation: According to the second algorithm of divide and conquer approach, if the
input signal is stored in row wise, then we calculate the L point DFT of each column and
we multiply the resulting array by the factor WNpm.
1. If we split the N point data sequence into two N/2 point data sequences f 1(n) and f2(n)
corresponding to the even numbered and odd numbered samples of x(n), then such an
FFT algorithm is known as decimation-in-time algorithm.
a) True
b) False

Answer: a
Explanation: Let us consider the computation of the N=2 v point DFT by the divide and
conquer approach. We select M=N/2 and L=2. This selection results in a split of N point
data sequence into two N/2 point data sequences f 1(n) and f2(n) corresponding to the
even numbered and odd numbered samples of x(n), respectively, that is
f1(n)=x(2n)
f2(n)=x(2n+1), n=0,1,2…N/2-1
Thus f1(n) and f2(n) are obtained by decimating x(n) by a factor of 2, and hence the
resulting FFT algorithm is called a decimation-in-time algorithm.
2. If we split the N point data sequence into two N/2 point data sequences f 1(n) and f2(n)
corresponding to the even numbered and odd numbered samples of x(n) and F 1(k) and
F2(k) are the N/2 point DFTs of f1(k) and f2(k) respectively, then what is the N/2 point DFT
X(k) of x(n)?
a) F1(k)+F2(k)
b) F1(k)-WNk F2(k)
c) F1(k)+WNk F2(k)
d) None of the mentioned

Answer: c
Explanation: From the question, it is given that
f1(n)=x(2n)
f2(n)=x(2n+1), n=0,1,2…N/2-1
X(k)=∑N−1n=0x(n)WknN, k=0,1,2..N-1
=∑nevenx(n)WknN+∑noddx(n)WknN
=∑(N2)−1m=0x(2m)W2kmN+∑(N2)−1m=0x(2m+1)Wk(2m+1)N
=∑(N2)−1m=0f1(m)WkmN/2+WkN∑(N/2)−1m=0f2(m)Wkm(N2)
X(k)=F1(k)+ WNk F2(k).
3. If X(k) is the N/2 point DFT of the sequence x(n), then what is the value of X(k+N/2)?
a) F1(k)+F2(k)
b) F1(k)-WNk F2(k)
c) F1(k)+WNk F2(k)
d) None of the mentioned

Answer: b
Explanation: We know that, X(k) = F1(k)+WNk F2(k)
We know that F1(k) and F2(k) are periodic, with period N/2, we have F1(k+N/2) = F1(k)
and F2(k+N/2)= F2(k). In addition, the factor WNk+N/2 = -WNk.
Thus we get, X(k+N/2)= F1(k)- WNk F2(k).
4. How many complex multiplications are required to compute X(k)?
a) N(N+1)
b) N(N-1)/2
c) N2/2
d) N(N+1)/2

Answer: d
Explanation: We observe that the direct computation of F 1(k) requires (N/2)2 complex
multiplications. The same applies to the computation of F2(k). Furthermore, there are N/2
additional complex multiplications required to compute W Nk. Hence it requires N(N+1)/2
complex multiplications to compute X(k).
5. The total number of complex multiplications required to compute N point DFT by
radix-2 FFT is?
a) (N/2)log2N
b) Nlog2N
c) (N/2)logN
d) None of the mentioned

Answer: a
Explanation: The decimation of the data sequence should be repeated again and again
until the resulting sequences are reduced to one point sequences. For N=2v, this
decimation can be performed v=log 2N times. Thus the total number of complex
multiplications is reduced to (N/2)log 2N.
6. The total number of complex additions required to compute N point DFT by radix-2
FFT is?
a) (N/2)log2N
b) Nlog2N
c) (N/2)logN
d) None of the mentioned

Answer: b
Explanation: The decimation of the data sequence should be repeated again and again
until the resulting sequences are reduced to one point sequences. For N=2 v, this
decimation can be performed v=log 2N times. Thus the total number of complex additions
is reduced to Nlog2N.
7. The following butterfly diagram is used in the computation of __________

a) Decimation-in-time FFT
b) Decimation-in-frequency FFT
c) All of the mentioned
d) None of the mentioned
Answer: a
Explanation: The above given diagram is the basic butterfly computation in the
decimation-in-time FFT algorithm.
8. For a decimation-in-time FFT algorithm, which of the following is true?
a) Both input and output are in order
b) Both input and output are shuffled
c) Input is shuffled and output is in order
d) Input is in order and output is shuffled

Answer: c
Explanation: In decimation-in-time FFT algorithm, the input is taken in bit reversal order
and the output is obtained in the order.
9. The following butterfly diagram is used in the computation of __________

a) Decimation-in-time FFT
b) Decimation-in-frequency FFT
c) All of the mentioned
d) None of the mentioned

Answer: b
Explanation: The above given diagram is the basic butterfly computation in the
decimation-in-frequency FFT algorithm.
10. For a decimation-in-time FFT algorithm, which of the following is true?
a) Both input and output are in order
b) Both input and output are shuffled
c) Input is shuffled and output is in order
d) Input is in order and output is shuffled

Answer: d
Explanation: In decimation-in-frequency FFT algorithm, the input is taken in order and
the output is obtained in the bit reversal order.
1. FFT algorithm is designed to perform complex operations.
a) True
b) False

Answer: a
Explanation: The FFT algorithm is designed to perform complex multiplications and
additions, even though the input data may be real valued. The basic reason for this is
that the phase factors are complex and hence, after the first stage of the algorithm, all
variables are basically complex valued.
2. If x1(n) and x2(n) are two real valued sequences of length N, and let x(n) be a complex
valued sequence defined as x(n)=x1(n)+jx2(n), 0≤n≤N-1, then what is the value of x1(n)?
a) x(n)−x∗(n)2
b) x(n)+x∗(n)2
c) x(n)−x∗(n)2j
d) x(n)+x∗(n)2j

Answer: b
Explanation: Given x(n)=x1(n)+jx2(n)
=>x*(n)= x1(n)-jx2(n)
Upon adding the above two equations, we get x 1(n)=x(n)+x∗(n)2.
3. If x1(n) and x2(n) are two real valued sequences of length N, and let x(n) be a complex
valued sequence defined as x(n)=x1(n)+jx2(n), 0≤ n≤ N-1, then what is the value of x2(n)?
a) x(n)−x∗(n)2
b) x(n)+x∗(n)2
c) x(n)+x∗(n)2j
d) x(n)−x∗(n)2j

Answer: d
Explanation: Given x(n)=x1(n)+jx2(n)
=>x*(n) = x1(n)-jx2(n)
Upon subtracting the above two equations, we get x 2(n)=x(n)−x∗(n)2j.
4. If X(k) is the DFT of x(n) which is defined as x(n)=x 1(n)+jx2(n), 0≤ n≤ N-1, then what is
the DFT of x1(n)?
a) 12[X∗(k)+X∗(N−k)]
b) 12[X∗(k)−X∗(N−k)]
c) 12j[X∗(k)−X∗(N−k)]
d) 12j[X∗(k)+X∗(N−k)]

Answer: a
Explanation: We know that if x(n)=x1(n)+jx2(n) then x1(n)=x(n)+x∗(n)2
On applying DFT on both sides of the above equation, we get
X1(k)=12DFT[x(n)]+DFT[x∗(n)]
We know that if X(k) is the DFT of x(n), the DFT[x*(n)]=X*(N-k)
=>X1(k)=12[X∗(k)+X∗(N−k)].
5. If X(k) is the DFT of x(n) which is defined as x(n)=x 1(n)+jx2(n), 0≤ n≤ N-1, then what is
the DFT of x1(n)?
a) 12[X∗(k)+X∗(N−k)]
b) 12[X∗(k)−X∗(N−k)]
c) 12j[X∗(k)−X∗(N−k)]
d) 12j[X∗(k)+X∗(N−k)]

Answer: c
Explanation: We know that if x(n)=x1(n)+jx2(n) then x2(n)=x(n)−x∗(n)2j.
On applying DFT on both sides of the above equation, we get
X2(k)=12jDFT[x(n)]−DFT[x∗(n)]
We know that if X(k) is the DFT of x(n), the DFT[x*(n)]=X*(N-k)
=>X2(k)=12j[X∗(k)−X∗(N−k)].
6. If g(n) is a real valued sequence of 2N points and x 1(n)=g(2n) and x2(n)=g(2n+1), then
what is the value of G(k), k=0,1,2…N-1?
a) X1(k)-W2kNX2(k)
b) X1(k)+W2kNX2(k)
c) X1(k)+W2kX2(k)
d) X1(k)-W2kX2(k)

Answer: b
Explanation: Given g(n) is a real valued 2N point sequence. The 2N point sequence is
divided into two N point sequences x1(n) and x2(n)
Let x(n)=x1(n)+jx2(n)
=> X1(k)=12[X∗(k)+X∗(N−k)] and X2(k)=12j[X∗(k)−X∗(N−k)]
We know that g(n)=x1(n)+x2(n)
=>G(k)=X1(k)+W2kNX2(k), k=0,1,2…N-1.
7. If g(n) is a real valued sequence of 2N points and x 1(n)=g(2n) and x2(n)=g(2n+1), then
what is the value of G(k), k=N,N-1,…2N-1?
a) X1(k)-W2kX2(k)
b) X1(k)+W2kNX2(k)
c) X1(k)+W2kX2(k)
d) X1(k)-W2kNX2(k)

Answer: d
Explanation: Given g(n) is a real valued 2N point sequence. The 2N point sequence is
divided into two N point sequences x1(n) and x2(n)
Let x(n)=x1(n)+jx2(n)
=> X1(k)=12[X∗(k)+X∗(N−k)] and X2(k)=12j[X∗(k)−X∗(N−k)]
We know that g(n)=x1(n)+x2(n)
=>G(k)=X1(k)-W2kNX2(k), k=N,N-1,…2N-1.
8. Decimation-in frequency FFT algorithm is used to compute H(k).
a) True
b) False

Answer: a
Explanation: The N-point DFT of h(n), which is padded by L-1 zeros, is denoted as H(k).
This computation is performed once via the FFT and resulting N complex numbers are
stored. To be specific we assume that the decimation-in frequency FFT algorithm is
used to compute H(k). This yields H(k) in the bit-reversed order, which is the way it is
stored in the memory.
9. How many complex multiplications are need to be performed for each FFT algorithm?
a) (N/2)logN
b) Nlog2N
c) (N/2)log2N
d) None of the mentioned

Answer: c
Explanation: The decimation of the data sequence should be repeated again and again
until the resulting sequences are reduced to one point sequences. For N=2v, this
decimation can be performed v=log 2N times. Thus the total number of complex
multiplications is reduced to (N/2)log 2N.
10. How many complex additions are required to be performed in linear filtering of a
sequence using FFT algorithm?
a) (N/2)logN
b) 2Nlog2N
c) (N/2)log2N
d) Nlog2N

Answer: b
Explanation: The number of additions to be performed in FFT are Nlog 2N. But in linear
filtering of a sequence, we calculate DFT which requires Nlog 2N complex additions and
IDFT requires Nlog2N complex additions. So, the total number of complex additions to be
performed in linear filtering of a sequence using FFT algorithm is 2Nlog 2N.
11. How many complex multiplication are required per output data point?
a) [(N/2)logN]/L
b) [Nlog22N]/L
c) [(N/2)log2N]/L
d) None of the mentioned

Answer: b
Explanation: In the overlap add method, the N-point data block consists of L new data
points and additional M-1 zeros and the number of complex multiplications required in
FFT algorithm are (N/2)log2N. So, the number of complex multiplications per output data
point is [Nlog22N]/L.
1. If the desired number of values of the DFT is less than log 2N, a direct computation of
the desired values is more efficient than FFT algorithm.
a) True
b) False

Answer: a
Explanation: To calculate a N point DFT using FFT algorithm, we need to perform (N/2)
log2N multiplications and N log 2N additions. But in some cases where desired number of
values of the DFT is less than log 2N such a huge complexity is not required. So, direct
computation of the desired values is more efficient than FFT algorithm.
2. What is the transform that is suitable for evaluating the z-transform of a set of data on
a variety of contours in the z-plane?
a) Goertzel Algorithm
b) Fast Fourier transform
c) Chirp-z transform
d) None of the mentioned

Answer: c
Explanation: Chirp-z transform algorithm is suitable for evaluating the z-transform of a
set of data on a variety of contours in the z-plane. This algorithm is also formulated as a
linear filtering of a set of input data. As a consequence, the FFT algorithm can be used
to compute the Chirp-z transform.
3. According to Goertzel Algorithm, if the computation of DFT is expressed as a linear
filtering operation, then which of the following is true?
a) yk(n)=∑Nm=0x(m)W−k(n−m)N
b) yk(n)=∑N+1m=0x(m)W−k(n−m)N
c) yk(n)=∑N−1m=0x(m)W−k(n+m)N
d) yk(n)=∑N−1m=0x(m)W−k(n−m)N

Answer: d
Explanation: Since WN-kN = 1, multiply the DFT by this factor. Thus
X(k)=WN-kN∑N−1m=0x(m)W−kmN=∑N−1m=0x(m)W−k(N−m)N
The above equation is in the form of a convolution. Indeed, we can define a sequence
yk(n) as
yk(n)=∑N−1m=0x(m)W−k(n−m)N
4. If yk(n) is the convolution of the finite duration input sequence x(n) of length N, then
what is the impulse response of the filter?
a) WN-kn
b) WN-kn u(n)
c) WNkn u(n)
d) None of the mentioned

Answer: b
Explanation: We know that yk(n)=∑N−1m=0x(m)W−k(n−m)N
The above equation is of the form y k(n)=x(n)*hk(n)
Thus we obtain, hk(n)= WN-kn u(n).
5. What is the system function of the filter with impulse response h k(n)?
a) 11−W−kNz−1
b) 11+W−kNz−1
c) 11−WkNz−1
d) 11+WkNz−1

Answer: a
Explanation: We know that h k(n)= WN-kn u(n)
On applying z-transform on both sides, we get
Hk(z)=11−W−kNz−1
6. What is the expression to compute y k(n) recursively?
a) yk(n)=WN-kyk(n+1)+x(n)
b) yk(n)=WN-kyk(n-1)+x(n)
c) yk(n)=WNkyk(n+1)+x(n)
d) None of the mentioned

Answer: b
Explanation: We know that h k(n)=WN-kn u(n)=yk(n)/x(n)
=> yk(n)=WN-kyk(n-1)+x(n).
7. What is the equation to compute the values of the z-transform of x(n) at a set of points
{zk}?
a) ∑N−1n=0x(n)znk, k=0,1,2…L-1
b) ∑N−1n=0x(n)z−n−k, k=0,1,2…L-1
c) ∑N−1n=0x(n)z−nk, k=0,1,2…L-1
d) None of the mentioned

Answer: c
Explanation: According to the Chirp-z transform algorithm, if we wish to compute the
values of the z-transform of x(n) at a set of points {zk}. Then,
X(zk)=∑N−1n=0x(n)z−nk, k=0,1,2…L-1
8. If the contour is a circle of radius r and the z k are N equally spaced points, then what
is the value of zk?
a) re-j2πkn/N
b) rejπkn/N
c) rej2πkn
d) rej2πkn/N

Answer: d
Explanation: We know that, if the contour is a circle of radius r and the z k are N equally
spaced points, then what is the value of z k is given by rej2πkn/N
9. How many multiplications are required to calculate X(k) by chirp-z transform if x(n) is
of length N?
a) N-1
b) N
c) N+1
d) None of the mentioned

Answer: c
Explanation: We know that yk(n)=WN-kyk(n-1)+x(n).Each iteration requires one
multiplication and two additions. Consequently, for a real input sequence x(n), this
algorithm requires N+1 real multiplications to yield not only X(k) but also, due to
symmetry, the value of X(N-k).
10. If the contour on which the z-transform is evaluated is as shown below, then which of
the given condition is true?

a) R0>1
b) R0<1
c) R0=1
d) None of the mentioned
Answer: a
Explanation: From the definition of chirp z-transform, we know that V=R0ejθ.
If R0>1, then the contour which is used to calculate z-transform is as shown below.
11. How many complex multiplications are need to be performed to calculate chirp z-
transform?(M=N+L-1)
a) log2M
b) Mlog2M
c) (M-1)log2M
d) Mlog2(M-1)

Answer: b
Explanation: Since we will compute the convolution via the FFT, let us consider the
circular convolution of the N point sequence g(n) with an M point section of h(n) where
M>N. In such a case, we know that the first N-1 points contain aliasing and that the
remaining M-N+1 points are identical to the result that would be obtained from a linear
convolution of h(n) with g(n). In view of this, we should select a DFT of size M=L+N-1.
Thus the total number of complex multiplications to be performed are Mlog 2M.
1. The effect of round off errors due to the multiplications performed in the DFT with
fixed point arithmetic is known as Quantization error.
a) True
b) False

Answer: a
Explanation: Since DFT plays a very important role in many applications of DSP, it is
very important for us to know the effect of quantization errors in its computation. In
particular, we shall consider the effect of round off errors due to the multiplications
performed in the DFT with fixed point arithmetic.
2. What is the model that has been adopt for characterizing round of errors in
multiplication?
a) Multiplicative white noise model
b) Subtractive white noise model
c) Additive white noise model
d) None of the mentioned

Answer: c
Explanation: Additive white noise model is the model that we use in the statistical
analysis of round off errors in IIR and FIR filters.
3. How many quantization errors are present in one complex valued multiplication?
a) One
b) Two
c) Three
d) Four

Answer: d
Explanation: We assume that the real and imaginary components of {x(n)} and {W Nkn} are
represented by ‘b’ bits. Consequently, the computation of product x(n). W Nkn requires four
real multiplications. Each real multiplication is rounded from 2b bits to b bits and hence
there are four quantization errors for each complex valued multiplication.
4. What is the total number of quantization errors in the computation of single point DFT
of a sequence of length N?
a) 2N
b) 4N
c) 8N
d) 12N

Answer: b
Explanation: Since the computation of single point DFT of a sequence of length N
involves N number of complex multiplications, it contains 4N number of quantization
errors.
5. What is the range in which the quantization errors due to rounding off are uniformly
distributed as random variables if Δ=2 -b?
a) (0,Δ)
b) (-Δ,0)
c) (-Δ/2,Δ/2)
d) None of the mentioned

Answer: c
Explanation: The Quantization errors due to rounding off are uniformly distributed
random variables in the range (-Δ/2,Δ/2) if Δ=2-b. This is one of the assumption that is
made about the statistical properties of the quantization error.
6. The 4N quantization errors are mutually uncorrelated.
a) True
b) False

Answer: a
Explanation: The 4N quantization errors are mutually uncorrelated. This is one of the
assumption that is made about the statistical properties of the quantization error.
7. The 4N quantization errors are correlated with the sequence {x(n)}.
a) True
b) False

Answer: b
Explanation: According to one of the assumption that is made about the statistical
properties of the quantization error, the 4N quantization errors are uncorrelated with the
sequence {x(n)}.
8. How is the variance of the quantization error related to the size of the DFT?
a) Equal
b) Inversely proportional
c) Square proportional
d) Proportional

Answer: d
Explanation: We know that each of the quantization has a variance of Δ 2/12=2-2b/12.
The variance of the quantization errors from the 4N multiplications is 4N. 2 -2b/12=2-
2b
(N/3).
Thus the variance of the quantization error is directly proportional to the size of the DFT.
9. Every fourfold increase in the size N of the DFT requires an additional bit in
computational precision to offset the additional quantization errors.
a) True
b) False

Answer: a
Explanation: We know that, the variance of the quantization errors is directly proportional
to the size N of the DFT. So, every fourfold increase in the size N of the DFT requires an
additional bit in computational precision to offset the additional quantization errors.
10. What is the variance of the output DFT coefficients |X(k)|?
a) 1N
b) 12N
c) 13N
d) 14N

Answer: c
Explanation: We know that the variance of the signal sequence is (2/N) 2/12=13N2
Now the variance of the output DFT coefficients |X(k)|=N.13N2=13N.
11. What is the signal-to-noise ratio?
a) σX2.σq2
b) σX2/σq2
c) σX2+σq2
d) σX2-σq2

Answer: b
Explanation: The signal-to-noise ratio of a signal, SNR is given by the ratio of the
variance of the output DFT coefficients to the variance of the quantization errors.
12. How many number of bits are required to compute the DFT of a 1024 point
sequence with a SNR of 30db?
a) 15
b) 10
c) 5
d) 20

Answer: a
Explanation: The size of the sequence is N=210. Hence the SNR is
10log10(σX22/σq2)=10 log1022b-20
For an SNR of 30db, we have
3(2b-20)=30=>b=15 bits.
Note that 15 bits is the precision for both addition and multiplication.
13. How many number of butterflies are required per output point in FFT algorithm?
a) N
b) N+1
c) 2N
d) N-1

Answer: d
Explanation: We find that, in general, there are N/2 in the first stage of FFT, N/4 in the
second stage, N?8 in the third state, and so on, until the last stage where there is only
one. Consequently, the number of butterflies per output point is N-1.
14. What is the value of the variance of quantization error in FFT algorithm, compared to
that of direct computation?
a) Greater
b) Less
c) Equal
d) Cannot be compared

Answer: c
Explanation: If we assume that the quantization errors in each butterfly are uncorrelated
with the errors in the other butterflies, then there are 4(N-1) errors that affect the output
of each point of the FFT. Consequently, the variance of the quantization error due to
FFT algorithm is given by
4(N-1)(Δ2/12)=N(Δ2/3)(approximately)
Thus, the variance of quantization error due to FFT algorithm is equal to the variance of
the quantization error due to direct computation.
15. How many number of bits are required to compute the FFT of a 1024 point sequence
with a SNR of 30db?
a) 11
b) 10
c) 5
d) 20

Answer: a
Explanation: The size of the FFT is N=2 10. Hence the SNR is 10 log1022b-v-1=30
=>3(2b-11)=30
=>b=21/2=11 bits.